- Email: [email protected]
Study of air pollution in the proximity of a waste incinerator
Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Study of air pollution in the proximity of a waste incinerator V. Barrera a, G. Calzolai a,b, M. Chiari b, F. Lucarelli a,b,⇑, S. Nava b, M. Giannoni b, S. Becagli c, D. Frosini c a
Department of Physics and Astronomy – University of Florence, Via G. Sansone 1, 50019 Sesto Fiorentino (Fi), Italy National Institute of Nuclear Physics (INFN) – Florence, Via G. Sansone 1, 50019 Sesto Fiorentino (Fi), Italy c Department of Chemistry – University of Florence, Via della Lastruccia 3, 50019 Sesto Fiorentino (Fi), Italy b
a r t i c l e
i n f o
Article history: Received 15 March 2015 Received in revised form 15 July 2015 Accepted 9 August 2015 Available online 18 September 2015 Keywords: PIXE Atmospheric aerosol PIXE Positive Matrix Factorization Biomass burning
a b s t r a c t Montale is a small town in Tuscany characterised by high PM10 levels. Close to the town there is a waste incinerator plant. There are many concerns in the population and in the press about the causes of the high levels of pollution in this area. Daily PM10 samples were collected for 1 year by the FAI Hydra Dual sampler and analysed by different techniques in order to obtain a complete chemical speciation (elements by PIXE and ICP-MS, ions by Ion Chromatography, elemental and organic carbon by a thermo-optical instrument); hourly fine (<2.5 lm) and coarse (2.5–10 lm) PM samples were collected for shorter periods by the Streaker sampler and hourly elemental concentrations were obtained by PIXE analysis. Positive Matrix Factorization identified and quantified the major aerosol sources. Biomass burning turned out to be the most important source with an average percentage contribution to PM10 of 27% of and even higher percentages during the winter period when there are the highest PM10 concentrations. The contribution of the incinerator source has been estimated as about 6% of PM10. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Particle Induced X-ray Emission (PIXE) is a suitable technique for analysing aerosol samples [1,2] due to its ability to carry out a multi-elemental analysis of the particulate deposited on the filter surface without any solubilization procedure, therefore shortening the analysis time and reducing the sample contamination risk. In particular it is unrivalled for the detection of the crustal elements. However, it provides only part of the desired information with regard to the chemical composition therefore it is mandatory to perform also at least measurements for important ionic species (e.g., ammonium, nitrate), for organic carbon (OC) and elemental carbon (EC). Finally Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES) or Mass Spectroscopy (ICP-MS) and PIXE can be used as complementary techniques and their combined use is able to give powerful information on chemical speciation of metals in the particulate matter (PM). By using the appropriate extraction conditions (extracting solution composition, pH, temperature, pressure, contact time) the metal fraction which is more ‘‘available” for the natural systems is obtained and in this way it is possible to evaluate better the impact of heavy metals on the environment and the human health [3]. ⇑ Corresponding author at: via Sansone 1, I-50019 Sesto Fiorentino (Firenze), Italy. Tel.: +39 055 4572274; fax: +39 055 4572641. E-mail address: [email protected] (F. Lucarelli). http://dx.doi.org/10.1016/j.nimb.2015.08.015 0168-583X/Ó 2015 Elsevier B.V. All rights reserved.
Standard daily sampling allows the study of aerosol composition for a long period covering all the seasonal changes in aerosol composition and tracing sources which may be present only in specific periods; however some particulate emissions change within a few hours and source apportionment receptor models need a series of samples containing material from the same set of sources in differing proportions. Increasing the time resolution of the measurements produces samples that have greater between-sample variability in the source contributions than samples integrated over longer time periods. Therefore time-resolved measurement can give a valuable help in source identification. Montale is a small town in Tuscany characterised by high PM10 levels. There are many concerns in the population and in the press about the causes of the high levels of pollution in this area, mainly because close to the town there is a waste incinerator plant. However many other sources can give a relevant contribution to PM10: many fields are present in the surroundings therefore a relevant crustal contribution could be present; biomass burning is used as heating system and also open fires are often present for the combustion of pruning; traffic, too, could give a relevant contribution. Therefore a multidisciplinary approach involving different analytical techniques is necessary to obtain a complete chemical speciation which can be used as a starting point for the application of multivariate statistical methods like Positive Matrix Factorization (PMF) to identify the PM sources and their contribution to the PM mass.
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
The Regional Government has promoted an extensive field campaign for the aerosol characterization in Montale, to give to policymakers the knowledge and the tools for a significant reduction of the main anthropogenic emissions. We will present here some of the results obtained for the first period of the campaign both for the daily and the hourly resolution samples.
2. Methods 2.1. Sampling Montale is a small town located 35 km W of Florence (Italy) with about 10,000 inhabitants. The sampling site is close to a garden and adjacent to a public car park, mainly for residential use. On a wider scale, the site can be classified as a ‘‘suburban” site, as it is in a built area but close to vast not urbanised areas. PM10 samples were collected on a daily base (from midnight to midnight), every second day, for 1 year (from December 2013 to December 2014) by the low-volume (2.3 m3/h, EU rule EN 12341) FAI Hydra Dual sampler equipped with two inlets so that aerosol can be simultaneously collected on Teflon and Quartz fibre filters (47 mm diameter), thus allowing the application of different analytical techniques. During shorter periods (2 weeks in winter and 2 weeks in summer) the aerosol was also collected by a low volume (1 lpm) Streaker sampler. In this device particles are separated on two different stages: the coarse fraction (2.5 lm < Dae < 10 lm) and the fine fraction (Dae < 2.5 lm), collected on a Kimfol foil and on a Nuclepore filter respectively. The two collecting substrata are paired on a cartridge which, rotating at constant speed for 1 week, produces a circular continuous deposition of particulate matter (‘‘streak”) on both stages [4].
2.2. Measurements and data analysis PM10 daily mass concentrations were obtained by weighing the Teflon filters by an analytical balance in controlled conditions of temperature (20 ± 1 °C) and relative humidity (50 ± 5%). Samples on Teflon filters are cut in three parts. On one half of the filter PIXE is used to measure the concentrations of all the elements with atomic number Z > 10. PIXE analyses were performed at the 3MV Tandetron accelerator of the INFN-LABEC laboratory, with the new external beam set-up similar to the one extensively used so far [5] but improved by the duplication of the SDD detector used for medium–high energy X-rays [6]; the measuring time was therefore reduced to only 90 s/sample with better statistics. Each sample was irradiated for 90 s with a 3.0 MeV proton beam (2 mm2 spot, 10–150 nA intensity). A filter scanning was carried out to analyse most of the deposit area. PIXE spectra were fitted using the GUPIX code [7] and elemental concentrations were obtained by a calibration curve from a set of thin standards of known areal density (Micromatter Inc.). The water-soluble fraction (sample extraction in ultra-pure MilliQ water in ultrasonic bath) of inorganic cations, inorganic anions and low molecular weight organic anions is measured by Ion Chromatography (IC) on one quarter of the filter [8]; ICP methods are used to determine the ‘‘soluble fraction” (in the acidic extraction conditions) of several major (ICP-AES) and trace (ICP-MS) metals on the remaining quarter of the filter [3]. 0.1% sub-boiled distilled HNO3 (pH = 1.5) in an ultrasonic bath for 15 min at room temperature was used as extraction method. This fraction represents the most ‘‘available” metal fraction (including free metal, labile complexes, carbonate and bicarbonate salts), considering the pH = 1.5 as the lowest limit for ‘‘natural” pH values [3].
113
A 1.5 cm2 punch of the Quartz fibre filters is used for Organic and Elemental Carbon assessment by Thermo Optical Transmission analysis (using a Sunset Lab analyser [9]) at LABEC. Streaker samples were analysed by PIXE, with the same experimental set-up used for daily samples. A properly collimated proton beam (2.7 MeV, 50–300 nA) was used to scan the deposit in steps corresponding to 1 h of aerosol sampling, thus providing the elemental concentrations with hourly time resolution. Thanks to the improved efficiency of the experimental set-up, 90 s. of measuring time for each spot corresponding to 1 h of sampling was enough to obtain a good statistics [6]. We report here the first results obtained for the winter and spring season. Positive Matrix Factorization (PMF) has been applied to this preliminary data set (daily and hourly samples separately) aiming at a first identification and quantification of the major aerosol sources, using the EPA PMFv5 software. PMF is an advanced factor analysis technique based on a weighted least square fit approach [10]; it uses realistic error estimates to weigh data values and imposes non-negativity constraints in the factor computational process. Briefly, the PMF factor model may be written as X = GF + E, where X is a known n by m matrix of the m measured chemical species in n samples; G is an n by p matrix of source contributions to the samples (i.e. time variations of the p factor scores); F is a p by m matrix of factors composition (often called source profiles). G and F are factor matrices to be determined and E is defined as a residual matrix. Input data were prepared using the procedure suggested by Polissar [11] and PMF results for different number of factors and multiple values of FPEAK were systematically explored to find out the most reasonable solution. For daily samples, to obtain absolute source profiles and contributions the aerosol mass was introduced as a variable with a 400% error. For hourly data, only elemental concentrations are measured and no information about the PM mass concentration is available: in this case, only relative source profiles (elemental ratios within the composition of the identified sources) can be obtained and source time series are in arbitrary units. 3. Results 3.1. PM10 mass PM10 concentrations are reported in Fig. 1. Lower levels were recorded in spring, with values between 10 and 20 lg/m3, except for a peak close to the PM10 daily limit threshold on 22 May (44.7 lg/m3) in correspondence of an episode of Saharan dust transport (see below). In winter levels were far higher with many concentration peaks around 100 lg/m3 (up to 174 lg/m3 on December 20th). This is due to the typical winter weather conditions with greater atmospheric stability, with a reduced height of the boundary layer and a poor dispersion of the pollutants themselves 3.2. Chemical composition As an example of the results obtained from the analysis of the daily samples, in Fig. 2 the Na, Si, K and S concentrations (obtained by PIXE) are reported. Na is a typical marker of sea-salt aerosol (together with Mg and Cl). Several episodes with a strong increase of Na, always together with Mg, sometimes with sometimes without a corresponding increase in Cl concentration, are present, pointing to the transport of marine aerosol. In all the cases the Mg/Na ratio (0.14) is very similar to the one typical of bulk sea-water (Mg/Na = 0.12, [12]) but often there is a strong Cl depletion probably due to reactions
114
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
Fig. 1. PM10 mass concentrations (lg/m3) during the first period of the sampling campaign.
Fig. 2. Concentrations (lg/m3) of Na, Si, K and S obtained by PIXE.
with sulphuric and nitric acids producing gaseous HCl during air masses transport. The sea-salt component was obtained summing up Na (corrected for the crustal contribution), Mg, Cl and ss-S (obtained as 0.253*ssNa, ssNa = Na - (Al * 0.348)); it represents a negligible contribution that, in general, is below 1.5 lg/m3 except in particular cases concentrated in the first half of February, in which higher concentrations (3.3 lg/m3 on February 8th, 7.8 lg/m3 on February 14th, and 2.9 lg/m3 on February 16th) were observed; in these days there was always a strong wind coming from the Tyrrhenian Sea that brought the marine aerosol to the interior of the region. Si is a typical marker of crustal dust both of local origin and longrange transport from desert regions, such as the Sahara. In Fig. 2 events of probable Saharan origin on days in which air masses trajectories are coming from the Sahara region (calculations with the transport model HYSPLIT NOAA’s Air Resources Laboratory [13]) show high Si concentrations and also a simultaneous increase of all the other typical crustal elements. Since PIXE is a quantitative technique, it allowed obtaining a further proof of the Saharan origin by the observation of the inter-elemental ratios, whose values are different for the desert and the local dust. The elemental ratios that show the most significant differences between Saharan and nonSaharan days are those involving either Fe or Ca, due to a Fe and Ca enrichment of local dust with respect to the Saharan dust as
already found in other studies [14–16]. For example, the Fe/Ca, Fe/Al, Ca/Al ratios changed from 1.17, 3.12, 3.44 during normal days to 0.7, 0.91, 1.45 during these Saharan episodes. An estimate of the soil dust component concentration can be calculated considering the crustal elements (obtained by PIXE) as oxides [17]; corrections were however applied to this formula to take into account sea-salt contributions to Na and Mg and possible anthropogenic contributions to the other elements [14]. Fig. 3 shows the time trend of the soil dust component; it contributes, in percent, more than twice in the spring compared to winter. In some Saharan dust episodes the concentration reaches values of 6–9 lg/m3 with a maximum of 10.5 lg/m3 (corresponding to approximately 25% of the PM10 mass) on April 3rd. In the rest of the days not affected by these particular events, the concentration of the crustal component remains below 5 lg/m3. Between 8 and 22 March it is possible to notice a plateau due to some Saharan dust intrusions. Sulphur, which is essentially a secondary aerosol component, has instead higher values in spring. The best conditions of atmospheric circulation on a regional scale (e.g. transport of SO2 and sulfate from thermo-electric plants located outside the city to the sampling site) and the increased efficiency of photochemical oxidation of SO2 to sulfate, may explain the increase of the sulfur in the warmer period.
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
115
Fig. 3. Concentration (lg/m3) of the soil dust component obtained from the crustal elements measured by PIXE. The most intense episodes related to Saharan dust transport are indicated.
K is present in the soil dust but it is also a typical marker of biomass burning. The remarkable concentration of K in winter points to a non negligible impact of biomass burning in the area. Fig. 4 shows the average percentage contributions of the main components of the aerosol for all the analysed samples and for only winter and spring samples separately. The carbon component Particulate Organic Matter (POM) + EC gives the main contribution, followed by the secondary inorganic and the crustal components. The marine contribution is negligible, as expected. The percentage of POM is particularly high in the winter period, when the values of PM10 are greater. This may be probably due to an increase in emissions (heating by BB). In the spring season, the percentage of crustal component and secondary inorganics increase. The high value of the ratio OC/EC (on average about 7), especially in winter, is symptomatic of a strong contribution from sources not related to traffic such as biomass combustion. A similar trend was already observed in another area in Tuscany characterised by high PM10 concentrations caused mainly by the impact of biomass burning [18]. Within the secondary inorganic component, nitrates show higher values in winter, while the sulfates have the highest concentrations during the warmer seasons. The seasonality of nitrates may be due to an increase in emissions of gaseous precursors (for the presence of domestic heating or for an increase of vehicular traffic) but also to the weather conditions (less dispersion of local pollutants in winter and increased volatility in summer). The increase of the sulphates in the warmer period was already explained above.
Mn, Fe, Ni, Cu, Zn, Se, Br, Sr, Zr by PIXE, Na+, Cl , NH+4, NO-3, SO24 by IC, V, Ba, Mo, Cd by ICP) and source apportionment by PMF allowed identifying 10 aerosol sources;
3.3. Aerosol source apportionment
3.3.4. Local dust Characterised by crustal elements (Mg, Al, Si, K, Ca, Ti, Fe) with enrichment of EC, OC, Ca and Fe, it is associated to an ‘urban’ dust (EC, OC and Fe may be produced by pneumatics, brakes and asphalt
The combination of a complete PM chemical speciation (EC/OC by Thermo Optical Transmission analysis, Mg, Al, Si, K, Ca, Ti, Cr,
3.3.1. Incinerator Mostly composed by EC, OC, NO3 and traced by specific elements (Cl , Pb, Cd, Zn) that can be associated to incinerator plant emissions [19–21]; this contribution is present during all the campaign (Fig. 5). 3.3.2. Traffic Mainly composed by OC and EC and traced by EC, Fe, Cu, Ba and Mo; it has a maximum contribution in winter when there is atmospheric stability. The ratio OC/EC of 4 reveals a significant proportion of secondary organic aerosols from oxidation of primary volatile organic compounds (VOCs) from fuel combustion and from enhanced anthropogenic transformation of biogenic VOCs by NOx. 3.3.3. Saharan dust (SD) Mainly characterised by crustal elements (Mg, Al, Si, K, Ca, Ti, Fe). The source profile is very close to the average crust composition, with enrichment factors (EF), calculated with respect to Al using the average continental crust composition reported by [22] close to 1. The time trend is characterised by concentration peaks during days classified as Saharan intrusions (on the basis of both models and satellite observations) and it is close to zero in all other days (Fig. 5).
Fig. 4. Average percentage contributions of the main components of the aerosol for all the analysed samples and for only winter and spring samples separatedly.
116
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
Fig. 5. Time trend of the contribution (lg/m3) of some of the sources identified by the PMF analysis of the daily filters: Saharan dust, biomass burning, incinerator.
wear, Ca by construction works). As opposed to the Saharan dust, this source contributes all the year long. 3.3.5. Sea salt Characterised by Na+, Mg and Cl , and, to a lesser extent, Mg and Br, with inter-elemental ratios in accordance with those reported in literature for seawater. The time pattern is characterised by short episodic peaks, occurring when air masses are directly transported by strong winds from the Tyrrhenian Sea. 3.3.6. Aged sea salt Characterised by Na+ and Mg, it is contaminated by NO3 and 2 SO4 , and depleted in Cl . It is known that different heterogeneous reactions between airborne sea-salt particles and gaseous pollutants, like nitric and sulfuric acid, may lead to Cl volatilization and formation of nitrates and sulfates [23,24]. Thus, this source is a mixture of anthropogenic and natural source contributions. 3.3.7. Heavy oil combustion Composed by OC and sulphate, and characterised by tracers as V and Ni. This source is due to residual oil combustion processes, mostly from activities that are located outside the area, like energy production, refinery, industrial plants and shipping. 3.3.8. Secondary nitrates and organics Mostly composed by NO3 and OC, it is to be associated to a secondary local component. NO3 are, indeed, produced by oxidation in atmosphere of NOx that is derived from local combustion processes as heating, biomass burning and traffic. The time trend is characterised by an important seasonal periodicity with higher values until March (see Section 3.2). 3.3.9. ‘Secondary sulphates Mainly composed by SO24 and NH+4, it is associated to a regional secondary component, which originates from atmospheric SO2 (emitted mainly from thermo-electric plants, and quickly distributed on a regional area). The seasonal trend has higher values in spring.
3.3.10. Biomass burning (BB) It is mainly correlated with OC and EC and tracers as K, Zn, Br and Pb. The OC/EC ratio in this profile (7) is within the ranges reported in literature for this source [25]). As expected, the contributions of this source are highly season-dependent with maxima in winter (Fig. 5) when, as already observed, there are most of the exceedances of the limits (with a maximum contribution up to 59 lg/m3), and very low in spring, when it is mainly due to the open fires. The pie chart of the identified source contributions to PM10 is reported in Fig. 6. The BB source is the one which gives the most relevant contribution (27%) followed by secondary nitrates (17%). Their contribution increases even more during the days in which the daily limit was exceeded, up to 34% and 27%. 3.4. Hourly resolution data The use of the Streaker sampler allowed obtaining high time resolved and size segregated data, i.e. the elemental concentrations with hourly time resolution, in the fine and coarse aerosol fractions, for about two weeks during the winter season (January21– February 8). PMF identified 7 sources for the fine fraction and 5 for the coarse one. All the sources already identified from the analysis of daily data, have been found, except the secondary nitrate source (which was not identified since PIXE cannot detect nitrogen), all with similar source profiles as those obtained for daily data. The results confirmed and reinforced the results obtained by the daily samples. As an example the hourly time pattern of the SD, BB, and incinerator sources are reported in Fig. 7. The SD source is only present during episodes of transport of Saharan aerosol (confirmed by the back-trajectories of the air masses calculated with Hysplit model) and zero outside. The time trend of the BB source supports its identification as biomass burning for domestic heating: it is characterised by a periodic pattern with peaks starting in the evening and lasting several hours. The absence of the evening-night peak on some days is explained by the meteorological conditions. Some peaks during the day are probably due to open fires.
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
117
Fig. 6. Pie chart of the identified source contributions to PM10.
Fig. 7. Hourly temporal patterns of factor contributions (arbitrary units) obtained by the PMF analysis of hourly data in the fine fraction: Saharan dust, biomass burning, incinerator.
The incinerator source does not show any particular daily pattern but it is characterised by sporadic peaks. 4. Conclusions A preliminary characterization of particulate matter in Montale (Tuscany) has been carried out and the main aerosol sources have been identified. Biomass burning turned out to be the most important source with an average percentage contribution of 27% of PM10 and very high level during the winter period. It is worth noting that the cold season is characterised by stagnant atmospheric conditions that favour the accumulation of pollutants in the
boundary layer, thus producing an increase in contributions also from other sources. The contribution of the incinerator source has been estimated as about 6% of PM10. Hourly data usefully supported and reinforced results obtained by the analysis of daily samples. Data elaboration is still in progress and PMF analysis of the whole 1 year long data set will complete this study. Acknowledgements This study was supported by the Regional Government of Tuscany, in the framework of PATOS2.2 project. V. Barrera acknowledges CONACYT program (Grant 208131) for financial support.
118
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
References [1] W. Maenhaut, K. Malmqvist, in: R.V. Grieken (Ed.), Handbook of X-ray Spectrometry, Marcel Dekker, Antwerp, 2001, pp. 761–771. [2] F. Lucarelli, S. Nava, G. Calzolai, M. Chiari, R. Udisti, F. Marino, X-Ray Spectrom. 40 (2011) 162–167. [3] R. Traversi, S. Becagli, G. Calzolai, M. Chiari, M. Giannoni, F. Lucarelli, S. Nava, F. Rugi, M. Severi, R. Udisti, Nucl. Instr. Meth. B 318 (2014) 130–134. [4] Website:. [5] F. Lucarelli, G. Calzolai, M. Chiari, M. Giannoni, D. Mochi, S. Nava, L. Carraresi, Nucl. Instr. Meth. Phys. Res. B 318 (2014) 55–59. [6] G. Calzolai, F. Lucarelli, M. Chiari, S. Nava, M. Giannoni, L. Carraresi, P, Prati, R. Vecchi, Nucl. Instr. Meth. Phys. Res. B 363 (2015) 99–104. [7] A. Maxwell, W.J. Teesdale, J.L. Campbell, Nucl. Instr. Meth. B 95 (1995) 407. [8] S. Becagli, C. Ghedini, S. Peeters, A. Rottiers, R. Traversi, R. Udisti, M. Chiari, A. Jalba, S. Despiau, U. Dayan, A. Temara, Atmos. Environ. 45 (2011) 6788–6801. [9] M.E. Birch, R.A. Cary, Aerosol Sci. Technol. 25 (1996) 221. [10] P. Paatero, U. Tapper, Environmetrics 5 (1994) 111–126. [11] A.V. Polissar, P.K. Hopke, P. Paatero, J. Geophys. Res. (1998) 19045–19057. [12] J.H. Seinfeld, S.N. Pandis, Atmospheric Chemistry and Physics: From Air Pollution to Climate Change, first ed., J. Wiley, New York, USA, 1998. [13] R.R. Draxler, G.D. Rolph, HYSPLIT Model, NOAA ARL READY, 2003 Website: .
[14] S. Nava, S. Becagli, G. Calzolai, M. Chiari, F. Lucarelli, P. Prati, R. Traversi, R. Udisti, G. Valli, R. Vecchi, Atmos. Environ. 60 (2012) 444–452. [15] M. Chiari, F. Lucarelli, F. Mazzei, S. Nava, L. Paperetti, P. Prati, G. Valli, R. Vecchi, X-Ray Spectrom. 34 (2005) 323–329. [16] F. Marenco, P. Bonasoni, F. Calzolari, M. Ceriani, M. Chiari, P. Cristofanelli, A. D’ Alessandro, P. Fermo, F. Lucarelli, F. Mazzei, S. Nava, A. Piazzalunga, P. Prati, G. Valli, R. Vecchi, J. Geophys. Res. 111 (2006) D24202, http://dx.doi.org/10.1029/ 2006JD007145. [17] D.D. Cohen, J. Crawford, E. Stelcer, V.T. Bac, Atmos. Environ. 44 (2010) 320. [18] S. Nava, F. Lucarelli, F. Amato, S. Becagli, G. Calzolai, M. Chiari, M. Giannoni, R. Traversi, R. Udisti, Sci. Total Environ. 511 (2015) 11–20. [19] M. Morishita, G.J. Keeler, A.S. Kamal, J.G. Wagner, J.R. Harkema, A.C. Rohr, Atmos. Environ. 45 (2011) 1627–1637. [20] M. Morishita, G.J. Keeler, J.G. Wagner, J.R. Harkema, Atmos. Environ. 40 (2006) 3823–3834. [21] J.I. Yoo, K.H. Kim, H.N. Jang, Y.C. Seo, K.S. Seok, J.H. Hong, M. Jang, Atmos. Environ. 36 (2002) 5057–5066. [22] B. Mason, Principles of Geochemistry, third ed., Wiley, New York, 1966. [23] J.H. Seinfeld, Atmospheric Chemistry and Physics of Air Pollution, Wiley, New York, 1986. [24] H.B. Singh, Composition, Chemistry and Climate of the Atmosphere, Van Nostrand Reinhold, New York, 1995. [25] V. Bernardoni, R. Vecchi, G. Valli, A. Piazzalunga, P. Fermo, Sci. Total Environ. 409 (2011) 4788–4795.
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Study of air pollution in the proximity of a waste incinerator V. Barrera a, G. Calzolai a,b, M. Chiari b, F. Lucarelli a,b,⇑, S. Nava b, M. Giannoni b, S. Becagli c, D. Frosini c a
Department of Physics and Astronomy – University of Florence, Via G. Sansone 1, 50019 Sesto Fiorentino (Fi), Italy National Institute of Nuclear Physics (INFN) – Florence, Via G. Sansone 1, 50019 Sesto Fiorentino (Fi), Italy c Department of Chemistry – University of Florence, Via della Lastruccia 3, 50019 Sesto Fiorentino (Fi), Italy b
a r t i c l e
i n f o
Article history: Received 15 March 2015 Received in revised form 15 July 2015 Accepted 9 August 2015 Available online 18 September 2015 Keywords: PIXE Atmospheric aerosol PIXE Positive Matrix Factorization Biomass burning
a b s t r a c t Montale is a small town in Tuscany characterised by high PM10 levels. Close to the town there is a waste incinerator plant. There are many concerns in the population and in the press about the causes of the high levels of pollution in this area. Daily PM10 samples were collected for 1 year by the FAI Hydra Dual sampler and analysed by different techniques in order to obtain a complete chemical speciation (elements by PIXE and ICP-MS, ions by Ion Chromatography, elemental and organic carbon by a thermo-optical instrument); hourly fine (<2.5 lm) and coarse (2.5–10 lm) PM samples were collected for shorter periods by the Streaker sampler and hourly elemental concentrations were obtained by PIXE analysis. Positive Matrix Factorization identified and quantified the major aerosol sources. Biomass burning turned out to be the most important source with an average percentage contribution to PM10 of 27% of and even higher percentages during the winter period when there are the highest PM10 concentrations. The contribution of the incinerator source has been estimated as about 6% of PM10. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Particle Induced X-ray Emission (PIXE) is a suitable technique for analysing aerosol samples [1,2] due to its ability to carry out a multi-elemental analysis of the particulate deposited on the filter surface without any solubilization procedure, therefore shortening the analysis time and reducing the sample contamination risk. In particular it is unrivalled for the detection of the crustal elements. However, it provides only part of the desired information with regard to the chemical composition therefore it is mandatory to perform also at least measurements for important ionic species (e.g., ammonium, nitrate), for organic carbon (OC) and elemental carbon (EC). Finally Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES) or Mass Spectroscopy (ICP-MS) and PIXE can be used as complementary techniques and their combined use is able to give powerful information on chemical speciation of metals in the particulate matter (PM). By using the appropriate extraction conditions (extracting solution composition, pH, temperature, pressure, contact time) the metal fraction which is more ‘‘available” for the natural systems is obtained and in this way it is possible to evaluate better the impact of heavy metals on the environment and the human health [3]. ⇑ Corresponding author at: via Sansone 1, I-50019 Sesto Fiorentino (Firenze), Italy. Tel.: +39 055 4572274; fax: +39 055 4572641. E-mail address: [email protected] (F. Lucarelli). http://dx.doi.org/10.1016/j.nimb.2015.08.015 0168-583X/Ó 2015 Elsevier B.V. All rights reserved.
Standard daily sampling allows the study of aerosol composition for a long period covering all the seasonal changes in aerosol composition and tracing sources which may be present only in specific periods; however some particulate emissions change within a few hours and source apportionment receptor models need a series of samples containing material from the same set of sources in differing proportions. Increasing the time resolution of the measurements produces samples that have greater between-sample variability in the source contributions than samples integrated over longer time periods. Therefore time-resolved measurement can give a valuable help in source identification. Montale is a small town in Tuscany characterised by high PM10 levels. There are many concerns in the population and in the press about the causes of the high levels of pollution in this area, mainly because close to the town there is a waste incinerator plant. However many other sources can give a relevant contribution to PM10: many fields are present in the surroundings therefore a relevant crustal contribution could be present; biomass burning is used as heating system and also open fires are often present for the combustion of pruning; traffic, too, could give a relevant contribution. Therefore a multidisciplinary approach involving different analytical techniques is necessary to obtain a complete chemical speciation which can be used as a starting point for the application of multivariate statistical methods like Positive Matrix Factorization (PMF) to identify the PM sources and their contribution to the PM mass.
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
The Regional Government has promoted an extensive field campaign for the aerosol characterization in Montale, to give to policymakers the knowledge and the tools for a significant reduction of the main anthropogenic emissions. We will present here some of the results obtained for the first period of the campaign both for the daily and the hourly resolution samples.
2. Methods 2.1. Sampling Montale is a small town located 35 km W of Florence (Italy) with about 10,000 inhabitants. The sampling site is close to a garden and adjacent to a public car park, mainly for residential use. On a wider scale, the site can be classified as a ‘‘suburban” site, as it is in a built area but close to vast not urbanised areas. PM10 samples were collected on a daily base (from midnight to midnight), every second day, for 1 year (from December 2013 to December 2014) by the low-volume (2.3 m3/h, EU rule EN 12341) FAI Hydra Dual sampler equipped with two inlets so that aerosol can be simultaneously collected on Teflon and Quartz fibre filters (47 mm diameter), thus allowing the application of different analytical techniques. During shorter periods (2 weeks in winter and 2 weeks in summer) the aerosol was also collected by a low volume (1 lpm) Streaker sampler. In this device particles are separated on two different stages: the coarse fraction (2.5 lm < Dae < 10 lm) and the fine fraction (Dae < 2.5 lm), collected on a Kimfol foil and on a Nuclepore filter respectively. The two collecting substrata are paired on a cartridge which, rotating at constant speed for 1 week, produces a circular continuous deposition of particulate matter (‘‘streak”) on both stages [4].
2.2. Measurements and data analysis PM10 daily mass concentrations were obtained by weighing the Teflon filters by an analytical balance in controlled conditions of temperature (20 ± 1 °C) and relative humidity (50 ± 5%). Samples on Teflon filters are cut in three parts. On one half of the filter PIXE is used to measure the concentrations of all the elements with atomic number Z > 10. PIXE analyses were performed at the 3MV Tandetron accelerator of the INFN-LABEC laboratory, with the new external beam set-up similar to the one extensively used so far [5] but improved by the duplication of the SDD detector used for medium–high energy X-rays [6]; the measuring time was therefore reduced to only 90 s/sample with better statistics. Each sample was irradiated for 90 s with a 3.0 MeV proton beam (2 mm2 spot, 10–150 nA intensity). A filter scanning was carried out to analyse most of the deposit area. PIXE spectra were fitted using the GUPIX code [7] and elemental concentrations were obtained by a calibration curve from a set of thin standards of known areal density (Micromatter Inc.). The water-soluble fraction (sample extraction in ultra-pure MilliQ water in ultrasonic bath) of inorganic cations, inorganic anions and low molecular weight organic anions is measured by Ion Chromatography (IC) on one quarter of the filter [8]; ICP methods are used to determine the ‘‘soluble fraction” (in the acidic extraction conditions) of several major (ICP-AES) and trace (ICP-MS) metals on the remaining quarter of the filter [3]. 0.1% sub-boiled distilled HNO3 (pH = 1.5) in an ultrasonic bath for 15 min at room temperature was used as extraction method. This fraction represents the most ‘‘available” metal fraction (including free metal, labile complexes, carbonate and bicarbonate salts), considering the pH = 1.5 as the lowest limit for ‘‘natural” pH values [3].
113
A 1.5 cm2 punch of the Quartz fibre filters is used for Organic and Elemental Carbon assessment by Thermo Optical Transmission analysis (using a Sunset Lab analyser [9]) at LABEC. Streaker samples were analysed by PIXE, with the same experimental set-up used for daily samples. A properly collimated proton beam (2.7 MeV, 50–300 nA) was used to scan the deposit in steps corresponding to 1 h of aerosol sampling, thus providing the elemental concentrations with hourly time resolution. Thanks to the improved efficiency of the experimental set-up, 90 s. of measuring time for each spot corresponding to 1 h of sampling was enough to obtain a good statistics [6]. We report here the first results obtained for the winter and spring season. Positive Matrix Factorization (PMF) has been applied to this preliminary data set (daily and hourly samples separately) aiming at a first identification and quantification of the major aerosol sources, using the EPA PMFv5 software. PMF is an advanced factor analysis technique based on a weighted least square fit approach [10]; it uses realistic error estimates to weigh data values and imposes non-negativity constraints in the factor computational process. Briefly, the PMF factor model may be written as X = GF + E, where X is a known n by m matrix of the m measured chemical species in n samples; G is an n by p matrix of source contributions to the samples (i.e. time variations of the p factor scores); F is a p by m matrix of factors composition (often called source profiles). G and F are factor matrices to be determined and E is defined as a residual matrix. Input data were prepared using the procedure suggested by Polissar [11] and PMF results for different number of factors and multiple values of FPEAK were systematically explored to find out the most reasonable solution. For daily samples, to obtain absolute source profiles and contributions the aerosol mass was introduced as a variable with a 400% error. For hourly data, only elemental concentrations are measured and no information about the PM mass concentration is available: in this case, only relative source profiles (elemental ratios within the composition of the identified sources) can be obtained and source time series are in arbitrary units. 3. Results 3.1. PM10 mass PM10 concentrations are reported in Fig. 1. Lower levels were recorded in spring, with values between 10 and 20 lg/m3, except for a peak close to the PM10 daily limit threshold on 22 May (44.7 lg/m3) in correspondence of an episode of Saharan dust transport (see below). In winter levels were far higher with many concentration peaks around 100 lg/m3 (up to 174 lg/m3 on December 20th). This is due to the typical winter weather conditions with greater atmospheric stability, with a reduced height of the boundary layer and a poor dispersion of the pollutants themselves 3.2. Chemical composition As an example of the results obtained from the analysis of the daily samples, in Fig. 2 the Na, Si, K and S concentrations (obtained by PIXE) are reported. Na is a typical marker of sea-salt aerosol (together with Mg and Cl). Several episodes with a strong increase of Na, always together with Mg, sometimes with sometimes without a corresponding increase in Cl concentration, are present, pointing to the transport of marine aerosol. In all the cases the Mg/Na ratio (0.14) is very similar to the one typical of bulk sea-water (Mg/Na = 0.12, [12]) but often there is a strong Cl depletion probably due to reactions
114
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
Fig. 1. PM10 mass concentrations (lg/m3) during the first period of the sampling campaign.
Fig. 2. Concentrations (lg/m3) of Na, Si, K and S obtained by PIXE.
with sulphuric and nitric acids producing gaseous HCl during air masses transport. The sea-salt component was obtained summing up Na (corrected for the crustal contribution), Mg, Cl and ss-S (obtained as 0.253*ssNa, ssNa = Na - (Al * 0.348)); it represents a negligible contribution that, in general, is below 1.5 lg/m3 except in particular cases concentrated in the first half of February, in which higher concentrations (3.3 lg/m3 on February 8th, 7.8 lg/m3 on February 14th, and 2.9 lg/m3 on February 16th) were observed; in these days there was always a strong wind coming from the Tyrrhenian Sea that brought the marine aerosol to the interior of the region. Si is a typical marker of crustal dust both of local origin and longrange transport from desert regions, such as the Sahara. In Fig. 2 events of probable Saharan origin on days in which air masses trajectories are coming from the Sahara region (calculations with the transport model HYSPLIT NOAA’s Air Resources Laboratory [13]) show high Si concentrations and also a simultaneous increase of all the other typical crustal elements. Since PIXE is a quantitative technique, it allowed obtaining a further proof of the Saharan origin by the observation of the inter-elemental ratios, whose values are different for the desert and the local dust. The elemental ratios that show the most significant differences between Saharan and nonSaharan days are those involving either Fe or Ca, due to a Fe and Ca enrichment of local dust with respect to the Saharan dust as
already found in other studies [14–16]. For example, the Fe/Ca, Fe/Al, Ca/Al ratios changed from 1.17, 3.12, 3.44 during normal days to 0.7, 0.91, 1.45 during these Saharan episodes. An estimate of the soil dust component concentration can be calculated considering the crustal elements (obtained by PIXE) as oxides [17]; corrections were however applied to this formula to take into account sea-salt contributions to Na and Mg and possible anthropogenic contributions to the other elements [14]. Fig. 3 shows the time trend of the soil dust component; it contributes, in percent, more than twice in the spring compared to winter. In some Saharan dust episodes the concentration reaches values of 6–9 lg/m3 with a maximum of 10.5 lg/m3 (corresponding to approximately 25% of the PM10 mass) on April 3rd. In the rest of the days not affected by these particular events, the concentration of the crustal component remains below 5 lg/m3. Between 8 and 22 March it is possible to notice a plateau due to some Saharan dust intrusions. Sulphur, which is essentially a secondary aerosol component, has instead higher values in spring. The best conditions of atmospheric circulation on a regional scale (e.g. transport of SO2 and sulfate from thermo-electric plants located outside the city to the sampling site) and the increased efficiency of photochemical oxidation of SO2 to sulfate, may explain the increase of the sulfur in the warmer period.
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
115
Fig. 3. Concentration (lg/m3) of the soil dust component obtained from the crustal elements measured by PIXE. The most intense episodes related to Saharan dust transport are indicated.
K is present in the soil dust but it is also a typical marker of biomass burning. The remarkable concentration of K in winter points to a non negligible impact of biomass burning in the area. Fig. 4 shows the average percentage contributions of the main components of the aerosol for all the analysed samples and for only winter and spring samples separately. The carbon component Particulate Organic Matter (POM) + EC gives the main contribution, followed by the secondary inorganic and the crustal components. The marine contribution is negligible, as expected. The percentage of POM is particularly high in the winter period, when the values of PM10 are greater. This may be probably due to an increase in emissions (heating by BB). In the spring season, the percentage of crustal component and secondary inorganics increase. The high value of the ratio OC/EC (on average about 7), especially in winter, is symptomatic of a strong contribution from sources not related to traffic such as biomass combustion. A similar trend was already observed in another area in Tuscany characterised by high PM10 concentrations caused mainly by the impact of biomass burning [18]. Within the secondary inorganic component, nitrates show higher values in winter, while the sulfates have the highest concentrations during the warmer seasons. The seasonality of nitrates may be due to an increase in emissions of gaseous precursors (for the presence of domestic heating or for an increase of vehicular traffic) but also to the weather conditions (less dispersion of local pollutants in winter and increased volatility in summer). The increase of the sulphates in the warmer period was already explained above.
Mn, Fe, Ni, Cu, Zn, Se, Br, Sr, Zr by PIXE, Na+, Cl , NH+4, NO-3, SO24 by IC, V, Ba, Mo, Cd by ICP) and source apportionment by PMF allowed identifying 10 aerosol sources;
3.3. Aerosol source apportionment
3.3.4. Local dust Characterised by crustal elements (Mg, Al, Si, K, Ca, Ti, Fe) with enrichment of EC, OC, Ca and Fe, it is associated to an ‘urban’ dust (EC, OC and Fe may be produced by pneumatics, brakes and asphalt
The combination of a complete PM chemical speciation (EC/OC by Thermo Optical Transmission analysis, Mg, Al, Si, K, Ca, Ti, Cr,
3.3.1. Incinerator Mostly composed by EC, OC, NO3 and traced by specific elements (Cl , Pb, Cd, Zn) that can be associated to incinerator plant emissions [19–21]; this contribution is present during all the campaign (Fig. 5). 3.3.2. Traffic Mainly composed by OC and EC and traced by EC, Fe, Cu, Ba and Mo; it has a maximum contribution in winter when there is atmospheric stability. The ratio OC/EC of 4 reveals a significant proportion of secondary organic aerosols from oxidation of primary volatile organic compounds (VOCs) from fuel combustion and from enhanced anthropogenic transformation of biogenic VOCs by NOx. 3.3.3. Saharan dust (SD) Mainly characterised by crustal elements (Mg, Al, Si, K, Ca, Ti, Fe). The source profile is very close to the average crust composition, with enrichment factors (EF), calculated with respect to Al using the average continental crust composition reported by [22] close to 1. The time trend is characterised by concentration peaks during days classified as Saharan intrusions (on the basis of both models and satellite observations) and it is close to zero in all other days (Fig. 5).
Fig. 4. Average percentage contributions of the main components of the aerosol for all the analysed samples and for only winter and spring samples separatedly.
116
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
Fig. 5. Time trend of the contribution (lg/m3) of some of the sources identified by the PMF analysis of the daily filters: Saharan dust, biomass burning, incinerator.
wear, Ca by construction works). As opposed to the Saharan dust, this source contributes all the year long. 3.3.5. Sea salt Characterised by Na+, Mg and Cl , and, to a lesser extent, Mg and Br, with inter-elemental ratios in accordance with those reported in literature for seawater. The time pattern is characterised by short episodic peaks, occurring when air masses are directly transported by strong winds from the Tyrrhenian Sea. 3.3.6. Aged sea salt Characterised by Na+ and Mg, it is contaminated by NO3 and 2 SO4 , and depleted in Cl . It is known that different heterogeneous reactions between airborne sea-salt particles and gaseous pollutants, like nitric and sulfuric acid, may lead to Cl volatilization and formation of nitrates and sulfates [23,24]. Thus, this source is a mixture of anthropogenic and natural source contributions. 3.3.7. Heavy oil combustion Composed by OC and sulphate, and characterised by tracers as V and Ni. This source is due to residual oil combustion processes, mostly from activities that are located outside the area, like energy production, refinery, industrial plants and shipping. 3.3.8. Secondary nitrates and organics Mostly composed by NO3 and OC, it is to be associated to a secondary local component. NO3 are, indeed, produced by oxidation in atmosphere of NOx that is derived from local combustion processes as heating, biomass burning and traffic. The time trend is characterised by an important seasonal periodicity with higher values until March (see Section 3.2). 3.3.9. ‘Secondary sulphates Mainly composed by SO24 and NH+4, it is associated to a regional secondary component, which originates from atmospheric SO2 (emitted mainly from thermo-electric plants, and quickly distributed on a regional area). The seasonal trend has higher values in spring.
3.3.10. Biomass burning (BB) It is mainly correlated with OC and EC and tracers as K, Zn, Br and Pb. The OC/EC ratio in this profile (7) is within the ranges reported in literature for this source [25]). As expected, the contributions of this source are highly season-dependent with maxima in winter (Fig. 5) when, as already observed, there are most of the exceedances of the limits (with a maximum contribution up to 59 lg/m3), and very low in spring, when it is mainly due to the open fires. The pie chart of the identified source contributions to PM10 is reported in Fig. 6. The BB source is the one which gives the most relevant contribution (27%) followed by secondary nitrates (17%). Their contribution increases even more during the days in which the daily limit was exceeded, up to 34% and 27%. 3.4. Hourly resolution data The use of the Streaker sampler allowed obtaining high time resolved and size segregated data, i.e. the elemental concentrations with hourly time resolution, in the fine and coarse aerosol fractions, for about two weeks during the winter season (January21– February 8). PMF identified 7 sources for the fine fraction and 5 for the coarse one. All the sources already identified from the analysis of daily data, have been found, except the secondary nitrate source (which was not identified since PIXE cannot detect nitrogen), all with similar source profiles as those obtained for daily data. The results confirmed and reinforced the results obtained by the daily samples. As an example the hourly time pattern of the SD, BB, and incinerator sources are reported in Fig. 7. The SD source is only present during episodes of transport of Saharan aerosol (confirmed by the back-trajectories of the air masses calculated with Hysplit model) and zero outside. The time trend of the BB source supports its identification as biomass burning for domestic heating: it is characterised by a periodic pattern with peaks starting in the evening and lasting several hours. The absence of the evening-night peak on some days is explained by the meteorological conditions. Some peaks during the day are probably due to open fires.
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
117
Fig. 6. Pie chart of the identified source contributions to PM10.
Fig. 7. Hourly temporal patterns of factor contributions (arbitrary units) obtained by the PMF analysis of hourly data in the fine fraction: Saharan dust, biomass burning, incinerator.
The incinerator source does not show any particular daily pattern but it is characterised by sporadic peaks. 4. Conclusions A preliminary characterization of particulate matter in Montale (Tuscany) has been carried out and the main aerosol sources have been identified. Biomass burning turned out to be the most important source with an average percentage contribution of 27% of PM10 and very high level during the winter period. It is worth noting that the cold season is characterised by stagnant atmospheric conditions that favour the accumulation of pollutants in the
boundary layer, thus producing an increase in contributions also from other sources. The contribution of the incinerator source has been estimated as about 6% of PM10. Hourly data usefully supported and reinforced results obtained by the analysis of daily samples. Data elaboration is still in progress and PMF analysis of the whole 1 year long data set will complete this study. Acknowledgements This study was supported by the Regional Government of Tuscany, in the framework of PATOS2.2 project. V. Barrera acknowledges CONACYT program (Grant 208131) for financial support.
118
V. Barrera et al. / Nuclear Instruments and Methods in Physics Research B 363 (2015) 112–118
References [1] W. Maenhaut, K. Malmqvist, in: R.V. Grieken (Ed.), Handbook of X-ray Spectrometry, Marcel Dekker, Antwerp, 2001, pp. 761–771. [2] F. Lucarelli, S. Nava, G. Calzolai, M. Chiari, R. Udisti, F. Marino, X-Ray Spectrom. 40 (2011) 162–167. [3] R. Traversi, S. Becagli, G. Calzolai, M. Chiari, M. Giannoni, F. Lucarelli, S. Nava, F. Rugi, M. Severi, R. Udisti, Nucl. Instr. Meth. B 318 (2014) 130–134. [4] Website:
[14] S. Nava, S. Becagli, G. Calzolai, M. Chiari, F. Lucarelli, P. Prati, R. Traversi, R. Udisti, G. Valli, R. Vecchi, Atmos. Environ. 60 (2012) 444–452. [15] M. Chiari, F. Lucarelli, F. Mazzei, S. Nava, L. Paperetti, P. Prati, G. Valli, R. Vecchi, X-Ray Spectrom. 34 (2005) 323–329. [16] F. Marenco, P. Bonasoni, F. Calzolari, M. Ceriani, M. Chiari, P. Cristofanelli, A. D’ Alessandro, P. Fermo, F. Lucarelli, F. Mazzei, S. Nava, A. Piazzalunga, P. Prati, G. Valli, R. Vecchi, J. Geophys. Res. 111 (2006) D24202, http://dx.doi.org/10.1029/ 2006JD007145. [17] D.D. Cohen, J. Crawford, E. Stelcer, V.T. Bac, Atmos. Environ. 44 (2010) 320. [18] S. Nava, F. Lucarelli, F. Amato, S. Becagli, G. Calzolai, M. Chiari, M. Giannoni, R. Traversi, R. Udisti, Sci. Total Environ. 511 (2015) 11–20. [19] M. Morishita, G.J. Keeler, A.S. Kamal, J.G. Wagner, J.R. Harkema, A.C. Rohr, Atmos. Environ. 45 (2011) 1627–1637. [20] M. Morishita, G.J. Keeler, J.G. Wagner, J.R. Harkema, Atmos. Environ. 40 (2006) 3823–3834. [21] J.I. Yoo, K.H. Kim, H.N. Jang, Y.C. Seo, K.S. Seok, J.H. Hong, M. Jang, Atmos. Environ. 36 (2002) 5057–5066. [22] B. Mason, Principles of Geochemistry, third ed., Wiley, New York, 1966. [23] J.H. Seinfeld, Atmospheric Chemistry and Physics of Air Pollution, Wiley, New York, 1986. [24] H.B. Singh, Composition, Chemistry and Climate of the Atmosphere, Van Nostrand Reinhold, New York, 1995. [25] V. Bernardoni, R. Vecchi, G. Valli, A. Piazzalunga, P. Fermo, Sci. Total Environ. 409 (2011) 4788–4795.