lable at ScienceDirect

Atmospheric Environment 88 (2014) 275e284

Contents lists avai

Atmospheric Environment

journal homepage: www.elsevier .com/locate/atmosenv

Long-term particle measurements in Finnish Arctic: Part I e Chemicalcomposition and trace metal solubility

James R. Laing a, Philip K. Hopke a,*, Eleanor F. Hopke a, Liaquat Husain b,c,Vincent A. Dutkiewicz b,c, Jussi Paatero d, Yrjö Viisanen d

aCenter for Air Resources Engineering and Science, Clarkson University, Potsdam, NY 13699, USAbNew York State Department of Health, Wadsworth Center, Empire State Plaza, Albany, NY 12201-0509, USAcAtmospheric Sciences Research Center, and Department of Environmental Health Sciences, School of Public Health, SUNY, Albany, NY 12201-0509, USAd Finnish Meteorological Institute, Helsinki, Finland

h i g h l i g h t s

� Forty-seven years of week-long Finnish Arctic PM chemical compositions are given.� Seasonality of the chemical species is presented.� Trace metals were analyzed providing near-total and soluble metal concentrations.� Trace metal solubilities in Arctic aerosol are presented.

a r t i c l e i n f o

Article history:Received 10 October 2013Received in revised form25 February 2014Accepted 1 March 2014Available online 4 March 2014

Keywords:Arctic aerosolCompositionArctic hazeTrace metal solubility

DOI of original article: http://dx.doi.org/10.1016/j.* Corresponding author.

E-mail address: phopke@clarkson.edu (P.K. Hopke

http://dx.doi.org/10.1016/j.atmosenv.2014.03.0021352-2310/� 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

Week-long total suspended particle filter samples collected between 1964 and 2010 from Kevo, Finlandwere analyzed for trace metals, soluble trace metals and major ions. Ion chromatography was used tomeasure major ions. Inductively coupled plasma mass spectrometry was used to measure the trace metaland soluble trace metal concentrations. Species of anthropogenic origin (V, Co, Cu, Ni, As, Cd, Pb, SO4

2�)have significantly higher concentrations compared with other Arctic locations. A clear seasonal trendwith winter/spring maxima and summer minima is observed for most species, although it is less pro-nounced than those found in the high Arctic due to the relative proximity to Eurasian pollution sources.High concentrations of Cu (14.1 ng/m3), Ni (0.97 ng/m3), and Co (0.04 ng/m3) indicate the influence ofnon-ferrous metal smelters on the Kola Peninsula, although Cu unexpectedly did not correlate with Ni orCo. Ni and Co were highly correlated. Cu, Re, Tl, As, W, and V had high solubilities (61%e87%), Co and Nihad solubility’s of w33%, and Pb had a solubility of 22.9%. The solubility of metals can help determine ifthe source is natural or anthropogenic. It also dictates the bioavailability of metals once introduced to theenvironment.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Arctic haze has been actively studied for over 40 years. Highconcentrations of ground level particulate matter (PM) as well aslayers aloft are observed in the late winter and early spring, whilelow concentrations are typically observed during the rest of theyear. Extensive reviews characterizing PM and the related transportphenomenon have been published (Barrie, 1986; Shaw, 1995; Law

atmosenv.2014.01.015.

).

and Stohl, 2007; Quinn et al., 2007). Long-range transport of mid-latitude anthropogenic and wild fire emissions drive the season-ality observed in the Arctic. The ground-level winterespringmaxima in the Arctic are mainly due to enhanced long-rangetransport from Eurasia. Eurasian sources are higher in latitudethan North American or Asian sources and therefore are morereadily encompassed by the polar front in winter. Atmosphericblocking and the development of anticyclones over Eastern Europeand Siberia also enhance rapid transport into the Arctic from Eur-asia (Iversen and Joranger, 1985; Raatz and Shaw, 1984; Raatz,1989). During winter, stable, dry, and cold conditions enhancepollutant’s atmospheric lifetime by restricting vertical mixing anddilution, minimizing wet scavenging, and slowing the oxidation of

Fig. 1. Map of the sampling location, Kevo, Finland.

J.R. Laing et al. / Atmospheric Environment 88 (2014) 275e284276

SO2 (Barrie and Barrie, 1990; Raatz and Shaw,1984). Transport fromNorth America or Asia to the Arctic are along cyclonic paths wherewet scavenging is very effective at ground level (Raatz, 1991).Transport from these regions tends to be at higher elevations alongsloping isentropes into the free troposphere, entering the Arcticfrom above by radiative cooling (Stohl, 2006).

There are only a few long-term studies of Arctic aerosolcomposition. Measurements of PM compositions have been madeat Alert, Canada by the Meteorological Service of Canada starting in1980 (Barrie and Hoff, 1985; Gong and Barrie, 2005; Sirois andBarrie, 1999; Xie et al., 1999a,b); Barrow, Alaska by the NationalOceanic and Atmospheric Administration (NOAA) and seven Na-tional Park Service locations in Alaska starting in 1986 (Polissaret al., 1998a, 1998b, 1999, 2001; Quinn et al., 2009); and theZeppelin Observatory at Ny-Ålesund by the Norwegian Institute forAir Research (NILU) starting in 1994 (Berg et al., 2004; Hirdmanet al., 2010a,b). Decreasing concentrations of anthropogenic com-ponents have been found with a sharp decrease in the early 1990’scoinciding with the economic collapse of the Soviet Union andEastern Europe.

The solubility of trace metals is important because of its envi-ronmental implications, and also because it can provide informa-tion about its emission source. Solubility depends on the source ofemissions, pH of the particle, and particle size (Colin et al., 1990;Desboeufs et al., 2001, 2005). Trace metals on particles of anthro-pogenic or marine origins are more soluble than crustal particles(Desboeufs et al., 2005; Giusti et al., 1993). Oxides are usuallyformed during combustion, which are highly soluble. Crustalderived elements are bound to insoluble alumino-silicates. Thesolubility of trace metals determines their bioavailability andtoxicity. A particle consisting of a highly soluble metal will be muchmore bioactive and more readily taken up by plants and animals(Qureshi et al., 2006).

For this study, the chemical composition of weekly PM sam-ples from 1964 to 2010 collected in the Finnish Arctic weremeasured, providing a 47-year dataset of trace metals, water-soluble trace metals, and ionic species. This dataset will provide

valuable information about trace metal solubility in the EuropeanArctic, long-term changes in the chemical composition of theEuropean Arctic aerosol, and their potential source areas. A pre-vious paper has analyzed non sea-salt SO4

2� and methane sulfonicacid (MSA) (Laing et al., 2013), and black carbon (BC) (Dutkiewiczet al., submitted for publication). Detailed trend analysis andsource identification of the trace metals and ionic species re-ported in this paper will be presented in a companion paper(Laing et al., 2014).

2. Methods

2.1. Sampling site

The Kevo Subarctic Research Institute (latitude 69�450N, longi-tude 27�020E, height 98 m above sea level) is located 350 km northof the Arctic Circle (Fig. 1). The sampling location is described inmore detail by Yli-Tuomi et al. (2003). The site is situated in thebirch sub-zone of the boreal coniferous forest. The topography ofthe surrounding area is characterized by gently sloping fell high-lands with river valleys. The elevation is mostly between 100 and400 m above sea level. The area is sparsely populated (0.4 inhab/km2).

The sun shines without setting from mid-May until the end ofJuly, and remains below the horizon from late November to mid-January. The mean temperature of the coldest month (January)is �14.0 �C and the warmest month (July) þ13.1 �C. The annualmean temperature is �1.3 �C. The ground is already covered withsnow in October, and on average, the snow cover remains untilmid-May. The mean annual precipitation is 433 mm of which 30e40% falls as snow (Pirinen et al., 2012).

When measurements started in 1964, there were four buildings.From 1968 to 1978, six additional buildings were built, and since1990 two newbuildings have been built. Themeteorological stationchanged from wood burning to electricity in 1971. All of the newbuildings are heated with electricity.

J.R. Laing et al. / Atmospheric Environment 88 (2014) 275e284 277

2.2. Sample collection

Week-long filter samples were collected by the Finnish Meteo-rological Institute (FMI) starting in October 1964. The automatedsampler was established to monitor atmospheric radioactivity. Airwas alternatively drawn through two filters switching every 4 h(Paatero et al., 1994). From October 1964 to March 1978 square12 cm � 14 cm Whatman cellulous filters (Grade 42) were usedwith a flowrate of approximately 7 m3/h, resulting in weeklysample volumes of about 1200 m3. Starting in January 1979, rect-angular 14 cm � 14 cm Whatman GF/A glass fiber filters were usedwith an average flowrate around 27 m3/h and a weekly samplevolume of approximately 4500 m3. No filters were collected fromMarch 1978 to January 1979 when the sampler was being changed.Samples from April to June 1986 were not available for analysis dueto the Chernobyl disaster. The samplers did not have size-segregating inlets so total suspended particles were collected. Af-ter the filters were removed from the device, they were sent to theFMI Air Quality Department in Helsinki where they were countedfor Pb-210 six months after sampling (Paatero et al., 1998). The Pb-210 activity concentrations (Paatero et al., 2000), Rn-222 (Aaltonenet al., 2001) and total beta activity (FinnishMeteorological Institute,1984) results have been reported (Paatero et al., 2000). Filters werestored at room temperature in an envelope with other filters fromthe same year. Coated paper was used to separate the weeklysamples. The filters through February 1978 were previouslyanalyzed by ion chromatography (IC) for major ions and MSA, andtrace elements by Instrumental Neutron Activation Analysis (INAA)(Yli-Tuomi et al., 2003).

The remaining 1/4th of the Whatman 42 filters was cut exactlyin half using an acid-washed Plexiglas filter template and Teflon�

forceps in a laminar flow hood (EPA method IO-3.1, USEPA, 1999).The Plexiglas filter template, Teflon� forceps, and scalpel werewiped with a dry Kimwipe� between samples. One half of theWhatman 42 filters was acid-digested and analyzed for tracemetals by Inductively Coupled Plasma Mass Spectrometry (ICP-MS). The other half was analyzed for black carbon. The archivedwater extracted Whatman 42 samples from 1964 to 1978 werefiltered, acidified to 1.8% HNO3, and analyzed by ICP-MS.

The glass fiber filters were cut into two pieces in a laminar flowhood at Finnish Meteorological Institute (FMI). Half of each filterwas sent to Clarkson University and half kept in the FMI archives.The filters were cut into three equal sections in the samemanner asthe Whatman 42 filters. One section of the filter was analyzed forblack carbon (BC) (Dutkiewicz et al., submitted for publication), onesection was acid digested and analyzed for trace metals by ICP-MS,and one section water extracted and analyzed for trace metals byICP-MS and ions by IC.

2.3. Chemical analysis

2.3.1. Acid digestionOne section of the Whatman 42 filters (1964e1978) and one of

the three sections of the glass fiber filters (1979e2010) were aciddigested following EPA Method IO-3.1. To theWhatman 42 samples10 mL of extraction solution (5.55% HNO3/16.75% HCl diluted withMilli-Q water) was added to the vessel. To the glass fiber samples,15 mL of the extraction solution was added and 20 samples wereextracted per run. Samples were digested in Teflon� PFA digestionvessels in a CEM MARSXpress� microwave using a programwith aramp of 23 mine185 �C. The extracted samples were centrifuged.An aliquot was taken and diluted 1:10 with Milli-Q water toanalyze. Prior to every use, the Teflon� PFA digestion vessels werethoroughly washed, rinsed, and acid washed with HNO3using aMilestone TraceClean system. One solution blank and one filter

blank were included in every microwave run. Blank filters were notcollected at the time of collection. NewWhatman 42 andWhatmanGF/A glass-fiber filters were used as blanks, respectively. The lack ofauthentic period blank filters increases the uncertainty in the blankcorrected values by of the order of 5%

2.3.2. Water extractionOne section of the glass fiber filters (1979e2010) were water

extractedwith 20mL ofMilli-Qwater for 24 h at room temperature.After 24 h, the sample was split into two samples, one prepared forIC analysis and one for ICP-MS analysis. For ICP-MS, a 2 mL aliquotwas transferred to a 15 mL polypropylene centrifuge tube, dilutedto 10 mL with Milli-Q water and acidified to 2% HNO3. To theremaining 18 mL, 9 mL chloroform was added to prepare for ICanalysis. Chloroform was used to prevent potential bacterial ac-tivity. The IC samples were stored in a refrigerator after extraction.Every 20 samples, there was a solution blank and filter blank.

The archived water extracted samples analyzed for IC by Yli-Tuomi et al. (2003) were diluted 1:10, acidified to 1.8% HNO3, andfiltered using polypropylene syringeless filter (PTFE 0.2 syringefilter) for ICP-MS analysis.

2.3.3. ICP-MS analysisThe acid digested samples and the water extracted samples

were analyzed using a Thermo X-series 2 ICP-MS following EPAMethod IO-3.5. Collision cell technology with kinetic energydiscrimination (CCTED) using 7% H2 in He at 3 mL/min was used toreduce polyatomic isobaric interferences. Calibration standardspiked blanks and SRM 1648a,urban particulate matter (NationalInstitute of Standards and Technology, Gaithersburg, MD) wereused to validate the methods. Three different samples matriceswere analyzed; acid digested samples with 1.675% HCl and 0.555%HNO3 (1964e2010), chloroform water extracted samples acidifiedto 1.8% HNO3 (1964e1978), and water extracted samples acidifiedto 1.8% HNO3 (1979e2010). The detection limit was determined as 3times the standard deviation of 7 method blanks. Multi-elementstandards, Spex CLMS-2, CLMS-3, and CLMS-4, were used for thecalibration curves (SPEX CertiPrep, Metuchen, NJ). High-Puritystandards, CRM-TMDW, ICP-MS-C, and ICP-MS-D, were used toroutinely verify the calibration curve (High-Purity Standards,Charleston, SC). Calibration checks using the calibration standardsand duplicates were performed every 40 samples. An internalstandard using Bi, Ge, 6Li, Sc, Tb, Y diluted to 50 ppb was used tocorrect drift in the instrument. For elements with concentrationswell above the detection limits, the uncertainties in the measured,blank corrected values are of the order of 15e20%

Two extraction methods were used to determine trace ele-ments, acid-digestion for near-total element concentration, andwater extraction for water soluble concentration. Glass fiber filtershave large blank concentrations, especially when acid digested(Berg et al., 1993). Some elements were rendered unusable due tohigh blank concentrations (Li, Cr, Al, Zn, In, Cl�). In addition to theblanks being very high for these elements, there was considerablevariability in the blank concentrations. That variability in additionto using new filters as blanks instead of blank filters collected at thetime of sampling made it impossible to estimate reasonable blankvalues for these elements. High blanks were not a significant issuefor the water extracted samples. Thus, there are some elements forwhich the water extracted concentrations are presented withoutthe accompanying acid digested concentrations (Mn, Fe, As, Se, Mo,W, Tl).

There are also some elements for which the acid digested con-centrations are presented without the water extracted (Ag, Au).These elements did not stay in solution for the calibration andcheck standards without HCl present. All of the acid digested

Table 1For the calculation of the means and standard deviations (ng/m3), values below detection were given a value of ½ the detection limit.

Constituents Analytical method Arithmetic mean Arithmetic std Geometric mean Geometric std Number of >DL samples Number of samples missing <DL %

51V-ad ICP-MS 0.79 1.27 0.39 3.57 2265 73 2.6259Co-ad ICP-MS 0.04 0.05 0.03 2.99 1918 73 17.0860Ni-ad ICP-MS 0.97 1.22 0.62 2.96 2328 72 0.0465Cu-ad ICP-MS 14.1 24.5 7.63 2.93 2328 73 0107Ag-ad ICP-MS 0.27 4.8 0.04 3.98 2035 73 12.2118Sn-ad ICP-MS 4.09 6.34 2.70 2.30 2291 73 1.54121Sb-ad ICP-MS 0.14 0.27 0.09 2.93 2309 73 0.79185Re-ad ICP-MS 0.001 0.001 0.00 2.22 1503 75 34.28197Au-ad ICP-MS 0.02 0.06 0.01 3.02 1263 73 44.36208Pb-ad ICP-MS 4.62 7.22 2.75 2.68 2328 73 051V-we ICP-MS 0.57 0.97 0.20 5.09 2163 89 6.2155Mn-we ICP-MS 0.44 0.78 0.18 4.33 2311 89 0.0456Fe-we ICP-MS 2.05 4.15 0.80 5.38 1674 89 26.5759Co-we ICP-MS 0.02 0.03 0.01 3.39 2229 89 3.4660Ni-we ICP-MS 0.4 0.52 0.23 2.92 2294 89 0.7565Cu-we ICP-MS 6.99 13 3.26 3.61 2311 89 0.0475As-we ICP-MS 0.43 0.68 0.19 4.32 2255 89 2.3778Se-we ICP-MS 0.07 0.07 0.04 4.31 1877 89 18.1295Mo-we ICP-MS 0.03 0.04 0.01 4.22 1832 89 19.99111Cd-we ICP-MS 0.07 0.21 0.02 4.72 2305 89 0.29118Sn-we ICP-MS 0.13 0.32 0.05 4.23 1996 89 13.16121Sb-we ICP-MS 0.05 0.1 0.02 3.73 2304 89 0.33182W-we ICP-MS 0.01 0.01 0.00 3.46 1965 89 14.45185Re-we ICP-MS 0.001 0.001 0.00 1.73 1668 89 26.82205Tl-we ICP-MS 0.01 0.02 0.00 3.64 1863 89 18.7208Pb-we ICP-MS 0.92 1.92 0.33 4.70 2311 89 0.04BC 212 249 135 2.64 2310 91 0Naþ IC 216 131 194 1.58 2326 75 0Kþ IC 91.6 74.5 68.6 2.22 2326 75 0Mg2þ IC 11.2 10.5 7.59 2.73 2318 75 0.33Ca2þ IC 32.5 29.2 20.1 4.04 2291 75 1.46MSA IC 20.3 27.5 4.31 11.67 1955 75 15.45SO2�

4 IC 1469 1325 1066 2.26 2326 74 0nss� SO2�

4 IC 1415 1310 1007 2.33 2326 73 0

ad e acid digested.we e water extracted.nss� SO2þ

4 ¼ SO2þ4 ��0:2529*Na.

J.R. Laing et al. / Atmospheric Environment 88 (2014) 275e284278

elements will be designated “ad”, and thewater extracted elements“we”. For some elements and ions (Cu-ad, Sn-ad, Pb-ad, Na, K, Mg,Ca), the blank corrected concentrations for the glass fiber filtersamples were visibly higher than the cellulose filters. These higherblank values increased the uncertainties in these values byapproximately 15%. To correct for these differences and make acongruent dataset, the glass fiber filter concentrations werenormalized by setting the average of the first year of the glass fiberfilters equal to the last year of the cellulose filters.

2.3.4. IC analysisThe cellulose filters were previously extracted and analyzed

(Yli-Tuomi et al., 2003). The glass fiber filters were analyzed formajor anions (Cl� and SO4

2�), MSA and cations (Naþ, Kþ, Mg2þandCa2þ) using a DX-500 ion chromatograph with AS50 autosampler,GP50 gradient pump, and ED50 electrochemical detector. Prior toanalysis an aliquot of the sample was filtrated with GHP syringefilter for anion analysis and IC syringe filter for cation analysis. Toreduce the baseline conductivity and therefore the backgroundnoise ASRS 300 and CSRS 300 suppressors were used for anions andcations, respectively. The MSA and major anions were separatedwith AS4A-SC column using a gradient of 5e28 mM Na2B4O7. Thecations were analyzed with a CS12A column and 22 mM H2SO4eluent. The detection limit was determined by following the EPA 40CFR 136, Appendix B method (USEPA, 1985). The calibration stan-dards were Spex ICMIX 1 for anions, and Spex ICMIX3 and SpexICMIX4 for cations (SpexCertiPrep, Metuchen, NJ). Duplicate in-jections of 50 mL were used for all samples to ensure the results.Check standards from a different batch of stock standard were used

to verify the calibration curve. Calibration checks were performedevery 20 samples. Typical uncertainties in the measured, blankcorrected IC ion concentrations were 20%.

3. Results and discussion

3.1. Trace metal and cations

Table 1 shows the chemical species analyzedwith the arithmeticmean and standard deviation, geometric mean and standard de-viation, number of missing samples and number of samples belowdetection limit. For the calculations in Table 1, the samples belowdetection were given a value of ½ of the detection limit. Non-seasalt SO4

2�was calculated by subtracting 0.25 times Na fromSO4

2�(Li and Barrie, 1993). The acid digested samples do notrepresent the total concentration. Hydrofluoric acid was not used,and therefore silica-based minerals were not completely digested.Sulfate was the dominant constituent measured at Kevo with amean concentration of 1469 ng/m3, followed by BC (212 ng/m3), Na(216 ng/m3), and K (91.6 ng/m3). The most predominant metalswere Cu, Pb, and Sn. Sulfate and MSA, and BC have been discussedelsewhere (Dutkiewicz et al., submitted for publication; Laing et al.,2013). Time-series plots with a 1-year moving average of selectedconstituents are presented in Figs. 2e3 and Figs. S1eS4.

Constituents associated with anthropogenic emissions (V, As,SO4

2�) were 2- to 13-times higher than concentrations at Alert(Cheng et al., 1993; Gong and Barrie, 2005; Xie et al., 1999a) orvarious sites in Alaska (Polissar et al., 1998a). Concentrations oftrace metals (V, Co, Ni, Cu, Cd, and Pb) at Kevo were higher than

Fig. 2. Time-series (black) with 1-year moving average (red) for acid-digested V, Co, Ni, Cu, and Sn. For the moving averages, <DL values were given a value of ½ DL; gaps in thedataset were filled using the average of the value of the same time periods the year prior and after.

J.R. Laing et al. / Atmospheric Environment 88 (2014) 275e284 279

those found at Ny-Ålesund, Svalbard (Berg et al., 2004, 2008). Kconcentrations, which have anthropogenic sources as well asbiomass burning sources, were 10-times higher at Kevo comparedto Alert. These differences are to be expected since Kevo is muchcloser to anthropogenic sources, forest fires, and agricultural fires inEurope than the other Arctic sites.

In the scatter plot with BC, an edge is observed with a similarratio (K/BC ¼ 0.108) (Figure S5) to those found from burningdifferent types of wood (Fine et al., 2001, 2004a, 2004b), and offresh smoke plumes from wild-fires in Europe (Saarikoski et al.,2007). Moderate Resolution Imaging Spectroradiometer (MODIS)data, started in 2001, has been used to determine the geographiclocation of forest and agricultural fires (Justice et al., 2002). Most ofthe fire activity is from agricultural fires in Ukraine and EuropeanRussia, and forest fires in Russia (Hao et al., 2012). The Russianfederation and Eastern Europe were the largest contributors toagricultural fires (31e36%) globally (Korontzi et al., 2006), and allfires regionally (69% of Eurasia) (Hao et al., 2012). The fires inUkraine and European Russia coincide with the harvest of winterand spring grains and subsequent burning of the fields (Korontzi

et al., 2006; McCarty et al., 2012). Stohl et al. (2007) observed se-vere pollution events at Ny-Ålesund as a result of agricultural firesin the Baltic countries, Belarus, Ukraine, and Russia. There werevery few fires in Western European countries, which have enforcedburn bans since the 1980s (Korontzi et al., 2006). Niemi et al. (2006)used the MODIS data and transport modeling to estimate half of all“pollution” episodes captured by air quality monitoring stations insouthern Finland from 2001 to 2007 were caused by wildfires and/or agricultural fires in Belarus, Ukraine, and European Russia, theremainder being anthropogenic.

Most of the anthropogenic trace elements (V, Se, Pb) and sea saltspecies (Na, Mg) are highly seasonal with maxima in February andMarch, and minima in the summer (Figs. 4 and 5, S6eS8). Theseasonal trends at Kevo are not as well defined as those at Alert.Kevo is closer to known anthropogenic sources in Europe, thuspolluted air masses are more likely to be transported to Kevo in thesummer than the high Arctic. The industrial complexes on the KolaPeninsula in Russia heavily influence northern Finland and Norway(Berg and Steinnes, 1997; Paatero et al., 2008; Sivertsen et al., 1992).The majority of emissions on the Kola Peninsula come from two

Fig. 3. Same as Fig. 2 with water extracted As, Se, and Cd, and acid-digested Sb and Pb.

J.R. Laing et al. / Atmospheric Environment 88 (2014) 275e284280

non-ferrous ore roasting and metal smelting facilities, the Sever-onickel plant in Monchegorsk and the Pechenganickel plant inZapolyarnyy/Nikel (Tuovinen et al., 1993). In the early 1970’s theseplants started to process ore shipped from Noril’sk in addition tolocal ores. The Noril’sk ore has similar Ni and Cu content, but muchhigher sulfur content (w30% S) compared to the local ore (w6.5% S)(Bulatovic, 2007; Paatero et al., 2008; Stebel et al., 2007). All of theores processed are predominantly CueNi sulfide, composed ofpentlandite ((Fe,Ni)9S8), chalcopyrite (CuFeS2) and pyrrhotite(Fe1�xS). The local ores also contain a significant amount of cobalt(Bulatovicet al., 2007). The majority of Ni, Cu, and Co emissionscome from the Severonickel plant in Monchegorsk, but the emis-sions from the Pechenganickel plant in Zapolyarnyy/Nikel are alsosignificant (Paatero et al., 2008). Boyd et al. (2009) estimated that1994 emissions from the Kola Peninsula of Ni, Cu, and Co were1,916, 1,097, and 92.1 tons/year, respectively. The estimated totalemission of Ni, Cu, and Co in Europe in 1979 was 16,000, 15,500,and 2000 tons/year, respectively (Pacyna and Pacyna, 2001).Because of the emissions from these plants, the Kola Peninsula isone of highest areas of heavy metal deposition in Europe (Harmenset al., 2010). Ni, Cu, and As concentrations near the border betweenNorway and Russia were found to be 5e10 times higher than

concentrations measured at Birkenes in southern Norway(Sivertsen et al., 1992).

At Kevo Ni and Co concentrations are highly correlated (Fig. 6).The slope of the Ni vs Co scatter plot (18.8) is very similar to theestimates of Barcan (2002)’s estimates of Ni/Co yearly emissionsfrom the Severonickel smelter complex (19.7e29.28), and Boyd et al.(2009)’s emissions estimates from the Kola Peninsula’s copperenickel industry (20.8). The results for Cu unexpectedly did notcorrelate well with either Ni or Co (Fig. 6). The edges marked in theCovs Cu, andNi vs Cu scatter plots (Fig. 6) are close to the ratios fromthe Kola Peninsula’s copperenickel industry (Barcan, 2002; Boydet al., 2009), but there is clearly an additional source of Cu. Theseasonal trend for Cu shows nomaxima or minima (Fig. 4). The lackof seasonality along with the high concentrations indicates a localsource, most likely in the Kola Peninsula. The relative proximity(130e300 km) allows transport during the summer, so that theconcentrations are not as dependent on the atmospheric conditionsthat promote long range transport. The ratios of average concen-trations of Cu, Ni, and Co are similar to those found in mosses innorthern Finland (Harmens et al., 2010; Poikolainen et al., 2004).

Marine-derived particles are prevalent as Kevo is close to theNorwegian and Barents Seas. The majority of Mg seems to be sea

Fig. 4. Monthly concentrations for acid digested V, Co, Ni, Cu, Sn, Sb. The line inside the box represents the median, the top and bottom of the box are the 25th and 75th percentiles,and the 5th and 95th percentiles are the error bars.

J.R. Laing et al. / Atmospheric Environment 88 (2014) 275e284 281

salt, the Mg/Na ratio is 0.155, close to the ratio in seawater (0.13)(Mason,1966). Na, K and Ca have anthropogenic and crustal sourcesthat are larger than the sea salt concentrations, evidence of sea-saltratios were not obvious on the scatter plots.

V, As, Pb, Mo, Sb,W, Th, and Se have very strong seasonality withmaxima in FebruaryeMarch. Most emissions of these elements areassociated with the combustion of fossil fuels. The higher winter

Fig. 5. Same as Fig. 4 for water extracted

concentrations are caused bymore effective transport in the winterfrom the mid-latitudes. A clear edge on the V vs Ni scatter plot canbe seen, which is indicative of oil combustion (Fig. 7). The slope ofthe edge (0.38) is close to the Ni/V ratio from stationary fossil fuelcombustion (0.36) (Pacyna and Pacyna, 2001) and the combustionof oil (0.28) (Pacyna et al., 1984). Vanadium is also correlated withnss-SO4

2� (Pearson correlation ¼ 0.76), mainly coming from oil

As, Se, and Cd and acid digested Pb.

Fig. 6. Scatter plots of acid digested Co, Cu, and Ni.

J.R. Laing et al. / Atmospheric Environment 88 (2014) 275e284282

combustion in continental Europe (Pacyna et al., 1984; Pacyna andPacyna, 2001).

Mo, Se, and Sb are markers from coal burning (Pacyna et al.,1984; Pacyna and Pacyna, 2001). The scatter plots for these ele-ments do not show any distinct correlation. Pb is predominantlyemitted from cars using unleaded gasoline (Poikolainen et al.,2006; von Storch et al., 2003). Over the past five decades, therehave been significant changes in the use of fuels and regulatoryemission control policies. The long-term trends and source iden-tification over this time period will be discussed in a companionpaper (Laing et al., 2014).

3.2. Solubility of trace metals

Comparing the acid digested concentrations with the waterextracted concentrations gives an estimate of water solubility. Themean solubility of each element was determined from the slope ofthe Deming regression, which accounts for measurement errors, ofthe water extracted concentrations against acid digested concen-trations (Fig. 8). Some of the acid digested glass fiber filter samples

Fig. 7. Scatter plot of acid-digested Ni and V.

had significant blank values that rendered the values unusable. Forthese elements, the acid digested cellulose filter samples (1964e1978) are still valid, and the solubility of only the cellulose filters areprovided.

Table 2 summarizes the mean solubility and standard error forthe given trace metals for the cellulose filters, glass fiber filters, andthe complete dataset. The acid digestion was not necessarily acomplete digestion in that the alumino-silicates were notcompletely digested. The water extracted cellulose filters (1964e1978) were extracted in 2002 with chloroform saturated water andstored in a refrigerator. Eleven years passed between extraction andacidification and analysis by ICP-MS. The glass fiber filter sampleswere acid preserved 24 h after extraction and they were analyzedby ICP-MS within 6 months of extraction. These differences couldcause some of the discrepancy in the apparent solubility of certainmetals in cellulose and glass fiber filters. V and Pb show lowersolubility for the cellulose filters, while Co, Ni, Sb, and Re showhigher solubility for the glass fiber filter samples. Cu had similarsolubilities for both filter types.

Fig. 8. Scatter plot of water extracted Vanadium concentrations against acid digestedVanadium concentrations with regression lines for cellulose filter samples and glassfiber filter samples.

Table 2Solubility of selected elements for the cellulose filters, the glass fiber filters, and thecomplete dataset using Deming regression. Only values greater than the detectionlimit were used.

Cellulose filters Glass fiber filters 1964e2010

1964e1978 1979e2010

Solubility Std. error Solubility Std. error Solubility Std. error

V 59.1 1.36 69.7 1.16 67.9 0.92Mn 56.5 1.20Fe 11.1 0.78Co 54.5 4.36 31.1 1.01 32.8 0.91Ni 43.3 1.90 33.7 1.07 34.0 0.75Cu 61.2 2.72 62.1 2.01 60.8 1.20As 73.8 0.99Se 27.5 2.08Mo 32.3 2.85Cd 55.9 3.31Sn 3.00 0.62 0.47 0.10 0.56 0.10Sb 42.2 1.27 19.0 1.19 25.4 1.03W 66.8 73.52Re 102.5 6.64 72.4 1.73 74.7 1.63Tl 86.6 1.87Pb 24.4 0.83 57.6 2.45 22.9 0.42

J.R. Laing et al. / Atmospheric Environment 88 (2014) 275e284 283

Compared to the solubility of an urban aerosol, the solubility ofmost species (V, Mn, Fe, Co, Ni, As, Se, Sb, and Pb) analyzed weresignificantly lower at Kevo (Qureshi et al., 2006). Re, Tl, As, W, and Vhave the highest solubility (67e87%). The solubility of Co and Niwere similar, 33% and 34% for Co and Ni, respectively. Cu had anoverall solubility of 60.8%. The low solubility for Fe (11.1%) mayindicate a mostly crustal source. The solubility for Pb in this study(22.9%) is comparable to previously reported values (10e39%)(Qureshi, 2004 and references within).

4. Conclusions

Forty-seven years of weekly filters have been analyzed for tracemetals, soluble tracemetals, andmajor ions collected in the FinnishArctic. Concentrations of chemical species with anthropogenic or-igins were high compared to other Arctic sites due to Kevo’sproximity to Eurasian sources, specifically the Kola Peninsula’s in-dustrial area. Most of the chemical species showed seasonal max-ima in late winter/spring and minima in the summer, although theseasonality was weaker than that seen at sites further north in theArctic. This seasonality is due to enhanced long-range trans-portation in the winter and increased particle preservation due tothe stable, dry, and cold conditions.

Compared to the solubility of typical urban aerosol particles(Qureshi et al., 2006), the solubility of trace elements (V, Mn, Fe, Co,Ni, As, Se, Sb, and Pb) found at Kevo were significantly lower. Re, Tl,As, W, and V have the highest solubility (67e87%). The solubility ofCo and Ni were similar, 33% and 34% for Co and Ni, respectively. Cuhad an overall solubility of 60.8%. The low solubility for Fe (11.1%)may indicate a mostly crustal source. The solubility for Pb in thisstudy (22.9%) is comparable to previously reported values (10e39%)(Qureshi, 2004 and references therein).

Acknowledgements

This work was supported by the National Science Foundationunder grants no. AGS-1007329 (Clarkson University) and AGS-1007261 (SUNY-Albany).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atmosenv.2014.03.002.

References

Aaltonen, V., Paatero, J., Hatakka, J., Viisanen, Y., 2001. Airborne Cesium-137 inNorthern Finland in the early 1960’s based on the measurement of archived airfilter samples. In: The Eighth Nordic Seminar on Radioecology, Rovaniemi,Finland 25.e28.2.2001.

Barcan, V., 2002. Nature and origin of multicomponent aerial emissions of thecopper-nickel smelter complex. Environment International 28, 451e456.

Barrie, L.A., 1986. Arctic air pollution: an overview of current knowledge. Atmo-spheric Environment 20, 643e663.

Barrie, L.A., Barrie, M.J., 1990. Chemical components of lower tropospheric aerosolsin the high arctic: six years of observations. Journal of Atmospheric Chemistry11, 211e226.

Barrie, L.A., Hoff, R.M., 1985. Five years of air chemistry observations in the CanadianArctic. Atmospheric Environment 19, 1995e2010.

Berg, T., Steinnes, E., 1997. Recent trends in atmospheric deposition of trace ele-ments in Norway as evident from the 1995 moss survey. The Science of theTotal Environment 208, 197e206.

Berg, T., Røyset, O., Steinnes, E., 1993. Blank values of trace elements in aerosolfilters determined by ICP-MS. Atmospheric Environment 27A, 2435e2439.

Berg, T., Kallenborn, R., Manø, S., 2004. Temporal trends in atmospheric heavy metaland organochlorine concentrations at Zeppelin, Svalbard. Arctic, Antarctic, andAlpine Research 36, 284e291.

Berg, T., Aas, W., Pacyna, J., Uggerud, H.T., Vadset, M., 2008. Atmospheric trace metalconcentrations at Norwegian background sites during 25 years and its relationto European emissions. Atmospheric Environment 42, 7494e7501.

Boyd, R., Barnes, S.-J., De Caritat, P., Chekushin, V.A., Melezhik, V.A.,Reimann, C., Zientek, M.L., 2009. Emissions from the copper-nickel industryon the Kola Peninsula and at Noril’sk, Russia. Atmospheric Environment 43,1474e1480.

Bulatovic, S.M., 2007. Handbook of Flotation Reagents: Chemistry, Theory andPractice, vol. 1. Elsevier, Amsterdam, Netherlands.

Cheng, M.D., Hopke, P.K., Barrie, L., Rippe, A., Olson, M., Landsberger, S., 1993.Qualitative determination of source regions of aerosol in Canadian high Arctic.Environmental Science & Technology 27, 2063e2071.

Colin, J.L., Jaffrezo, J.L., Gros, J.M., 1990. Solubility of major species in precipitation:factors of variation. Atmospheric Environment 24A, 537e544.

Desboeufs, K.V., Losno, R., Colin, J.L., 2001. Factors influencing aerosol solubilityduring cloud processes. Atmospheric Environment 35, 3529e3537.

Desboeufs, K.V., Sofikitis, A., Losno, R., Colin, J.L., Ausset, P., 2005. Dissolution andsolubility of trace metals from natural and anthropogenic aerosol particulatematter. Chemosphere 58, 195e203.

Dutkiewicz, V.A., DeJulio, A.M., Ahmed, T., Laing, J., Hopke, P.K., Skeie, R.B.,Viisanen, Y., Paatero, J., Husain, L., 2014. Atmospheric black carbon concentra-tions in Finnish Arctic: a comparison between measurements and model basedpredictions. Journal of Geophysical Research Atmospheres (submitted forpublication).

Fine, P.M., Cass, G.R., Simoneit, B.R.T., 2001. Chemical characterization of Fine par-ticle emissions from fireplace combustion of woods grown in the NortheastUnited States. Environmental Science & Technology 35, 2665e2675.

Fine, P.M., Cass, G.R., Simoneit, B.R.T., 2004a. Chemical characterization of Fineparticle emissions from the fireplace combustion of wood types grown in themidwestern and Western United States. Environmental Engineering Science 21,387e409.

Fine, P.M., Cass, G.R., Simoneit, B.R.T., 2004b. Chemical characterization of Fineparticle emissions from the wood stove combustion of prevalent United Statestree species. Environmental Engineering Science 21, 705e721.

Finnish Meteorological Institute, 1984. Observations of Radioactivity 1982. Ob-servations of Radioactivity No. 22. Finnish Meteorological Institute, Helsinki,p. 72.

Giusti, L., Yang, Y.-L., Hewitt, C.N., Hamilton-Taylor, J., Davison, W., 1993. The solu-bility and partitioning of atmospherically derived trace metals in artificial andnatural waters: a Review. Atmospheric Environment 10, 1567e1578.

Gong, S.L., Barrie, L.A., 2005. Trends of heavy metal components in the Arcticaerosols and their relationship to the emissions in the Northern Hemisphere.Science of the Total Environment 342, 175e183.

Hao, W.M., Petkov, A., Nordgren, B., Corley, R., Urbanski, S., 2012. Daily and seasonalvariability of fire activity in Northern Eurasia from 2002 to 2010 & validation ofFS MODIS burned area algorithm. In: EPA 2012 International Emission In-ventory Conference, Tampa, Florida e August 13e16, 2012.

Harmens, H., et al., 2010. Mosses as biomonitors of atmospheric heavy metaldeposition: spatial patterns and temporal trends in Europe. EnvironmentalPollution 158, 3144e3156.

Hirdman, D., Sodemann, H., Eckhardt, S., Burkhart, J.F., Jefferson, A., Mefford, T.,Quinn, P.K., Sharma, S., Ström6, J., Stohl, A., 2010a. Source identification of short-lived air pollutants in the Arctic using statistical analysis of measurement dataand particle dispersion model output. Atmospheric Chemistry and Physics 10,669e693.

Hirdman, D., Burkhart, J.F., Sodemann, H., Eckhardt, S., Jefferson, A., Quinn, P.K.,Sharma, S., Ström, J., Stohl, A., 2010b. Long-term trends of black carbon andsulphate aerosol in the Arctic: changes in atmospheric transport and sourceregion emissions. Atmospheric Chemistry and Physics 10, 9351e9368.

Iversen, T., Joranger, E., 1985. Arctic air pollution and large scale atmospheric flows.Atmospheric Environment 19, 2099e2108.

J.R. Laing et al. / Atmospheric Environment 88 (2014) 275e284284

Justice, C.O., Giglio, L., Korontzi, S., Owens, J., Morisette, J.T., Roy, D., Descloitres, J.,Alleaume, S., Petitcolin, F., Kaufman, Y., 2002. The MODIS fire products. RemoteSensing of Environment 83, 244e262.

Korontzi, S., McCarty, J., Loboda, T., Kumar, S., Justice, C., 2006. Global distribution ofagricultural fires in croplands from 3 years of Moderate Resolution ImagingSpectroradiometer (MODIS) data. Global Biogeochemical Cycles 20. http://dx.doi.org/10.1029/2005GB002529.

Laing, J.R., Hopke, P.K., Hopke, E.F., Husain, L., Dutkiewicz, V.A., Paatero, J.,Viisanen, Y., 2013. Long-term trends of biogenic sulfur aerosol and its rela-tionship with sea surface temperature in Arctic Finland. Journal of GeophysicalResearch Atmospheres 118, 11,770e11,776.

Laing, J.R., Hopke, P.K., Hopke, E.F., Husain, L., Dutkiewicz, V.A., Paatero, J., Viisanen, Y.,2014. Long-term particle measurements in Finnish Arctic: part IIe trend analysisand source identification. Atmospheric Environment 88, 285e296.

Law, K.S., Stohl, A., 2007. Air pollution: origins and impacts. Science 315, 1537e1540.Li, S.-M., Barrie, L.A., 1993. Biogenic sulfur aerosol in the Arctic troposphere: 1.

Contributions to total sulfate. Journal of Geophysical Research Atmospheres 98(D11), 20613e20622.

Mason, B., 1966. Principles of Geochemistry. John Wiley & Sons, Inc., New York.McCarty, J.L., Ellicott, E.A., Romanenkov, V., Rukhovitch, D., Koroleva, P., 2012. Multi-

year black carbon emissions from cropland burning in the Russian Federation.Atmospheric Environment 63, 223e238.

Niemi, J.V., Saarikoski, S., Aurela, M., Tervahattu, H., Hillamo, R., Westphal, D.L.,Aarnio, P., Koskentalo, T., Makkonen, U., Vehkamäki, H., Kulmala, M., 2006.Changes in background aerosol composition in Finland during polluted andclean periods studied by TEM/EDX individual particle analysis. AtmosphericChemistry & Physics 6, 5049e5066.

Paatero, J., Hatakka, R., Mattsson, R., Lehtinen, I., 1994. A comprehensive station formonitoring atmospheric radioactivity. Radiation ProtectionDosimetry 54, 33e39.

Paatero, J., Hatakka, R., Mattsson, R., Viisanen, Y., 1998. Analysis of daily 210Pb airconcentrations in Finland, 1967e1996. Radiation Protection Dosimetry 77 (3),191e198.

Paatero, J., Hatakka, R., Mattson, R., Aaltonen, V., Viisanen, Y., 2000. Long-termvariations of Lead-210 concentrations in ground-level air in Finland: effects ofthe North-Atlantic Oscillation. In: Transport and Chemical Transformation inthe Troposphere: Proceedings of the EUROTRAC-2 Symposium. Springer-Verlag,Berlin, Germany.

Paatero, J., Dauvalter, V., Derome, J., Lehto, J., Pasanen, J., Vesala, T., Miettinen, J.,Makkonen, U., Kyrö, E.-M., Jernström, J., Isaeva, L., Derome, K., 2008. Effects ofKola Air Pollution on the Environment in the Western Part of the Kola Peninsulaand Finnish Lapland e Final Report. Finish Meteorological Institute, Helsinki,Finland. No. 2008:6. UDC 504.054.

Pacyna, J.M., Pacyna, E.G., 2001. An assessment of global and regional emissions oftrace metals to the atmosphere from anthropogenic sources worldwide. Envi-ronmental Reviews 9, 269e298.

Pacyna, J.M., Semb, A., Hanssen, J.E., 1984. Emission and long-range transport oftrace elements in Europe. Tellus 36B, 163e178.

Pirinen, P., Simola, H., Aalto, J., Kaukoranta, J.-P., Karlsson, P., Ruuhela, R., 2012.Climatological Statistics of Finland 1981e2010. Reports 2012:1. Finnish Mete-orological Institute, Helsinki.

Poikolainen, J., Kubin, E., Piispanen, J., Karhu, J., 2004. Atmospheric heavy metaldeposition in Finland during 1985e2000 using mosses as bioindicators. TheScience of the Total Environment 318, 171e185.

Poikolainen, J., Kubin, E., Piispanen, J., Karhu, J., 2006. Estimation of the long-rangetransport of mercury, cadmium, and lead to northern Finland on the basis ofmoss surveys. Arctic, Antarctic, and Alpine Research 36, 292e297.

Polissar, A.V., Hopke, P.K., Malm, W.C., Sisler, J.F., 1998a. Atmospheric aerosol overAlaska 1.Spatial and seasonal variability. Journal of Geophysical Research At-mospheres 103, 19,035e19,044.

Polissar, A.V., Hopke, P.K., Paatero, P., Malm, W.C., Sisler, J.F., 1998b. Atmosphericaerosol over Alaska 2.Elemental composition and sources. Journal ofGeophysical Research Atmospheres 103, 19,045e19,057.

Polissar, A.V., Hopke, P.K., Paatero, P., Kaufmann, Y.J., Hall, D.K., Bodhaine, B.A.,Dutton, E.G., Harris, J.M., 1999. The aerosol at Barrow, Alaska: long-term trendsand source locations. Atmospheric Environment 33, 2441e2458.

Polissar, A.V., Hopke, P.K., Harris, J.M., 2001. Source regions for atmospheric aerosolmeasured at Barrow, Alaska. Environmental Science & Technology 35, 4214e4226.

Quinn, P.K., Shaw, G., Andrews, E., Button, E.G., Ruoho-Airola, T., Gong, S.L., 2007.Arctic haze: current trends and knowledge gaps. Tellus 59B, 99e114.

Quinn, P.K., Bates, T.S., Schulz, K., Shaw, G.E., 2009. Decadal trends in aerosolchemical composition at Barrow, Alaska: 1976e2008. Atmospheric Chemistryand Physics 9, 8883e8888.

Qureshi, S., 2004. Characterization of Chemical Composition and Sources of PM2.5Aerosols (Ph.D. dissertation). State University of New York at Albany, School ofPublic Health, Department of Environmental Health Sciences.

Qureshi, S., Dutkiewicza, V.A., Khan, A.R., Swami, K., Yang, K.X., Husain, L.,Schwab, J.J., Demerjian, K.L., 2006. Elemental composition of PM2.5 aerosols inQueens, New York: solubility and temporal trends. Atmospheric Environment40, S238eS251.

Raatz, W.E., 1989. An anticyclonic point of view on low level tropospheric long-range transport. Atmospheric Environment 23, 2501e2504.

Raatz, W.E., 1991. The climatology and meteorology of Arctic air pollution. Pollutionof the Arctic atmosphere, 13e42.

Raatz, W.E., Shaw, G.E., 1984. Long-range tropospheric transport of pollutionaerosols into the Alaskan Arctic. Journal of Climate and Applied Meteorology 23,1052e1064.

Saarikoski, S., Sillanpää, M., Sofiev, M., Timonen, H., Saarnio, K., Teinilä, K.,Karppinen, A., Kukkonen, J., Hillamo, R., 2007. Chemical composition of aerosolsduring a major biomass burning episode over northern Europe in spring 2006:experimental and modelling assessments. Atmospheric Environment 41, 3577e3589.

Shaw, G.E., 1995. The Arctic haze phenomenon. Bulletin of the American Meteo-rological Society 76 (12), 2403e2413.

Sirois, A., Barrie, L.A., 1999. Arctic lower tropospheric aerosol trends and compo-sition at Alert, Canada: 1980e1995. Journal of Geophysical Research Atmo-spheres 104 (D9), 11599e11618.

Sivertsen, B., Makarova, T., Hagen, L.O., Baklanov, A.A., 1992. Air Pollution in theBorder Areas of Norway and Russia. NILU-Norwegian Institute for Air Research,Lillestrom, pp. 1e14.

Stebel, K., Christensen, G., Derome, J., Grekelä, I., 2007. State of the Environment inthe Norwegian, Finnish and Russian Border Area. The Finnish Environment.

Stohl, A., 2006. Characteristics of atmospheric transport into the Arctic troposphere.Journal of Geophysical Research 11, D11306.

Stohl, A., Burkhart, J.F., Fjæraa, A.M., Forster, C., Herber, A., Hov, Ø., Lunder, C.,McMillan, W.W., Oltmans, S., Shiobara, M., Simpson, D., Solberg, S., Stebel, K.,Ström, J., Tørseth, K., Treffeisen, R., Virkkunen, K., Yttri, K.E., 2007. Arctic smokee record high air pollution levels in the European Arctic due to agricultural firesin Eastern Europe in spring 2006. Atmospheric Chemistry & Physics 7, 511e534.

Tuovinen, J.P., Laurila, T., Lättilä, H., Ryaboshapko, A., Brukhanov, P., Korolev, S., 1993.Impact of the sulphur dioxide sources in the Kola Peninsula on air quality innorthernmost Europe. Atmospheric Environment 27A, 1379e1395.

USEPA, 1985. 49 FR 43430, Oct. 26, 1984; 50 FR 694, 696, Jan. 4, 1985, as Amended at51 FR 23703, June 30, 1986.

USEPA (United States Environmental Protection Agency), 1999. Compendiummethod IO-3.1 selection, preparation and extraction of filter material. Com-pendium of Methods for the Determination of Inorganic Compounds inAmbient Air. OH: Center for Environmental Research Information.

von Storch, H., Costa-Cabral, M., Hagner, C., Feser, F., Pacyna, J., Pacyna, E., Kolb, S.,2003. Four decades of gasoline lead emissions and control policies in Europe: aretrospective assessment. The Science of the Total Environment 311, 151e176.

Xie, Y., Hopke, P.K., Paatero, P., Barrie, L.A., Li, S.-M., 1999a. Identification of sourcenature and seasonal variations of Arctic aerosol by multilinear engine. Atmo-spheric Environment 33, 2549e2562.

Xie, Y., Hopke, P.K., Paatero, P., Barrie, L.A., .Li, S.-M., 1999b. Identification of sourcenature and seasonal variations of Arctic aerosol by positive matrix fractioni-zation. Journal of Atmospheric Sciences 56, 249e260.

Yli-Tuomi, T., Venditte, L., Hopke, P.K., Basunia, M.S., Landsberger, S., Viisanen, Y.,Paatero, J., 2003. Composition of the Finnish Arctic aerosol: collection andanalysis of historic filter samples. Atmospheric Environment 37, 2355e2364.