1 23

Environmental Science and PollutionResearch ISSN 0944-1344 Environ Sci Pollut ResDOI 10.1007/s11356-013-2067-1

Seasonal variations in the chemicalcomposition of particulate matter: a casestudy in the Po Valley. Part I: macro-components and mass closure

C. Perrino, M. Catrambone, S. DallaTorre, E. Rantica, T. Sargolini &S. Canepari

1 23

Your article is protected by copyright and

all rights are held exclusively by Springer-

Verlag Berlin Heidelberg. This e-offprint is

for personal use only and shall not be self-

archived in electronic repositories. If you wish

to self-archive your article, please use the

accepted manuscript version for posting on

your own website. You may further deposit

the accepted manuscript version in any

repository, provided it is only made publicly

available 12 months after official publication

or later and provided acknowledgement is

given to the original source of publication

and a link is inserted to the published article

on Springer's website. The link must be

accompanied by the following text: "The final

publication is available at link.springer.com”.

HEAVY METALS IN THE ENVIRONMENT : SOURCES, INTERACTIONS AND HUMAN HEALTH

Seasonal variations in the chemical composition of particulatematter: a case study in the Po Valley. Part I: macro-componentsand mass closure

C. Perrino &M.Catrambone & S. Dalla Torre & E. Rantica &

T. Sargolini & S. Canepari

Received: 22 April 2013 /Accepted: 9 August 2013# Springer-Verlag Berlin Heidelberg 2013

Abstract The seasonal variability in the mass concentrationand chemical composition of atmospheric particulate matter(PM10 and PM2.5) was studied during a 2-year field studycarried out between 2010 and 2012. The site of the study wasthe area of Ferrara (Po Valley, Northern Italy), which is char-acterized by frequent episodes of very stable atmospheric con-ditions in winter. Chemical analyses carried out during thestudy allowed the determination of the main components ofatmospheric PM (macro-elements, ions, elemental carbon, or-ganic matter) and a satisfactory mass closure was obtained.Accordingly, chemical components could be grouped into themain macro-sources of PM: soil, sea spray, inorganic com-pounds from secondary reactions, vehicular emission, organicsfrom domestic heating, organics from secondary formation, andother sources. The more significant seasonal variations wereobserved for secondary inorganic species in the fine fraction ofPM; these species were very sensitive to air mass age and thusto the frequency of stable atmospheric conditions. During thewinter ammonium nitrate, the single species with the highestconcentration, reached concentrations as high as 30 μg/m3. Theintensity of natural sources was fairly constant during the year;increases in natural aerosols were linked to medium and long-range transport episodes. The ratio of winter to summer con-centrations was roughly 2 for combustion product, close to 3 forsecondary inorganic species, and between 2 and 3 for organics.The winter increase of organics was due to poorer atmosphericdispersion and to the addition of the emission from domestic

heating. A similar winter to summer ratio (around 3) wasobserved for the fine fraction of PM.

Keywords Particulate matter . Chemical speciation .

Po Valley .Mass closure . Atmospheric stability

Introduction

Seasonal variations in the mass concentration and chemicalcomposition of atmospheric particulate matter (PM) are, inprinciple, due to two main factors: meteorological variationsthrough the year and the consequent variation in the strength ofsome PM sources which are linked to the weather (e.g., domes-tic heating). The meteorology of the Mediterranean Basin isgenerally characterized by warm to hot, dry summers and mildto cool, wet winters. During summer, the region is dominatedby subtropical high pressure cells, with little precipitation;winter is influenced by the polar jet stream and its associatedstorms which brings rain or snow at higher latitudes and eleva-tions. Seasonal meteorological variations are linked to changesin the strength of some PM emission sources: soil particles aremost likely re-suspended during arid summers, secondary pol-lutants, produced by precursor reactions in the atmosphere,rapidly increase during periods of atmospheric stability, anddomestic heating adds to other PM sources during cold periods.

Within this general framework, the Po Valley, a 46,000-km2 large area of Northern Italy that runs from the WesternAlps to the Adriatic Sea (Fig. 1), has some characteristicswhich are quite distinct. In this area, the local circulation isoften weak and this leads to foggy and damp winters; duringthe summer, sunny days with comfortable temperatures andclear skies alternate with very hot, humid days. Particularlyduring the winter, the almost enclosed nature of the Po Valleyfavors the development of temperature inversions, leading toextended periods of high atmospheric stability (Carbone et. al.2010; Pernigotti et al. 2012). As a consequence of both limited

Responsible editor: Céline Guéguen

C. Perrino (*) :M. Catrambone : S. Dalla Torre : E. Rantica :T. Sargolini : S. CanepariCNR Institute of Atmospheric Pollution Research, Via Salaria,Km 29,300, Monterotondo St., Rome 00015, Italye-mail: perrino@iia.cnr.it

S. CanepariDepartment of Chemistry, Sapienza University of Rome,P.le Aldo Moro, 5, Rome 00185, Italy

Environ Sci Pollut ResDOI 10.1007/s11356-013-2067-1

Author's personal copy

atmospheric mixing and the high density of local sources ofPM, severe atmospheric pollution episodes are frequentlyexperienced in the cities of this area during the winter. Thisresults in numerous exceedances of the PM limits stated in theair quality directive of the European Union 2008/50/EC (EC2008) and produces a considerable health impact (Saarikoskiet al. 2012; Squizzato et al. 2013). It is worth noting that thePo Valley has been indicated by the CAFE baseline scenarioas one of the most polluted sites in Europe, with a significantloss in statistical life expectancy due to anthropogenic contri-butions to PM air concentrations (EEA 2007).

Studying the seasonal variability of atmospheric particlemass concentration and chemical composition requires a longtime series of observations, preferably carried out in areas wheresummer–winter differences are enhanced (Carbone et al. 2010).We report in this paper the results of a 2-year study of the massconcentration and chemical composition of PM10 and PM2.5

(particulate smaller than 10 or 2.5 μm in aerodynamic diameterrespectively) carried out in the area of Ferrara, in the eastern partof the Po Valley. Seasonal differences in the strength of the mainmacro-sources of PM (soil, sea, secondary production of inor-ganic aerosol, combustion processes, and the primary and sec-ondary production of organic species) have been evaluated. Theresults of a parallel study concerning seasonal differences in theconcentration and solubility of micro- and trace elements arereported in a companion paper (Canepari et al. Seasonal varia-tions in the chemical composition of particulate matter: a casestudy in the Po Valley. Part II: concentration and solubility ofmicro- and trace- elements, this issue).

Experimental

The field study was carried out in a suburban area of Ferrara(44°50′N 11°37′E), a city of roughly 130,000 inhabitants

located in the eastern Po Valley. The area of the study, shownin Fig. 1, is influenced by several PM sources: the urban area,the highway A13, and a major industrial area containing apower plant, a urban waste incinerator and many small andmedium size enterprises (SMEs). The study was performedbetween October 2010 and September 2012, with four inten-sive 1-month Special Observation Periods (SOPs) conductedduring winter and summer. Three locations were selected forthe study: an industrial site (A), close to the power plant, thewaste incinerator and the SMEs, a rural site (B), located as faras feasible from the main emission sources, a residential site(C), located in the hamlet of Cassana, about 6 km from thecenter of Ferrara. Distances between the sites are: A–B3.7 km, A–C 1 km; and B–C 2.8 km.

During the whole study, the mass concentration of PM10

and PM2.5 was measured daily at all three sites, using auto-mated dual channel beta attenuation monitors (SWAM 5a DualChannelMonitor, FAI Instruments, Fonte Nuova, Rome, Italy)equipped with Teflon membrane filters (TEFLO, 47 mm,2.0-μm pore size, PALL Life Sciences). During the sameperiod, the mixing properties of the lower atmosphere weredetermined hourly at site C, using an automated monitor of thenatural radioactivity due to Radon progeny (PBL MixingMonitor, FAI Instruments, Fonte Nuova, Rome, Italy).Knowledge of the mixing properties of the lower atmosphereallows uncoupling of PM concentration variations due tochanges in the emission/transformation rate of pollutants fromthose due to changes in the dilution properties of the boundarylayer. Details of this easy technique and its use for understand-ing pollution events can be found in Kataoka et al. (2001),Perrino et al. (2001, 2008, 2009, 2010), Sesana et al. (2003),Vecchi et al. (2004, 2007).

The overall chemical characterization of the aerosol wasperformed during the four SOPs. During these periods, threeadditional dual channel samplers (HYDRA Dual Sampler,

INDUSTRIAL AREA

URBAN AREA

HIGHWAY

B

C

A

Po Valley

Fig. 1 Map of the sampling area and location of the sampling sites. Site A industrial; site B rural; site C residential

Environ Sci Pollut Res

Author's personal copy

FAI Instruments, Fonte Nuova, Rome, Italy) were deployed atthe three sites in order to collect daily PM10 and PM2.5

samples also on quartz fiber filters (Tissuquartz 2500QAT,47 mm, PALL Life Sciences). The simultaneous samplingon Teflon and quartz filters allowed the subsequent chemicaldetermination of all the macro-components of PM, accordingto the method reported in Canepari et al. 2009 and Perrinoet al. 2009. A total of 720 daily PM samples were analyzed.Briefly, macro-elements were determined on Teflon filters byx-ray fluorescence (X-Lab2000, SPECTRO AnalyticalInstruments, Kleve); then the filters were extracted in acetatebuffer (Canepari et al. 2006, 2010): a part of the solution wasanalyzed by ion chromatography (ICS1000, Dionex Co., CA,USA), the rest of the solution and the residue were kept forsubsequent analysis of micro- and trace elements (see Part II);elemental and organic carbon were detected on quartz filtersby thermo-optical analysis (OCEC Carbon Aerosol Analyzer,Sunset Laboratory, OR, USA; NIOSH-quartz temperatureprotocol). This overall analytical procedure allows the deter-mination of each individual component which typically makesup more than 1 % of the mass amount of PM10 and PM2.5: Si,Al, Fe, Na, K, Mg, Ca, chloride, nitrate, sulfate, ammonium,elemental carbon, and organic species as a whole.

One quarter of each quartz filter was used for the determi-nation of levoglucosan, a reliable tracer of biomass combus-tion. Two series of 15 samples were pooled in order to obtaintwo average values per SOP. The filters were extracted in de-ionized water. High-performance anion-exchange chromatog-raphy with pulsed amperometric detection was used forlevoglucosan determination.We used a DionexDX-500 seriesion chromatograph equipped with a DC ICS-3000 oven, aGP40 Gradient Pump, and a CarboPac™ PA10 analyticalcolumn and guard column. The electrochemical detector wasa Dionex ED50/ED50A electrochemical cell, utilizing dispos-able gold electrodes.

Results and discussion

Mass concentration

The pattern of PM10 and PM2.5 mass concentrations duringthe 2-year study is shown in Fig. 2 (upper and lower panel,respectively). The data shows that a clear increase in PMconcentration occurred during the winter: average PM10

values were as high as 48.5 μg/m3 during the colder period(October–March) and 29.0 μg/m3 during the warmer period(April–September); average PM2.5 values were 38.5 and17.7 μg/m3, respectively. The average ratio between PM10

concentration during the winter and the summer was as highas 1.7 (2.2 in the case of PM2.5).

This increase may, in principle, be due both to the lessefficient atmospheric mixing during the cold season, and to

the increase in the strength of some seasonal PM sources. Thehigher relative increase that was observed for PM2.5 withrespect to PM10 is compatible with both explanations. It iswell known that a significant fraction of the coarser particles,contained in PM10, derives from natural sources, while sec-ondary pollutants and combustion products are contained inthe fine fraction of PM (PM2.5). It is thus conceivable that boththe winter increase of secondary pollutants, which are sensi-tive to air mass aging which occurs during periods of atmo-spheric stability, and the increase in the strength of combus-tion sources, which includes domestic heating, affected PM2.5

more than PM10 concentration. As a consequence, the PM2.5

contribution to PM10 also showed a clear seasonal difference:on average, PM2.5 accounted for about 80 % of PM10 duringthe winter months and about 60 % during the warmer months.

Another interesting observation from the data in Fig. 2 ishow similar the PM concentrations were over time at the threesites. This result would indicate that the concentration of PMwas influenced relatively little by local sources and supportthe hypothesis that it was mainly driven by larger-scale phe-nomena, such as the meteorological conditions. Also, the areaof the study and, in general, the Po Valley is characterized by avery high background pollution, which makes the contribu-tion of local sources less discernible.

Natural radioactivity

More information about the role played by atmospheric sta-bility in determining PM concentrations can be obtained fromFig. 3, where the temporal variations of PM10 concentration atsite B during one summer and one winter SOP are comparedwith the variation in natural radioactivity. To interpret thisparameter, it is necessary to consider that radon gas is emittedfrom the ground at a rate that can be assumed to be constant onthe temporal and spatial scale of the observations (a fewweeks, several kilometers) and that its only transformation isthe radioactive decay. It follows that the concentration ofRadon daughters in the atmosphere is driven by the mixingof the lower atmosphere: it increases when the atmosphericmixing is slow and decreases in situations of advection or ofefficient atmospheric mixing. This behavior matches and re-flects the behavior of unreactive atmospheric pollutants emit-ted with a constant rate (Kataoka et al. 2001; Perrino et al.2001, 2008, 2009, 2010; Sesana et al. 2003; Vecchi et al.2004, 2007).

During the warm period (Fig. 3, upper graph, referring tothe SOP of May–June 2012) natural radioactivity generallyshowed a clear diurnal pattern, with maxima during the night,indicative of the stable nocturnal boundary layer, and minimaduring the day, due to the convective atmospheric mixing. Thecomparison with the pattern of PM concentration shows thatthe increase of PM was driven by the increase of nighttimestability and, particularly, by the combination of nighttime

Environ Sci Pollut Res

Author's personal copy

stability and poor daytime mixing (high nighttime maximaand high daytime minima) lasting for a number of consecutivedays. For example, during the week 15th–22nd June, PMconcentration increased only slightly from the 15th to the17th, and this period was characterized by nocturnal atmo-spheric stability but good daytime mixing. Instead, it in-creased distinctly from the 18th to the 21st, when daytimemixing was weaker and this situation held up for many sub-sequent days. On the 22nd, the decrease in both nighttime anddaytime radioactivity values was accompanied by a cleardecrease in PM concentration. It is worth noting that duringthe warm period, in general, the convective daytime mixing ofthe atmosphere maintained PM concentration at rather lowvalues.

During the cold period (Fig. 3, lower graph, referring to theSOP of January–February 2011) the diurnal pattern of naturalradioactivity was, mostly, no longer apparent. In the area ofthe Po Valley winter is very often characterized by very weakatmospheric mixing during both night and day, which

corresponds to constantly high radioactivity values. In theseconditions, the worse PM pollution episodes occur whenmany consecutive days are characterized by high daytimestability or during periods characterized by high daytime andnighttime radioactivity values. Also, a sharp and significantincrease of PM concentration is typically observed when re-stabilization of the atmosphere occurs during daytime hours (atemporary decrease of the mixing height occurring during themorning, a period of the day when a fast increase is generallyobserved). A general coherence between natural radioactivityand PM concentration patterns was observed also during thisSOP. Particularly interesting are the cases of January 31st andFebruary 10th, when a daytime re-stabilization of the atmo-sphere occurred, with natural radioactivity values continuous-ly increasing from early morning up to noon. This relativelyinfrequent situation causes the trapping of pollutants releasedduring the morning, a period when the height of the mixedlayer generally increases, into a more and more stagnant airlayer, causing a sharp increase in PM concentration. It is more

0

20

40

60

80

100

120

140

g m

-3

OCTOBER 2010 - SEPTEMBER 2012

PM10

SITE A

SITE B

SITE C

0

20

40

60

80

100

120

140

g m

-3

OCTOBER 2010 - SEPTEMBER 2012

PM2.5

SITE A

SITE B

SITE C

Fig. 2 Time variation of PM10

and PM2.5 mass concentration atthe three sites during the 2-yearstudy. Yellow areas indicate theSpecial Observation Periods

Environ Sci Pollut Res

Author's personal copy

difficult to explain, only on the basis of atmospheric stability,the very high PM value recorded on January 26th.

During the first week of the winter SOP, PM10 concentra-tion were much lower than expected on the basis of stabilityconditions, highlighted by the natural radioactivity pattern.During this week, as discussed in detail in part II, a thick fogepisode occurred in the area of Ferrara, which resulted in anincrease in the average diameter of the atmospheric particles.This likely caused both a higher deposition velocity (andconsequent reduction in PM concentration) and the exclusionof some particles, originally in the size fraction below 10 μm,from the sampling head.

Chemical composition and mass closure

Table 1 reports the mean concentration, 10 and 90 ° percentileof the main PM components; the data have been averaged for

the two SOPs carried out during the same season and for thethree sites (the differences among the three sites were notstatistically significant). The data shows that the concentra-tions of Al, Si, Mg, and Ca (soil components) were higherduring summer than during winter, and that all these elementswere found in the coarse fraction of PM. Conversely, ammo-nium, elemental carbon, and organic carbon were higherduring the winter and mainly contained in the fine PMfraction.

The other components showed a more complex behavior.Fe was almost completely in the coarse fraction but showedrather high winter concentration values, probably because ofthe influence of dust re-suspension (enhanced by atmosphericstability conditions). Nitrate seems to have two main sources:one producing particles in the fine fraction, much more activeduring the winter (most likely, secondary formation of ammo-nium nitrate), and another one producing particles in the

0

500

1000

1500

2000

2500

13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9 10 11

Co

un

ts/min

ute

JANUARY - FEBRUARY 2011

0

20

40

60

80

100

120

g/m

3

0

500

1000

1500

2000

2500

30 31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

Co

un

ts/min

ute

MAY - JUNE 2012

0

20

40

60

80

100

120

g/m

3

Fig. 3 Time patterns of naturalradioactivity, expressed in countsper minute (red area , right axis),and PM10 concentration (blackline with diamonds, left axis)during the summer SOP 2012 andthe winter SOP 2011 at site B

Environ Sci Pollut Res

Author's personal copy

coarse fraction, active during both winter and summer (naturalsources).

Potassium showed two main sources, too: one leading tohigh concentration of fine particles during the winter (possi-bly, domestic heating), the other one producing coarse parti-cles during the summer (soil). During the winter, in fact, thiselement correlated well with organic carbon (OC), both inPM10 and PM2.5 (R2=0.93 and 0.91, respectively), whileduring the summer, a good correlation was observed amongthe insoluble fraction of potassium in PM10 and soil compo-nents (R2 in the range 0.81–0.93).

Unexpectedly, the concentration level of sulfate did notshow significant seasonal variations. In fact, although sulfate,in the form of ammonium sulfate and bisulfate, is mostlyproduced during the summer as a result of photochemicalprocesses, its winter concentration was much influenced bythe transport of polluted air masses from the Balkans area (theseepisodes are discussed in detail in the companion paper).

To obtain the mass closure, that is the correspondencebetween the mass concentration of PM and the sum of theindividual chemical determinations, we applied a correctionfactor for oxygen in metal oxides (Chan et al. 1997;Marcazzan et al. 2001), calculated carbonate as the sum ofcalcium multiplied by 1.5 and magnesium multiplied by 2.5,and multiplied organic carbon by a factor α , to take intoaccount non-C atoms in organic molecules. The value of αwas set to 1.8 for suburban sites A and C and to 2.1 for the

rural site B, as suggested in Turpin and Lim (2001) andViidanoja et al. (2002).

Figure 4 shows the scatter plot of measured and recon-structed PM10mass during the four SOPs (average of the threesites). Satisfactory results were obtained in all cases, withPearson’s coefficient between 0.89 and 0.99. Similar resultswere obtained for PM2.5.

On average, during the summer SOPs, the chemical anal-yses allowed the identification of 94.8 % (year 2011) and92.6 % (year 2012) of the PM10 mass; during the two winterSOPs, the percentages were 85.5 and 86.4, respectively; sim-ilar results were obtained for PM2.5. The unidentified massdetected in both PM10 and PM2.5, which was much higherduring the winter season, was likely due to water retained onatmospheric particles, as discussed in Perrino et al. (2013) andCanepari et al. (2013). It is worth noting, however, that ap-proximations in the determination of the OC/OM and metal/metal oxide conversion factors can also be relevant sources ofuncertainty (Terzi et al. 2010). The value of α depends on thecomposition of the organic fraction at the observation pointand is highly variable with the period of the year, time of theday, distance from the anthropogenic sources, amount andtype of vegetation, and particle aging. Currently, knowledgeof the chemical composition of atmospheric organic matter isquite poor, (about 15 % of compounds are identified, seeMüller et al. 2005) and the estimation of an average α valuesuffers from a high level of uncertainty.

Table 1 Seasonal variability of the main components in PM10 and PM2.5: mean concentration, 10 and 90 ° percentile (μg/m3); average of three sitesand two 1-month SOPs for each season

WINTER SOPs—PM10 WINTER SOPs—PM2.5 SUMMER SOPs—PM10 SUMMER SOPs—PM2.5

Mean 10 ° 90 ° Mean 10 ° 90 ° Mean 10 ° 90 ° Mean 10 ° 90 °

Al 0.14 0.041 0.27 0.059 0.029 0.087 0.21 0.087 0.38 0.076 0.047 0.11

Si 0.54 0.12 1.1 0.15 0.040 0.26 0.92 0.37 1.7 0.27 0.15 0.41

Fe 0.25 0.064 0.57 0.059 <0.002 0.15 0.20 0.094 0.36 0.022 <0.002 0.053

Na 0.66 0.13 1.1 0.18 0.054 0.30 0.40 0.19 0.71 0.14 0.078 0.20

K 0.62 0.34 0.92 0.53 0.26 0.83 0.23 0.15 0.35 0.10 0.056 0.17

Mg 0.26 0.010 0.55 0.034 0.005 0.085 0.41 0.22 0.64 0.076 0.025 0.14

Ca 0.90 0.15 1.8 0.20 0.062 0.33 1.3 0.59 2.2 0.27 0.11 0.43

Cl− 0.55 0.12 1.0 0.21 0.072 0.39 0.16 0.041 0.37 0.045 0.024 0.067

NO3− 9.3 2.1 20 6.6 1.4 14 2.0 0.81 3.6 0.53 0.095 1.2

SO4= 3.5 1.8 5.7 2.8 1.2 4.6 3.1 1.5 5.0 2.8 1.2 4.5

Na+ 0.39 0.063 0.83 0.12 0.035 0.24 0.42 0.091 0.88 0.090 0.035 0.17

NH4+ 3.8 1.2 7.5 2.9 0.9 5.5 1.0 0.39 1.7 0.86 0.30 1.6

K+ 0.49 0.27 0.74 0.42 0.21 0.66 0.14 0.067 0.25 0.11 0.054 0.17

Mg++ 0.11 0.014 0.22 0.026 0.006 0.049 0.10 0.036 0.18 0.029 0.010 0.053

Ca++ 0.55 0.12 1.17 0.14 0.042 0.24 0.90 0.38 1.6 0.22 0.093 0.35

OC 10 5.7 17 9.3 5.0 16 4.3 3.0 5.9 3.5 2.2 4.7

EC 1.2 0.46 2.0 1.0 0.42 1.7 0.61 0.34 0.96 0.52 0.29 0.79

Environ Sci Pollut Res

Author's personal copy

Daily and seasonal variations of macro-source intensity

In general, atmospheric PM include a number of macro-sources: two main natural sources (soil and sea), the produc-tion of both inorganic and organic secondary species fromreactions in the atmosphere, and two main combustion pro-cesses (traffic and domestic heating). Other sources, oftenwith lower strength, include the release of primary bioaerosol,the re-suspension of road dust, and different types of industrialemission. In this study, we used the results of the chemicalanalysis to estimate the intensity of the main PM macro-sources and to trace their time variations.

The soil contribution was calculated by summing the con-centration of elements (as metal oxides) generally associatedwith mineral dust: Al, Si, and Fe plus the insoluble fractions ofK, Mg, and Ca, calculated as the difference between the XRFand the ion chromatography (IC) determinations, plus calciumand magnesium carbonate. The regression analysis for PM10

shows that the correlations among these elements were satis-factory during the winter SOPs (R2 was in the range of 0.70–0.98) and excellent during the summer SOPs (R2 in the range0.83–0.99).

The estimation of the sea salt contribution was relativelycomplex. The main components of sea spray, in fact, are alsoproduced by other sources: sodium and magnesium are minorcomponents of soil, and ammonium chloride is a product ofatmospheric reactions; in addition, in aged air masses, thereaction of NaCl with nitric acid produces NaNO3 and reducesthe NaCl concentration. In the literature, the sea spray contri-bution to PM is calculated using a number of different proce-dures: some authors (e.g., Harrison et al. 2003) used chloridemultiplied by 1.65, others (e.g. Theodosi et al. 2011) used

sodium ion concentration multiplied by 3.27, others (e.g.Putaud et al. 2004; Pey et al. 2009) used the sum of sodiumand chloride, and others (e.g. Terzi et al. 2010) used differentapproaches for summer and winter data.

In this paper, we preferred to use two different approachesfor the two size fractions and two seasons: sea salt contribu-tion to PM10 and to summer PM2.5 was calculated from theconcentration of soluble sodium and chloride, determined byIC, multiplied by 1.176 in order to take into account minor seawater components (sulfate, magnesium, calcium, potassium),as described in Perrino et al. (2009, 2010). Given the very lowcontribution of sea salt to fine particles and the enhancedformation of secondary reaction products in fine PM insidestable air masses, for winter PM2.5 we preferred to includechloride among inorganic secondary products, while the seasalt contribution was calculated from sodium only, accordingto Theodosi et al. (2011). This choice was supported by themuch higher correlation between chloride and ammonium(R 2=0.64) than between chloride and sodium (R 2=0.09)obtained for PM2.5 during the winter SOP.

The secondary inorganic fraction was calculated as the sumof non-sea salt sulfate, nitrate, and ammonium. The regressionanalysis showed that the highest correlation was obtained byammonium and nitrate during the winter (R2 was 0.93 inPM10 and 0.90 in PM2.5) and by ammonium and sulfateduring the summer (R2 was 0.81 and 0.85, respectively).

The most difficult task was the distinction between primaryorganics (pOC), mostly released by incomplete combustion,and secondary organics (sOC), produced by gas/particle con-version of volatile organic compounds. A possible strategy forestimating pOC relies on the use of elemental carbon (EC),which is generally considered as a reliable tracer of combus-tion emission. The method, reported by Castro et al. (1999),consists in the measurement of the OC/EC minimum ratiooccurring during periods of reduced photochemical activity. Itis assumed that EC and pOC share the same sources and thatthe organic carbon exceeding the minimum OC/EC ratio canbe considered of secondary origin. Other assumptions for areliable application of this method are that the contribution ofnon-combustion pOC (e.g., biogenic aerosol) is low, and thatpOC/EC ratio is a constant at the sampling site. This secondassumption may not be verified in areas when significantemission from residential wood combustion is experienced;incomplete combustion of wood is in fact expected to releasemuch more organics with respect to vehicular exhausts.Minimum OC/EC ratios of 1.1–1.3 have been reported byCastro et al. for urban sites where vehicular emissions are thedominant OC source, but much higher values have been foundin rural areas, characterized by residential wood burning(Castro et al. 1999; Wang et al. 2005 and references therein).

In the area under study, the main combustion processesreleasing PM in the atmosphere are vehicular emission (thehighway is at a few hundreds of meters from the three sampling

y = 0.83x + 0.64R² = 0.988

y = 0.72x + 5.2R² = 0.889

y = 0.72x + 6.9R² = 0.930

y = 0.78x + 3.3R² = 0.973

0

20

40

60

80

100

120

0 20 40 60 80 100 120

MA

SS

CO

NC

EN

TR

AT

ION

(g

/m3 )

SUM OF CHEMICAL DETERMINATIONS ( g/m3)

Fig. 4 Scatter plot of the mass concentration and sum of chemicalcomponents for PM10 during the four SOPs. Red squares summer2011; orange squares summer 2012; blue squares winter 2011; paleblue squares winter 2012

Environ Sci Pollut Res

Author's personal copy

sites) and wood combustion for domestic heating, very com-mon in Northern Italy. Our results show that the minimum OC/EC ratio in PM10 during the winter was 5.8, indicating a highprevalence of the wood combustion source. For this reason, wechose to use the minimum OC/EC value reported by Castroet al. to obtain an estimate of pOC from vehicular emission only(organic species that condense from the exhausts gases and coatthe surface of elemental carbon particles). The traffic sourcewas then calculated as EC amount plus the same amountmultiplied by 1.1.

For the estimation of pOC due to residential wood com-bustion, we used the measurement of levoglucosan, ananhydrosugar produced by the thermal degradation of cellu-lose and hemicelluloses. Levoglucosan concentration, mea-sured on a 15-day basis, varied between 0.61 and 0.75 μg/m3

during the winter SOPs and was always below 0.02 μg/m3

during the summer SOPs. Emission factors for levoglucosan,expressed as the amount of tracer per unit mass of OC, varysubstantially with the type of biomass and type of fire. Toobtain an estimate of the contribution of pOC from residential

wood combustion we used a conversion factor of 0.15, suit-able for European areas (Leithead et al. 2006; Engling et al.2006; Szidat et al. 2006).

Daily variations

Figure 5 shows the daily variation in the strength of the mainPM10 macro-sources during one summer and one winter SOP.For the study of the daily variations, we considered fivemacro-sources: soil, sea, secondary inorganics, vehicularemission, and organic matter (including sOC, biomass burn-ing, and other primary sources such as bioaerosol). For theconversion of organic carbon to organic matter we applied thesame α value used in the mass closure calculations.

The upper graph (May–June 2012) clearly indicates theoccurrence of two episodes of sea spray advection (June 8–13th and June 26th) and one episode in which soil componentsincreased notably (from June 19th to 22nd, having maximumintensity on June 21st). The output of the model BSC-DREAM8b (Dust REgional Atmospheric Model, Basart et al.

0

5

10

15

20

25

30

35

40

45

50

30 31 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

CO

NC

EN

TR

AT

ION

(g

/m3 )

MAY - JUNE 2012

VEHICULAR EMISSIONORGANICSSECONDARY INORGANICSSEASOIL

0

10

20

30

40

50

60

70

80

90

100

13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 1 2 3 4 5 6 7 8 9 10 11

CO

NC

EN

TR

AT

ION

(g

/m3 )

JANUARY - FEBRUARY 2011

VEHICULAR EMISSIONORGANICSSECONDARY INORGANICSSEASOIL

Fig. 5 Daily variation of thechemical composition of PM10 atsite B during the summer SOP2012 (upper graph) and thewinter SOP 2011 (lower graph)

Environ Sci Pollut Res

Author's personal copy

2012, available on the web at http://www.bsc.es/projects/earthscience/BSC-DREAM/), which simulates the extensionof dust plumes originating from the desert regions of NorthAfrica, indicate advection of desert dust from North Africanareas towards Northern Italy during this period. This is a typicalexample of an increase in PM concentration, which cannot beexpected on the basis of the natural radioactivity pattern, as it isdue to the arrival of an external PM source.

The lower graph in Fig. 5 (January–February 2011) showshigher contributions of secondary inorganic species with re-spect to the summer SOP. During the winter SOP, we recordednitrate concentration levels which were often higher than20 μg/m3 and sometimes higher than 30 μg/m3; these latterremarkable values occurred during January 26th and February10th, and were presumably associated with a high amount ofwater, which was likely responsible for the unaccounted mass

0

10

20

30

40

50

60

A B C A B C A B C A B C

WINTER 2011

g/m

3

MASS CONCENTRATION

PM fine

PM coarse

0

3

6

9

12

15

18

21

A B C A B C A B C A B C

g/m

3

INORGANIC SECONDARY SPECIES

PM fine

PM coarse

0

1

2

3

4

5

6

7

8

9

A B C A B C A B C A B C

WINTER 2011 .SUMMER 2011

g/m

3

SOIL

PM fine

PM coarse

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

A B C A B C A B C A B C

WINTER 2011

g/m

3

SEA-SPRAY

PM fine

PM coarse

0.0

0.5

1.0

1.5

2.0

2.5

3.0

A B C A B C A B C A B C

WINTER 2011

g/m

3

VEHICULAR EMISSION

PM fine

PM coarse

0

3

6

9

12

15

18

21

A B C A B C A B C A B C

WINTER 2011

g/m

3

ORGANICS

PM fine (wood combustion)PM fine (secondary sources)PM coarse

a) b)

c) d)

e) f)

WINTER 2012 SUMMER 2012

SUMMER 2012WINTER 2012SUMMER 2011WINTER 2011SUMMER 2012WINTER 2012SUMMER 2011

SUMMER 2011 WINTER 2012 SUMMER 2012 SUMMER 2011 WINTER 2012 SUMMER 2012

SUMMER 2012WINTER 2012SUMMER 2011

Fig. 6 Seasonal variations in the mass concentration (a) and contributions of the main macro-sources (b–f) to PM10 at the three sites. Pale colorsindicate the contribution of the fine fraction (PM2.5), dark colors indicate the contribution of the coarse fraction (PM10-2.5)

Environ Sci Pollut Res

Author's personal copy

in the mass closure. As a consequence of atmospheric stabilityand the significant emission rate of ammonium nitrate precur-sors (ammonia, emitted by agricultural and breeding activitiesand nitrogen oxides, emitted by traffic, domestic heating, andindustries), in the Po Valley, ammonium nitrate is confirmedby far to be the species with the highest concentration duringthe winter period (Putaud et al. 2004; Squizzato et al. 2013).

Seasonal variations

Figure 6 summarizes the seasonal variations in the concentra-tion of PMmass and of chemical components released from themain PM macro-sources during the four SOPs. In this case oforganics, an estimate of the contribution of residential woodcombustion, based on levoglucosan determination, is reported.Each bar includes the contribution of the fine fraction (PM2.5,pale colors) and of the coarse fraction (PM10-2.5, dark colors).

Figure 6a shows the seasonal variations in the total mass.Winter mass concentrations of PM10 were about twice thesummer values, but the winter increase was totally due tothe fine fraction (winter/summer ratio of the fine fractionwas about 3), while the coarse contribution remained almostconstant. On average, during the winter, the fine fraction ofPMwas about four times higher than the coarse fraction, whileduring the summer, the fine/coarse ratio was about 1.3. Thesevariations are the results of the seasonal changes in the inten-sity of the main macro-sources.

Natural sources, i.e., soil (Fig. 6b) and sea spray (Fig. 6c),generate particles mainly in the coarse fraction, with a fine/coarse ratio slightly higher than 0.3. The intensity of thesemacro-sources is very variable among years, as it depends onrelatively infrequent transport episodes. However, PM fromsea spray tends to be higher during the winter because of thehigher frequency of the advection needed to bring these par-ticles from the sea to inland areas. PM from soil, on the otherhand, is generally high during the summer because of both theenhanced dryness of local soil and the higher frequency oflong-range transport of desert dust from North Africa.

Anthropogenic macro-sources (vehicular emission, Fig. 6d,organics, Fig. 6e, and secondary inorganics, Fig. 6f), asexpected from their formation processes, were mainly in thefine fraction, with fine/coarse ratios between 5 and 10. Seasonalvariations show a winter/summer ratio of about 2 for trafficemission and close to 3 for secondary inorganics. For organics,the winter/summer ratio was also close to 3; however, whenconsidering separately sOC and pOC from domestic woodcombustion, this ratio became about 1.5 and about 50, respec-tively. All these winter increases are likely linked to the stabilityof the air masses, which favors both the accumulation ofemission products and the production of secondary species.Both sOC and secondary inorganics are particularly sensitiveto air mass aging, and their increase under stable atmosphericconditions is substantial (Perrino et al. 2013). As the latter two

classes of compounds constitute around one half of the PM10

mass, this winter increase in their concentration was reflected ina clearly higher PMmass concentration.Moreover, the estimateof the contribution of residential wood burning indicate thatduring the winter, this source was responsible for roughly halfof the organic matter and for a significant fraction of fineatmospheric PM (about 20 %).

Conclusions

The results of this study show that significant seasonal varia-tions occurred in the mass concentration and chemical compo-sition of atmospheric particles in the area of Ferrara, in NorthernItaly. These variations were mainly due to inefficient atmo-spheric mixing experienced in the Po valley region during thewinter, which could be successfully traced by monitoring nat-ural radioactivity due to Radon progeny. Atmospheric stabilityinfluenced, above all, inorganic secondary components of PM;the increase in the concentration of these species, and particu-larly of ammonium nitrate, is one of the main factors responsi-ble for the numerous exceedances of the EU concentrationlimits recorded in this area. From this perspective, it would beinteresting to evaluate the water content of PM, which alsoseemed to increase significantly during winter conditions ofatmospheric stability.

When considering the main PM macro-sources, we ob-served that the winter/summer concentration ratio was around2 for combustion products and roughly 3 for secondary inor-ganic species. For the latter group, the winter increase ofammonium nitrate and the transport episodes of sulfate fromthe Balkan area overcame the summer increase of sulfate dueto photochemical activity. Among organic species, a roughestimate of the contribution from traffic and from residentialwood heating showed winter to summer concentration ratioabout 1.5 and 50, respectively, indicating a main contributionof wood burning to winter pollution in the study area. Naturalsources of PM, i.e., soil and sea spray, show seasonal varia-tions mainly linked to the frequency of medium-range andlong-range transport events.

Acknowledgments The authors are grateful to L. Tofful and S. Paretifor their technical assistance during the SOPs. This study has been fundedby HERA s.p.a.

References

Basart S, Pérez C, Nickovic S, Cuevas E, Baldasano JM (2012)Development and evaluation of the BSC-DREAM8b dust regionalmodel over Northern Africa, theMediterranean and theMiddle East.Tellus 64:1–23

Canepari S, Cardarelli E, Giuliano A, Pietrodangelo A (2006)Determination of metals and non volatile ions in airborne particulate

Environ Sci Pollut Res

Author's personal copy

matter by a new two-step sequential leaching procedure. Part A:experimental design and optimisation. Talanta 69:581–587

Canepari S, Perrino C, Astolfi ML, Catrambone M, Perret D (2009)Determination of soluble ions and elements in ambient airsuspended particulate matter: inter-technique comparison of XRF,IC and ICP for sample-by-sample quality control. Talanta 77:1821–1829

Canepari S, Astolfi ML, Moretti S, Curini R (2010) Comparison ofextracting solutions for elemental fractionation in airborne particu-late matter. Talanta 82:834–844

Canepari S, Farao C, Marconi E, Giovannelli C, Perrino C (2013)Qualitative and quantitative determination of water in airborneparticulate matter. Atmos Chem Phys 13:1193–1202

Carbone C, Decesari S, Mircea M, Giulianelli L, Finessi E, Rinaldi M,Fuzzi S, Marinoni A, Duchi R, Perrino C, Sargolini T, Vardè M,Sprovieri F, Gobbi GP, Angelini F, Facchini MC (2010) Size-resolved aerosol chemical composition over the Italian peninsuladuring typical summer and winter conditions. Atmos Environ44:5269–5278

Castro LM, Pio CA, Harrison RM, Smith DJT (1999) Carbonaceousaerosol in urban and rural European atmospheres: estimation ofsecondary organic carbon concentrations. Atmos Environ33:2771–2781

Chan YC, Simpson RW,McTainsh GH, Vowles PD (1997) Characterisationof chemical species in PM2.5 and PM10 aerosols in Brisbane, Australia.Atmos Environ 31:3773–3785

EC: EU Air Quality Directive 2008/50/EC, 2008.EEA (2007) Air pollution in Europe 1990–2004. EEAReport No 2/2007.

European Environment Agency, Copenhagen. ISSN 1725–9177.Engling G, Carrico CM, Kreidenweis SM, Collett JL, Day DE, Malm

WC, Lincoln E, Hao WM, Iinuma Y, Herrmann H (2006)Determination of levoglucosan in biomass combustion aerosol byhigh-performance anion-exchange chromatography with pulsed am-perometric detection. Atmos Environ 40:S299–S311

Harrison RM, Jones AM, Lawrence RG (2003) A pragmatic mass closuremodel for airborne particulate matter at urban background androadside sites. Atmos Environ 37:4927–4933

Kataoka T, Yunoki E, Shimizu M, Mori T (2001) A study of the atmo-spheric boundary layer using radon and air pollutants as tracer.Bound-Layer Meteorol 101:131–155

Leithead A, Li SM, Hoff R, Cheng Y, Brook J (2006) Levoglucosan anddehydroabietic acid: evidence of biomass burning impact on aero-sols in the Lower Fraser Valley. Atmos Environ 40:2721–2734

Marcazzan GM, Vaccaro S, Valli G, Vecchi R (2001) Characterisation ofPM10 and PM2.5 particulate matter in the ambient air of Milan(Italy). Atmos Environ 35:4639–4650

Müller K, van Pinxteren D, Plewka A, Svrcina B, Kramberger H,Hofmann D, Bächmann K et al (2005) Aerosol characterisation atthe FEBUKO upwind station Goldlauter (II): detailed organic chem-ical characterization. Atmos Environ 39:4219–4231

Pernigotti D, Georgieva E, Thunis P, Bessagnet B (2012) Impact ofmeteorology on air quality modeling over the Po valley in northernItaly. Atmos Environ 51:303–310

Perrino C, Pietrodangelo A, Febo A (2001) An atmospheric stabilityindex based on radon progeny measurements for the evaluation ofprimary urban pollution. Atmos Environ 35:5235–5244

Perrino C, Catrambone M, Pietrodangelo A (2008) Influence of atmo-spheric stability on the mass concentration and chemical composi-tion of atmospheric particles: a case study in Rome, Italy. EnvironInt 34:621–628

Perrino C, Canepari S, Catrambone M, Dalla Torre S, Rantica E,Sargolini T (2009) Influence of natural events on the concentrationand composition of atmospheric particulate matter. Atmos Environ43:4766–4779

Perrino C, Canepari S, Pappalardo S, Marconi E (2010) Time-resolvedmeasurements of water-soluble ions and elements in atmosphericparticulate matter for the characterization of local and long-rangetransport events. Chemosphere 80:1291–1300

Perrino C, Canepari S, CatramboneM (2013) Comparing performance ofTeflon and quartz membrane filters collecting atmospheric PM:influence of atmospheric water. Aerosol Air Qual Res 13:137–147

Pey G, Querol X, Alastuey A (2009) Variations of levels and compositionof PM10 and PM2.5 at an insular site in the Western Mediterranean.Atmos Res 94:285–299

Putaud JP et al (2004) A European aerosol phenomenology 2: chemicalcharacteristics of particulate matter at kerbside, urban, rural andbackground sites in Europe. Atmos Environ 38:2579–2595

Saarikoski S, Carbone S, Decesari S, Giulianelli L, Angelini F,Canagaratna M, Ng NL, Trimborn A, Facchini MC, Fuzzi S,Hillamo R, Worsnop D (2012) Chemical characterization of spring-time submicrometer aerosol in Po Valley, Italy. Atmos Chem Phys12:8401–8421

Sesana L, Caprioli E, Marcazzan GM (2003) Long period study ofoutdoor radon concentration in Milan and correlation between itstemporal variation and dispersion properties of the atmosphere. JEnviron Radioact 65:147–160

Squizzato S, Masiol M, Brunelli A, Pistollato S, Tarabotti E, RampazzoG, Pavoni B (2013) Factors determining the formation of secondaryinorganic aerosol: a case study in the Po Valley (Italy). Atmos ChemPhys 13:927–1939

Szidat S, Jenk TM, Synal H-A, Kalberer M, Wacker L, Hajdas I, Kasper-Giebl A, Baltensperger U (2006) Contributions of fossil fuel, bio-mass burning, and biogenic emissions to carbonaceous aerosols inZurich as traced by 14C. J Geophys Res 111(D7)

Terzi E, Argyropoulos G, Bougatioti A, Mihalopoulos N, Nikolaou K,Samara C (2010) Chemical composition and mass closure of ambi-ent PM10 at urban sites. Atmos Environ 44:2231–2239

Theodosi C, Grivas G, Zarmpas P, Chaloulakou A, Mihalopoulos N(2011) Mass and chemical composition of size-segregated aerosols(PM1, PM2.5, PM10) over Athens, Greece: local versus regionalsources. Atmos Chem Phys 11:11895–11911

Turpin BJ, Lim H (2001) Species contribution to PM2.5 mass concentra-tion: revisiting common assumptions for estimating organic mass.Aerosol Sci Technol 35:602–610

Vecchi R, Marcazzan G, Valli G, Ceriani M, Antoniazzi C (2004) Therole of atmospheric dispersion in the seasonal variation of PM1 andPM2.5 Concentration and composition in the urban area of Milan(Italy). Atmos Environ 38:4437–4446

Vecchi R, Marcazzan G, Valli G (2007) A study on nighttime–daytimePM10 concentration and elemental composition in relation to atmo-spheric dispersion in the urban area of Milan (Italy). Atmos Environ41:2136–2144

Viidanoja J, Sillanpää M, Laakia J, Kerminen VM, Hillamo R, Aarnio P,Koskentalo T (2002) Organic and black carbon in PM2.5 andPM10: 1 year of data from an urban site in Helsinki, Finland.Atmos Environ 36:3183–3193

Wang H, Kawamura K, Shooter D (2005) Carbonaceous and ioniccomponents in wintertime atmospheric aerosols from two NewZealand cities: implications for solid fuel combustion. AtmosEnviron 39:5865–5875

Environ Sci Pollut Res

Author's personal copy