Chemosphere 62 (2006) 947–956

www.elsevier.com/locate/chemosphere

Chemical characterisation of PM episodes in NE Spain

M. Viana *, X. Querol, A. Alastuey
Institute of Earth Sciences ‘‘Jaume Almera’’, C/Lluis Solé i Sabarı́s s/n, 08028 Barcelona, Spain

Received 13 January 2005; received in revised form 17 May 2005; accepted 30 May 2005
Available online 3 August 2005

Abstract

The chemical composition of ambient particulate matter (PM) varies widely as a function of its main emission
sources and of the chemical reactions which take place in the atmosphere. The aim of this study is to obtain the chem-
ical profile of PM10 and PM2.5 during peak PM episodes, thus identifying the main emission sources and/or atmo-
spheric processes which originate the PM episodes. To this end, cluster analysis was applied to a set of PM10 and
PM2.5 data collected throughout 2001 in two urban and industrialised areas in NE Spain. As a result of this analysis,
five clusters were identified for each site, and the interpretation of their chemical profiles lead to the identification of five
types of peak PM episodes for each site: industrial, traffic and regional re-circulation episodes at both sites, plus crustal
episodes in Barcelona, and peak traffic and industrial episodes (T + I) in Tarragona. Traffic episodes are characterised
by daily means of 23 and 10 lg/m3 of OM + EC in Barcelona and Tarragona in PM10. Levels of secondary inorganic
aerosols reach average daily means of 19 and 11 lg/m3 in Barcelona and Tarragona in PM10 during industrial episodes.
High levels of sulphate (>5 lg/m3) and ozone (up to 77 lg/m3 daily mean) are good tracers of regional re-circulation
episodes. During crustal episodes daily means of crustal elements reach up to 34 lg/m3 in Barcelona. Special attention
has been drawn to the composition of the mineral matter during the different PM episodes.

2005 Published by Elsevier Ltd.

Keywords: Cluster analysis; Source contribution; Chemical profile; Sahara dust; Mineral matter

1. Introduction The study of this atmospheric pollutant has increased

Particulate matter (PM) is a complex mixture of sub-
stances suspended in the atmosphere in solid or liquid
state with different properties (e.g. size distribution or
chemical composition amongst others) and origins
(anthropogenic and natural). Owing to this mixture of
substances the chemical composition of PM may vary
widely as a function of its major emission sources and
the subsequent chemical reactions which take place in
the atmosphere.
*
Corresponding author. Fax: +34 934110012.
E-mail address: mviana@ija.csic.es (M. Viana).
in the past years due to the understanding of its adverse
effects on health (Dockery and Pope, 1996; Künzli et al.,
2000; WHO, 2003) and the consequent implementation
in the year 2001 of the European directive 1999/30/
EU, which establishes annual and daily limit values for
PM10 (the PM mass which passes through a size selec-
tive impactor inlet with a 50% efficiency cut-off at
10 lm aerodynamic diameter).
The chemical mass balance is the most commonly
used method for assessing particulate matter source con-
tributions (Wilson et al., 2002), and statistical methods
such as factor analysis and multi-linear regression
(Thurston and Spengler, 1985) have also produced inter-
esting results regarding source identification. The results

0045-6535/$ - see front matter
2005 Published by Elsevier Ltd.

doi:10.1016/j.chemosphere.2005.05.048
948 M. Viana et al. / Chemosphere 62 (2006) 947–956
obtained from the use of these techniques are usually ex-
pressed as the contributions (in % and/or lg/m3) of the
sources to the PM mass. However, the main objective of
the present study is not to identify the contributions of
the sources but to obtain the chemical profile of PM dur-
ing peak concentration episodes in the North-Eastern
Iberian Peninsula, and thus a different approach has
been chosen: the use of cluster analysis. Statistical clus-
tering of the PM samples classifies the samples in a
straightforward manner according to their chemical pro-
files, and yields a number of clusters characterised by
their specific chemical composition (Peña and Rivera,
1992). The resulting compositional profile is then a
reflection of the major emission source and/or the atmo-
spheric process involved in the production of the peak
PM episodes.
2. Methodology
Two urban monitoring sites were selected in Barce-
lona and Tarragona, on the North-Eastern coast of
Spain (Fig. 1). The Barcelona site is located in a park,
under the influence of dense road traffic (Meridiana
Avenue, located 150 m to the North of the station)
and a large industrial area including two fuel-oil power
plants. The Tarragona site is an urban background
site located on the rooftop of a seven-storey build-
ing, under the influence of PM emissions from road
Fig. 1. Study area and location of monitoring sites Barcelona and Tarragona.
traffic and a highly industrialised area (oil chemistry
and transformation, fuel-oil power plant, chemical
industries).
The sampling of PM10 and PM2.5 was carried out
from 09/01/2001 to 27/12/2001 by means of the EN-
1234-1 reference high volume method for PM10 (Ander-
sen, 68 m3/h) and high volume (30 m3/h) MCV samplers
for PM2.5. It is important to consider that this sampling
technique may be subject to the loss of a certain fraction
of carbonaceous species and NO
3 by volatilisation (Tur-
pin et al., 2000; Eatough et al., 2003). Three PM10 and
two PM2.5 samples were collected per week, and 2
PM10 and 1 PM2.5 samples per week were selected for
chemical analysis. The methodology (Querol et al.,
2001) is based on the acid digestion of 1/2 of the filter
for the determination of major and trace elements by
ICP-AES and ICP-MS, water leaching of 1/4 of the filter


to determine soluble ion concentrations ðSO2
4 , NO3 ,
þ

NH4 , Cl ) by ion chromatography and FIA colorime-
try, and elemental analysis of total C in the remaining
1/4 of the filter. The results from these chemical analyses
were then grouped into four main components: crustal
ðCO2
3 , SiO2, Al2O3, Ca, K, Mg, Fe, P2O5), OM + EC
(organic matter + elemental carbon), SIA (secondary

þ
inorganic aerosols, nonmarine SO2
4 , NO3 , NH4 Þ and

2

marine (Na, Cl , marine SO4 ). A variable percentage
of undetermined mass was observed. Relative analytical
errors were between 3% and 10% for the elements
studied.
 M. Viana et al. / Chemosphere 62 (2006) 947–956
Cluster analysis was applied to the PM10 samples by
means of the software package STATISTICA 4.2. The
algorithm selected was K-means clustering by single
linkage, taking Euclidean distances as the distance mea-
sure. The total number of samples was 88 for each site,
although the final clustering was reduced to 85 cases for
Barcelona and 81 for Tarragona as any cases with miss-
ing variables were excluded from the analysis. The same
analysis was then applied to the PM2.5 samples, but
owing to the lower number of samples available (44
and 43 samples respectively) the results were not
concluding.
3. Chemical profiles of PM episodes
The cluster analysis performed on the chemical data-
set produced five independent clusters of samples for
each site (Tables 1 and 2). Main tracer species were iden-
tified in the resulting chemical composition of each clus-
ter, and these tracers were compared to those presented
in available source contribution literature (Pacyna, 1986;
Chueinta et al., 2000; Kim and Henry, 2000; Querol
et al., 2001; Rodrı́guez et al., 2002; Sternbeck et al.,
2002; Pakkanen et al., 2003; Adachi and Tainosho,
2004; Rodrı´guez et al., 2004). As a result, the clusters ob-
tained for Barcelona were identified as industrial, traffic,
regional, crustal and undetermined (Table 1), whereas
those obtained for Tarragona were identified as indus-
trial, traffic, traffic and industrial peak episodes
(T + I), regional and undetermined (Table 2). The tracer
species which helped interpret the origin of each of the
clusters are discussed in detail below.
The undetermined cluster represents samples in
which the PM mass results from a mixture of sources.
Traffic and industrial episodes are considered to have a
local origin, as they result from the impact of local emis-
sions on PM levels under low dispersive atmospheric
scenarios. The T + I cluster in Tarragona was identified
as peak industrial and/or traffic episodes. The regional
clusters refer to regional re-circulation episodes as de-
fined by Millán et al. (1997), which are originated by
the re-circulation and ageing of air masses during the
summer months due to specific orographic and meteoro-
logical conditions in the Western Mediterranean basin.
Finally, in the case of Barcelona a cluster defined as
crustal was identified, whereas it was not identified in
the Tarragona analysis. In Barcelona this cluster repre-
sents the contribution of mineral matter from a number
of anthropogenic sources such as construction and
demolition dust and re-suspension from the park in
which the station is located. It sporadically includes dust
transported during African dust outbreaks, although
these cases only represent a minor percentage during
the sampling period (3 days out of 10 in the cluster).
The fact that this cluster is not present in Tarragona
949
may be due to the location of the site at a considerable
height (seven-storey building), which limits the influ-
ence of mineral dust sources such as road dust re-sus-
pension. The local influence of the park is also absent
in Tarragona. African dust contributions are not identi-
fied as an isolated cluster in Tarragona as the percentage
of the variability of the dataset represented by these
samples was too low for them to be isolated as one
independent cluster by the statistical analysis (consider-
ing that the analysis identifies no other direct crustal
sources).
The average chemical composition of PM10 and
PM2.5 for each cluster is summarised in Tables 1 and 2.
It is important to note that the cluster identified as traffic
in Barcelona in PM2.5 includes only one sample, and thus
the results may not be fully representative. Out of the 88
PM10 samples and the 44 and 43 PM2.5 samples avai-
lable for Barcelona and Tarragona respectively, 85 and
81 PM10 and 43 and 36 PM2.5 samples were considered
for the cluster analysis.
Industrial episodes: the number of cases in this cluster
is similar in Barcelona (n = 15, 18% of the PM10 cases)
and Tarragona (n = 20, 25%). The mean daily PM10
levels obtained during these episodes are 57.5 lg/m3 in
Barcelona (9.9 lg/m3 standard deviation, s.d.) and
37.1 lg/m3 in Tarragona (9.6 lg/m3 s.d.), and they were
mainly registered between April and May and in Sep-
tember, showing no specific seasonal trend. The main
tracers of the industrial origin of these episodes are the
secondary inorganic aerosols (SIA, Querol et al., 2001;
Rodrı´guez et al., 2004), which present average values
of 18.6 lg/m3 in Barcelona (6.5 lg/m3 s.d.) and
10.9 lg/m3 in Tarragona (5.4 lg/m3 s.d.). Due to the
influence of traffic at both sites, the levels of the OM +
EC and crustal fractions are also relatively high (13.7
lg/m3 and 17.0 lg/m3 in Barcelona, and 7.9 lg/m3 and
10.6 lg/m3 in Tarragona for OM + EC and the crustal
fraction, respectively). The composition of the SIA frac-
tion has also been analysed, with the result that in
Tarragona 55% of the total SIA mass is constituted by
nonmarine sulphate (6.0 lg/m3) and 31% by nitrate
(3.4 lg/m3), whereas in Barcelona both species represent
a similar fraction (39% nitrate, 7.2 lg/m3, and 40%
sulphate, 7.4 lg/m3). The prevalence of sulphate over
nitrate in Tarragona during industrial episodes responds
to the higher relative weight of industrial emissions with
respect to traffic emissions at this site, while in Barcelona
the traffic and industrial influences seem to have a simi-
lar relative weight. Due to the fact that industrial emis-
sions in the study areas generally do not affect the
mineral load, the composition of this fraction is very
similar to the annual average (Al2O3/Ca = 0.92 vs 0.91
annual mean in Barcelona, and 0.45 vs 0.43 in
Tarragona).
With respect to the trace elements, these episodes are
characterised in Barcelona and Tarragona by high levels
950 M. Viana et al. / Chemosphere 62 (2006) 947–956

Table 1
Average chemical composition of PM10 and PM2.5 at a traffic site with industrial influence in Barcelona, as a function of the 5 PM
episodes identified: industrial, traffic, regional re-circulation, crustal and mixed sources
Industrial Traffic Regional Crustal Mixed Industrial Traffic Regional Crustal Mixed
PM10 PM10 PM10 PM10 PM10 PM2.5 PM2.5 PM2.5 PM2.5 PM2.5
n = 15 n=8 n = 26 n = 10 n = 26 n=9 n=1 n = 12 n=8 n = 13
(18%) (9%) (31%) (12%) (31%) (21%) (2%) (28%) (19%) (30%)
lg/m3
T 19 14 25 19 16 20 13 24 16 15
SO2 5 7 5 7 5 4 5 4 4 2
NO 31 87 8 20 24 35 89 7 21 32
NO2 45 54 25 35 37 46 45 21 34 39
O3 21 8 44 29 21 23 6 46 19 16.4
PM 57.5 66.3 36.1 64.7 36.1 36.9 48.0 18.8 38.1 24.7
OM + EC 13.7 22.7 6.6 10.5 10.8 15.1 26.2 7.5 14.1 13.6
CO2

3 4.2 5.4 3.0 6.7 3.2 1.0 0.4 0.7 2.0 0.7
SiO2 6.3 5.2 4.3 15.0 3.2 1.1 0.4 0.8 4.1 1.0
Al2O3 2.1 1.7 1.4 5.0 1.1 0.4 0.1 0.3 1.4 0.3
Ca 2.3 3.0 1.4 3.3 1.7 0.4 0.2 0.3 0.9 0.4
K 0.5 0.6 0.4 1.0 0.4 0.2 0.3 0.1 0.4 0.2
Na 0.9 0.8 1.6 1.5 0.8 0.1 0.2 0.3 0.3 0.2
Mg 0.3 0.3 0.3 0.7 0.2 0.2 <0.1 0.1 0.2 0.1
Fe 1.2 1.6 0.7 2.2 0.8 0.3 0.3 0.2 0.5 0.2
nmSO2
4 7.4 5.5 5.3 3.8 3.3 6.3 6.3 3.9 4.0 3.4
mSO2
4 0.2 0.2 0.4 0.4 0.2 <0.1 <0.1 0.1 0.1 0.1
NO
3 7.2 5.7 2.5 3.4 2.7 5.2 6.4 0.5 3.1 2.0
Cl 0.6 1.3 1.1 1.4 1.0 0.3 0.9 0.3 0.4 0.5
NHþ4 3.9 2.0 1.5 1.8 1.4 3.7 4.1 1.4 2.0 1.7
%acc. 89.7 86.4 86.5 89.3 87.6 93.0 95.8 88.0 90.6 101.6
Undet. 6.0 9.1 5.1 7.3 4.9 3.0 2.0 3.8 5.5 1.2
Crustal 17.0 18.0 11.6 34.0 10.7 3.6 1.9 2.4 9.6 3.0
Marine 1.8 2.4 3.2 3.3 2.0 0.5 1.0 0.7 0.7 0.7
SIA 18.6 13.3 9.3 8.9 7.4 15.2 16.8 5.8 9.2 7.1
OM + EC 13.7 22.7 6.6 10.5 10.8 15.1 26.2 7.5 14.1 13.6
ng/m3
Ti 83 92 51 229 64 18 7 15 69 19
V 25 24 11 16 11 19 13 5 9 7
Cr 12 13 6 9 7 4 1.9 2 3 3
Co 0.5 0.5 0.3 0.7 0.3 0.3 0.1 0.3 0.2 0.3
Ni 13 11 5 7 6 9 6 3 4 4
Cu 67 87 34 42 45 36 48 26 24 37
Zn 122 180 60 104 96 61 67 42 61 55
As 2 1.9 1.0 2 1.2 1.9 1.1 0.7 0.9 0.8
Rb 1.9 2 1.3 5 1.3 0.6 0.5 0.5 1.4 0.5
Sr 7 8 5 11 5 1.5 1.0 1.2 3 2
Cd 0.8 1.5 0.5 0.5 0.7 0.9 1.2 0.3 0.5 0.7
Sn 6 7 4 3 4 6 8 2 4 4
Sb 12 21 6 10 8 4 8 3 4 4
Ba 50 74 25 57 38 6 8 4 11 14
Pb 95 107 29 69 46 60 54 20 49 42
n: Number of samples; OM + EC: organic matter + elemental carbon; nmSO2
2

4 : nonmarine sulphate; mSO4 : marine sulphate; %acc.:
PM mass recovery through analysis; undet.: undetermined; crustal: CO2

3 , SiO 2 , Al2 O3 , Ca, K, Mg, Fe, P 2O 5; SIA: secondary inorganic

þ

aerosols (nmSO2
2

4 , NO3 , NH4 ); marine: Na, Cl , mSO4 ; T: temperature (C).

of Ni (13 ng/m3 and 6 ng/m3), V (25 ng/m3 and 10 and/or petroleum refinery emissions (Pacyna, 1986).
ng/m3) and Cr (12 ng/m3 and 4 ng/m3). These tracers Both PM sources are found in the study areas (one
typically reflect the presence of fuel-oil power plant power plant and a refinery in Tarragona, two power
 M. Viana et al. / Chemosphere 62 (2006) 947–956 951

Table 2
Average chemical composition of PM10 and PM2.5 at an urban background site with industrial influence in Tarragona, as a function
of the 5 PM episodes identified: industrial, traffic, T + I (traffic + industrial peak episodes), regional re-circulation and mixed sources
Industrial Traffic T+I Regional Mixed Industrial Traffic T+I Regional Mixed
PM10 PM10 PM10 PM10 PM10 PM2.5 PM2.5 PM2.5 PM2.5 PM2.5
n = 20 n = 20 n=8 n = 16 n = 17 n=8 n=6 n=6 n=6 n = 10
(25%) (25%) (10%) (20%) (21%) (22%) (17%) (17%) (17%) (28%)
lg/m3
T 23 15 18 26 18 18 17 17 18 16
SO2 4 4 8 3 2 4 4 7 1.9 1.8
NO 33 63 136 11 16 36 67 131 11 16
NO2 38 42 49 31 27 44 40 47 24 27
O3 55 28 19 77 49 52 26 18 75 48
PM 37.1 42.6 47.1 31.1 29.5 27.3 26.9 24.5 19.6 16.2
OM + EC 7.9 10.4 14.2 6.4 6.6 8.1 8.3 11.3 5.2 6.0
CO2

3 3.9 3.0 4.6 2.5 2.9 0.9 0.5 0.6 0.5 0.4
SiO2 2.7 2.2 3.3 1.5 1.9 0.8 0.2 0.2 0.5 0.3
Al2O3 0.9 0.8 1.2 0.5 0.7 0.3 0.1 0.1 0.2 0.1
Ca 2.0 1.7 2.8 1.2 1.8 0.4 0.3 0.4 0.2 0.2
K 0.3 0.4 0.4 0.2 0.3 0.2 0.2 0.2 0.1 0.1
Na 1.3 0.9 0.9 1.5 1.3 0.6 0.3 0.3 0.3 0.2
Mg 0.3 0.3 0.4 0.3 0.3 0.1 0.1 0.1 0.1 <0.1
Fe 0.4 0.4 0.6 0.3 0.3 0.1 0.1 0.1 0.1 0.1
nmSO2
4 6.0 4.7 4.3 5.7 3.3 6.3 3.4 3.1 4.9 2.8
mSO2
4 0.3 0.2 0.2 0.4 0.3 0.1 0.1 0.1 0.1 0.1
NO
3 3.4 6.0 4.9 2.3 2.2 1.7 3.8 3.2 0.5 1.1
Cl 0.8 0.8 0.6 0.9 1.2 0.2 0.3 0.3 0.1 0.1
NHþ4 1.5 2.2 1.2 1.3 0.9 2.5 1.6 1.6 1.9 1.4
%acc. 86.1 79.7 85.0 80.7 81.5 82.4 74.9 91.5 80.4 89.7
Undet. 5.8 8.6 7.2 5.9 5.9 5.5 7.5 3.6 6.2 5.8
Crustal 10.6 8.7 13.4 6.5 8.2 2.9 1.5 1.7 1.7 1.3
Marine 2.4 1.9 1.8 2.8 2.7 0.9 0.7 0.7 0.5 0.4
SIA 10.9 12.9 10.4 9.4 6.3 10.5 8.8 7.9 7.3 5.2
OM + EC 7.9 10.4 14.2 6.4 6.6 8.1 8.3 11.3 5.2 6.0
ng/m3
Ti 22 23 35 16 22 8 4 6 6 5
V 10 8 12 6 4 9 6 6 4 3
Cr 4 2 3 1.9 1.8 4 1.9 3 2 0.8
Co 0.2 0.2 0.3 0.2 0.1 0.1 0.1 0.1 0.1 0.1
Ni 6 4 6 3 2 5 3 5 3 1.7
Cu 25 34 44 22 23 40 40 39 22 18
Zn 31 44 50 24 22 19 24 30 14 12
As 0.7 1.0 1.1 0.4 0.5 0.4 0.6 0.7 0.2 0.3
Rb 0.9 0.9 1.3 0.6 0.7 0.6 0.4 0.4 0.2 0.2
Sr 6 4 6 4 5 2 1.0 0.9 1.0 0.6
Cd 0.3 0.4 0.3 0.1 0.2 0.2 0.4 0.4 0.2 0.1
Sn 1.7 2 2 2 1.2 1.3 1.5 0.9 0.7 0.4
Sb 4 5 7 2 3 1.7 1.8 5 0.9 4
Ba 13 12 17 9 10 3 4 4 2 3
Pb 19 36 40 13 17 15 26 23 10 12
n: Number of samples; OM + EC: organic matter + elemental carbon; nmSO2
2

4 : nonmarine sulphate; mSO4 : marine sulphate; % acc.:
PM mass recovery through analysis; undet.: undetermined; crustal: CO2

3 , SiO 2 , Al2 O3 , Ca, K, Mg, Fe, P 2 O5; SIA: secondary inorganic

þ

aerosols (nmSO2
2

4 , NO3 , NH4 ); marine: Na, Cl , mSO4 ; T: temperature (C).

plants in Barcelona). The levels of gaseous pollutants do Barcelona (n = 8, 9%), despite the marked traffic influ-
not show marked increases during these episodes. ence at the latter site. The fact that the Tarragona traffic
Traffic episodes: the number of cases in this cluster is cluster is more populated than in Barcelona does not
higher in Tarragona (n = 20, 25% of the cases) than in indicate that the number of traffic episodes is higher in
952 M. Viana et al. / Chemosphere 62 (2006) 947–956
Tarragona, but that the traffic contributions are not as
clearly differentiated in Barcelona. The higher degree of
mixture between emissions from different sources in Bar-
celona is probably related to the location of the main
industrial and traffic sources in the same direction (N-
NE, Besòs river valley and Meridiana Ave.) with respect
to the monitoring site. The PM10 levels registered reach
high levels at both sites (the highest for all clusters in
Barcelona, 66.3 lgPM10/m3, 7.0 lg/m3 s.d. and 42.6
lgPM10/m3, 11.0 lg/m3 s.d. in Tarragona). These
episodes are mainly registered from October to March,
coinciding with lower atmospheric dispersive conditions.
The chemical composition of PM during these episodes
in Barcelona is mainly determined by OM + EC (22.7
lg/m3, 4.6 lg/m3 s.d.) as a result of the road traffic
exhaust emissions, followed by the mineral fraction
(18.0 lg/m3, 4.6 lg/m3 s.d.) originated from road dust
erosion and re-suspension (Kim and Henry, 2000; Querol
et al., 2001). Note that the mean OM + EC levels during
traffic episodes double the annual average obtained in
Barcelona (11 lg/m3). In Tarragona the strong influence
of the industrial emissions generates high SIA background
levels during all episodes. Thus, during traffic episodes the
highest OM + EC levels are registered (10.4 lg/m3, except
for peak T + I episodes) and in parallel high SIA levels are
registered (12.9 lg/m3, 6.9 lg/m3 s.d.). During traffic epi-
sodes the nitrate fraction prevails over sulphate in Tarrag-
ona (47% nitrate, 36% sulphate) whereas in Barcelona the
ratios are once again balanced (43% nitrate, 42% sulphate).
The composition of the mineral road dust re-suspended
by traffic is in both cases dominated by Ca (3.0 lg/m3 in
Barcelona, 1.7 lg/m3 in Tarragona) with respect to
Al2O3 (1.7 lg/m3 in Barcelona, 0.8 lg/m3 in Tarragona),
indicating that construction work dust is a significant
contributor to mineral dust deposited on roads at urban
sites. The Ti/Fe ratio may constitute a good tracer of road
dust re-suspension as the results obtained for Barcelona
(Ti/Fe = 59) and Tarragona (Ti/Fe = 53) are significantly
different to that obtained in Barcelona for the crustal
cluster (Ti/Fe = 106).
The emissions from road traffic are typically charac-
terised by high levels of Ba, Sb, Cd, Pb, Zn and Cu
(Sternbeck et al., 2002; Pakkanen et al., 2003; Adachi
and Tainosho, 2004). These results are confirmed during
traffic episodes in Barcelona, where the average concen-
trations of these trace elements reach levels of Cu (87 ng/
m3), Zn (180 ng/m3), Cd (1.5 ng/m3), Sb (21 ng/m3), Ba
(74 ng/m3) and Pb (107 ng/m3) which are maximal when
compared to the other clusters. In the case of Tarragona
the most outstanding elements are Cu (34 ng/m3), Zn
(43 ng/m3), Sb (5 ng/m3) and Pb (36 ng/m3), although
it must be noted that the traffic cluster in Tarragona
does not include the peak traffic episodes, included in
the T + I cluster.
Finally, traffic episodes are also characterised by
the highest levels of gaseous compounds such as NO
(87 lg/m3, 63 lg/m3 in Barcelona and Tarragona) and
NO2 (54 lg/m3, 42 lg/m3 in Barcelona and Tarragona).
Consequently, O3 levels are relatively low (8 lg/m3 and
28 lg/m3, respectively).
Traffic + Industrial peak episodes (T + I): this cluster
was only identified in the Tarragona analysis, and in-
cludes 8 PM10 samples (10% of the PM10 cases). The
meteorological analysis of these days showed that the
synoptic scale scenario was dominated by high pressure
systems, mainly in winter, which favoured the stagna-
tion of the air masses and the increase of pollutants
emitted by local sources. The mean chemical composi-
tion of these local anthropogenic pollution episodes is
dominated by OM+EC (14.2 lg/m3, 3.0 lg/m3 s.d.)
and crustal components (13.4 lg/m3, 6.6 lg/m3 s.d.) as
a result of the traffic influence, but also by SIA
(10.4 lg/m3, 5.5 lg/m3 s.d.) originating from the indus-
trial emissions. As a result of the source mixture present
in this cluster, the chemical composition is characterised
by high levels of elements such as V (12 ng/m3) and Ni
(6 ng/m3), tracers of oil combustion emissions, but also
by Cu (44 ng/m3), Ba (17 ng/m3) or Sb (7 ng/m3), tracers
of road traffic emissions. Furthermore, levels of mineral
trace elements such as Ti (35 ng/m3) or Sr (6 ng/m3) are
also found in relatively high concentrations during these
episodes. Regarding gaseous compounds, these episodes
present the highest mean concentrations of SO2 (8
lg/m3), NO (136 lg/m3) and NO2 (49 lg/m3) and the
lowest of O3 (19 lg/m3) in Tarragona.
Crustal episodes: crustal episodes (n = 10, 12% of the
PM10 cases) were only detected in Barcelona, and they
include a number of anthropogenic sources such as con-
struction and demolition dust and re-suspension from
the park in which the station is located, and sporadically
mineral dust transported during African dust outbreaks
(3 out of 10 cases in the cluster). In Tarragona the sta-
tion is located at a higher altitude than the Barcelona
site, and this may account for a reduced influence of
anthropogenic re-suspended dust. These episodes were
detected in Barcelona in February–April, October and
November, showing no seasonal trend. Average PM10
levels obtained in Barcelona are the second highest after
the traffic episodes (64.7 lg/m3). As expected, the chem-
ical composition is dominated by the levels of crustal
components (34.0 lg/m3, 10.4 lg/m3 s.d., Chueinta
et al., 2000; Kim and Henry, 2000; Querol et al.,
2001), but it is important to note that in this fraction
the levels of Al2O3 are predominant (5.0 lg/m3) with
respect to Ca (3.3 lg/m3). This composition contrasts
with that obtained during traffic episodes in which Ca
prevailed both in Barcelona and Tarragona. Apart from
the higher Al2O3/Ca ratio (1.5) during crustal episodes
with respect to traffic episodes (Al2O3/Ca = 0.6 in
Barcelona, 0.4 in Tarragona), the Al2O3 levels are also
higher in this cluster (5.0 lg/m3 vs 1.7 lg/m3 during traf-
fic episodes in Barcelona), evidencing the influence of
 M. Viana et al. / Chemosphere 62 (2006) 947–956
clay minerals from African dust. This influence may also
be observed regarding trace elements such as Ti (229
ng/m3) Rb (5 ng/m3) or Sr (11 ng/m3), which present
maximal concentrations during crustal episodes.
Regional re-circulation episodes: the mean daily
PM10 levels obtained during regional re-circulation epi-
sodes at both study sites are lower than under crustal,
traffic or industrial episodes. On average, PM10 levels
reach 36.1 lg/m3 in Barcelona (n = 26, 31%) and
31.1 lg/m3 in Tarragona (n = 16, 20%). The main
PM10 constituents during these days are SIA and crustal
components in Barcelona (9.3 lg/m3, 4.0 lg/m3 s.d. and
11.6 lg/m3, 3.0 lg/m3 s.d.) and Tarragona (9.3 lg/m3,
5.1 s.d. and 6.5 lg/m3, 2.4 s.d.). As defined by Millán
et al. (1997) regional re-circulation episodes are charac-
terised by the ageing of air masses and they mainly occur
during the spring and summer months, and it is a pro-
cess which favours the formation of secondary aerosols
(Rodrı´guez et al., 2002). Thus, owing to the high atmo-
spheric stability of ammonium sulphate with respect to
ammonium nitrate in warm periods, 57–61% of the
SIA mass in PM10 is comprised of sulphate in Barcelona
and Tarragona (mean levels of 5.3 lg/m3 and 5.7
lg/m3), and only 25–27% by nitrate (2.4–2.5 lg/m3).
Furthermore, the summer convective circulations in-
crease the re-suspension of soil particles and thus result
in an increase of the crustal load in PM10 (mainly
made up of Ca compounds, especially in Tarragona).
The composition of PM10 during these episodes is not
significantly characterised by the levels of any of the
trace elements analysed, which are usually found in
lower concentrations than under the other clusters
identified. Conversely, gaseous pollutants are very good
tracers of this type of episode as O3 levels reach the
highest values (77 lg/m3 in Tarragona and 44 lg/m3 in
Barcelona) and NO the lowest (11 lg/m3 and 8 lg/m3,
respectively). Owing to the fact that these episodes
mainly occur in summer (>80% in this study), tempera-
ture may constitute another good indicator of the occur-
rence of regional re-circulation episodes (highest mean
levels of 25 C and 26 C registered in Barcelona and
Tarragona for this type of episode, respectively).
Undetermined (mixed) episodes: these clusters group
the samples in which a mixture of sources was identified,
and thus do not present a specific chemical profile or
gaseous tracers. They consist of 26 cases (31%) in Barce-
lona and 17 (21%) in Tarragona for PM10, and are de-
tected throughout the year.
The cluster analysis was also applied to PM2.5, but
the results were not representative due to the low num-
ber of cases available for each site (36 in Tarragona,
43 in Barcelona). However, in order to characterise the
chemical composition of the finer grain size fraction
the clustering obtained for PM10 was applied to the
PM2.5 samples. For this approach it was assumed that
the same sources which dominate in PM10 have a signif-
953
icant influence on the finer fractions. This assumption
may only be considered valid if all the sources identified
in this study (industry, traffic, regional re-circulation,
crustal material) have an impact on the levels of both
PM10 and PM2.5, even though this impact may be
clearly distinct for each fraction. The results here pre-
sented for PM2.5 (extrapolated from PM10) should thus
not be considered as fully conclusive. This approach for
PM2.5 has implications mainly for the days in which
coarser sized emission sources dominate in PM10 (e.g.,
crustal material). Taking into account that the impact
of this source is lower in PM2.5 than in PM10 (although
not negligible), the resulting PM2.5 chemical profile
would overestimate the contribution of the finer grain
size sources, as their relative weight would be higher
than that of the coarser grain size sources. A minimum
number of 80–100 cases would be necessary for the opti-
mal differentiation of independent clusters.
PM2.5 levels under traffic episodes reach mean daily
values of 48.0 lg/m3 in Barcelona (n = 1, 2%) and
27.0 lg/m3 in Tarragona (n = 6, 17%). Owing to the fine
grain size distribution of OM + EC, it is this component
that predominates in the composition of this PM frac-
tion (26.2 lg/m3 in Barcelona, 8.3 lg/m3 in Tarragona).
The relative weight of the mineral fraction decreases sig-
nificantly in PM2.5 (20–27% in PM10 to 4–6% in PM2.5
at both sites). In absolute values the levels are very sim-
ilar in Barcelona (1.9 lg/m3) and Tarragona (1.5 lg/m3),
although higher levels were expected in Barcelona as the
PM2.5 crustal fraction is significantly influenced by road
traffic (Querol et al., 2004). This might be due to the fact
that there is only one PM2.5 sample available in
Barcelona.
Levels of PM2.5 are maximal during industrial epi-
sodes in Tarragona (27.3 lgPM2.5/m3, n = 8, 22%)
and high in Barcelona (36.9 lgPM2.5/m3, n = 9, 21%).
The major contributor to the PM2.5 mass are the SIA,
with 15.2 lg/m3 in Barcelona and 10.5 lg/m3 in Tarra-
gona. The OM + EC contribution is also significant
(15.1 lg/m3 in Barcelona and 8.1 lg/m3 in Tarragona).
During regional re-circulation episodes PM2.5 levels
reach 18.8 lg/m3 in Barcelona (n = 12, 28%) and
19.6 lg/m3 in Tarragona (n = 6, 17%), and the main
components are OM + EC and SIA (7.5 lg/m3 and
5.8 lg/m3 respectively in Barcelona and 5.2 lg/m3 and
7.3 lg/m3 in Tarragona). The prevalence of sulphate
with respect to nitrate and ammonia observed in PM10
is even more evident in PM2.5, as 67–68% of the SIA
mass is made up of nonmarine sulphate (3.9 lg/m3 and
4.9 lg/m3 in Barcelona and Tarragona) whereas nitrate
constitutes only 7–9% (0.5 lg/m3 at both sites). The
composition of the crustal load is determined by Ca in
Tarragona (Al2O3/Ca = 0.7) owing to the geology of
the area, whereas in Barcelona the ratio is higher
(Al2O3/Ca = 1.1). These results were also observed for
PM10.
954 M. Viana et al. / Chemosphere 62 (2006) 947–956
Fig. 2. Chemical composition of PM10 and PM2.5 in Barcelona and Tarragona as a function of the different PM episodes identified.
Finally, the mean daily PM2.5 values registered dur-
ing crustal episodes reach 38.1 lg/m3 in Barcelona
(n = 8, 19%), with a mineral load of 9.6 lg/m3. This re-
sult is significantly high, as it represents 35% of the
PM2.5 mass. The influence of clay minerals is still evi-
dent in the PM2.5 fraction, as the Al components prevail
over Ca (Al2O3/Ca = 1.4).
Accordingly to the results described above, Fig. 2
summarises the mean chemical composition of PM10
and PM2.5 during the different PM episodes identified
in Barcelona and Tarragona. The figure shows how
the percentages of OM + EC and SIA increase in
PM2.5 with respect to PM10 at both sites, whereas the
mineral and the marine fractions decrease significantly
due to their coarser grain size distributions.
4. Composition of the crustal load
The chemical composition of the crustal load may
vary as a function of the different PM episodes. These
variations are presented in Table 3, expressed as the rela-
tive weight of Al2O3, Ca, Fe and K with P respect to the
sum of these elements and compounds [ (Al2O3, Ca,
Fe, K)]. This group of elements has been selected be-
cause they are major components of the mineral fraction
and their concentrations were determined by chemical
analyses (direct measurements).
The largest variations are observed at the Barcelona
site. For this station, Ca is the dominant element in
PM10 and PM2.5 during industrial and traffic episodes
(37% and 44% for PM10 and 29% and 24% for PM2.5,
respectively), mainly due to the re-suspension of road
dust by traffic. Road dust may contain high levels of Ca
originated from construction works in the city area
and/or from road pavement and paint (Adachi and
Tainosho, 2004). During crustal episodes, on the other
hand, the Al2O3 component prevails (44% in PM10 and
42% in PM2.5) as the African and local soil dust is made
up by a relatively high proportion of clay materials. Re-
gional re-circulation episodes favour the transport of re-
gional as well as local dust, and thus the resulting
chemical composition of the mineral fraction is balanced
between Al2O3 (36% in PM10, 33% in PM2.5) and Ca
(37% in PM10, 32% in PM2.5). The relative weight of
Fe and K is rather constant in the PM10 fraction for all
PM episodes (18–23% and 8–9%, respectively), but it pre-
sents larger variations in PM2.5. In the fine fraction,
maximal levels of these elements are detected during
traffic episodes (29% Fe, 32% K), indicating that they
are directly linked to road traffic emissions, probably to
abrasion of brake lining. The presence of high levels of
Fe in road dust has been observed by a number of authors
(Sternbeck et al., 2002; Adachi and Tainosho, 2004).
The results obtained for Tarragona are much more
constant, as the Ca component is the highest for all of
 M. Viana et al. / Chemosphere 62 (2006) 947–956 955

Table 3
P
Relative weight of Al2O3, Ca, Fe and K with respect to the sum of the same elements and compounds [ (Al2O3, Ca, Fe, K)] for each
of the clusters identified at the Barcelona and Tarragona study sites
Industrial Traffic Regional Crustal Mixed Average
PM10 PM2.5 PM10 PM2.5 PM10 PM2.5 PM10 PM2.5 PM10 PM2.5 PM10 PM2.5
Barcelona
Al2O3 35 29 25 15 36 33 44 42 27 27 33 29
Ca 37 29 44 24 37 32 29 29 43 34 38 30
Fe 20 24 23 29 18 20 19 16 20 20 20 22
K 8 18 9 32 9 15 9 12 9 19 9 19
Traffic Industrial T+I Regional Mixed Average
PM10 PM2.5 PM10 PM2.5 PM10 PM2.5 PM10 PM2.5 PM10 PM2.5 PM10 PM2.5
Tarragona
Al2O3 24 11 26 27 24 12 23 27 23 20 24 19
Ca 52 41 55 41 55 48 53 42 58 48 55 44
Fe 13 15 12 13 13 17 13 15 11 15 12 15
K 11 33 8 19 9 24 10 16 8 17 9 22

the PM episodes in PM10 and PM2.5, ranging from 52–
58% in PM10 and from 41–48% in PM2.5. The relative
weights of the remaining elements are also rather con-
stant in PM10 (Al2O3 23–26%, Fe 11–13%, K 8–11%),
but not in PM2.5. As in Barcelona, the relative weights
of Fe and K increase significantly in the fine fraction (Fe
17% during peak traffic episodes, K 33% during traffic
episodes).
On average, in PM10 Ca has a stronger relative
weight within the crustal load in Tarragona than in Bar-
celona, even though it is dominant at both sites (38% in
Barcelona, 55% in Tarragona). Road traffic emissions
are the main cause of the higher relative weight of Fe
in Barcelona (traffic site, 20%) than in Tarragona (urban
background, 12%), and the results obtained for K are the
same at both sites (9%). As regards PM2.5, at both sites
the fine fraction is enriched in Fe and K with respect to
PM10 (22% and 19% respectively in Barcelona, 15%
and 22% in Tarragona) and depleted in Ca (30% in Bar-
celona, 44% in Tarragona). Al2O3 also shows a decrease
in PM2.5 at both sites (29% in Barcelona, 19% in Tarra-
gona), although it is not as marked as in the case of Ca.
The coarser Ca particles are originated from pavement
abrasion and road dust re-suspension, whereas the Fe
and K particles result from finer grain size sources such
as brake lining abrasion. These results coincide with
CE (2004), where it is stated that PM10 particles are
mainly produced by pavement abrasion whereas when
automobile parts such as brake lining are worn out, finer
grain size particles (PM2.5) are produced.
5. Conclusions
The main objective of the present study is to identify
the chemical profiles of PM10 and PM2.5 episodes in the
industrialised urban areas of North-Eastern Spain. To
this end, a cluster analysis was performed on PM10
and PM2.5 samples. The differentiation between clusters
in Tarragona was relatively more complicated than in
Barcelona due to the high industrial background in this
city, which results in high SIA levels during all episodes.
Thus, the variations of the levels of OM + EC and gas-
eous pollutants (NO, SO2) played a key role in the iden-
tification of traffic and T + I episodes.
• Traffic episodes are characterised by mean daily val-
ues of 22.7 lg/m3 of OM + EC in Barcelona and
14.2 lg/m3 of OM + EC in Tarragona in PM10.
Good tracers of traffic emissions are Cu, Zn, Sb
and Pb as a result of exhaust emissions and break lin-
ing abrasion.
• During industrial episodes the composition of the
PM10 mass is determined by the secondary inorganic
aerosols (SIA), as they add up to daily mean levels of
18.6 lg/m3 in Barcelona and 10.9 lg/m3 in Tarrag-
ona. Due to the fact that the main industrial activities
at both sites are power generation by fuel-oil and oil
transformation (in Tarragona), the levels of Ni and V
are good tracers of industrial emissions.
• Regional re-circulation episodes, typical of the Wes-
tern Mediterranean (Millán et al., 1997; Rodrı́guez
et al., 2002), are characterised by high levels of
O3 and SIA, especially sulphates (5.3 lg/m3 in
Barcelona, 5.7 lg/m3 in Tarragona) as a consequence
of the enhanced photochemistry of the aged air
masses.
• Crustal episodes are only identified at the Barcelona
site, as they mainly represent the influence from differ-
ent anthropogenic sources (construction, demolition,
re-suspension from a nearby park) and sporadically
African dust contributions. The composition of this
956 M. Viana et al. / Chemosphere 62 (2006) 947–956
mineral dust (Al2O3/Ca = 1.5) differs clearly from
that of city road dust (Al2O3/Ca = 0.6).
• T + I episodes are only observed in Tarragona, and
represent peak PM episodes registered under strong
anticyclonic conditions. The chemical profile of these
episodes is characterised by high levels of OM + EC
(14.2 lg/m3), SIA (10.4 lg/m3) and gaseous pollu-
tants (136 lgNO/m3, 8 lgSO2/m3).
The composition of the mineral load in PM10 varies
as a function of the PM episodes. On average, Ca has a
stronger relative weight in Tarragona than in Barcelona,
even though it is dominant at both sites (38% of
P 102, 8811–8823.
(Al2O3, Ca, Fe, K) in Barcelona, 55% in Tarragona).
Road traffic emissions are the main cause of the higher
relative weight of Fe in Barcelona (traffic site, 20%) than
in Tarragona (urban background, 12%). At both sites
the fine fraction is enriched in Fe and K with respect
to PM10 and depleted in Ca and Al2O3 as the Fe and
K particles result from finer grain size sources such as
brake lining abrasion. The coarser Ca particles are
originated from pavement abrasion and road dust re-
suspension.
Acknowledgements
The authors would like to thank the Spanish Minis-
tries of Education and Science and Technology
(CGL2004-05984-C07-02/CLI) and the Autonomous
Government of Catalonia for their funding. The data
on gaseous pollutants were kindly provided by the Envi-
ronmental Department of the Autonomous Government
of Catalonia. The authors would like to acknowledge
Ferrán Clua and Joan Miró for performing the Tarra-
gona sampling.
References
Adachi, K., Tainosho, Y., 2004. Characterization of heavy
metal particles embedded in tire dust. Environ. Int. 30 (8),
1009–1017.
CE, 2004. II Position Paper on PM, CAFE Programme,
Environment D.G., European Commission, 214 pp.
Chueinta, W., Hopke, P.K., Paatero, P., 2000. Investigation of
sources of atmospheric aerosol at urban and suburban
residential areas in Tailand by positive matrix factorization.
Atmos. Environ. 34, 3319–3329.
Dockery, D., Pope, A., 1996. Epidemiology of acute health
effects: Summary of time-series studied. In: Spengler, J.D.,
Wilson, R. (Eds.), Particles in Our Air: Concentration and
Health Effects. Harvard University Press, pp. 123–147.
Eatough, D.J., Long, R.W., Modey, W.K., Eatough, N.L.,
2003. Semi-volatile secondary organic aerosol in urban
atmospheres: meeting a measurement challenge. Atmos.
Environ. 37, 1277–1292.
Kim, B.M., Henry, R.C., 2000. Application of SAFER model
to the Los Angeles PM10 data. Atmos. Environ. 34, 1747–
1759.
Künzli, N., Kaier, R., Medina, S., Studnicka, M., Chanel, O.,
Filliger, P., Herry, M., Jr, F.H., Puybonnieux-Texier, V.,
Quénel, P., Schneider, J., Seethaler, R., Vergnaud, J.C.,
Sommer, H., 2000. Public health impact of outdoor and
traffic related air pollution: a European assessment. The
Lancet 356, 795–801.
Millán, M., Salvador, R., Mantilla, E., Kallos, G., 1997. Photo-
oxidant dynamics in the Mediterranean basin in summer:
results from European research projects. J. Geophys. Res.
Pacyna, J.M., 1986. Emission factors of atmospheric elements.
In: Nriagu, J.O., Davidson, C.I. (Eds.), Toxic Metals in the
Atmosphere. Wiley, New York, pp. 33–52.
Pakkanen, T.A., Kerminen, V., Loukkola, K., Hillamo, R.E.,
Aarnio, P., Koskentalo, T., Maenhaut, W., 2003. Size
distributions of mass and chemical components in street-
level and rooftop PM1 particles in Helsinki. Atmos.
Environ. 37, 1673–1690.
Peña, D., Rivera, G.S.D., 1992. Estadı́stica. Modelos y méto-
dos. 1. Fundamentos. Alianza Universidad Textos.
Querol, X., Alastuey, A., Rodrı́guez, S., Plana, F., Ruiz, C.R.,
Cots, N., Massagué, G., Puig, O., 2001. PM10 and PM2.5
source apportionment in the Barcelona Metropolitan Area,
Catalonia, Spain. Atmos. Environ. 35–36, 6407–6419.
Querol, X., Alastuey, A., Viana, M.M., Rodrı́guez, S.,
Artı́ñano, B., Salvador, P., Santos, S.G.D., Patier, R.F.,
Ruiz, C.R., Rosa, J.D.L., Campa, A.S.D.L., Menedez, M.,
Gil, J.I., 2004. Speciation and origin of PM10 and PM2.5 in
Spain. J. Aerosol. Sci. 35 (9), 1151–1152.
Rodrı́guez, S., Querol, X., Alastuey, A., Mantilla, E., 2002.
Origin of high summer PM10 and TSP concentrations at
rural sites in Eastern Spain. Atmos. Environ. 36 (19), 3101–
3112.
Rodrı́guez, S., Querol, X., Alastuey, A., Viana, M.M., Alarcón,
M., 2004. Comparative PM10–PM2.5 source contribution
study at rural, urban and industrial sites during PM
episodes in Eastern Spain. Sci. Total Environ. 328, 95–113.
Sternbeck, J., Sjödin, A., Andréasson, K., 2002. Metal emis-
sions from road traffic and the influence of resuspension—
results from two tunnel studies. Atmos. Environ. 36, 4735–
4744.
Thurston, G.D., Spengler, J.D., 1985. A quantitative assess-
ment of source contribution to inhalable particulate matter
pollution in Metropolitan Boston. Atmos. Environ. 19 (1),
9–25.
Turpin, B.J., Saxena, P., Andrews, E., 2000. Measuring and
simulating particulate organics in the atmosphere: problems
and prospects. Atmos. Environ. 34, 2983–3013.
WHO, 2003. Health aspects of air pollution with particulate
matter, ozone and nitrogen dioxide. World Health Organi-
zation, 98 pp.
Wilson, W.E., Chow, J.C., Claiborn, C., Fusheng, W., Enge-
lbrecht, J., Watson, J.G., 2002. Monitoring of particulate
matter outdoors. Chemosphere 49, 1009–1043.