www.elsevier.com/locate/scitotenv

Science of the Total Environment 334–335 (2004) 337–346

Characterisation of aerosol particulate matter from urban

and industrial environments: examples from Cardiff

and Port Talbot, South Wales, UK

Teresa Morenoa,*, Tim P. Jonesb, Roy J. Richardsc

a Instituto de Ciencias de la Tierra ‘‘Jaume Almera’’, Consejo Superior de Investigaciones Cientıficas (CSIC),

C/ Lluis Sole i Sabaris, s/n, 08028 Barcelona, SpainbSchool of Earth, Ocean and Planetary Sciences, Cardiff University, Cardiff CF10 3YE, Wales, UK

cSchool of Biosciences, Cardiff University, Cardiff CF10 3US, Wales, UK

Accepted 1 April 2004

Abstract

A high-volume cascade impact collector (1100 l/min air flow) was used to collect air samples in an industrial (Port Talbot)

and an urban (Cardiff) site with the purpose of characterising both coarse (PM10–2.5) and fine (PM2.5) fractions comprising the

total sample. PM10–2.5 and PM2.5 samples were collected by cascading air through two polyurethane foams on which particles

impact and become deposited. Air sample collection rates are to some extent dependent on weather conditions, notably rainfall,

humidity, and especially, wind direction, but samples show a very different and distinctive air particle composition between the

two collection sites. Thus, although both Cardiff and Port Talbot are coastal sites and therefore have high contents in chlorides,

Port Talbot is extremely rich in tiny Fe spherules (>30%, in both coarse and fine fractions) from a nearby steel plant.

Mineralogical characterisation using SEM-EDX shows a clear fractionation between the particle composition in the PM

fractions, with the coarse fraction being dominated by chlorides, sulphates (gypsum), and silicates, and the fine fraction having

high proportions of ammonium sulphates and elemental and organic carbon compounds, most of the latter being linked to traffic

pollution.

D 2004 Elsevier B.V. All rights reserved.

Keywords: Urban and industrial particulate matter; PM10– 2.5; PM2.5; High-volume collector; Iron spherules

1. Introduction

State-of-the-art epidemiological research has found

strong links between air pollution and morbidity and

mortality rates, with air pollution potentiating a wide

0048-9697/$ - see front matter D 2004 Elsevier B.V. All rights reserved.

doi:10.1016/j.scitotenv.2004.04.074

* Corresponding author. Tel.: +34-934095410; fax: +34-

934110012.

E-mail address: tmoreno@ija.csic.es (T. Moreno).

range of human cardiorespiratory health problems

especially in children, old, and ill people (USEPA,

1996; Wilson and Spengler, 1996; Holgate et al.,

1999; Peters et al., 2000). Such health problems

include cardiac arrhythmias, reducing lung function,

asthma, chronic bronchitis, and increasing respiratory

symptoms, such as sinusitis, sore throat, dry and wet

cough, and hay fever. Recent research suggests that

these problems may be related to specific components

in the particulate matter rather than mass (Ghio and

T. Moreno et al. / Science of the Total Environment 334–335 (2004) 337–346338

Devlin, 2001). However, there is still uncertainty over

the characteristics that make some airborne particles

more bioreactive than others. Because recent research

tends to agree that particle size is the most influential

factor, most of the recent work on this subject is

focused primarily on the finest air particle fraction

(Schwartz and Meas, 2000), although other factors

including chemical composition (Richards, 1997;

Adamson et al., 2000) and particle morphology may

have an important role in the possible lung damage. In

response to the link between high particulate matter

concentrations and health problems, the Expert Panel

on Air Quality Standards (EPAQS) has established a

limit of 50 Ag/m3 of PM10 (all particles passing

through an inlet which allows through 50% of 10-

Am aerodynamic diameter particles) in a 24-h period

for the UK, although this mass limit is commonly

exceeded in urban and industrial areas.

Airborne particles can have an anthropogenic or a

natural origin. Natural particles include for example

background sea salt, mineral dust, and biological

particles (pollen, spores, viruses) as well as transient

loadings from specific pollution events, such as vol-

canic emissions and forest fires. However, a majority

of aerosol particles in urban and industrial areas are a

result of human actions, notably from vehicle and

chimney exhausts, hydrocarbon combustion for heat

and electricity, quarrying, solid waste incineration,

biomass burning, and various industrial processes

such as iron and steel milling and metal smelting.

Sometimes it can be difficult to differentiate between

natural and anthropogenic particles, as is the case for

example with many of the sea salt particles analysed

in the UK. Whereas some of these are naturally blown

in from marine areas, many are from chlorides used to

prevent road icing, and with these particles having

identical composition to those from sea salt making, it

is impossible to recognise their anthropogenic origin.

Airborne particles are generally subdivided into

coarse (10–2.5 Am) and fine ( < 2.5 Am) fractions,

named as PM10–2.5 and PM2.5, respectively. In gen-

eral, the PM2.5 fraction of a sample (all particles

passing through an inlet which allows through 50%

of 2.5-Am aerodynamic diameter particles) can stay in

the atmosphere for a longer period (days or even

weeks) and travel very long distances (ten to

thousands of kilometres). In contrast the PM10–2.5

fraction stays in the atmosphere only for minutes or

hours and is normally deposited very close to its

source area (metres). The concentration of particles

in the air at any given time is also dependent on

weather conditions, notably temperature and rainfall

(with highest concentrations occurring during cold

and dry days) and wind strength and direction.

Both PM10–2.5 and PM2.5 can be emitted directly

into the atmosphere (primary) or formed there as a

result of chemical reactions between other particles

(secondary). Most of the coarse particles are primary

in origin and include biological particles, those pro-

duced by agricultural and construction works, marine

aerosols, and some traffic-related particles (e.g., from

brakes and tyres). The fine fraction is in contrast

mainly dominated by secondary particles like nitrates

and sulphates produced in the atmosphere as a result

of condensation of nitrogen and sulphur oxides. Other

major sources of fine particles are the primary emis-

sions from power industrial plants (fly-ash) and the

combustion of hydrocarbon fuels (particles from ve-

hicle exhausts).

The composition of these airborne particles can be

very variable from one site to another; hence, it can be

reasonably predicted that aerosol samples from differ-

ent locations will interact differently with the lung

environment. Therefore, the chemical and morpholog-

ical characterisation of different air samples is an

essential first step in any investigation of the possible

effects that these particle mixtures can have on human

health. This study compares two such aerosol mix-

tures, one from a city centre and the other from a town

noted for its heavy industry.

2. Methodology

Air samples were collected for a month period with

a high-volume impactor in two populated areas,

Cardiff as the urban location, and Port Talbot as the

industrial site. Both locations, separated by about 30

miles, lie on the south coast of Wales with variable

winds but a prevailing SW wind direction.

The Cardiff site is located just below ground level

in a partially covered (roof but open sides) car parking

area of the Biosciences department, in Cardiff Uni-

versity, less than 500 m from the city centre (Fig. 1A).

The site is within the Civic Centre and is not close to a

major road, although there is light traffic throughout

Fig. 1. Location maps of the selected sites for collection of air particles in (A) Cardiff as an urban site, and (B) Port Talbot as an industrial site

(black star marks the specific collection point; maps provided by http://www.streetmap.co.uk).

T. Moreno et al. / Science of the Total Environment 334–335 (2004) 337–346 339

the day in the adjacent road (Museum Avenue) lying

to the west. Other busier roads include the moderately

busy Park Place (100 m northeast), the busy A470

(250 m southwest) and the main thoroughfare A48

(1600 m north). The site lies around 300 m north from

the sea.

The collecting site in Port Talbot is located in the

grounds of the Groeswen Hospital (Fig. 1B). This

hospital is situated 100 m southwest of the extremely

busy M4 motorway, 100 m north of a moderately busy

road, 800 m northeast of a major steel producing

factory and less than 2500 m away from the sea. The

traffic flow of the surrounding area is estimated to be

50,000 to 55,000 vehicles per day (DETR, 2002).

2.1. Sample collection

A high-volume impactor (HVI) designed by Har-

vard University (Demokritou et al., 2002) and mod-

ified in-house was used to collect the samples.

Mobility of the collector is achieved by transporting

it by car towing a small trailer. It proved necessary to

build an air-cooling soundproofing box to meet all

UK noise regulations, and to protect the equipment in

a metal cage. The collector, nicknamed ‘‘the super-

sucker’’ is capable of collecting exceptionally large

amounts of sample within a foam substrate. The

aerosol sample can then be relatively easily removed

from the substrate with minimum sample alteration.

The HVI runs on a 40-Amp electrical supply, sucking

air at 1100 l/min through the head containing the foam

substrates (Fig. 2A). This head contains three different

cylindrical inlets through which the air passes (Fig.

2B). In the first step, particles bigger than 10 Am, such

as insects, fragments of plants or hairs, impact on a

polyurethane foam (PUF) substrate, with the air then

moving to the next step by passing through a 10-Aminlet with the 10- to 2.5-Am particles (PM10–2.5)

impacting in a second PUF. Finally, in the last step,

the air, which by now is travelling more rapidly,

passes through a 2.5-Am inlet depositing 2.5- to 0.1-

Am particles (PM2.5) in a third PUF. The size distri-

bution of the PM10–2.5 and PM2.5 fractions obtained

has been studied with image analyses, and the results

show only a small number of particles bigger than 10

Am getting into the coarse fraction and no particles

bigger than 2.5 Am in the fine fraction, indicating that

the size-fractionation mechanism designed for the

collector is working efficiently.

The HVI was left running for periods of 43 and 25

days in Cardiff and Port Talbot, respectively. The

Cardiff collection was obtained from January 22 to

March 6, 2001, whereas Port Talbot air samples were

collected from the April 26 to May 25, 2001. PUF

substrates, previously weighed in a five-decimal bal-

ance were changed every 4 to 11 days, with the

longest periods of collection under wet conditions

(when less sample was being collected per day).

These substrates were then stored in sterile petri

dishes and weighed again to determine the mass of

airborne samples collected using the same balance to

check the amount of samples obtained.

Fig. 2. (A) Sketch showing the three impactor levels of the high-volume impactor where airborne particles are collected, with a closeup (B) of

one of the levels.

T. Moreno et al. / Science of the Total Environment 334–335 (2004) 337–346340

The polyurethane foam [H2N–C( = O)O–CH2

CH3] was selected as a substrate because of its

unreactive chemistry and its porous texture, allowing

the collection of higher quantities of particles. Other

filters tried in the HVI during an initial phase of

experimentation at Cardiff include petroleum gel

(vaseline) and polycarbonate, both of which were

destroyed because of the enormous amount of air

passing through, and fiber glass and aluminium disks

that did not collect enough sample, because most of

the particles rebounded from them.

2.2. Sample characterisation

Characterisation of the particles was performed

using scanning electron microscopy. Size and shape

were observed using a Field Emission Scanning

Electron Microscope (XL30-FEG SEM, Philips Elec-

tron Optics, NL). PUF substrates were directly ‘flat-

mounted’ onto aluminium SEM stubs using epoxy-

resin (Araldite) as an adherent between the PUF and

the stub. The samples were then gold/palladium-

coated using a 208HR Sputter Coater (Cressington,

UK) and an MTM20 Thickness Controller (Cressing-

ton). Microscope conditions were accelerating voltage

of 5 kV, working distance of 5 mm, beam spot size of

3 and with the gold foil aperture inserted. Particle

chemical analyses were obtained with a low-vacuum

JEOL5900LV Scanning Electron Microscope (SEM)

via an energy-dispersive X-ray microanalysis system

(EDX). Microscope working distance was 10 mm,

accelerating voltage of 20 kV, and the beam spot size

was 2. Sample preparation was as described for the

FESEM studies, although samples were carbon-coated

(instead of gold/palladium).

3. Results

3.1. Sample collection

A total of 697.05 mg were collected in the Cardiff

location, with an average sample of up to 38 mg per

day in dry conditions (Table 1). The total sample was

fractionated into 539.21 mg of PM2.5 and 157.84 mg

of PM10–2.5. Weather conditions were variable during

the collection period, with the first 28 days being

much wetter than the last 15. Higher amounts of

Table 1

PM collection rates for Port Talbot and Cardiff locations

Period of time PM (weights in mg) PM2.5/ Amount

10–2.5 Am 2.5–0.1 Am PM10– 2.5

ratio

total PM

(mg/day)

Port Talbot

26/04/01–

02/05/01

57.35 61.84 1.08 19.86

02/05/01–

09/05/01

58.42 79.64 1.36 19.72

09/05/01–

14/05/01

51.12 99.85 1.95 30.19

18/05/01–

25/05/01

52.04 93.12 1.79 20.73

Total in

25 days:

553.38 mg

Cardiff

22/01/01–

01/02/01

26.63 75.73 2.84 9.23

01/02/01–

12/02/01

26.00 75.74 2.91 9.25

12/02/01–

19/02/01

41.34 134.53 3.25 25.12

19/02/01–

26/02/01

19.80 78.11 3.94 13.98

26/02/01–

02/03/01

14.49 48.85 3.37 15.83

02/03/01–

06/03/01

29.58 126.25 4.27 38.95

Total in

43 days:

697.05 mg

T. Moreno et al. / Science of the Total Env

particles were collected under cold, dry and lower

humidity conditions, so that the highest values (par-

ticularly of the PM2.5) were collected in the last

sample. The Cardiff sample is clearly dominated by

the finer fraction, with the PM2.5/PM10–2.5 ratio lying

within the range 2.8–4.2.

In Port Talbot, 553.38 mg of total sample (334.45

mg PM2.5 and 218.93 mg PM10–2.5) was collected in

only 25 days (Table 1). Weather conditions in the site

during May were mostly dry with very variable wind

direction. Highest average sample amount per day

was up to 30 mg. The PM2.5/ PM10–2.5 ratio in this

sample was not as wide as in Cardiff, varying from 1

to 2. The apparently higher (or heavier) amount of

PM10–2.5 in the Port Talbot sample is due at least in

part to the large amount of fine ( < 2.5 Am) but dense

Fe spherules impacted on, and becoming trapped

within, the PM10–2.5 foam.

3.2. Sample characterisation

A wide range of particles was observed and iden-

tified in both the Cardiff and Port Talbot samples.

Whereas the PM10–2.5 fraction shows clear individual

particles, with sometimes smaller particles nucleating

on bigger ones, the PM2.5 fraction presents a much

more homogeneous texture, with very commonly a

mass of agglomerated amorphous particles covering

larger ones. A brief description of all particles ana-

lysed is presented below.

Elemental and organic carbon compounds com-

prise all those particles whose EDX analyses showed

only carbon and oxygen and include biological par-

ticles (mostly spherical pollen and spores; Fig. 3A), as

well as those particles produced by the internal

combustion of oil and petrol in vehicles (soot par-

ticles; Fig. 3B). Such combustion produces organic

compounds that condense in the air to form spherules

smaller than 100 nm in diameter (Colvile et al., 2001),

these subsequently aggregate to form larger particles.

Because of their very small size and low density, these

agglomerates can travel in air for long distances, and

they are preferentially collected in the finest, PM2.5

fraction.

Most chlorides analysed are sea salt particles

(NaCl), either directly blown in from the coast or

secondarily sourced from anthropogenic use of salt on

roads. Such particles can travel for many kilometres

so that even primary, marine-derived sea salt can be

found in UK cities well away from the coast. Sea salt

particles are frequently square in shape (Fig. 3C) and

typically lie in the 10- to 2.5-Am range. Irregular

ammonium chloride (NH4Cl) particles are also com-

mon, especially in the PM2.5 fraction.

Iron particles are very common in both industrial

and urban sites. These particles are commonly the

result of the condensation of iron released into the

atmosphere during high-temperature processes, such

as metal smelting industries (e.g., steel factories).

Most of the iron particles identified are perfect

spherules, although some irregular iron particles have

also been analysed. Iron spherules are normally

smaller than 2 Am and form both clusters and

individual particles on the foam surface. Individual

spherules commonly serve as a nucleation point for

even smaller particles (some of these also being iron

spherules; Fig. 3D).

ironment 334–335 (2004) 337–346 341

Fig. 3. FESEM images of (A) organic particles (small white spherules in the centre of the image); (B) soot particles produced by the internal

combustion of diesel and petrol in vehicles; (C) perfect cubes of sea salt; (D) Fe spherule from a steel factory with smaller particles nucleating

on it (foam substrate in the background); (E) several subhedral silicate particles; and (F) euhedral crystals of calcium sulphate (gypsum).

T. Moreno et al. / Science of the Total Environment 334–335 (2004) 337–346342

Nitrogen-bearing compounds are also common

particles in both coarse and fine fractions (being

more abundant in the fine fraction). They form in

the atmosphere by the oxidation of nitrogen oxide.

Very random in shape, these compounds are inter-

preted as mostly ammonium nitrate (NH4NO3) or,

less frequently, sodium nitrate (NaNO3). Silicates

(commonly referred in other works as mineral dust)

can be very abundant in the coarse fraction of

airborne samples (Fig. 3E). In most cases, they

preserve their crystalline shape, making them easily

recognisable in the FESEM images. Silicate particles

include minerals, such as quartz (SiO2), clays as

kaolinite (Al4(Si4O10)(OH)8), illite (KAl4(Si7A-

l1O20)(OH)4) and montmorillonite ((Ca, Na)(Al,Mg,-

Fe ) 4 [ (S i ,A l ) 8 O2 0 ] (OH) 4 . nH2O) , f e l d spa r s

((K,Na)[AlSi3O8]) and plagioclase (Na[AlSi3O8]–

Ca[Al2Si2O8]). These particles normally originate

from the erosion of local geological formations,

quarrying and construction work.

Sulphates are very common particles in UK air,

especially in the PM2.5 fraction. Within the fine

T. Moreno et al. / Science of the Total Environment 334–335 (2004) 337–346 343

fraction, the most abundant sulphate particles are

ammonium sulphate, these being formed as secondary

particles in the atmosphere by the oxidation of SO2.

The coarser fraction is relatively richer in gypsum

(CaSO4. 2H2O) and sea-related sodium and potassium

sulphates. Morphologically, most of the sulphates

analysed are euhedral crystals (Fig. 3F), needles (in

the case of gypsum), or cubic ammonium sulphate

[(NH4)2SO4]. Sulphate particles have been observed

to grow diagenetically within the foam substrate after

impacting on it (process enhanced by the sample

having been collected under humid conditions). This

secondary sulphate growth produces new crystals that

sometimes reach 40 Am in size.

Other particles found in the Port Talbot sample

include Fe–Ca compounds with variable amounts of

sulphur and zinc and commonly irregular Fe–S par-

ticles. These particles are interpreted as artificial

products from the reaction of iron and other elements

from the coking (a potential source of S) and sintering

(a potential source of Ca) plants and steel convertors

in the steelworks.

A total of 949 analyses were obtained from par-

ticles from the Cardiff sample (459 PM10–2.5 and 490

PM2.5), as compared to a total of 1442 from the Port

Talbot sample (768 PM10–2.5 and 674 PM2.5). A

Fig. 4. Cardiff and Port Talbot particle characterisation de

greater number of analyses was obtained from Port

Talbot sample because of its more heterogeneous

composition (Mamane et al., 2000).

3.2.1. Cardiff samples

In the Cardiff sample, both fractions (PM10–2.5 and

PM2.5) show several similarities. Both of them are

clearly influenced by the proximity of the sea (>20%

chlorides), and both include sulphates, nitrates, ele-

mental and organic compounds, silicates, Fe spherules

and carbonates (probably derived from works on

buildings with limestone facade). The main differ-

ences between the coarser and finer fractions are

found in the relative proportions of particle types

and therefore the amount of the various elements

present. Thus, the dominant elements in Cardiff

PM10–2.5 are C, O, Fe, N, S, Si, Cl and Ca in

decreasing order, with particles being dominated by

chlorides (26%), sulphates (23%) and elemental and

organic carbon compounds (20%; Fig. 4A). Other

particles (31%) include nitrates, silicates, Fe spherules

and less than 1% of carbonates. In comparison, the

dominant elements in the Cardiff PM2.5 sample are C,

O, S and N followed by Fe, Cl and Si. This fraction

has a more amorphous texture, with most of the

particles covered with sulphates and wet soot. It is

rived from c. 500 SEM-EDX analyses per sample.

T. Moreno et al. / Science of the Total Environment 334–335 (2004) 337–346344

dominated by soluble particles comprising sulphates

(35%, mainly ammonium sulphate), nitrates (21%)

and chlorides (20%, ammonium chloride predomi-

nantly). The remaining particles include elemental

and organic carbon compounds, silicates, Fe spherules

and again less than 1% carbonates (Fig. 4B).

3.2.2. Port Talbot samples

The Port Talbot sample is characterised by a more

marked contrast between the PM10–2.5 and PM2.5

fractions. Although both of them are overwhelmingly

dominated by mostly Fe spherules (>30%) emitted to

the atmosphere by the steel plant, the relative con-

centrations of other elements are very different. In

the PM10–2.5 fraction, apart from C, Fe and O, the

commonest elements in decreasing order of abun-

dance are Ca, Si and Cl, followed by smaller

amounts of N, S, Mn, K, Al, Na, Mg, Zn, Br and

Pb (again in decreasing order). The coarser fraction

in Port Talbot is dominated by Fe spherules (31%;

Fig. 4C) and presumably deposited in the PM10–2.5

substrate due to their high density. A study on the

size distribution of 600 Fe spherules from Port Talbot

showed that the majority of them are smaller than 3

Am, with only 5% of them being bigger than that

size (Fig. 5). Other particles in the sample include

chlorides (21%, this is a coastal site), silicates,

sulphates, Ca–Fe-rich particles (again linked to the

emissions from the steel plant), elemental and organ-

Fig. 5. Histogram illustrating the size distribution of 600 iron spherules in Po

ic carbon compounds and nitrates. The large majority

of the sulphates are calcium-bearing, with the Ca

probably coming from the use of carbonates in the

steel works.

In the PM2.5 fraction, apart from C, Fe and O, the

dominant element is S followed by much smaller

amounts of (in decreasing order) Ca, N, Zn, K, Si,

Na, Al, Mn and Br. As in the coarse fraction, the

sample is also dominated by Fe spherules (34%; Fig.

4D), with the majority of them contaminated with

sulphur. This contamination in the analyses is proba-

bly related to the nucleation of tiny ( < 2 Am) sulphate

particles on the Fe spherules (Fig. 4D). Unlike the

PM10–2.5 sample, the Port Talbot PM2.5 does not

contain chlorides or Ca–Fe–rich particles. Other

particles include elemental and organic carbon com-

pounds (31%), sulphates (26%, mainly ammonium

sulphate), and smaller proportions of nitrates

(NH4NO3) and silicates. Primary deposited material

is partially obscured by an overgrowth of crystalline

to amorphous aggregates of secondary phases, mostly

sulphates and wet soot as described in Cardiff sam-

ples. The aggregate comprises relatively large euhe-

dral crystals of Ca sulphate (commonly >2.5 Am) that

must have grown in the foam and a much finer mass

of other sulphates (mostly ammonium) which can be

seen to overgrow the gypsum. This secondary over-

growth has also nucleated on the condensed iron

spherules.

rt Talbot airborne samples, with 95% of them being smaller than 3 Am.

T. Moreno et al. / Science of the Total Environment 334–335 (2004) 337–346 345

4. Discussion and conclusions

Very high amounts of airborne particles (up to 38

mg/day) can be now collected in polyurethane foams

using a high-volume impactor. Such amounts of

sample improve our ability to obtain a more repre-

sentative chemical characterisation of airborne par-

ticles from specific sites, to allow close observation of

morphological relationships between different types

of particles, and to conduct larger and more numerous

experiments on the collected aerosol (Greenwell et al.,

2002). The main controls on the amount of sample

collected are proximity to the pollution source, wind

direction (the Port Talbot sample was heaviest under a

prevailing southerly wind; Fig. 1B), and amount of

rainfall (dry periods enhance sample collection).

The two locations selected (Cardiff and Port Tal-

bot) to compare airborne particles in an urban and an

industrial site have revealed very different and char-

acteristic compositions. As an urban site, Cardiff city

has a high traffic volume and although the sampler

was not located adjacent to any of the main city roads,

the sample collected still has a high proportion of

sulphates, nitrates and elemental and organic carbon

compounds (including soot) that are mainly traffic

related. Cardiff is also a coastal city and this is

reflected in the abundance of chlorides (NaCl), espe-

cially in the coarse fraction (Fig. 4A). The iron

spherules found in this sample are likely to be related

to the nearby Port Talbot steel plant (f 50 km),

because such particles, produced as condensates from

the plant emissions, can stay in the atmosphere for

several days and travel for hundreds of kilometres.

However, the relatively minor amounts ( < 6%) of Fe

spherules on the Cardiff sample are much less than

those collected from the Port Talbot site. These Fe

condensates are the dominant type of particle in the

Port Talbot air, suggesting that local inhabitants are

breathing very high amounts (>30% of total number

of air particles) of < 3-Am iron spherules, as well as

more complex Ca–Fe–(S)-bearing particles. Port Tal-

bot is also bounded to the north by the busy M4 so

that a traffic-related chemical signature is recorded by

the presence of elemental and organic carbon com-

pounds, sulphates and nitrates (Fig. 4D). Finally, as

with Cardiff, the coastal location of Port Talbot is

reflected by the presence of more than 20% NaCl

particles in its coarse fraction.

The data described show an example of how air

particles from an industrial site can be discriminated

from a UK city centre site, using a high-volume

impact collector. Furthermore, the collecting method

allows a well-defined fractionation between the par-

ticles impacting in the coarser fraction (10–2.5 Am)

from those in the finer one ( < 2.5 Am). Thus, the

PM10–2.5 fraction of both Cardiff and Port Talbot is

clearly richer in chloride (NaCl), silicates and Ca-

bearing sulphate (gypsum), whereas the PM2.5 frac-

tion tends to be dominated by ammonium sulphate

and elemental and organic carbon compounds, and

the chloride (if present in the fine fraction) is

interpreted as predominantly ammonium chloride.

The size fractionation method can separate out pri-

mary and secondary aerosols. Air particles containing

ammonium are the result of chemical reactions in the

atmosphere (secondary particles) that normally pro-

duce particles smaller than 2.5 Am. This is consid-

ered to be the reason why ammonium sulphate and

ammonium chloride particles are preferentially pres-

ent in the PM2.5 fraction, in contrast with the more

abundant calcium sulphate or sodium chloride (pri-

mary particles) in the PM10–2.5 fraction. In the case

of the high-density particles related to the steel plant

emissions (Fe spherules), although mostly small

enough (most of them < 3 Am) to be collected in

the finer PM2.5 fraction, they are the most abundant

particle in both coarse and fine fractions in Port

Talbot. The most plausible reason for this lack of

efficient fractionation lies in the high density of such

particles, impacting at high speed onto the first foam

substrate and becoming trapped rather than rebound-

ing and moving onto the next step in the collecting

process. Alternatively, many of the coarse particles

are composites of smaller particles, which may result

from agglomeration/aggregation between different

particle types.

The much greater collection capacity of a high-

volume impactor, embedding billions of particles in a

foam substrate, allows a more effective and realistic

characterisation of air samples using this technique.

This method of aerosol collection, combined with

characterisation using high quality SEM-EDX, is

enabling toxicological research that was previously

impractical or impossible. The characterised samples

are currently being used by three laboratories in the

UK for a spectrum of bioreactivity studies: King’s

T. Moreno et al. / Science of the Total Environment 334–335 (2004) 337–346346

College (antioxidant effects) and NHLI (human lung

cell studies) in London and Cardiff Biosciences (ox-

idative potential, toxicogenomic studies).

Acknowledgements

We are very thankful to Martin Hooper (Neath Port

Talbot County Borough Council) and the Groeswen

Hospital in Port Talbot for their help while locating

and collecting with the high-volume sampler. We also

thank ‘‘streetmap’’ for allowing us to include their

maps in figure 1 for research purposes. This research

project was funded by MRC (G9901020).

References

Adamson IYR, Prieditis H, Hedgecock C, Vincent R. Zinc is the

toxic factor in the lung response to an atmospheric particulate

sample. Toxicol Appl Pharmacol 2000;166:111–9.

Colvile RN, Hutchinson EJ, Mindell JS, Warren RF. The transport

sector as a source of air pollution. Atmos Environ 2001;35:

1537–65.

Demokritou P, Kavouras IG, Ferguson ST, Koutrakis P. Develop-

ment of a high volume cascade impactor for toxicological and

chemical characterization studies. Aerosol Sci Technol 2002;

36:925–33.

DETR (Department of Environment Transport and the Regions).

Automatic urban and rural network site information archive;

2002. http://www.aeat.co.uk/netcen/airqual.

Ghio A, Devlin R. Inflammatory lung injury after bronchial instil-

lation of air pollution particles. Am J Respir Crit Care Med

2001;164:704–8.

Greenwell L, Moreno T, Jones TP, Richards RJ. Particle-induced

oxidative damage is ameliorated by pulmonary antioxidants

(particle-induced oxidative damage). Free Radic Biol Med

2002;32:898–905.

Holgate S, Samet J, Koren H, Maynard R. Air pollution and health.

San Diego: Academic Press; 1999.

Mamane Y, Willis RD, Conner TL. Evaluation of computer-con-

trolled scanning electron microscopy applied to an ambient ur-

ban aerosol sample. Aerosol Sci Technol 2000;34:97–107.

Peters A, Liu E, Verrier R, Schwartz J, Gold D, Mittleman M, et al.

Air pollution and incidence of cardiac arrhythmia. Epidemiolo-

gy 2000;11(1):11–7.

Richards RJ. Small particles, big problems. Biologist 1997;44

(1):249–51.

Schwartz J, Meas LM. Fine particles are more strongly associated

than coarse particles with acute respiratory health effects in

schoolchildren. Epidemiology 2000;11:6–10.

U.S. Environmental Protection Agency (USEPA). Office of Air

Quality Planning and Standards, Review of national ambient

air quality standards for particulate matter: policy assessment

of scientific and technical information, Report no. EPA-452/

R-96-013. Washington, DC: USEPA; 1996. pp. V-2–V-24,

V-27–V-28, V-71.

Wilson R, Spengler J. Particles in our air: concentrations and health

effects. Boston: Harvard University Press; 1996.