Characterization of indoor air quality in primary schools in Antwerp, Belgium

Abstract

The indoor air quality of 27 primary schools located in the city centre and suburbs of Antwerp,
Belgium, was assessed. The primary aim was to obtain correlations between the various
pollutant levels. Indoor:outdoor ratios and the building and classroom characteristics of each
school were investigated. This paper presents results on indoor and local outdoor PM2.5 mass
concentrations, its elemental composition in terms of K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, Br, Pb,
Al, Si, S, and Cl, and its black smoke content. In addition, indoor and local outdoor levels of the
gases NO2, SO 2,O 3, and BTEX (benzene, toluene, ethyl benzene, and xylene isomers) were
determined. Black smoke, NO2, SO 2 and O3, occurred at indoor:outdoor ratios below unity,
indicating their significant outdoor sources. No linear correlation was established between
indoor and outdoor levels for PM2.5 mass concentrations and BTEX; their indoor:outdoor ratios
exceeded unity except for benzene. Classroom PM2.5 occurred with a different elemental
composition than local outdoor PM2.5. The re-suspension of dust because of room occupation
is probably the main contributor for the I/O ratios higher than 1 reported for elements typically
constituting dust particles. Finally, increased benzene concentrations were reported for
classrooms located at the lower levels.

Practical Implications The elevated indoor PM2.5, and BTEX concentrations in primary school
classrooms, exceeding the ambient concentrations, raise concerns about possible adverse
health effects on susceptible children. This is aggravated by the presence of carpets and in the
case of classrooms at lower levels. Analysis of PM2.5’s elemental composition indicated a
considerable contribution of soil dust to indoor PM2.5 mass. In order to set adequate threshold
values and guidelines, detailed information on the health impact of specific PM2.5 composites
is needed. The results suggest that local outdoor air concentrations measurements do not
provide an accurate estimation of childrens personal exposures to the identified air pollutants
inside classrooms.

Introduction

Each indoor micro-environment is uniquely characterized, which is determined by the local

outdoor air, specific building characteristics and indoor activities (Stranger et al., 2007).
Consequently, each individuals personal exposure will be determined by a combination of the
local outdoor pollutant levels and the different indoor micro-environments to which he/she is
exposed to, and his/her residence time in each. Since Belgian primary school children spend on
average 7–11 h per weekday at school, mostly indoors, the indoor and outdoor air quality in
school buildings are expected to be key role players in the assessment of the effects of their
personal exposure to air pollution, especially since they are particularly vulnerable to pollutants
(Dockery et al., 1989; Ha et al., 2003; Pope and Dockery, 1992). Tranter (2005) emphasized the
clinical perspective of schools, by stating that the school environment in industrialized nations
has a lower potential for exposure than the home environment, except for allergic individuals.
From a public health perspective the author suggests that school environments should be
targeted for both primary and secondary prevention. Furthermore, Mendell and Heath (2005)
concluded in their critical review that evidence suggests a poor indoor environmental quality
(IEQ) in schools because of the influence that indoor pollutants and thermal conditions have on
the performance and attendance of students. The importance of classroom ventilation to
achieve a healthy indoor environment for students was highlighted by Daisey et al. (2003).
Several international case studies have been performed in order to establish the indoor air
quality in school environments. The indoor air quality is often discussed in terms of
Indoor/Outdoor ratios (I/O ratios) of which analysis are mostly performed in limited sets of
classrooms. In Hong Kong, carbon dioxide (CO2), sulphur dioxide (SO2), nitrogen monoxide (NO,)
nitrogen dioxide (NO2), PM10, formaldehyde (HCHO) and total bacteria count were monitored
in five schools (Lee and Chang, 1999, 2000). Particulate matter (PM) inside classrooms was
identified as being the major problem, since the maximum indoor PM10 levels exceeded 1000
ug/m3. In Germany different fractions of indoor PM were studied for their mass concentration
and particle number in 64 classrooms. In an explorative analysis, the authors identified that
increased PM concentrations correlated significantly with increased CO2 concentrations and
classrooms at lower levels (Fromme et al., 2007). Blondeau et al. (2005) studied the relationship
between indoor and outdoor ozone (O3), NO, NO2, and PM in eight French schools and identified
a strong correlation between indoor O3 and the air tightness of a building. Furthermore,
resuspension of previously deposited particles was found to have a considerable influence on
indoor PM. Bertoni et al. (2002) established the indoor and outdoor benzene, toluene, ethyl
benzene, and xylene isomers (BTEX) levels in 17 classrooms in Italy and identified increased
concentrations of benzene inside classrooms; the indoor and local outdoor toluene and xylene
levels were low and comparable. Other studies focussed on the relation between pollution and
health outcomes. Meininghaus et al. (2003) organized a study aimed at identifying compounds
(aldehydes, organic acids, volatile organic compounds, SO2 and NH3), which could be responsible
for the reported adverse health effects, but no cause could be identified. Janssen et al. (2003)
identified an association between truck traffic and its related air pollutants with chronic
respiratory symptoms in Dutch schoolchildren living close to motorways. Also in the Netherlands
it was found in a study of 54 children from four different schools that the personal exposure of
children attending schools located within 100 m from a major road had a 30% increase of soot
exposure, and a 37% increase of NOx exposure. The differences appeared to be smaller and
insignificant for PM2.5 and NO2 (Van Roosbroeck et al., 2007).

In spite of the various studies performed worldwide to assess the indoor air quality in school
environments, the characterization of PM2.5 in indoor air remained limited to the
determination of mass concentrations and particle counts. Up to now no information was
available in open literature on the elemental composition of PM2.5 in classrooms and its relation
to the ambient air, gaseous indoor pollutants, and building and classroom characteristics. The
main goal of this study was to assess the indoor air quality in 27 Antwerp primary schools, by
determining indoor and outdoor PM2.5 mass concentrations and its elemental composition and
the gaseous compounds NO2, SO2, O3 and BTEX in parallel measurements. These pollutants were
correlated by means of statistical methods to the building and classroom characteristics in an
attempt to explain the concentration profiles and identify possible sources aiming at an
improvement of the indoor environment of classrooms.

Experimental procedure

Sampling

Indoor and local outdoor particulate matter and gaseous samples were collected from 27
primary schools that were a subset of the participants of the ISAAC III project (International
Survey of Asthma and Allergy in Childhood, in 56 countries; performed in 2001 in Antwerp
amongst school children aged from 6 to 7 and 13 to 14 years) in order to relate health outcomes
to the indoor air quality at a later stage. Criteria considered were as follows: location, respective
distances from each other, and proximity to high traffic emissions. In the urban area of Antwerp,
which is defined as the region enclosed by the Antwerp Ring and the river Scheldt, 15 schools
were selected, equally distributed over the district. The area (southern suburban) is situated 20
km southwards from Antwerp, and includes the districts Niel, Schelle, Aartselaar, and Boom. In
this area, 12 schools were selected. In both areas, schools at background as well as hotspot
locations were chosen. For the purpose of this study, a hotspot location was defined as a place
situated in close proximity to a main road, characterized by a higher traffic density than the
other, background sites. Indoor samples were taken in the classroom of the second school year
level in each school, in order to assess the exposure of the children from whom health
parameters were collected during ISAAC III. The 27 schools were assigned with the symbols S1
to S27 with both S27 and S11 positioned the closest to traffic emissions. Two sampling
campaigns (SP1 and SP2) were conducted (in December 2002 and June 2003), being
representative for winter and early summertime respectively. In each sampling campaign, air
samples were collected over five consecutive days in every school. The urban area was sampled
in the first half of the campaigns and the suburban campaign in the second halves. The study
focussed on the fine fraction PM2.5, since it is classified as the (respirable fraction) of PM (ISO
7708 – 1995) and hence can penetrate the respiratory tract. Parallel indoor and outdoor PM2.5
samples were collected in each school using Harvard-type impactors (MS&T Area Sampler, Air
Diagnostics and Engineering) with sharp PM2.5 cut-off points, in a combination with quasi
noiseless oil free pumping units (Air Diagnostics and Engineering, model SP-280E). The aerosol
samples were collected on pre-weighed 37-mm Teflon membrane filters (Anderson Teflon
Membrane filters, 2 lm pore-size), placed on a support ring, supported by a drain disk (Polyolefin
drain disks, Whatman International) to maintain integrity. The samplers operated at a flow rate
of 10 L/min. With the aid of timers, 12-h daytime samples (08:00 a.m. up to 08:00 p.m.) were
collected from Monday to Friday, since a pilot study in classrooms showed that 12-hsampled
PM2.5 concentrations exceeded those of 24-h by 40% and the 12-h results would therefore be
more representative of the real-time exposure of the school children. Filters were replaced in
the middle of the week to avoid overloading and consequently around 18 m3 of air was sampled
per filter. After collecting the PM samples, all filters were stored at )10
C. The gaseous compounds NO2, SO 2, BTEX and O3 were
sampled on three types of cartridges in parallel and continuously (24 h for five consecutive days)
for five schooldays, by means of
radiello

(radiello, Fondazione Salvatore Maugeri) radial passive samplers. Radiello diffusion tubes have
been studied, validated and used by various researchers (Pennequin-Cardinal et al., 2005a,b;
Saarela et al., 2003). The detection limits of these samples were 0.9 ug/m3, 0.4 ug/m3, 0.75
ug/m3and below 0.1 ug/m3for NO2, SO 2,O 3, and BTEX respectively. Safe and childproof
sampling sites were ensured and complied with the rules as prescribed by ISO (Scheider, 1999).
Indoor samples were collected at a height of about 1–1.5 m above the floor, which is the
breathing zone inside the classroom. The selected place was not allowed to be closer than 1 m
to a wall, a door or an active heating system. Furthermore, the indoor sampling site was selected
as far away as possible from the blackboard, taking into account a minimum distance of 3 m.
Outdoor sampling sites were selected at places which provided electricity and a tamper-free
environment. Outdoor samples were collected at heights of 1.5–2 m above the ground; at a
minimum distance of 1 m from the closest building. Class titulars filled out questionnaires
concerning the school building characteristics and the indoor characteristics of the classrooms.
Ventilation habits (reported as the number of times that a window was opened to ventilate the
classroom, during SP2 supplemented with the time span of the ventilation), the number of pupils
attending the lessons, the interior composition of the classroom and the position of the
classroom towards the closest street were assessed and used in the interpretation of the data.
Each classroom was equipped with a blackboard and chalk; none of the rooms was air
conditioned. An overview of these data is shown in Tables 1 and 2.

Analysis

All parameters analyzed were preceded by validation and optimization of sampling and
measurement techniques as a part of a quality control program. PM mass concentrations were
determined gravimetrically. In addition, the black smoke (BS) concentration was determined to
estimate the elemental carbon content as a marker for traffic (diesel fuel sources), by measuring
particle reflectance. Subsequently, the aerosol filter samples were analyzed for their elemental
content (S, Cl, K, Ca, Ti, V, Mn, Fe, Ni, Zn, and Pb) by means of energy dispersive X-ray
fluorescence (ED-XRF) and for its water-soluble content (Cl), NO2 ), NO3 ), SO42), Na+, NH 4 +, Ca2+
and Mg2+ by means of ion chromatography (IC). The black smoke determinations based on
reflectometry, must be regarded as relative values only, because although the Black Smoke
Method involves a transformation from reflectance units into mass concentrations (BS [ug/m3]
=3.663 + 987a; whereas a = absorption coefficient Edwards et al., 1983), these are considered
as unreliable when considered as absolute values (Brunekreef et al., 2005). The aqueous extracts
from the passive samplers were analyzed for the collected NO2, SO2, and O3 by means of IC, UV-
VIS spectrophotometry and gas chromatography – mass spectrometry (GC-MS). BTEX was
recovered from the passive sampler by CS2 desorption and analysis was performed by a Varian
3800 GC coupled to a Varian Saturn 2000 ion trap MS (Varian, Walnut Creek, CA, USA), equipped
with a AT-624 column (Alltech Associates Inc., Lokeren, Belgium), using 2-fluorotoluene as
internal standard.

Results and discussion Particulate mass concentrations

Table 3 summarizes the local outdoor and indoor PM2.5 collected during the two sampling
campaigns (SP1 and SP2, which took place in December 2002 and June 2003). All the average
local outdoor PM2.5 mass concentrations exceeded the proposed EU 24-h limit value of 35
ug/m3) (II Proposition Paper, EU), except for the second sampling period in the southern
suburbs. Furthermore, the maximum values in the city centre exceeded the EU limit by four
times. From the data in Table 3 it is clear that the average indoor PM2.5 mass concentrations
were in general higher than the local outdoor ones for both sampling periods and areas
investigated. The average indoor PM2.5 levels ranged from 54 to 72 ug/m3 and exceeded the
Norwegian indoor guideline value of 20 ug/m3 averaged over 24 h, the Belgian indoor guideline
value of 15 ug/m3, averaged over 1 year, and the proposed EU annual ambient guideline value
of 35 ug/m3 as described by Stranger et al.(2007). No linear correlation between indoor and
outdoor levels could be found. Figure 1 represents the spatial distribution, with values, of the
respective indoor:outdoor relationships (I/O ratios). The southern suburban areas were
dominated by I/O ratios that were on average 80% higher than in the urban area and in certain
schools I/O ratios reached values up to 8.8. This is a strong indication of the presence of a
significant indoor source, however, it also has to be kept in mind that the local outdoor PM2.5
mass concentrations for SP2 were much lower than for SP1 in the south and a consequent
increase in the ratio is to be expected. Since the absolute mass concentrations did not show an
obvious correlation with any of the building and classroom characteristics, the re-suspension of
dust was suspected to cause the increased indoor PM2.5 levels, which is in agreement with the
work carried out by Blondeau et al. (2005). After investigating the I/O ratios and time-invariable
building characteristics (as listed in Table 1), a significant difference (P < 0.01) between I/O ratios
of classrooms with and without the presence of carpets was identified (Student-t test: I/O ratios
with carpets were on average 2.53 and 2.72 for SP1 and SP2, respectively and with-out 1.05 and
1.00, respectively). These findings indicate that the re-suspension of dust in the schools
investigated is primarily an issue of concern in classrooms with carpets, although resuspension
of dust from other types of floor covering is not ruled out.

Particulate elemental concentrations

The absolute results for the elemental composition of the indoor and outdoor PM2.5 as
determined by ED-XRF analysis during both sampling periods are presented in Table 4. From the
data in Table 4 it is seen that the elements S, Si, Fe, Ca, and Al had the highest contributions to
local outdoor PM and Ca, S, Si, and Al to indoor PM in schools. The heavy metals such as Cu, Pb,
Zn, Ni, Cr, and V occurred at minor concentrations, both indoors and outdoors and none
exceeded the ambient and indoor guideline values (Stranger et al., 2007). None, or very weak
correlations were found between the indoor and outdoor elemental composition of the
particles. The potential health implications of this important finding need further investigation.
An overview of the average I/O ratios per element, averaged over the 2 selected areas (i.e.,
urban and suburban) is displayed in Figure 2. Five samples were excluded as outliers (applying
the Q-test for the identification of outliers). According to Figure 2, the average I/O ratios could
be grouped in two; those indicative of the absence of a significant indoor source (V, Ni, Zn, Pb,
Br, and Mn, with I/O < 1 = Group 1) and those indicative of the presence of an indoor source (Cl,
Ca, Al, Si, K, Ti, and Fe, with I/O >1 =Group 2). Pearson correlations for the I/O ratios for the
Group 1 and 2 elements ranged from 0.748 to 0.848 (99% significance level) and 0.725 to 0.908
(99% significance level), respectively. The use of Principal Cluster Analysis (PCA - SIMCA 3.01
software package), in addition to the Pearson correlations gave conclusive evidence that the
increased indoor PM2.5 mass concentrations were mainly particles of which the elemental
concentrations were representative of dust. Since Cl is known to be an important composite of
detergents, cleaning activities inside the classroom might be an indoor source of this compound.
Ca occurred at significantly higher levels (Independent Samples student t-test, 95% confidence
limit) than the other natural elements (K, Ti, Fe, Si, and Al) during SP1 in both areas and during
SP2 in the southern suburbs (except towards Al in the latter case). This lead to the idea that Ca
could be attributed to a real indoor source, probably originating from the chalk (mainly CaSO4)
used on the blackboards and/or the gypsum walls and plasters used as construction materials.
Although 66% of the schools showed a linear correlation (R = 0.599 up to 0.987 at 95 and 99%
significance level) for the elemental PM2.5 I/O ratios with-in a sampling period, no correlation
could be established between the two sampling periods. This is an indication of the considerable
influence of time-dependant indoor circumstances of the classrooms, like ventilation habits,
which was unfortunately not quantified satisfactorily during the investigation. None of these
characteristics showed, however, a significant correlation, when tested.

Particulate black smoke concentrations

A summary of the results is presented in Table 5. Indoor and local outdoor BS concentrations
were comparable during SP1 in both areas, but during SP2 considerably higher BS levels were
registered in the urban area compared to the southern suburbs.

Since BS is known to be a proxy for elemental carbon, the BS enrichment of PM2.5 can be applied
as a marker for influences of traffic. In order to estimate to which extent the indoor PM2.5 was
BS-enriched, the ratios between BS and PM2.5 mass concentrations were determined in each
classroom sample. According to the results presented in Table 5, indoor and local outdoor PM2.5
in the city centre and the southern suburbs were similarly BS-enriched within and between both
sampling periods. Furthermore, local outdoor BS concentrations were highly significantly (P <
0.001) correlated with the local outdoor PM2.5 mass concentrations during the two sampling
periods (R = 0.577 during SP1 and R = 0.829 during SP2). This linear correlation was neither
completely dependent on seasonal variations nor on the attributes of the two different
geographical locations investigated. This also explains the insignificant variation of the BS. On
the other hand, no linear correlation could be established between the indoor BS concentrations
and the indoor PM2.5 mass concentrations during the two sampling periods, indicating the
insignificant contribution of indoor BS to the PM2.5 inside classrooms. In contrast to the findings
of Fromme et al. (2007), no significant difference was found between BS/PM2.5 present in
classrooms on different levels in the school building. The values of the I/O ratios indicate the
absence of significantly contributing indoor sources, in contrast to the PM2.5. No linear
correlation could be established between the I/O ratios of BS and PM2.5, contradicting the
findings of Brunekreef et al. (2005) in a small population of elderly cardiovascular patients.

Gaseous concentrations NO2 – indoor and outdoor. Table 6 summarizes the indoor and outdoor
NO2 levels for the studied population.

It is evident that all the local outdoor concentrations exceeded on average the current EU annual
guideline value of 40 ug/m3, with up to three times for registered maxima. The increased values
for SP2 were related to elevated O3 levels in the warmer season, promoting increased NO to NO2
conversion and with the consequent O3 formation in the presence of VOCs and sunlight. The
strong positive correlation between the paired indoor and outdoor NO2 levels (Pearson
correlation coefficient of 0.561; 99% confidence limit) supported the assumption that indoor
NO2 mainly originated from outdoor sources.

SO2 – indoor and outdoor. An overview of the average SO2 levels recorded is presented in Table
6. Both indoor and local outdoor levels were on average very low, did not exceed indoor or
outdoor guideline values and were comparable in both areas, indicating the absence of any
contribution because of different geographical positions. All indoor concentrations were on
average 70% lower than the corresponding outdoor levels, with a very low average I/O ratio of
0.3 ± 0.1, confirming the outdoor origins of indoor SO2. For SP1 this assumption was supported
by the positive correlation between the indoor and outdoor levels of SO2 (Pearson correlation
coefficient of 0.529; 99% significance level). However, because of the low concentration levels
during SP2, no correlation could be identified in this sampling period.

O3 – indoor and outdoor. The average indoor and local outdoor O3 levels for SP1 and SP2 are
presented in Table 6. The seasonal dependence of local outdoor O3 is reflected by the
considerably higher values recorded for SP2. As expected, the indoor O3 in this data set occurred
at very low concentrations, often below the detection limit of the
radiello

diffusion tubes (0.75 ug/m3. No correlation between the indoor and the outdoor O3 levels could
be established, despite the absence of obvious indoor sources, like photocopying machines, and
could be ascribed to the very low concentrations registered.

BTEX – indoor and outdoor. An overview of the average results is reported in Table 7. The annual
ambient EU limit value of 5 ug/m3for benzene was almost equalled at schools S2 and S5, but in
general, measured benzene concentrations were well below this threshold value. The local
outdoor concentrations of all BTEX compounds strongly correlated with each other (R = 0.612–
0.857 at the 99% significance level), which supports the assumption that the outdoor BTEX
originates from one common source, namely traffic emissions, of which the higher urban
concentrations (high traffic density) are proof. Indoor benzene appeared to be the only BTEX
component, which occurred strongly linearly correlated with the local outdoor concentrations
(R = 0.814, 99% significance level). Furthermore, depending on the floor level of the classroom,
a significant difference between indoor benzene levels was established (applying the
Independent Samples Student t-test), characterized by average concentrations of 1.08 ± 0.9
ug/m3and 0.45 ± 0.30 ug/m3, on the ground floor and higher floor levels respectively (P < 0.005).
No linear correlations could be established between the indoor and outdoor concentrations of
toluene, ethylbenzene and the xylene isomers. Furthermore, the indoor microenvironment
indicated levels of ethylbenzene and the three xylene isomers that were strongly linearly
correlated to each other (R = 0.851–0.966, 99% significance
level).Indoortoluene,however,showednoor onlypoor correlations to the other indoor BTEX
compounds. In contrast to the lower indoor concentrations of the NO2, SO 2 and O3, higher
indoor BTEX levels were reported in comparison to the outdoor levels. I/O ratios close to unity
confirm the absence of any significant indoor benzene source, in agreement with the correlation
previously determined. Average I/O ratios for toluene, ethylbenzene, and the xylene isomers
and the absence of any correlation between the indoor and the outdoor levels jointly indicate
that indoor sources of these compounds were present. The suburban area was characterized by
higher average I/O levels than the urban area, however, only for toluene a significant difference
was identified (applying the Independent Samples Student t-test), with average values of 2.5 ±
0.7 and 1.2 ± 0.2 in the southern suburbs and city centre, respectively. The most elevated I/O
ratios were identified in the southern suburbs where low local outdoor BTEX levels were
recorded. In the urban area the elevated local outdoor concentrations contributed
proportionally more to the I/O ratio and dominated the effect of indoor sources and sinks. In
Hong Kong Guo et al. (2003) identified I/O ratios equalizing unity for all BTEX compounds in a
case study of schools. However, other studies contradict these findings. For instance, Bertoni et
al. (2002) identified elevated benzene concentrations inside Italian classrooms and comparable
concentrations for the rest of the BTEX components.

Conclusions

This study assessed selected aspects of the indoor air quality in relation to the local ambient
(outdoor) air of 27 primary schools in the urban and suburban areas of Antwerp, Flanders,
Belgium, with the specific aim to investigate the real-time daily exposure of 7–8 year old school
children. For this purpose a set of particulate and gaseous air pollutants was monitored inside a
classroom and on the outside playground of each school. Attention was paid to potential health
triggers such as NO2, SO2, O3 and BTEX and respirable particulate matter (PM2.5). Only in the
cases of BS, NO2, SO2, and O3, correlations between indoor and outdoor concentrations could
be established. Since those compounds occurred at I/O ratios below unity, no significant indoor
sources of these pollutants were identified or suspected. For this reason the seasonal influences
on ambient NO2 and O3 could be registered inside the classrooms as well, causing a slight
increase of indoor levels during early summertime. However, because of their different
deposition velocities and decomposition rates in indoor air, the respective I/O ratios of the gases
varied considerably, decreasing in the sequence NO2, SO2, and O3. Concerning PM2.5 mass
concentrations and BTEX, no linear correlation was established between the indoor and the
outdoor levels, except for benzene. I/O ratios exceeding unity in the order toluene > ethyl
benzene > xylene were found for these organic components. Classroom PM2.5 occurred at
higher mass concentrations and with a different elemental composition than local outdoor
PM2.5. The elemental concentrations with I/O ratios higher than 1 were reported for those
typically constituting dust particles and are probably because of the re-suspension of dust as a
consequence of the room occupation. Despite a thorough investigation into possible
correlations between the building and indoor characteristics and pollutant concentrations only
two showed a significant dependence. The indoor benzene concentrations decreased with an
increase classroom level and an increased I/O ratio of PM2.5 mass concentrations were
observed with the presence of a carpet. From the results it was evident that the presence of
carpets, causing a re-suspension of dust particles and classrooms at lower levels with an
increased benzene concentration are a potential concern. The use of local outdoor air pollutant
concentrations as estimators of the personal exposures of the children inside classrooms would
then lead to an overestimation of the exposures to benzene, NO2, SO 2 and O3 and an
underestimation of the exposure to PM2.5, toluene, ethyl benzene and the xylene isomers.
Consequently, local outdoor air concentrations measurements do not provide an accurate
estimation of the childrens personal exposures to the identified air pollutants. Specifically, the
difference in the elemental composition of PM2.5 indoors is therefore a real concern in this
regard and warrants further investigation.

Table 1 Overview of the school building characteristics collected by means of questionnaires

Table 2 Overview of the classroom characteristics collected by means of questionnaires

Table 3 Overview of the PM2.5 mass concentration values and their I/O ratios (composite of a
week of 12-h daytime samples) for both sampling periods [December 2002 (SP1) and June 2003
(SP2)] for 27 study schools in Antwerp, Flanders, Belgium.

Fig. 1 Spatial distribution of the indoor/outdoor (I/O) ratios of the PM2.5 mass concentrations
in the city centre (upper part of the figure) and the southern suburbs (average SP1 and SP2) for
27 study schools in Antwerp, Flanders, Belgium
Table 4 Elemental concentrations (ng m)3) of PM2.5 in 27 classrooms environments in Antwerp
and its suburbs during SP1 and SP2
Fig. 2 Average elemental I/O ratios (composite of a week of 12-h daytime samples) for both
sampling periods [December 2002 (SP1) and June 2003 (SP2)] for 27 study schools in Antwerp,
Flanders, Belgium

Table 5 Overview of the local outdoor and indoor 12 h PM2.5 BS (a.u.) and the average BS I/O
in classroom environments in the urban and suburban area for both sampling periods
[December 2002 (SP1) and June 2003 (SP2)] for 27 study schools in Antwerp, Flanders, Belgium
Table 6 Overview of the average indoor and outdoor NO2 SO2 and O3 levels (lgm )3) for both
sampling periods [December 2002 (SP1) and June 2003 (SP2)] for 27 study schools in Antwerp,
Flanders, Belgium

Table 7 Overview of the average indoor and outdoor BTEX levels (lgm )3) for both sampling
periods [December 2002 (SP1) and June 2003 (SP2)] for 27 study schools in Antwerp, Flanders,
Belgium