Comfort Parameters and Particulate Matter (PM10 and PM2.5) in School Classrooms and Outdoor Air

In January 2012, one kindergarten and eight elementary school classrooms were monitored. The campaign included simultaneous measurements, indoors and outdoors, of comfort parameters, CO, CO2 and particles. Automatic monitors using a light scattering technique were employed to measure PM10 continuously. During occupied periods, low volume samplers were used to daily collect PM2.5 samples, which were subsequently analysed for carbonates, organic carbon (OC), elemental carbon (EC) and water soluble inorganic ions. With regard to comfort, the schools did not meet the recommended levels in many rooms. Indoor-outdoor CO2 ratios between 3 and 12, and indoor levels much higher than 1000 ppm during the occupied periods, indicate the highly inadequate ventilation in these locations. The results clearly demonstrate that there is a high level of exposure to particulate matter in these schools. The continuous measurements of PM10 suggest that the physical activity of pupils, which is assumed to be more marked in younger children, contributes to a constant process of resuspension of sedimented particles. In addition, peak PM10 concentrations coincident with cleaning activities suggest the need to change certain practices to improve cleanliness. Around 40% of the PM2.5 mass is composed of carbonaceous matter, with 4–5 times higher OC mass fractions than EC. It was observed that both OC and EC were significantly influenced by indoor sources. Water-soluble inorganic ions represented around 10–20% of the PM2.5 mass measured in classrooms. Excluding calcium, in general the ionic species were present at indoor-outdoor ratios of less than 1, suggesting the major origin in the outdoor air.


INTRODUCTION
Studies carried out by the USA Environmental Protection Agency (EPA) of human exposure to air pollutants indicate that indoor air levels of many pollutants may be 2-5 times, and occasionally more than 100 times, higher than outdoor levels.These levels of indoor air pollutants are of particular concern because it is estimated that most people, including children, spend as much as 90% of their time indoors (Bronsema et al., 2004).Over the past several decades, the exposure to indoor air pollutants is believed to have increased due to a variety of factors, including the construction of more tightly sealed buildings, reduced ventilation rates to save energy, the use of synthetic building materials and furnishings, and the use of chemically formulated personal care products, household cleaners, etc.
Unlike other buildings, managing schools involves not only maintenance related concerns, but also child safety issues.Many aspects are unique to schools (EPA, 2005): (i) classrooms have about four times as many occupants as office buildings for the same floor area; (ii) in general, maintenance suffers from financial cuts; (iii) the presence of a wide variety of emission sources, including art and science supplies, and gyms; (iv) highly variable heating and ventilation systems; (v) sometimes portable classrooms or buildings that were not originally designed to meet the educational requirements are used.
Failure to prevent or respond promptly to indoor air quality (IAQ) problems can increase short-and long-term health outcomes (Norbäck et al., 2000;Simoni et al., 2006;Tillet, 2010;Simoni et al., 2010), impact both teacher and student's attendance, comfort and performance (Gilliland et al., 2001;Shaughnessy et al., 2006;Mohai et al., 2011), and accelerate the deterioration of school buildings and equipments (Lohbeck, 2008).Among the indoor air pollutants, there is an increasing interest in the measurement of particulate matter concentrations.Aerosol exposure via the inhalation route is of prime importance, representing a major potential source of hazard for human health of a greater or lesser severity, depending on the airborne concentrations, duration of exposure and chemical composition.Therefore, it is of utmost significance to measure particle concentrations in the size fractions known to affect the human health (Abdel-Salam, 2006).There are some recent studies reporting indoor and outdoor aerosol behaviour and its chemical composition (e.g., Chao and Wong, 2002;Na et al., 2004;Gemenetzis et al., 2006;Martuzevicius et al., 2008;Olson et al., 2008;Smolík et al., 2008Lai et al., 2010;Saraga et al., 2010;Zhu et al., 2010;Huang et al., 2012), but only a few refer to elementary schools (Fromme et al., 2008;Almeida et al., 2012;Oeder et al., 2012;Pegas et al., 2012;Smolík et al., 2012).In the present study, the simultaneous characterisation of inhalable (< 10 µm, PM 10 ) and thoracic (< 2.5 µm, PM 2.5 ) suspended particles, along with measurements of comfort parameters, were carried out.This aims to deepen the knowledge of indoor air quality in schools and childcare settings and, thus, to contribute to the elaboration of a prolific set of proposals leading to new policies, guidelines and practices.

Selected Schools
The selected schools are located in the municipality of Aveiro, central Portugal.This coastal municipality has a population of about 78,500 inhabitants within its administrative limits and a total area of 200 km².A total of nine classrooms were monitored, eight integrated in elementary schools and one in a kindergarten (Table 1).All the educational settings started their activities in the sixties or seventies, although, in the meantime, were somewhat refurbished and routine maintenance works were carried out.All the school buildings are naturally ventilated through openable windows and doors, and have white painted concrete walls, wooden window frames, wood or cork flooring and portable electric heaters.The number of pupils per class ranged from 20 to 25, either for kindergarten or primary levels.

Sampling Campaign
The sampling campaign took place between 9 and 29 January, 2012.Continuous measurements of temperature, relative humidity (RH), CO 2 , CO and total volatile organic compounds (TVOCs) were performed with an Indoor Air IQ-610 Quality Probe (Gray Wolf® monitor) in the classrooms of every school.The same measurements, excepting TVOCs, were continuously carried out outside on the playgrounds using an IAQ-CALC monitor (model 7545) from TSI.These equipments were supplied with a factory calibration certificate, but were further checked prior to their use with appropriate calibration kits.
Three automatic monitors using a light scattering technique were employed to measure PM 10 continuously in the three classrooms of each school: (i) a TSI Model 8533 DustTrak DRX, (ii) an AIR-AIDE model AA-3500, and (iii) a Topas monitor from Turnkey Instruments.This latter monitor is designed to continuously record environmental total suspended particles (TSP), PM 10 , PM 2.5 and PM 1 .A "demonstration of equivalence" exercise was carried out before the monitoring campaign.These three instruments were intercompared with gravimetric PM 10 measurements, following the EN 12341 (the reference procedure).The automatic monitors were initially calibrated in the factory using a "standard" test dust.The side-by-side intercomparison allowed the verification or correction of the calibration factors, which were then used to scale the readings of each monitor to give "true" results.
From Monday to Friday, during the occupancy periods, simultaneous sampling of PM 2.5 was performed in 3 classrooms and one outdoor site per school.The daily PM 2.5 samples were collected onto pre-baked (6 h at 500°C) 47 mm diameter quartz filters using Echo TCR Tecora samplers, following the EN 14907 norm.The four low-volume samplers were intercompared before the IAQ evaluation programme.The regression slopes for the results obtained from parallel measurements over one week differed by less than 5%.All the correlation coefficients were higher than 0.95.
Following the teachers' request in order to allow the free movement within the classrooms, the sampling equipments were installed in a central location at the bottom of the rooms.This location avoids direct contamination of the filters with chalk dust, principally when blackboards on the front wall are being used.The sampling heads or inlets were positioned at a height of about 1.2 m above floor and at a distance of around 1 m from the bottom wall.

Analytical Methodologies
Before weighing, the filters were conditioned in a desiccator at least for 24 h in a temperature-and humidity-controlled room.Before and after sampling, gravimetric determination was performed with a microbalance Mettler Toledo AG245 (readability 0.1 mg/0.01 mg).Filter weights were obtained from the average of 10 measurements, with weight variations less than 5%.After gravimetric determination of the PM 2.5 mass, filter punches were analysed by a homemade thermo-optical transmission system in order to obtain the carbonaceous content.After exposure of filter punches to HCl vapours to remove carbonate carbon, controlled heating in anoxic and oxic conditions was performed to separate, respectively, OC into two fractions of increasing volatility and EC.The first fraction corresponds to the volatilisation at T < 200°C of lower molecular weight organics (OC 1 ).The second fraction is related to the decomposition and oxidation of higher molecular weight species at temperatures ranging from 150 to 600°C (OC 2 ).The last fraction of OC is identified by transmittance and corresponds to pyrolysed organic carbon (PC) produced in the previous heating steps.Separation between OC and EC was achieved by initially heating the filter punches under an inert atmosphere to evaporate first the OC fraction.The remaining fraction is sequentially evaporated/burnt under a gas flow containing O 2 .This last carbon fraction contains initial EC plus OC that has pyrolysed during heating under an inert atmosphere.The interference between PC and EC can be controlled by continuous evaluation of the blackening of filter using a laser beam and a photodetector measuring the filter light transmittance.Carbonates present in PM 2.5 samples were analysed through the release of CO 2 , and measurement by the same non-dispersive infrared analyser coupled to the thermo- ; ceiling height 2.8 m; blackboard with chalk and whiteboard with markers; occupied 7.25 h per day optical system, when a punch of each filter was acidified with orthophosphoric acid (20%) in a free CO 2 gas stream.
It should be mentioned that results of an international round robin test on the analysis of carbonaceous particles positioned the University of Aveiro within the group with "best estimates".In addition, the comparison between the methodology of the University of Aveiro and the "European Supersites for Atmospheric Aerosol Research" (EUSAAR-2) protocol (Cavalli et al., 2009) for the quantification of the different carbon fractions in atmospheric particulate matter samples gave similar results for OC and EC, without significant differences at a 95% confidence level (Nunes et al., 2010).In the present study, two replicate carbon analyses were carried out.The analyses were repeated when the coefficient of variation was higher than 5%.
For the determination of water soluble inorganic ions, small parts of the filters were extracted by sonication during 15 min with 5 mL of ultra pure Milli-Q water.The liquid extracts were filtered through a 13 mm PVDF syringe filter (pore size 0.2 μm) and analysed by ion chromatography.An AS4 (4 mm) column with an AG4 guard column were used coupled to a Dionex AMMS II suppressor for Cl -, NO 3 and SO 4 2-determination.Cations (Na + , NH 4 + , K + , Mg 2+ and Ca 2+ ) were analysed using a CS12 (4 mm) column with a CG12 guard column.A Dionex CMMS III suppressor was used in these analyses.The anionic system comprised a Model 401 Dilutor (injector pump), a Gilson sampling injector Model 231, a Shimadzu oven CTO-6A, a Shimadzu conductivity detector CDD-6A, a Dionex interface UCI-100 and a Jasco pump PU-980.The cationic system integrates a Gilson injector 234 and a Dionex DX-100 chromatograph.Sets of multi-cation and anion standards were prepared from a stock solution with 1000 mg/L for each ion.The ion concentrations in the samples were obtained through individual calibration curves (r 2 ≥ 0.99).For the final results, blank values were taken into account.Samples were analysed by two different analysts on two different days.The inter-day precision, expressed as coefficient of variation, was ≤ 3%.

Ventilation Rates
The ventilation Eq. ( 1) has been used to calculate the fresh air ventilation rate (Griffiths and Eftekhari, 2008).For a well-mixed space the change in CO 2 concentration with time is given by: 2 2 6 6 0 10 10 where C t is the indoor concentration of CO 2 at time t (ppm), C ext is the outdoor concentration of CO 2 (ppm), C 0 is the concentration of CO 2 in the indoor air at time 0 (ppm), Q is the volume flow rate of air entering the space (m 3 /s), q CO 2 is the volumetric indoor emission rate of CO 2 (m 3 /s), V is the volume of the classroom (m 3 ) and t is the interval since t = 0 (s).When the classroom is unoccupied there is no CO 2 emission from the occupants, and q CO 2 = 0. Thus, Eq. ( 1) can be rearranged to give the following expression, which allows the ventilation rate (Q) to be calculated from measured concentration values time t apart:

Comfort Parameters and Ventilation Rates
The human sensation of warmth depends on the thermal balance of the body as a whole.Thermal comfort occurs when body temperatures are held within narrow ranges, and the physiological effort of regulation is minimised.A number of variables interact to determine whether people are comfortable with the temperature and relative humidity of the indoor air.Factors such as clothing, activity level, age, and physiology of people in schools vary widely, so the thermal comfort requirements differ for each individual.The American Society of Heating, Refrigerating, and Air-Conditioning Engineers (ASHRAE) Standard 55-1992 describes the temperature and humidity ranges that are comfortable for 80% of people engaged in largely sedentary activities, assuming "normal indoor clothing".At winter conditions, the recommended temperature values for a relative humidity of 30% range from 20.3 to 24.2°C.An interval between 19.7 and 23.3°C is suggested when the relative humidity value is 60%.For many rooms, the schools did not meet ASHRAE recommended levels for winter conditions (Table 2).The average relative humidity values varied depending on the school and room, some of them exceeding the recommended upper limit of 60%.Previous studies suggest that thermal abnormal conditions can cause discomfort or distraction among students, and have consequent effects on performance (Mendell and Heath, 2005).Elevated relative humidity can be responsible for IAQ problems resulting from mould and other microbiological amplification, and subsequent allergic incidents.
At concentrations occurring in most indoor environments, CO 2 build-up can be considered as a surrogate for other occupant-generated pollutants, particularly bioeffluents, and for ventilation rate per occupant, but not as a direct causal factor in human health responses (Apte and Erdmann, 2002).The outdoor air in most locations contains down to about 380 parts per million carbon dioxide.While levels below 5,000 ppm are considered to pose no serious health threat, experience indicates that individuals in schools with elevated CO 2 concentrations tend to report drowsiness, lethargy and a general sense that the air is stale (Mahyuddin and Awbi, 2012).Researchers have been looking for links between elevated CO 2 concentrations and reduced attendance, productivity or achievement (Shendell et al., 2004;Wargocki and Wyon, 2006).
CO 2 concentrations during school hours at a given day and classroom varied from 1300 ppm up to 7000 ppm (Fig. 1), largely exceeding the limit of 1000 ppm stipulated by the Portuguese legislation (RCESE, 2006).Regardless of classroom, the air exchange rates were not higher than 0.2 h -1 , showing an effective building tightness with closed windows that is responsible for the high levels of CO 2 and the extremely low outdoor air infiltration of not more than 38 L/min per person (Fig. 2).In Europe, depending on the country, the required outdoor airflow rates range from 300 to 500 L/min per person (Jones, 2011), while the recommended value in the US was set at 426 L/min per person (ASHRAE, 2004).

Particulate Matter Levels and Chemical Composition
The daily average PM 2.5 concentrations obtained during the occupancy periods ranged from 44 ± 3.2 μg/m 3 to 117 ± 16 μg/m 3 .The highest values were obtained in the school with the highest occupancy rates.In addition, this school, located in a residential area, is the only one surrounded by an unpaved playground.Soil particles are tracked in on shoes or clothing from the outdoors, contributing to indooroutdoor ratios (I/O) in the range 1.3-5 (average = 2.7).Lower average I/O ratios of 1.3 and 1.7 were, respectively, obtained in the city centre and rural schools, where comparable PM 2.5 concentrations have been measured, while an I/O value of 2.3 was determined for the establishment located in the residential area.I/O ratios greater than one indicate the presence of indoor sources, activities/movements of the occupants and reduced ventilation rates as a result of closed doors and windows, so causing accumulation.No significant differences were observed in outdoor PM 2.5 concentrations among schools with averages in the range 37-42 μg/m 3 .
In classrooms (# 2.2 and 2.3) of the school in the periphery of the city, the PM 10 concentrations (Table 3 and Fig. 3) frequently exceeded the limit value of 150 μg/m 3 established by the Portuguese legislation for indoor air (RSECE, 2006).In the classroom with highest occupancy rates (# 2.2), the average PM 10 levels for daytime and night-time periods were, respectively 362 ± 84 and 111 ± 37 μg/m 3 .PM 2.5 represented, on average, 12.7 and 28.0% of the PM 10 concentrations during the teaching hours and vacant periods, respectively (Table 4).This suggests that high PM 10 levels found in classrooms are probably due to resuspension of soil material or other coarse particles.The very high TSP concentrations, up to 3227 μg/m 3 , registered in classroom # 2.2, confirm the presence of resuspended coarse particles.In this classroom, PM 10 accounted for 12.7 ± 1.9% (period of school activities) and 28.0 ± 4.6% (night-time) of the TSP concentrations.Slightly higher averages for daytime and lower at night-time were obtained in the other classrooms, where a Topas monitor has been used.Human activity is seemingly the most important factor to account for the indoor levels of coarse particles.Without this source, at night the indoor particulate material is composed predominantly of PM 2.5 .Significant differences between night and day were also observed for the finest particle size fraction.The settling of coarser particles after school hours contributed to increased submicron particles PM 1 -to-PM 2.5 ratios during the night.
A wide range of PM 2.5 /PM 10 and PM 1 /PM 2.5 ratios has been reported worldwide for different schools, depending on season, meteorology, occupancy rates, physical characteristics of buildings, activities inside the classrooms and ventilation.
In all schools of the present study, the maximum concentrations of PM 10 during teaching hours were registered when students were entering or leaving the classrooms, when the blackboard was being used or when craft activities involving paper or paperboard cutting were underway (Fig. 3).Intense physical activities of occupants in classrooms cause either resuspension of previously deposided particles or delayed deposition/settling of coarse particles indoors (hindered settling), enhancing the indoor concentrations (Goyal and Khare, 2009).However, the highest PM 10 levels coincided with the cleaning activities at the end of the afternoon.In fact, the patterns of diurnal variation depend, in addition to other features, on the cleaning practices.Concentrations tend to be higher in rooms where sweeping is carried out than in those that are vacuumed.Mopping contributes to the lowest levels during the cleaning periods and favours steeper decays during nigh-time.
The particle concentrations observed in the Portuguese schools are consistent with results reported for other countries.It has been pointed out that the high levels are likely associated with the physical activities of pupils, which contribute to a constant process of resuspension of sedimented material from the daily tasks, from skin desquamation or clothing, and from outdoor sources (Brunekreef et al., 1997;Glória (traffic,

Legal limit
Fig. 1.CO 2 levels recorded during night-time and one typical working day in the 9 classrooms.Lee and Chang, 2000;Janssen et al., 2003;Peacock et al., 2003;Pénard-Morand et al., 2005;Ekmekcioglu and Keskin, 2007;Fromme et al., 2007;Van Roosbroeck et al., 2007;Diapouli et al., 2008;Richmon-Bryant et al., 2009).Brunekreef et al. (1997)   particle concentrations lower in the summer and twice higher during the winter.The high occupants' number associated with the small size of classrooms and poor ventilation, were pointed out as contributors to increased concentrations of PM in the winter.Seven primary schools in Athens, Greece, have been reported to present higher indoor PM 10 (229 μg/m 3 ) than outdoor levels (166 μg/m 3 ) (Diapouli et al., 2008).In a study carried out in a school building near an urban traffic roadway in Delhi city, Goyal and Khare observed maximum PM 10 concentrations of 14,474 ± 637.5 μg/m 3 at 11:00 to 11:30 a.m., in winter, when recess hour started and finished and children left and entered to the classrooms.The carbonaceous fraction (TC = OC + EC) accounted for 43 ± 10% of the PM 2.5 mass indoors.Similar values (39 ± 12%) were found outdoors (Fig. 4).Higher carbonaceous contents (52-80%, averaging 62%) have been reported for PM 2.5 sampled in an American elementary school in Montana (Ward et al., 2007).Variable TC mass fractions, from about 23 to 60% of PM 2.5 , were observed in seven schoolrooms in Riverside, California (Sawant et al., 2004).A wide interval of values were obtained for the percent amount of TC in PM 10 (median = 27%) in two schoolrooms in Munich (Fromme et al., 2008).In the present study, while EC represented, on average, 10% of the PM 2.5 mass collected at the playground of the city centre school, around half of this percentage was obtained for the other two institutions.Lower OC/EC ratios (average = 3.31) for the outdoor air of the city centre school also denote a more pronounced influence of traffic emissions than those observed for the periphery (OC/EC = 4.18) and rural (OC/EC = 4.03) schools.Much higher OC/EC ratios (average values ranging from 4.2 to 9.7) were obtained in the classrooms of all schools (Fig. 5), suggesting that the organic carbonaceous fraction is produced more by indoor sources rather than outdoor pollutant penetration.These indoor OC/EC ratios are within the range (5.3-10.8)measured by Na and Cocker (2005) in four schoolrooms in Riverside, California.Crilley et al. (2012) reported OC/EC ratios from 2.72 to 13.6 in PM 2.5 sampled in 17 schools in different suburbs in the city of Brisbane, Australia.Higher ratios (OC/EC from 5.3 to 33.6) were observed in PM 2.5 samples collected in elementary and middle schools in Libby, Montana (Ward et al., 2007).To justify the high OC/EC values found in an Australian school, Morawska et al. (2009) suggested, in addition to the influence of primary sources, the presence o secondary organic aerosols (SOA).Several research groups have conducted investigations into the formation of SOA in indoor environments through reactions of O 3 with VOCs (Morawska et al., 2009;Waring et al., 2011;and references therein).It can be argued that, unless there is office equipment, the O 3 concentrations in  classrooms are very low and, thus, the SOA formation potential is also very low.However, Waring (2009) and Langer et al. (2008) showed that d-limonene, used in air fresheners, detergents and other consumer products, even at ozone-very limited conditions, yields large SOA formation events.It is also known that ozone reacts with indooremitted terpenoids to form SOA, process that contributes to the rapid decay of this oxidant inside buildings and to the detection of I/O ratios substantially lower than 1 (Waring et al., 2011).On the other hand, taking into account that one of the main primary contributors to indoor particles is probably resuspended dust, some of which associated with soil, high OC/EC ratios are not surprising.In fact, it has been reported that particles from dust resuspension present high OC/EC ratios (e.g., Amato et al., 2009;Chen et al., 2012).
The poor indoor-outdoor correlations of OC (r 2 between 0.22 and 0.30) indicate the presence of multiple carbon sources, and different contributions to indoor and outdoor carbonaceous particles.For the school located in the residential area, a modest correlation (r 2 = 0.30) between the EC concentrations measured inside and outside was obtained, whereas lower correlations were found for the other educational settings.This indicates that, probably, the use of graphite pencils and their sharpening may represent a much more important indoor source of EC than that associated with penetration of outdoor traffic emissions.Concerning the relationships between indoor and outdoor measurements, each intercept roughly reflects EC concentrations that originate exclusively from indoor emission sources because intercepts are the concentration values when outdoor EC contributions are zero (Cao et al., 2012).The percentage of the EC intercept in the average EC concentration indicates the contribution of indoor sources to measured indoor EC levels.For the school located in the residential area, where a better indoor-outdoor relationship was obtained, the intercept/average value was 0.25, indicating that about 25% of indoor EC results from the input of indoor sources.The same procedure applied to OC led to similar results across schools.This crude assessment pointed out that around 15% of the indoor OC has an origin in indoor sources, which we believe to be an underestimation, taking into account the very high I/O ratios.
The slopes of the relationships between anions (Cl -, NO 3 and SO 4 2-) and cations (Na + , NH 4 + , K + , Mg 2+ and Ca 2+ ) expressed by the sum of the equivalent concentration are lower than 1 (average = 0.8).In addition to the contribution of carbonates in some samples, the anion deficits are thought to have been affected by the absence of data on F − , PO 4 3− , NO 2 − , and organic acid salts (Huang et al., 2012).The water soluble inorganic ions represented, as a sum, 16.5 ± 1.5%, 11.0 ± 4.8% and 19.2 ± 5.0% of the PM 2.5 mass measured in classrooms of the city centre, residential and rural schools, respectively.Sulphate and nitrate were always the dominant ionic species (Fig. 6).These two secondary ions were highly correlated (r 2 = 0.82), suggesting that their precursors were released from similar emission sources.Furthermore, there was no correlation between Ca 2+ and SO 4 2− and between Ca 2+ and NO 3 − , suggesting there was sufficient ammonia (NH 3 ) to completely neutralise the 2-molar ratios obtained for indoor particles (1.74 ± 0.69) point out that the neutralisation process produced a mixture of (NH 4 ) 2 SO 4 and (NH 4 ) 3 H(SO 4 ) 2 (Alves et al., 2007).In addition, the global good correlation between NH 4 + and NO 3 -(r 2 = 0.79) also suggests aerosol in the form of NH 4 NO 3 .It has been argued that ammonium nitrate aerosol will partially vaporise as it enters a building, which contributes to I/O ratios lower than 1.The decay of concentrations inside buildings has been attributed to transformation of indoor ammonium nitrate to ammonia and nitric acid gases, which are further lost by sorption and deposition on indoor surfaces (Lunden et al., 2003).
The moderate global correlations between Mg 2+ and Ca 2+ (r 2 = 0.50), Mg 2+ and Cl − (r 2 = 0.53), and Na + and Mg 2+ (r 2 = 0.45), indicate that these species may be associated with crustal dust.Excepting calcium in all schools, the I/O ratios for the other ions were generally lower than 1, indicating that the sources of these species were mainly outdoor air penetration rather than indoor sources.The very weak indoor-outdoor relationships (r 2 < 0.20) indicated that the indoor and outdoor water-soluble ion components had different sources.The outdoor calcium could be related to the resuspension of road dust caused by moving vehicles and exhaust gases.In addition to the calcium delivered by outdoor air penetration, the detection of this species in classrooms may be related to the use of chalk in blackboards.This indoor source could also be responsible for the detection of carbonate and potassium in classrooms.Possible outdoor sources for these two species are, respectively, soil dust and biomass burning.Carbonates were only detected in some samples, especially those obtained in classrooms with

Rural
Na + SO 4 2- NH 4 + K + Mg 2+ Ca 2+ Cl -NO 3 - Na + SO 4 2- NH 4 + K + Mg 2+ Ca 2+ Cl -NO 3 - Fig. 6.Mass fractions (percentage) of water soluble ions in indoor PM 2.5 and indoor-outdoor ratios.use of chalk or in outdoor samples from the school with unpaved playground.In these samples, carbonates represented up to 9% of the PM 2.5 mass.

Correlations between Indoor PM 2.5 and Comfort or Chemical Parameters
The PM 2.5 concentrations were correlated either with their chemical constituents or with comfort parameters (Table 5).No significant correlations between meteorological parameters and PM 2.5 levels were observed.Although not statistically significant, the negative correlations between PM 2.5 and RH indicate that aerosol levels decrease with increasing water vapour in the air.Inconsistent results have been reported for other schools.While Fromme et al. (2007) also observed negative correlations for schools in the city of Munich and a neighbouring district, Braniš et al. (2005) presented positive associations between the two parameters.It has been argued that RH promotes the formation of ammonium nitrate, but an increase in precipitation causes a decrease in all PM 2.5 components through scavenging (Tai et al., 2010).Some days of rain throughout the sampling campaign in the Aveiro region may have contributed to particle scavenging in the outdoor air and, consequently, to the decrease of PM 2.5 levels indoors.
Statistically significant positive correlations between CO 2 and PM 2.5 were found in all schools, confirming that the human presence and related activities represent an important source of particles indoors.This also indicates, once again, that inadequate ventilation plays a major role in the establishment of poor IAQ.
The r-square of the correlations between PM 2.5 and OC indicates that this carbonaceous component can explain 44, 91 and 45% of the variation of the indoor particle concentrations in the city centre, residential and rural schools, respectively.The statistically significant positive correlations between these two variables highlight the relevance of organic matter sources or formation processes to the particulate matter in classrooms.
Significant positive associations were found between PM 2.5 and both K + and Ca 2+ , which may be related to the use of chalk.In addition, these two water soluble ions may be associated with crustal dust.Calcium may also derive from the gradual decay of construction materials in the interior of buildings.A very significant correlation between PM 2.5 and Mg 2+ was observed in the rural school.Besides being a crustal element, magnesium is part of the chemical  , etc.).No significant correlations between other chemical constituents and PM 2.5 levels were obtained.

CONCLUSIONS
In most indoor spaces, the average temperatures and relative humidity values were outside the comfortable ranges stipulated by international organisations.Very high CO 2 levels and extremely low outdoor air infiltration indicate excessive building tightness, very deficient ventilations and excessive room occupancy.Exposure to particulate matter in schools is high.I/O ratios greater than one for particles indicate the presence of indoor sources other than the contribution from outdoor.Indoor concentrations are strongly influenced by activities and movement of occupants, which may allow resuspension of previously deposited particles or their delayed deposition or settling.Implementation of more breaks and recesses between classes, decreasing the occupancy per room, increasing the exchange of indoor air with the outdoor, reducing human activities and improving cleanliness of facilities might alleviate the high CO 2 and particle levels.Despite there being a large literature on class-size effects on academic achievement, in times of austere budgets, legislative actions to increase the number of pupils per class have been undertaken.However, taking into account the IAQ problems observed in most schools, policymakers need to bear in mind the possible negative effects of any increase in class size.The need to judiciously weigh costs and benefits is imperative.To develop effective risk assessment and management strategies, toxicological evaluations of particles is highly recommended in future researches.Moreover, strategies to minimise the exposure must be implemented and assessed for effectiveness in intervention studies aiming at reducing health risks of children.It should be taken into account that the benefits resulting from the adoption of strategies to improve IAQ outweigh the investment.In fact, there is growing evidence that interventions induce significant health benefits, which are largely greater than their costs.

Table 1 .
School classrooms selected for the indoor air quality monitoring campaign in Aveiro, Portugal.

Table 2 .
Comfort parameters, total volatile organic compounds and carbon oxides measured in indoor and outdoor environments.

Table 4 .
Mass concentration ratios (%) between different particle size fractions in this study (classrooms where a Topas monitor was used) and mean values reported in the literature for teaching hours.

Table 5 .
Correlations (r)and p-values (in brackets) between PM 2.5 concentrations and chemical constituents or comfort parameters for the 3 schools.< 0.05; ** p < 0.01; *** p < 0.001 formulation of many consumer products and construction materials (e.g., inks, toners, sealants, alloys, some types of bricks