Ultrafine Particles in Indoor Air of aSchool: Possible Role of SecondaryOrganic AerosolsL I D I A M O R A W S K A , * , † C O N G R O N G H E , †

G R A H A M J O H N S O N , † H A I G U O , †

E R I K U H D E , ‡ A N D G O D W I N A Y O K O †

International Laboratory for Air Quality and Health,Queensland University of Technology, Brisbane, Australia,Fraunhofer Wilhelm-Klauditz-Institute (WKI), MaterialAnalysis & Indoor Chemistry, Braunschweig, Germany

Received August 13, 2009. Accepted October 28, 2009.

The aim of this work was to investigate ultrafine particles(<0.1 µm) in primary school classrooms, in relation to theclassroom activities. The investigations were conducted in threeclassrooms during two measuring campaigns, which togetherencompassed a period of 60 days. Initial investigationsshowed that under the normal operating conditions of theschool there were many occasions in all three classrooms whereindoor particle concentrations increased significantlycompared to outdoor levels. By far the highest increases inthe classroom resulted from art activities (painting, gluing, anddrawing), at times reaching over 1.4 × 105 particle cm-3.The indoor particle concentrations exceeded outdoorconcentrations by approximately 1 order of magnitude, with acount median diameter ranging from 20 to 50 nm. Significantincreases also occurred during cleaning activities, whendetergents were used. GC-MS analysis conducted on 4 samplesrandomly selected from about 30 different paints and glues,as well as the detergent used in the school, showed thatd-limonene was one of the main organic compounds of thedetergent, however, it was not detected in the samples of thepaints and the glue. Controlled experiments showed thatthis monoterpene, emitted from the detergent, reacted with O3

(at outdoor ambient concentrations ranging from 0.06 to 0.08ppm) and formed secondary organic aerosols. Furtherinvestigations to identify other liquids that may be potentialsources of the precursors of secondary organic aerosols wereoutside the scope of this project, however, it is expectedthat the problem identified by this study could be more widelyspread, since most primary schools use liquid materials forart classes, and all schools use detergents for cleaning. Furtherstudies are therefore recommended to better understand thisphenomenon and also to minimize exposure of school childrento ultrafine particles from these indoor sources.

IntroductionIn recent years, there has been an increased interest in thesources, concentration levels, and human exposure toairborne particles of lower submicrometer range, oftenreferred to as ultrafine particles (<0.1 µm) and generallymeasured in terms of number, rather than mass concentra-

tion. As concluded by the World Health Organization (1),there is mounting evidence of the toxicological effects ofthese particles on human health, and while the main sourceof these particles is outdoor combustion processes, inparticular vehicle emissions in urban environments, indoorprocesses leading to secondary organic aerosol (SOA) forma-tion can also contribute to indoor ultrafine particle con-centrations (2).

In general, there is considerably less information availableon indoor ultrafine particles, and the systematic quantifica-tion of various indoor source emission rates of these particles,including particle formation mechanisms, particle concen-tration levels, and human exposure, are still to be investigated.Of particular significance is that the nature of emissions fromindoor sources is different from that of the outdoor sourcesand the composition and toxicology of indoor ultrafineparticles may also be different from that of the outdoorparticles. For example, based on their short-term (11-20min) experiments on mice, Wolkoff et al. (3) reported thatSOAs were not the causative agent for adverse effects onmice exposed to a terpene/ozone byproduct mixture. How-ever, it should be kept in mind these studies were only basedon respiration rates and therefore, they do not reflect longer-term impacts (on mice or humans).

One type of environment for which only a handful ofstudies have reported on the existence of ultrafine particlesis schools. While many studies have shown increased particlemass (PM2.5 or PM10) (4), and/or number concentration forparticles larger than 0.3 µm (4, 5), few measurements havebeen conducted on ultrafine particles.

Two examples of particle number concentration timeseries measured by a condensation particle counter (CPC)in schools in Germany during winter are provided by Zollneret al. (5), showing that the indoor concentrations weresignificantly lower than the traffic-influenced outdoor con-centrations. In another German study, conducted in 36schools in Munich during the summer, size distributionmeasurements were taken using a scanning mobility particlesizer (SMPS) and it was found that the median classroomconcentration was 5660 particle cm-3 (6). The characteriza-tion of particle number concentration in 37 classrooms fromtwo Canadian schools during winter was presented byWeichenthal et al. (7). The combined average of classroomconcentrations in both schools was 5017 particle cm-3, andon average, outdoor concentrations were higher than indoorconcentrations by 8989 particle cm-3. Apart from the timeswhen an electrical oven operated in the school, the studyfound a good correlation between outdoor and indoorconcentrations and developed a model linking the two. Inaddition, the impact of ventilation scenario on air exchangerates and indoor particle number concentrations in an air-conditioned Australian primary school classroom was in-vestigated by Guo et al. (8). It was found that the relationshipbetween outdoor particle number concentration and theindoor/outdoor ratio at different air exchange rates followeda power trendline.

Thus, it can be seen that the information currentlyavailable on ultrafine particles in schools remains limited,and it mainly focuses on the impact of outdoor air on indoorconcentrations. Therefore, the aim of this work was toundertake comprehensive indoor measurements of particlenumber and size distribution as a function of time (as wellas reference measurements from outdoors), to relate themto school activities, and to draw conclusions on the existenceof any indoor sources or processes elevating ultrafine particlenumber concentrations in the classrooms.

* Corresponding author e-mail: l.morawska@qut.edu.au.† Queensland University of Technology.‡ Fraunhofer Wilhelm-Klauditz-Institute (WKI).

Environ. Sci. Technol. 2009 43, 9103–9109

10.1021/es902471a CCC: $40.75 2009 American Chemical Society VOL. 43, NO. 24, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 9103

Published on Web 11/18/2009

Experimental MethodsThe experimental design of this study included three steps:(1) monitoring of classroom and outdoor particle numberconcentrations, as well as other parameters; (2) investigationsinto the link between activities and particle concentrationlevels; and (3) experiments to quantify the rate of particleformation under controlled conditions. Steps 2 and 3 weremotivated by the results from step 1.

School. The school was located several kilometers southof the Brisbane CBD. The indoor investigations, which arethe focus of this paper, were conducted in three classrooms,C1, C2, and C3, while the outdoor sampling location was atthe school oval (large outdoor grass playing field) (seeSupplement 1, Figure S1 in the Supporting Information (SI)).

One of the classrooms (C1) was used predominantly forart classes, including activities such as painting, gluing,drawing, etc. Therefore, large numbers of bottles, tubes, andliquids used for art activities were present in many locationsaround the classroom. In addition, the classroom wassubjected to the same detergents and cleaning practices asthe rest of the school.

The Lecture Room. Measurements to investigate particleformation process under controlled conditions (step 3) wereconducted in a lecture theater at the Queensland Universityof Technology (QUT) since the School was not available forsuch tests. Its total volume was 140 m3 and it was naturallyventilated by windows and fans.

Instrumentation and Parameters MeasuredTwo TSI model 3934 SMPSs (TSI Incorporated, St. Paul, MN),incorporating TSI model 3010 CPC, were used to measureschool indoor and outdoor particle size distributions in therange from 0.015 to 0.737 µm (PN0.015-0.7). TSI CPCs model3022 and 3025A were used to measure total outdoor andindoor particle number concentrations in size ranges from0.007 to 3 µm (PN0.007), and 0.005-3 µm (PN0.005),respectively. A TSI model 3936 SMPS with a TSI model 3022CPC was used to measure particle size distribution in therange from 0.005 to 0.160 µm (PN0.05-0.16) in the lecturetheater. An approximation of PM2.5 mass concentration (notactual gravimetric mass as the instrument was not calibratedfor each specific aerosol studied) was measured by the TSImodel 8520 DustTrak. For comparison of trends in time seriesconcentrations (but not of absolute concentration values)outdoor PM2.5 was measured by a 1400A tapered elementoscillating microbalance (TEOM, Rupprecht & Patashnick).Outdoor temperature and relative humidity at the schoolwas measured using a portable weather station (MonitorSensors (Aust)). Indoor O3 concentrations were measuredby a model 400E photometric O3 analyzer. Indoor lecturetheater O3 concentrations were controlled by an O3 generator(HLO-800, HAILEA, Guangdong, China), with a maximumcapacity of 100 mg h-1 and a maximum output of 3.5 L min-1.All instruments were tested and calibrated in the laboratorybefore measurements were conducted. The SMPSs calibra-tion was verified prior to the measurements using NISTtraceable standard size PSL particles. Comparison of thereported overall particle number concentration across iden-tical size ranges for the two SMPS systems for ambient andlaboratory generated aerosols showed that the overall particlenumber concentration agreed to within 20% between thetwo instruments. Details of the settings for each instrumentare given in Supplement 2 in the SI.

To investigate indoor VOC concentrations, samples werecollected by trapping tubes packed with 50 mg of absorbentmaterial (Tenax-TA). The sample size was 5 L, collected ata sampling rate of 0.15 L min-1. The samples were analyzedat the Fraunhofer WKI, Material Analysis & Indoor Chemistry,Germany, using gas chromatography-mass spectrometry

(GC-MS). Analyses were performed on a thermal desorptionGC/MS system (Markes Unity/Ultra, Agilent 6890 GC, 5973NMS) using internal and external standards for calibration, inaccordance with ISO-16000-6 and ISO-17025 accreditedconditions. The method covers a substance range from C5to C22 with a typical limit of determination of 1 µg m-3. Alsosamples of the detergent normally used in the school, as wellas samples of selected paints and glues used for art classes,were collected and analyzed by GC-MS. Glues, detergent,and paints were analyzed on a static-headspace GC/MSsystem (Agilent 1888 Headspace Sampler, 7890 GC, 5975 MS)using an equilibration time of 60 min at 60 °C for the samplematerials.

Study DesignSchool. The measurements at the school were conductedbetween January 23 and February 17, 2006 (first run: C1, C2,and oval) and November 15 and December 15, 2006 (secondrun: C1, C3, and oval). The instruments at the school ovalwere operated in an air-conditioned trailer, while a stationarymonitoring system was setup in one classroom at a time.Outdoor O3 concentration was not monitored at the schoolbut at a Queensland Environmental Protection Agency (EPA)monitoring station located about 10 km from the school.Since O3 distribution in the airshed of Brisbane is highlyhomogeneous (9), these values are representative of theschool outdoor O3 concentrations, in general. During thefirst round there were no teachers diaries distributed to recordactivities occurring within the classrooms or the school.However if, upon inspection of the data, unusually highconcentration were observed, extensive investigation wasconducted immediately through the whole building toidentify any operating sources which could contribute tothese concentrations. During the second round the teachersfrom the classrooms under investigation were asked to keepdiaries of the activities conducted there. Ventilation condi-tions in classroom C1 differed between the measurementruns: during the first run it was naturally ventilated and duringthe second run, air conditioner was used from time to time.Classroom C2 was naturally ventilated, and in classroom C3an air conditioner was occasionally used.

To determine whether VOCs emitted from the detergentused in the classroom could lead to secondary particleformation, experiments were conducted during the weekend(in the absence of any other activities), which includedmopping the floor with water or diluted detergent and thenmonitoring particle number and mass concentrations, aswell as collecting VOC samples for analysis.

Controlled Experiments on Particle Formation Rate. Toquantify particle formation rates, controlled experimentsincluded three steps: (1) background indoor particle numberconcentration measurements and setting indoor O3 con-centrations at several stable levels between 10 and 20 ppb(controlled by a O3 generator); it was necessary to use thegenerator because the ambient O3 concentrations varied,due to large variation in solar radiation, with heavy overcaston some of the days; (2) particle concentration measurementsconducted for 35 min after placing in the room a differentnumber of trays of 759 cm2, containing the detergent usedin the school; and (3) particle decay measurements conductedfor 30-300 min after the trays were removed. During steps1 and 2, two air samples were also collected for VOC analysis.From the time series concentration data, in conjunction withair exchange rates, particle formation rates were calculatedusing the method developed previously (10, 11)(see Supple-ment 3 in the SI).

Data Processing and AnalysisAll statistical analyses (correlation, t test) were conductedusing a statistical analysis software package, SPSS for

9104 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 24, 2009

Windows version 10 (SPSS Inc.). A level of significance of p< 0.05 was used for all statistical procedures. When thedistribution of the concentration measurements was notnormal, a robust analysis, in which the maximum andminimum were trimmed, was employed (12). In addition,nonparametric tests were undertaken to confirm the para-metric results. That is, the corresponding nonparametric testsled to the same conclusions of significance/nonsignificanceas the parametric tests. Comparisons of peak particle sizedistribution for different tray experiments were performedusing the Kolmogorov-Smirnov (K-S) test which can provideinformation on the level of similarity of two particle sizedistributions.

ResultsThe average minimum and maximum temperatures were20.9 and 29.1 °C, respectively, while maximum 1-h ambientO3 concentrations were 0.071 ppm, during the first run inJanuary/February 2006, and 0.079 ppm during the secondround in November/December 2006. Daily maximum O3

concentrations for both measurement campaigns are pre-sented in Supplement 4, Figure S2.

Measurement in the Classrooms and at the Oval. Thetime series of total particle number concentrations measuredin classroom C1 and at the oval are presented in Supplement5, Figure S3a and b (for the first and the second round,respectively) and classrooms C2 and C3 are presented inFigure S4a and b, respectively. During the first run, the highestindoor particle number concentration exceeded 1.4 × 105

particle cm-3, while the average outdoor concentration was5.2 × 103 particle cm-3. During the second round maximumconcentrations in C1 reached 1.07 × 105 particle cm-3, whilethe average outdoor concentration was 4.8 × 103 particlecm-3 and was similar to the first run.

Inspection of the data showed that all major indoorincreases in particle concentration were not related tooutdoor increases. The most notable were those betweenJanuary 31 and February 2 in C1 and, as presented in Figure1, prevailing for a significant part of the day. An extensivesearch of the area, which was undertaken immediately whenthese high concentrations were observed (January 31) andcontinued throughout the following days, failed to identifyany obvious sources. However, an observation was madethat during this period children were painting large surfacesof paper during art classes, which were then drying for many

hours. Figure 1 shows that on most days indoor particlenumber concentrations started increasing in the morning toreach maximum between 13:00 and 16:00. Particle countmedian diameter during these periods ranged from 20 to 50nm. From January 26 to January 30, a peak was also seenbetween 15:00 and 18:00 each day in the classroom.Investigations showed that this was due to tobacco smokingin the classroom after school hours. Figure 1 presents outdoorO3 concentration time series for this period which show apeak in concentrations occurring around 13:00-14:00. Sincethe outdoor particle variation patterns were similar, onlyone day of outdoor data is shown in the Figure 1.

During the first run several episodes of significantlyelevated indoor concentrations occurred in C1 also early inthe mornings between February 14 and 17, with a maximumexceeding 105 particle cm-3. In C2 and C3 concentrationswere also occasionally significantly elevated, well above theoutdoor concentrations, reaching maxima of 7.95 × 104

particles cm-3 and 1.38 × 104 particles cm-3, in C2 and C3respectively.

Elevated indoor particle number concentrations wereagain present during the second round of the measurementswith the daytime average indoor concentrations exceedingthe outdoor by a factor of 2, which was shown to bestatistically significantly (p < 0.01). By checking the teachers’activity dairies, it was found that there was a relation betweensome of the students’ art activities in this classrooms andelevated concentrations of the particles. It was also foundthat the indoor particle number concentration increasedduring the morning, and during afternoon cleaning activities.

While only one type of detergent was used in the schoolduring both runs, the number and variety of materials usedfor art classes was extensive and it was usually not possibleto relate specific peaks in particle concentrations to a specificliquid used during art classes, as in most of the cases severaldifferent liquids were used (for example different paints and/or glues). Since the researchers were not able to interfereand take note of all of the materials used during these classes,the only information available was the time that the art classeswere conducted and the general type of activities undertakenduring that time. Figure 2 shows an example of indoorparticles generated in the classroom before school hoursand during art class activities, where the doors and windowswere closed and air conditioner was occasionally used. Itcan be seen from Figure 2 that indoor particle number

FIGURE 1. Diurnal variation of total particle number concentration in C1 and the oval between January 25 and February 2 (the firstpart of Round 1), as well as outdoor O3 concentrations.

VOL. 43, NO. 24, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 9105

concentrations, in terms of both PN0.015-0.7 and PN0.005, beganto increase from 04:00, while outdoor particle numberconcentration was still decreasing. The concentrationsreached a peak at about 07:00 and then decreased, beforestarting to increase again after 09:00. The first peak coincideswith the morning cleaning activities, while the second peak,which was much higher than the first, occurred during schoolhours, and therefore is likely to be related to the art activitiesconducted on that day. Of interest is that PN0.005 concentra-tions were one time higher than PN0.015-0.7 indicating thepresence of a large number of particles in the lowernanometer size range, which are hypothesized to be freshlynucleated secondary organic particles.

Secondary Organic Aerosol Formation Testing in theClassroom. Three paints and one glue were randomlyselected from about 30 different paints and glues used in theclassroom. GC-MS analysis conducted on samples of thesematerials, as well on the detergent used in the school, showedthat d-limonene was one of the main organic compounds inthe detergent, while the compound was not detected in thesamples of the paints and the glue. To confirm the formationof secondary organic particles from precursors evaporated

from the detergent under the classroom conditions, twoexperiments were conducted in C1, on two consecutiveweekends. The first experiment involved mopping the floorof the classroom for 7 min with diluted detergent andmonitoring of particle number and mass concentrations, aswell as collecting VOC samples every 120 min. At the timeof mopping, outdoor O3 concentration was 0.018 ppm, whilethe maximum on that day was 0.039 ppm (see time seriespresented in Supplement 6 Figure S5). The results of theexperiment are presented in Figure 3, where it can be seenthat after mopping, particle number concentration startedto increase almost immediately. It should be noted that whilethere was also an increase in PM2.5 concentration whichoccurred about 20 min after the increase in particle numberconcentration, it closely coincided with the increase inoutdoor concentrations (to avoid adding complexity to thisgraph, these concentrations have not been included there).Therefore it was concluded that while mopping increasedparticle number concentration in the classroom, it did notresult in a detectable increase in PM2.5 concentration. Whenthe same experiment was repeated with pure tap water therewas no increase in particle number concentration, indicating

FIGURE 2. Outdoor and indoor PN0.015-0.7 and indoor PN0.005 concentrations measured in C3 and at the oval November 23, 2006.

FIGURE 3. Indoor particle number and 30 min average limonene concentrations measured in C1 on Saturday December 2, 2006.

9106 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 24, 2009

that the d-limonene, which evaporated from the detergentduring its reaction with O3, led to the formation of thesecondary organic particles during the first experiment.During measurements with the highest d-limonene con-centration, traces of limonene oxide were also detectable(1-2 µg m-3), indicating that oxidation processes were alsooccurring in the room.

Figure 4 shows the evolution of particle size after moppingof the floor with detergent, in comparison with particle sizeoutdoors. It can be seen that initially particle count mediandiameter (CMD) was similar indoors and outdoors, howeverimmediately after the mopping (and particle formationsseeFigure 3), particle CMD decreased, indicating the presenceof large concentrations of smaller particles. These particlessubsequently grew, and significantly exceeded the size ofoutdoor particles. For example, average indoor CMD between10:00 and 13:00 hrs was 160 nm compared to 80 nm duringthe same time outdoors.

Quantification of Secondary Organic Particle FormationRates. Although the results of the mopping experiment furthersupported the hypothesis of SOA formation in the schoolenvironment, the formation rates could not be quantifiedduring those measurements. Therefore, additional experi-ments were conducted in a lecture theater at QUT, usingtrays of the same detergent that was used in the school. Onecondition for conducting the experiments was that outdoorparticle concentration remained relatively stable throughoutthe experiments. The average air exchange rate during thesetests was 1.08 h-1.

It was found that when indoor O3 concentration levelswere lower than 0.005 ppm, there was no clear increase inparticle number concentration, however rapid particleformation was found to occur above this O3 concentration.The peak values for PN0.005 and PN0.005-0.16 concentrationreached 5.89 × 104 particles cm-3 and 1.22 × 104 particlescm-3, respectively, using 4 trays of detergent, and 6.85 × 104

particles cm-3 and 1.33 × 104 particles cm-3, respectively,using 5 trays of detergent (further details are presented inSupplement 7, Figure S6). These peak values were ap-proximately 20 and 10 times higher than backgroundconcentration of PN0.005 and PN0.005-0.16, respectively, whiled-limonene concentration was 398 and 525 µg m-3, for the4 trays and 5 trays, respectively.

An example of particle size distribution during theexperiments is shown in Figure 5. In general, particle CMDdecreased immediately after the introduction of the trays,which was followed by a subsequent increase. It can be seenfrom Figure 5 that the particle size distribution for 5 trayswas larger than for 4 trays and the difference was found tobe statistically significant (p < 0.01) for particles in the sizerange 0.005-160 nm. This indicates that a large number ofnanosized secondary particles formed, which subsequentlygrew in size.

PM2.5 concentration in the lecture theater did not increaseabove the background level during the particle formationprocess, supporting the findings from the classroom experi-ment, which also showed no impact of increased particlenumber concentration on PM2.5 mass.

Table 1 presents a summary of the estimated averageparticle number (PN0.005) formation rates ( jQs). Statisticalanalysis (correlation) showed that there was a significantpositive correlation between jQs, tray number, and d-limonene concentration. In terms of surface area of the

FIGURE 4. Time series of count median diameter (CMD) in classroom C1 and the oval during and after the mopping experiment onDecember 2, 2006.

FIGURE 5. Example of particle size distribution variation duringthe particle formation process. The first set of data (left) is for 4trays while the second set (right) is for 5 trays of the detergent.

VOL. 43, NO. 24, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 9107

evaporating liquid, the range of jQs per square centimeterranged from 1.02 × 107 to 1.59 × 108 particles min-1 cm-2

with an average value of 5.16 ( 6.33 × 107 particles min-1

cm-2. This large variation in jQs per square centimeterindicates that there may also be other factors that play animportant role in the particle formation process.

DiscussionThis study focused on monitoring classroom and outdoorparticle number concentrations, as well as other parameters,in three classrooms of a primary school over 60 monitoringdays. The investigations showed highly elevated indoorparticle number concentrations during various school activi-ties that were not related to outdoor concentration patternsand, at times, were up to 2 orders of magnitude higher thanthe outdoor concentrations. The contribution from anycombustion processes within the school was also excluded.Some of these increases in concentration coincided withcleaning activities in the classrooms, while the largest ofincreases occurred in the classroom where art activities wereconducted. In both cases the presence of a large number ofparticles in the lower nanometer size range were observed,and during the later events, particle concentration was ashigh as 1.4 × 105 particle cm-3. These findings motivated thesecond stage of the study, which was designed to explainwhich sources or processes contributed to formation of theseparticles. The hypothesis developed was that the detectedparticles were SOAs. The experiments conducted showedthat the monoterpene emitted from the detergent used inthe school reacted with O3 (at ambient concentrations inBrisbane), which led to the formation of SOAs, at concentra-tions well above the normal background concentration. Theexperiments also quantified the rate of particle formationunder different experimental conditions.

Several research groups have conducted investigationsinto the formation of SOAs in indoor environments, throughreactions of O3 with volatile organic compounds (VOCs), mostcommonly terpenes (R-pinene and limonene) and terpene-related compounds (linalool, R-terpineol) (13, 14). Studieswere conducted whereby a controlled amount of a knownsubstance was introduced into an enclosed space, along withO3, which was introduced by an O3 generator. The enclosedspaces were: test chambers, unoccupied offices, occupiedoffices (15, 16), and homes (17, 18). The substances intro-duced included (i) detergents (e.g., orange oil-based degreaserand a pine oil-based general-purpose cleaner) (19); (ii) paints(20); (iii) pure chemicals (e.g., limonene) (21); (iv) a mix ofVOCs (including terpenes) (22); (v) building materials(including cedar board, Japanese Cyprus board, and waxedand unwaxed plastic tiles) (23); (vi) orange oil-based degreaser(19); (vii) scented oil air fresheners (19); and (viii) floor polish.

The studies were conducted under different concentra-tions of O3. For example, the concentration was about 60ppb in a study by Singer et al. (19), and within a range of85-327 ppb in a study by Sarwar and Corsi (21). This

compares to outdoor concentrations of 30-60 ppb in mostlocations, and up to 140 ppb in polluted areas. In addition,Sarwar et al. (24) tested five consumer products at two O3

levels (3-15 ppb and 95-219 ppb) for each product, Waringet al. (25) tested ultrafine particle removal and generation byportable air cleaners in chamber at an O3 concentration of1-117 ppb, and Alshawa et al. (26) studied aerosol particleremoval and generation by ionization air purifiers in asparingly furnished office at an O3 concentration from 5 to225 ppb. In this study, a steady-state kinetic model wasdeveloped and tested. In general, the studies showed thatwhen conditions favoring a reaction were achieved, rapidformation of large numbers of nanometer size particlesoccurred, followed by particle growth. The mechanisms ofthe reaction were discussed by Weschler and Shields (13),Lui et al. (27), Weschler (28), and Sarwar and Corsi (21), andincluded homogeneous nucleation, as well as condensation/absorption of reaction products onto smaller particles, withcoagulation contributing much less significantly to particlegrowth.

The common element of all these studies is that theywere undertaken with the aim of investigating whether SOAformation in enclosed spaces would occur and under whatconditions. While none of the studies investigated SOAformation in response to particle bursts discovered in a realenvironment (which were not intentionally generated), theydid show that such process can occur in real environments,since the conditions of many of the studies were comparableto typical indoor environmental conditions, including theenvironment of the school investigated here.

The second stage of this study focused on detergent,because the analysis showed that it contained d-limonene,known to be a common precursor of SOA. However, it wasclear that there were many other potential sources of theVOCs, in particular, paints and other liquids used for artactivities in the school, and it was when these substanceswere used when the highest concentrations of the particleswere measured. While the four of them randomly selecteddid not contain d-limonene, it was outside the scope of thisproject to attempt to identify the particular liquids whichpotentially were the sources of the precursors of the SOA,since there were over 30 of them used in the art classroom.

It is expected that the problem identified by this study isnot restricted to this school, but is widely spread, as mostschools use liquid materials for art classes, and all schoolsuse detergents for cleaning. Therefore, it is recommendedthat further studies are conducted to (1) provide an inventoryof the liquids and other substances with the potential forevaporating VOCs within the indoor environment of theschool; (2) conduct chemical analysis of the content of themore commonly used substances; (3) conduct controlledexperiments of secondary particle formation in the presenceof these substances; (4) provide recommendations regardingthe use of low emitting substances in the schools to preventindoor formation of secondary particles; and (5) put in placea process for the testing and approval of any new substancesand materials that become available on the market, beforethey are used in schools.

AcknowledgmentsMembers of the ILAQH, QUT, in particular, Ms. RachaelAppleby, Dr. McKenzie Lim, and Mr. Hao Wang, as well asMs. Nerina Foley, are thanked for their assistance with thisstudy.

Supporting Information AvailableSchematic diagram of the school, instrument settings,calculations of particle formation rates, maximum outdoorO3 values throughout both measurement campaigns, timeseries of total particle concentrations measured in the

TABLE 1. Summary of the Estimated Average Particle Number(PN0.005) Formation Rates ( jQs)

tray Q̄s (p min-1 × 109) CMD (nm) O3(ppm) d-limonene (ppb) T (°C) RH (%)

1 32.6 23 0.01 a 27 53.81 8.6 29 0.014 11 27.9 58.62 15.5 24 0.009 20 27.2 58.62 22.7 33 0.01 19 28.7 52.13 26.7 52 0.011 28 28.6 52.83 24.7 46 0.014 17 28.3 60.14 443.6 23 0.011 72 28.6 53.45 603.4 28 0.01 95 28.2 59.4

a No data available.

9108 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 24, 2009

classrooms and at the oval, time series of O3 concentration,example of particle formation rates during controlled ex-periments in the lecture theatre. This material is availablefree of charge via the Internet at http://pubs.acs.org.

Literature Cited(1) WHO. Guidelines for Air Quality; World Health Organization, 2005.(2) Morawska, L.; Ristovski, Z.; Jayaratne, E. R.; Keogh, D.; Ling, X.

Ambient nano and ultrafine particles from motor vehicleemissions: characteristics, ambient processing and implicationson human exposure. Atmos. Environ. 2008, 42, 8113–8138.

(3) Wolkoff, P.; Clausen, P. A.; Larsen, K.; Hammer, M.; Larsen,S. T.; Nielsen, G. D. Acute airway effects of ozone-initiatedd-limonene chemistry: Importance of gaseous products. Toxicol.Lett. 2008, 181, 171–176.

(4) Parker, J. L.; Larson, R. R.; Eskelson, E.; Wood, E. M.; Veranth,J. M. Particle size distribution and composition in a mechanicallyventialited school building during air pollution episodes. IndoorAir 2008, 18, 386–393.

(5) Zollner, I.; Gabrio, T.; Link, B. Concentrations of particulatematter in schools in South-West Germany. Inhalat. Toxicol.2007, 19 (S1), 245–249.

(6) Fromme, H.; Twardella, D.; Dietrich, S.; Heitmann, D.; Schierl,R.; Liebl, B.; Ruden, H. Particulate matter in the indoor air ofclassrooms - exploratory results from Munich and surroundingarea. Atmos. Environ. 2007, 41, 854–866.

(7) Weichenthal, S.; Dufresne, A.; Infante-Rivard, C.; Joseph, L.Characterizing and predicting ultrafine particle counts inCanadian classrooms during winter months: Model develop-ment and evaluation. Environ. Res. 2008, 106, 349–360.

(8) Guo, H.; Morawska, L.; He, C.; Gilbert, D. Impact of ventilationscenario on air exchange rates and on indoor particle numberconcentrations in an air-conditioned classroom. Atmos. Environ.2008, 42 (4), 757–768.

(9) Morawska, L.; Vishvakarman, D.; Mengersen, K.; Thomas, S.Spatial variation of airborne pollutant concentrations in Bris-bane, Australia and its potential impact on population exposureassessment. Atmos. Environ. 2002, 34, 2437–2450.

(10) Abt, E.; Suh, H.; Allen, G.; Koutrakis, P. Characterization of indoorparticle sources: A study conducted in the metropolitan Bostonarea. Environ. Health Perspect. 2000, 108 (1), 35–44.

(11) Long, C. M.; Suh, H. H.; Catalano, P. J.; Koutrakis, P. Using time-and size-resolved particulate data to quantify indoor penetrationand deposition behavior. Environ. Sci. Technol. 2001, 35 (10),2089–2099.

(12) Morawska, L.; He, C.; Hitchins, J.; Gilbert, D.; Parappukkaran,S.; Mengersen, K. Characteristics of particle number and massconcentrations in residential houses in Brisbane, Australia.Atmos. Environ. 2003.

(13) Weschler, C.; Shields, H. The influence of ventilation on reactionsamong indoor pollutants: Modelling and experimental obser-vations. Indoor Air 2000, 10, 92–100.

(14) Nazaroff, W. W.; Weschler, C. J. Cleaning products and airfresheners: exposure to primary and secondary air pollutants.Atmos. Environ. 2004, 38, 2841–2865.

(15) Weschler, C. J.; Shields, H. C. Indoor ozone/terpene reactionsas a source of indoor particles. Atmos. Environ. 1999, 33 (15),2301–2312.

(16) Toftum, J.; van Dijken, F. Ultrafine and fine particle formationin a naturally ventilated office as a result of reactions betweenozone and scented products. In 7th International Conferenceon Energy-Efficient Healthy Buildings; Singapore, 2003.

(17) Long, C. M.; Suh, H. H.; Koutrakis, P. Characterization of indoorparticle sources using continuous mass and size monitors. J.Air Waste Manage. Assoc. 2000, 50 (7), 1236–1250.

(18) Hubbard, H. F.; Coleman, B. K.; Sarwar, G.; Corsi, R. L. Effectsof an ozone-generating air purifier on indoor secondary particlesin three residential dwellings. Indoor Air 2005, 15, 432–444.

(19) Singer, B. C.; Coleman, B. K.; Destaillats, H.; Hodgson, A. T.;Lunden, M. M.; Weschler, C.; Nazaroff, W. W. Indoor secondarypollutants from cleaning products and air freshener use in thepresence of ozone. Atmos. Environ. 2006, 40, 6696–6710.

(20) Lamorena, R. B.; Jung, S.-G.; Bae, G.-N.; Lee, W. The formationof ultrafine particles during ozone-initiated oxidations withterpenes emitted from natural paint. J. Hazard. Mater. 2007,141, 245–251.

(21) Sarwar, G.; Corsi, R. The effects of ozone/limonene on indoorsecondary organic aerosols. Atmos. Environ. 2006, 41 (5), 959–973.

(22) Fan, Z. H.; Lioy, P.; Weschler, C. J.; Fielder, N.; Kipen, H.; Zhang,J. F. Ozone-initiated reactions with mixtures of volatile organiccompounds under simulated indoor conditions. Environ. Sci.Technol. 2003, 37, 1811–1821.

(23) Aoki, T.; Tanabe, S. Generation of sub-micron particles andsecondary pollutants from building materials by ozone reaction.Atmos. Environ. 2007, 41, 3139–3150.

(24) Sarwar, G.; Olson, D. A.; Corsi, R. L.; weschler, C. J. Indoor fineparticles: the role of terpene emissions from consumer products.J. Air Waste Manage. Assoc. 2004, 54 (3), 367–377.

(25) Waring, M. S.; Siegel, J. A.; Corsi, R. L. Ultrafine particle removaland generation by portable air cleaners. Atmos. Environ. 2008,42, 5003–5014.

(26) Alshawa, A.; Russell, A. R.; Nizkorodov, A. A. Kinetic analysis ofcompetition between aerosol particle removal and generationby ionization air purifiers. Environ. Sci. Technol. 2007, 41, 2498–2504.

(27) Liu, X.; Mason, M.; Krebs, K.; Sparks, L. Full scale chamberinvestigation and simulation of air freshener emissions in thepresence of ozone. Environ. Sci. Technol. 2004, 38, 2802–2812.

(28) Weschler, C. Chemical reactions among indoor pollutants: whatwe’ve learned in the new millennium. Indoor Air 2004, 14 (S7),184–194.

ES902471A

VOL. 43, NO. 24, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 9109