Accepted Manuscript

Investigation of formaldehyde sources in French schools using a passive flux sampler

G. Poulhet, S. Dusanter, S. Crunaire, N. Locoge, V. Gaudion, C. Merlen, P. Kaluzny,P. Coddeville

PII: S0360-1323(13)00287-4

DOI: 10.1016/j.buildenv.2013.10.002

Reference: BAE 3530

To appear in: Building and Environment

Received Date: 30 June 2013

Revised Date: 1 October 2013

Accepted Date: 3 October 2013

Please cite this article as: Poulhet G, Dusanter S, Crunaire S, Locoge N, Gaudion V, Merlen C, KaluznyP, Coddeville P, Investigation of formaldehyde sources in French schools using a passive flux sampler,Building and Environment (2013), doi: 10.1016/j.buildenv.2013.10.002.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service toour customers we are providing this early version of the manuscript. The manuscript will undergocopyediting, typesetting, and review of the resulting proof before it is published in its final form. Pleasenote that during the production process errors may be discovered which could affect the content, and alllegal disclaimers that apply to the journal pertain.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Investigation of formaldehyde sources in French schools using a passive flux sampler Poulhet G.1,2,4, Dusanter S.1,2,3, Crunaire S.1,2, Locoge N. 1,2, Gaudion V. 1,2, Merlen C. 1,2, Kaluzny P.4, Coddeville P.1,2 sebastien.dusanter@mines-douai.fr 1 Univ Lille Nord de France, F-59000 Lille, France 2 Mines Douai, CE, F-59508 Douai, France 3 School of Public and Environmental Affairs, Indiana University, Bloomington, IN, USA 4 Tera Environnement, 38926, Crolles, France

While indoor air quality issues have received increasing attention the past decades, detailed investigations of the primary sources of indoor pollution are still difficult to carry out. There is a lack of analytical tools and measurement procedures to identify sources of pollutants and to characterize their emissions. Formaldehyde is an ubiquitous pollutant in indoor environments, which is known to lead to adverse health effects. This study describes a measurement procedure to apportion formaldehyde emissions from building and furnishing materials and presents a source apportionment study performed in French public schools. More than 29 sources of formaldehyde were characterized in each investigated classroom, with higher emissions from building materials compared to furnishing materials. Formaldehyde emission rates measured using passive flux samplers (PFS) ranged from 1.2 to 252 µg/m2/h, highlighting several strong emitters made of wood products and foam. Interestingly, the ceiling was identified as the main source of formaldehyde in most classrooms. Measured emissions and air exchange rates were constrained in a mass balance model to evaluate the impact of formaldehyde reduction strategies. These results indicate that formaldehyde concentrations can be reduced by 87-98% by removing or replacing the main source of emission by a less emissive material and by increasing the air exchange rate to 1 h-1. In addition, an intercomparison of total emissions calculated from (1) PFS measurements and from (2) measured formaldehyde concentrations and air exchange rates indicate that an unidentified sink of formaldehyde may exist in indoor environments. Keywords: Indoor air, Formaldehyde, Public school, Passive sampling, Emission rates, mass balance model

1. Introduction People from industrialized countries spend approximately 90% of their time in

indoor environments such as housing and office buildings [1–3], being exposed to elevated concentrations of indoor air pollutants. In light of these observations, indoor air quality has received increasing attention over the past decades to characterize both the nature and the concentration levels of indoor pollutants. There is a general agreement within the scientific community that chronic exposures to indoor pollutants can lead to pathologies such as asthma [4] and cancers [5].

Indoor environments are enriched in Volatile Organic Compounds (VOCs) [6], with indoor concentrations being 2 to 100 times higher than outdoor [7]. Reasons usually highlighted to explain elevated concentrations of indoor VOCs are related to an increasing use of manmade materials for building and furnishing, the use of cleaning supplies, unvented combustion processes such as gas stoves, and low air exchange rates due to recent energy saving politics [8].

Formaldehyde (HCHO) is one of the most frequently detected VOC in indoor environments as well as the most abundant aldehyde. This compound has been classified as carcinogenic in 2004 by the International Agency for Research on

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Cancer (IARC) [5]. A study conducted from 2003 to 2005 by the French observatory of indoor air quality (OQAI) in 567 accommodations reports a mean concentration of formaldehyde of 19.6 µg/m3 [9], which is approximately 10 times higher than outdoor. While this value is lower than the upcoming French regulation threshold of 30 µg/m3

for long-term exposure [10], it was highlighted that 22% of these accommodations exhibit concentrations higher than 30 µg/m3. Studies performed in other countries indicate similar results in Japan [11] and Canada [12] with mean formaldehyde concentrations of 17.6 µg/m3 and 33.2 µg/m3 respectively, lower concentrations in Sweden with a mean value of 8.3 µg/m3 [11], and higher concentrations in the USA with a mean value of 55 µg/m3 [6].

To reduce indoor concentrations of formaldehyde, as well as other VOCs, there is a need to identify the main sources of emission and to quantify the contribution of each source to the emission budget. Building and furnishing materials are known to emit a large range of VOCs, including formaldehyde, and to significantly enhance their indoor concentrations [13]. In this context, several analytical devices were developed to perform in-situ measurements of VOC emissions from solid materials, including the Field and Laboratory Emission Cell (FLEC) and Passive Flux Samplers (PFS).

While FLEC apparatus can be used to directly measure emission rates in indoor environments, most studies involving a FLEC focused on laboratory measurements to characterize building and furnishing materials and to evaluate the impact of air velocity, temperature, and relative humidity on emission rates [14–20]. On-site studies are not as frequent because this technique requires cumbersome equipment such as cylinders of zero air, pumps, and flow controllers [15,17,21]. In addition, a FLEC device is limited to measuring emission rates from one material at a time while indoor environments are usually built using several tens of different materials. The use of several FLEC to perform an exhaustive characterization of indoor emissions would be too cumbersome, time consuming, and expensive.

Passive sampling is more suitable for in-situ measurements of emissions. This technique uses inexpensive samplers and allows multiple samplings at a time by multiplying the number of samplers. Recent studies showed the feasibility of measuring emission rates of a few VOCs using passive samplers [22], including formaldehyde [23–25], toluene and pinenes [26], and semi-volatile organic compound such as phtalates [27].

A PFS was developed at Mines Douai to quantify formaldehyde emissions from building and furnishing materials [23]. This PFS was deployed in 24 unoccupied student rooms [28], showing that most emissions were in the range 1.2-12.1 µg/m2/h. A few strong emitters were identified, with emissions as high as 21.3-131.3 µg/m2/h. This study highlighted the potential of this new analytical tool to apportion the main sources of formaldehyde in indoor environments, and therefore to improve indoor air quality.

A national campaign was conducted in France from 2009 to 2011 in 316 day-care centers and primary schools to provide a comprehensive picture of formaldehyde and benzene concentration levels in French buildings receiving children [29,30]. This study was mandated by the French ministry of ecology, sustainable development, transport and housing and included the technical support from the French national institute for industrial environment and risks (INERIS) and the French scientific and technical center for building (CSTB). An average

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

formaldehyde concentration of 15.7 µg/m3 was inferred from two measurement periods of 4.5 days (summer and winter). This study showed that formaldehyde concentrations are acceptable for 85 % of the buildings considering the upcoming French regulation value of 30 µg/m3 for long-term exposure [10]. However, this study also highlighted that eight buildings were exhibiting formaldehyde concentrations close to 50 µg/m3, requiring further measurements to identify emission sources.

The French central laboratory for air quality monitoring (LCSQA) and Mines Douai carried out a diagnostic of formaldehyde sources in these eight polluted buildings. This publication describes the basics of a methodology designed to identify formaldehyde emitters and to quantify their contribution to the formaldehyde budget, based on the use of the PFS mentioned above. Emissive materials were categorized into building and furnishing materials to differentiate these two classes of emission. This study also makes use of chemical mass balance equations to evaluate the relevance of strategies of formaldehyde reduction.

2. Measurements

2.1. Description of the sites Measurements were carried out from June-July 2011 inside four buildings

located in Picardie (school 1), Provence-Alpes-Côte d'Azur (schools 2-3), and Pays de la Loire (school 4), and during May 2012 inside four additional buildings in Limousin (school 5), Centre region (school 6), and Franche-Comté (schools 7-8). Only one of the most polluted rooms was investigated for each site.

Details about the sites are given in Table I and II, including formaldehyde concentrations measured during the 2009-2011 national campaign. As shown in table I, more than 29 potential sources of formaldehyde were identified in each room. In order to correlate measured emission rates to total emission rates calculated from measured concentrations of formaldehyde and air exchange rates (section 2.2.3), it was decided to remove school consumables such as books and drawing equipment, whose emissions could not be measured in this study. The rooms were left closed for at least 12 hours before the measurements to make sure that formaldehyde concentrations had reached a steady-state concentration.

2.2. Methodology designed to apportion emission sources of formaldehyde

2.2.1 Material The methodology designed to investigate the formaldehyde budget involves five steps:

(i) Visual identification of different materials & measurement of covered surface areas

(ii) Measurement of air exchange rates (iii) Measurement of indoor and outdoor formaldehyde concentrations (iv) Measurement of formaldehyde emission rates for each identified material (v) Evaluation of strategies of formaldehyde reduction through the use of

chemical mass balance equations

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Measurements of air exchange rates were performed according to ASTM E

741-00 [31]. Carbon dioxide (CO2) was injected at the center of the room using a gaz cylinder of pure CO2 (Air liquide) and a fan was used to speed-up the mixing. Initial mixing ratios of CO2 were in the range 3000-5000 ppm. Temporal decays of CO2 were recorded with a Testo datalogger Term 400 and analyzed using the following equation:

Ln (Ct-Cb) = -at + ln(C0-Cb) Eq. 1

Parameters in Eq. 1 are the air exchange rate, a (h−1), the time, t (h), the ambient mixing ratio of CO2 outside the building, Cb, and the CO2 mixing ratios inside the room at the starting time and at a subsequent time t, C0 and Ct respectively.

Doors leading to other rooms were closed and small openings around the doors were taped to avoid air exchange with adjacent rooms. Only exchanges with outdoor air were allowed. As a consequence, air exchange rates and formaldehyde concentrations reported in this study are not representative of regular operating conditions of the rooms. These conditions were chosen to simplify the calculations of total emission rates from measured concentrations of formaldehyde (section 2.2.3) and to simplify mass balance equations that are used to evaluate strategies of formaldehyde reduction (section 2.2.4). If the doors were not closed and taped, an analysis of the formaldehyde budget would require measurements of formaldehyde concentrations in all adjacent rooms and the determination of air exchange rates between these rooms.

Formaldehyde concentrations were measured by active sampling of ambient air through DNPH cartridges (Waters, WAT037500), according to ISO 16000-3 [32]. The cartridges were analyzed by High Performance Liquid Chromatography (HPLC, Waters 2695) coupled to a photodiode array detector (365 nm). The detection limit (3σ) determined for a sampling duration of 6 hours at a sampling flow rate of 1 L/min was 0.03 µg/m3. Two collocated measurements were made at the center of the room in five of the eight buildings. These measurements showed a relative difference lower than 4.5%, excepted for school 8 where the two measured concentrations were different by 44%. An average value of the two measurements has been used in the following analysis.

Formaldehyde emissions were measured using a PFS developed by A. Blondel and H. Plaisance [23]. This sampler consists of a Petri dish made of tinted glass with an inner diameter of 35.4 mm and a depth of 20 mm. A quartz filter (Whatman) is placed at the bottom of the sampler and 500 µL of a homemade solution of DNPH is spread on the filter before sampling. The sampler is setup with the open face on the targeted material for approximately 6 hours. Formaldehyde diffuses inside the sampler, reacts with DNPH, and is trapped as a dinitrophénylhydrazine species. Care is taken to get a good sealing between the PFS and the materials to insure that the measured emission rates are specific to the materials under investigation.

At the end of the sampling period, the filter is transferred into a tainted glass tube and stored at -20°C until analysis. PFS filters were an alyzed at the latest three weeks after sampling. Conservation tests performed previously have shown that these filters can be kept at -20°C for at least two weeks without alteration of the samples [23].

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

The dinitrophenylhydrazone species is extracted from the filter using acetonitrile (HPLC grade, Waters) and analyzed by HPLC (Waters 2695). A detailed description of this analytical procedure can be found elsewhere [23]. An emission rate is calculated from the mass of formaldehyde that was trapped in the sampler and a calibration curve determined previously [23]. While the PFS was calibrated at 23°C using an emission chamber, this sampler can also be used at different temperatures as shown by A. Blondel and H. Plaisance [23]. These authors report that the PFS response is not affected in an investigated range of temperature of 23-35°C. Published laboratory characterizations [23] indicate a detection limit of 1.2 µg/m2/h and a good linearity up to 400 µg/m2/h. Measurement precisions (1σ) measured at detection limit, 25 and 100 µg/m2/h are 33, 15 and 8 % respectively.

Fifty PFS were used in each classroom to perform an exhaustive apportionment of formaldehyde sources. Only one sampler was deployed on most of the materials but two to four replicates were made on a few materials to test the repeatability. Four to seven blank measurements were also performed at each site to track potential contaminations of the PFS. The mass of formaldehyde collected on these blank samplers was low for six of the eight buildings, consistent with laboratory characterizations [23]. These blank measurements indicate no contamination of the PFS during on-site measurements. However, blank measurements were 5-15 times higher for schools 2 and 4 due to the use of a contaminated DNPH solution. It is worth noting that to minimize any bias in the measurements, the blank value is always subtracted from exposed samplers. However, this contamination led to a four-fold increase of the detection limit for school 4.

Emission rate and concentration measurements of formaldehyde (steps iii and iv of the methodology described above) were performed simultaneously for a duration of 6 hours. Air exchange rate (step ii) measurements were conducted either before or after the formaldehyde measurements.

2.2.2 Source apportionment

The contribution of each material (CSi) to the formaldehyde budget was calculated from measured emission rates (Ti in µg/m2/h) and measured surface areas (Si in m²), N being the total number of identified sources:

∑ ××=

N

ii

iii

ST

STCS

1

Eq.2

2.2.3 Mass balance equations

A mass balance model is defined by Eq. 3:

[ ] [ ] [ ]indooroutdoorindoor COHaCOHPa

V

Q

dt

COHd22

2 ×−××+= Eq.3

[H2CO]indoor and [H2CO]outdoor are indoor and outdoor formaldehyde concentrations (µg/m3) respectively. P is the fraction of formaldehyde that is not lost on surfaces when outdoor air enters in the room (P = 1 for formaldehyde [33] [34]). a and V are

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

the air exchange rate (h-1) and the room volume (m3). Q is the total emission rate of formaldehyde (µg/h), i.e. ΣTixSi.

Assuming that formaldehyde is in steady state, a total emission rate can be derived from Eq. 4:

[ ] [ ]( ) VaCOHPCOHQ outdoorindoorestimated ×××−= 22 Eq.4

Measured values of [H2CO]indoor, [H2CO]outdoor, a, and V are used to estimate the total emission rate of formaldehyde. This value is then compared to ΣTixSi

in section 3.3 to get insights into both (1) the validity of Eq.3 and (2) the representativity of the PFS measurements. It is worth noting that the term P.[H2CO]outdoor in Eq. 4 can be neglected in this study since P.[H2CO]outdoor is always lower than 2 % of [H2CO]indoor, excepted for school 8 for which it is close to 8%.

2.2.4 Evaluation of strategies of formaldehyde reduction

In order to evaluate the relevance of strategies of formaldehyde reduction, Eq. 4 can be rearranged into Eq. 5:

[ ] [ ][ ] Qa

aQ

COH

COHCOHAbatement

indoor

indoorindoor

*

*

2

*

22 1−≈−= Eq.5

[H2CO]*indoor is the formaldehyde concentration calculated after remediation. An abatement value is calculated for a remediation scenario in which the total emission rate is decreased from Q to Q* or the air exchange rate is increased from a to a*.

Scenarios tested in this study involve an increase of the air exchange rate to a*=1 h-1 (scenario I) and the removal or replacement of the main source of emission (scenario II). The main source of emission is defined as the material exhibiting the largest contribution to the total emission rate. If the main source of emission is a building material (schools 1-2, 4-5, and 8), this emitter is replaced by a less emissive material (12.5 µg/m2/h). In contrast, if the main source of emission is a furnishing material (schools 3 and 6-7), this emitter is removed from the room.

3. Results and discussion

3.1. PFS Performances Limits of detection calculated from blank measurements (Table III and IV) are

consistent with laboratory findings and range from 0.6-3.0 µg/m2/h, with the exception of school 4 where a detection limit of 7.0 µg/m2/h was observed. This higher limit of detection is due to the use of a contaminated DNPH solution as mentioned above.

Two to four replicates were performed on materials exhibiting emission rates ranging from 3-260 µg/m2/h. Figure 1 shows a scatter plot made of 61 replicates from the 8 sites. The measurements are randomly scattered around the 1:1 line and are within

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

the 3σ precision determined from laboratory testing. These results suggest that the precision for in-situ measurements was as good as that observed in the laboratory and that large surface areas covered by a same material exhibit homogeneous formaldehyde emissions. However, it is worth noting that these measurements were performed in non-heated buildings and it is likely that inhomogeneous emissions may occur during colder periods when the use of internal heaters leads to gradients of temperature within the rooms.

3.2. Measured air exchange rates, concentrations and emission rates Air exchange rates are low and range from 0.03-0.22 h-1 (Table II). These low

values are in part due to the sealing of the doors as described in section 2.2.1. Additional air exchange rate measurements were performed in some of the rooms with the doors closed but not taped. These measurements are 1.2-3.0 times higher, but always lower than 0.5 h-1. It is clear that such low values of air exchange rate can lead to an accumulation of indoor pollutants and is likely one of the reasons of elevated concentrations of formaldehyde.

As shown in table II, formaldehyde concentrations measured during this study range from 40-350 µg/m3 and are 22-116 times higher than outdoor concentrations. While these elevated concentrations are partly due to low air exchange rates, these values also confirm the presence of formaldehyde emitters.

Emission sources characterized in each school are detailed in table III and IV for the 2011 and 2012 campaigns respectively. Measured emission rates range from 1.2-252 µg/m2/h and indicate the presence of several strong emitters (Ti > 50 µg/m2/h). However, some strong emitters are not reported in Tables III and IV for the reason that the surface areas covered by these material are too small to lead to significant contributions to the formaldehyde budget. Less than 3 % of the materials exhibit emission rates below detection limit, except for schools 4 and 5 where this fraction increases to 30 and 12 % respectively.

The contribution of each material to the total emission rate was calculated for each site using Eq. 2 (Tables III and IV). It should be noted that formaldehyde emissions usually exhibit strong dependences with ambient temperature and emission rates reported in this study are thus representative of environmental conditions experienced at the time of the measurements. It is not possible from this study to assess whether the contribution of each emitter changes with temperature. However, we believe that their contribution will not change drastically within the range of temperature observed in these environments.

The main source of emission was the ceiling for each site, with a contribution close to or higher than 50% for five of the eight sites. The ceilings were made of concrete, mineral fiber tiles, painted plaster tiles, melamine foam, and chipboard. The ceilings were not always made of the most emissive materials but the coverage of large surface areas enhanced the contribution of these materials to the formaldehyde budget. Other individual sources contribute less than 10 % of the total emission rate, except for schools 3 and 6. However, it is interesting to note that small emitters can contribute to 20-32% of the total emission when grouped together (see ‘rest of sources’ in Tables III and IV).

Potential reasons that could explain significant emissions from each material used to build the ceilings are discussed below. Chipboards are the main emitters for

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

schools 2 and 6. These wood products are known to be strong formaldehyde emitters since it is an engineered material containing a formaldehyde based resin [35,36]. At the request of the school manager, a sample of the ceiling from school 2 was also analyzed using the chamber method (ISO 16000-9). An emission rate of 247 µg/m2/h was measured under normalized conditions (23°C, 50% RH), consistent with in-situ PFS measurements and confirming that this material is a strong emitter.

The main source of formaldehyde for school 7 is an alveolar urea-formaldehyde foam. This foam is also one of the main sources of emission for school 3 (decorative items). This type of foam is expected to be a strong emitter since formaldehyde is part of its composition.

It is surprising that ceilings made of concrete (school 3), mineral fiber tiles (schools 4-5), or painted plaster tiles (school 1) are amongst the main sources of emission at several sites. To our knowledge, these materials have not yet been reported as potential formaldehyde emitters. However, some formaldehyde based resins such as sulfonated melamine-formaldehyde can be added to the concrete, the paint, or the plaster to act as a superplasticizer. Such admixture may lead to significant formaldehyde emissions. Unfortunately, no information was available about these materials to verify this hypothesis. Adsorption of formaldehyde onto these materials and its later re-emission could also explain larger-than-expected emission rates. It is interesting to note that highly emissive decorative items attached to the celling in school 3 (116-187 µg/m2/h) may have led to its contamination. Finally, a third possibility may involve secondary formation of formaldehyde from the oxidation of adsorbed VOCs. Chamber studies showed that formaldehyde emissions may be enhanced when a material such as plaster is exposed to ozone [37]. However, O3-induced formaldehyde emissions are unlikely since O3 concentrations are low in indoor air.

Emission sources were grouped into two classes to make a distinction between building and furnishing materials. The foam covering the ceiling in school 7 was classified as furnishings because this material was added in the classroom to reduce ambient noise levels, similar to the decorative items in school 3. According to figures 2 and 3, more than 60% of the emissions come from building materials for half of the classrooms. The other half is composed of three classrooms where both building and furnishing materials contribute to approximately 50% of the total emission, and one classroom where furnishing materials are more emissive than building materials. For the latter, the foam covering the ceiling contributes to half of the formaldehyde emissions.

In most cases, changing furnishings will only have a small impact on formaldehyde concentrations. Since the total emission rate is mainly driven by a couple of emitters, the removal of these emitters or their replacement by less emissive materials should lead to a significant decrease of formaldehyde concentrations. Impacts of strategies of formaldehyde reduction based on these actions are discussed in section 3.4.

3.3. Comparison of total rates of formaldehyde emission Total emission rates calculated for each site from individual measurements of

material emissions are reported in Table V. These values were calculated from ΣTixSi, Ti and Si being the measured emission rate and surface area respectively for

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

the ith identified material. Table V indicates that total emission rates span more than an order of magnitude with values ranging from 1611-21482 µg.h-1.

Total emission rates were also calculated from measured formaldehyde concentrations and air exchange rates using Eq. 4 and are reported in Table V. A scatter plot of measured (PFS) vs. estimated (Eq. 4) total emission rates is shown in Figure 4. This figure shows that total emissions derived from PFS measurements are systematically higher by a factor 1.4-10 than values calculated from Eq. 4. Two reasons could explain this disagreement: (1) PFS measurements may not be representative of ambient emissions and (2) an indoor sink of formaldehyde may be missing from the chemical mass balance equation (Eq. 3). These two possibilities are discussed below.

Ambient emissions from solid materials are driven by the first Fick’s law and are dependent on the bulk air concentration of formaldehyde. Emissions also likely depend on ambient relative humidity as shown in some studies [38]. In contrast, emission rates derived from PFS measurements depend on an experimental calibration curve that was built using an emission chamber under normalized conditions of temperature (23°C) and relative humid ity (50%) [23]. In addition, the loading factor of 0.6 m2/m3 used during calibration controlled the formaldehyde concentration inside the chamber, and as a consequence, the material emission rate. Therefore, emission rates provided by PFS measurements may not be representative of ambient emissions.

To test the representativity of PFS measurements, additional experiments were performed using a piece of chipboard inside an emission chamber. Formaldehyde emissions were measured under various loading factors (0.3-1.7 m2/m3), characteristic of a concentration change in the chamber ranging from 60 to 160 µg/m3. These values are on the same order of magnitude than concentrations measured during the two campaigns (Table II). The chamber measurements indicate a decrease by a factor 2.5 of the emission rate when the formaldehyde concentration increases from 60 to 160 µg/m3. As a consequence, we estimate that emission rates derived from PFS measurements may be 2-3 times different than real ambient emissions. The differences observed between measured and calculated total emission rates for schools 5 and 6 are in the range 4-10 and cannot be only explained by errors associated to PFS measurements.

It is worth noting that this misrepresentation of ambient emissions is not specific to PFS and also occurs for emission test chambers and FLEC apparatus. However, although the PFS measurements may not be representative of ambient emissions, these measurements can still be used to intercompare different materials with the objective to pinpoint the strongest emitters.

A missing sink of formaldehyde in indoor environments could also explain part of the disagreement observed between measured and calculated total emissions. It is interesting to note that this disagreement was found to correlate with air exchange rates as shown in Figure 5. This observation points towards an underestimation of the total loss rate of formaldehyde, which is solely due to air exchange in Eq. 4. An additional sink of formaldehyde was reported by Traynor et al. [39] from a chamber experiment, with a first order loss rate of 0.40±0.24 h-1. More recently, Blondel [40] reported a first order loss rate of 0.34 h-1 from experimental determinations made in real indoor environments. This first order loss rate was determined by comparing

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

simultaneous temporal decays of CO2 and formaldehyde in indoor environments where CO2 and formaldehyde were injected at the center of a room.

The missing first order loss rate k was tentatively parameterized as a deposition process using Eq. 6, where v is the formaldehyde deposition velocity (m/h), S the total surface area covered by materials (m²) and V the room volume (m3).

V

Sk ×= ν Eq. 6

Nazaroff et al. [41] reported a deposition velocity of 0.18±0.11 m.h-1 based on the determination of k made by Traynor et al.[39]. Using a first order loss rate of 0.34 h-1 together with values of S and V measured by Blondel [40], a value of 0.105 m.h-1 was derived for the deposition velocity, consistent with the determination of Nazaroff et al. [41]. This value was used to calculate k for each room using values of S and V measured in this study.

Including this additional sink of formaldehyde into Eq. 4 leads to replace “a”, the air exchange rate, by “(a+k)”. Total emissions calculated using v = 0.105 m/h are reported in Table V and Figures 4-5. Inspection of Figure 4 shows that the agreement with values of total emission derived from PFS measurements is greatly improved. The differences are within 36 % of the PFS measurements. In addition, a close inspection of Figure 5 shows that there is no trend anymore when the ratio of total emission rates is plotted as a function of air exchange rates. These results suggest that an additional sink of formaldehyde has to be included in the chemical mass balance equation to reconcile emissions measured by PFS with measured concentrations of formaldehyde.

Adding an additional sink in Eq. 4 leads to a large increase of calculated total emission rates (up to a factor 12), with a larger impact on schools exhibiting lower air exchange rates such as schools 5 and 6. It is clear that omitting loss processes other than air exchange in indoor environments may lead to significant uncertainties in the use of Eq. 4 to derive total emission rates of VOCs.

It is likely that both, a misrepresentation of ambient emissions by PFS measurements and a missing sink of formaldehyde in Eq. 4, contribute to the disagreement observed in Figure 4. However, it is not possible from this study to determine the extent of their contributions. Studies are needed to quantify indoor sinks of formaldehyde under different environmental conditions and to determine to what extent PFS measurements are representative of ambient emissions.

3.4. Evaluation of strategies of formaldehyde reduction Equation 5 was used to evaluate the impact of strategies of formaldehyde

reduction based on two scenarios: (1) an increase of the air exchange rate to 1 h-1 (scenario I, a* > a), and (2) the removal of the main source of emission (when possible) or its replacement using a less emissive material (scenario II, Q* < Q). An emission rate of 12.5 µg/m2/h was chosen for the replacement material, which

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

corresponds to the upper limit allowed for a ceiling to get the best grade (A+) of the French labeling system concerning building material emissions [42].

Formaldehyde abatements calculated from Eq. 5 are shown in Figures 6 and 7. A formaldehyde reduction of 78-97 % could be achieved by increasing the air exchange rate to 1 h-1. It is worth noting that increasing the air exchange rate would also lower concentrations of other VOCs and would help improving indoor air quality. The replacement of the main sources has a lower impact on formaldehyde concentrations with an abatement of 8-67 %. A combination of these two scenarios would allow a reduction of formaldehyde concentrations in the range 87-98%.

In order to check whether including an additional sink of formaldehyde into Eq. 5 would impact these conclusions, calculations of abatement values were also performed by replacing a by (a+k) and a* by (a*+k). Figure 6 shows that the impact is small and similar values of abatement are derived with and without an additional sink. To be conservative, the lower values of abatement have been used to calculate expected formaldehyde concentrations from scenarios 1 and 2.

Initial concentrations of formaldehyde shown in Figures 6 and 7 are values reported in Table I from the 2009-2011 national field campaign that are representative of a normal use of the rooms. These figures show that scenario I would lead to a concentration drop that is large enough for all schools to exhibit concentrations lower than 30 µg.m-3 (upcoming French regulation value). Scenario II could be used in schools exhibiting only one main source of emission to reach a formaldehyde concentration lower than 30 µg.m-3, i.e. schools 1-2 and 7-8. Applying both scenarios would lead to a formaldehyde concentration close to or lower than 10 µg.m-3 in all classrooms.

Note that abatements of formaldehyde concentrations have been calculated using values of air exchange rates that were measured during this study. As indicated in section 2.2.1, these values are not representative of a normal use of the classrooms and may underestimate real air exchange rates. As a consequence, abatement values calculated for scenario 1 may be an upper limit of what could be achieved. However, these higher air exchange rates are in part due to air exchange between adjacent rooms that are also contaminated by formaldehyde emissions. This additional exchange of air is therefore not expected to help lowering formaldehyde concentrations in the investigated classrooms. As a consequence, the use of air exchange rate values displayed in Table II seems more suitable to calculate abatement values of formaldehyde.

4. Conclusion A methodology relying on the use of a passive flux sampler was applied in 8

polluted classrooms to apportion formaldehyde sources from building and furnishing materials. These results were used to evaluate the relevance of indoor air improvement strategies.

PFS performances were found to be similar during on-site and laboratory measurements, with limits of detection in the range 0.6-3.0 µg.m-2.h-1 and a repeatability of approximately 8% at 100 µg/m2/h. This study confirmed that the PFS is a convenient and powerful tool to perform source apportionment studies of formaldehyde emissions in indoor environments.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

The ceiling was identified as the main contributor to the formaldehyde budget at each site, although this surface was made of different materials such as concrete, mineral fiber tiles, plaster tiles, and chipboard. An important contribution from the ceiling is in part due to the large surface area that is covered. In addition, moderate to elevated rates of formaldehyde emission were measured for materials used to build the ceilings. Significant emissions could be the result of the use of formaldehyde-based resins in the conception of these materials, formaldehyde sorption onto the materials and its later re-emission, or secondary emissions due to the oxidation of VOCs adsorbed at the material surface.

A comparison of total emission rates derived from (i) PFS measurements and (ii) measured concentrations of formaldehyde and air exchange rates indicate differences in the range 1.4-10, depending on the site. This disagreement is likely due to a misrepresentation of ambient emissions by PFS measurements and a missing sink of formaldehyde in the mass balance equation used to calculate total emissions from measured formaldehyde concentrations. Additional studies are needed to determine to what extent PFS, FLEC, and emission test chamber measurements are representative of ambient emissions and to quantify potential unidentified sinks of formaldehyde in indoor environments.

A mass balance model indicates that formaldehyde concentrations could be reduced by 87-98 % by removing the main source of emission and by increasing the air exchange rate to 1 h-1. These abatement values are large enough to reduce formaldehyde concentrations in the investigated classrooms below the upcoming French regulation value of 30 µg/m3 for long term exposure.

Acknowledgments

This work was funded by Tera Environment (Ph.D of G. Poulhet) and by the French ministry of ecology, sustainable development, transport and housing through the French central laboratory for air quality monitoring (LCSQA). The authors want to thank Dr. Hervé Plaisance for his contribution to the initial stage of this study and Isabelle Fronval for samples analysis. The authors are also grateful to the cities involved in this study and to the French associations for air quality monitoring (AASQA) for their contributions during on-site measurements.

References [1] Brasche S, Bischof W. Daily time spent indoors in German homes--baseline

data for the assessment of indoor exposure of German occupants. International Journal of Hygiene and Environmental Health 2005;208:247–53.

[2] Leech JA, Nelson WC, Burnett RT, Aaron S, Raizenne ME. It’s about time: a comparison of Canadian and American time-activity patterns. Journal of Exposure Analysis and Environmental Epidemiology 2002;12:427–32.

[3] Office of Radiation and Indoor Air, Consumer Product Safety Commission. The Inside Story: A Guide to Indoor Air Quality. U.S. EPA/Office of Air and Radiation, EPA 402-K-93-007 2013.

[4] Annesi-Maesano I, Hulin M, Lavaud F, Raherison C, Kopferschmitt C, de Blay F, et al. Poor air quality in classrooms related to asthma and rhinitis in primary schoolchildren of the French 6 Cities Study. Thorax 2012;67:682–8.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

[5] IARC. A review of human carcinogens: chemical agents and related occupations, monographs on the evaluation of carcinogenic risks to humans, 100F 2012.

[6] Hodgson AT, Levin H. Volatile organic compounds in indoor air: A review of concentrations measured in North America since 1990 - LBNL-51715 2003.

[7] Jia C, Batterman S, Godwin C. VOCs in industrial, urban and suburban neighborhoods, Part 1: Indoor and outdoor concentrations, variation, and risk drivers. Atmospheric Environment 2008;42:2083–100.

[8] Salthammer T, Mentese S, Marutzky R. Formaldehyde in the Indoor Environment. Chemical Reviews 2010;110:2536–72.

[9] OQAI. National survey, Indoor air quality in French dwellings: Final report 2006.

[10] Journal officiel de la république française n°0281. Décret n° 2011-1727 du 2 décembre 2011 relatif aux valeurs-guides pour l’air intérieur pour le formaldéhyde et le benzène 2011.

[11] Sakai K, Norbäck D, Mi Y, Shibata E, Kamijima M, Yamada T, et al. A comparison of indoor air pollutants in Japan and Sweden: formaldehyde, nitrogen dioxide, and chlorinated volatile organic compounds. Environmental Research 2004;94:75–85.

[12] Gilbert NL, Guay M, David Miller J, Judek S, Chan CC, Dales RE. Levels and determinants of formaldehyde, acetaldehyde, and acrolein in residential indoor air in Prince Edward Island, Canada. Environmental Research 2005;99:11–7.

[13] Brooks BO, Utter GM, DeBroy JA, Schimke RD. Indoor Air Pollution: An Edifice Complex. Clinical Toxicology 1991;29:315–74.

[14] Wolkoff P, Clausen PA, Nielsen PA, Gunnarsen L. Documentation Of Field And Laboratory Emission Cell “FLEC”: Identification Of Emission Processes From Carpet, Linoleum, Paint, And Sealant By Modeling. Indoor Air 1993;3:291–7.

[15] Wolkoff P, Clausen PA, Nielsen PA. Application of the Field and Laboratory Emission Cell “FLEC” - Performance Study, Intercomparison Study, and Case Study of Damaged Linoleum in an Office. Indoor Air 1995;5:196–203.

[16] Uhde E, Borgschulte A, Salthammer T. Characterization of the field and laboratory emission cell—FLEC: Flow field and air velocities. Atmospheric Environment 1998;32:773–81.

[17] Risholm-Sundman M. Determination of Formaldehyde Emission with Field and Laboratory Emission Cell (FLEC) - Recovery and Correlation to the Chamber Method. Indoor Air 1999;9:268–72.

[18] Lundgren B, Jonsson B, Ek-Olausson B. Materials Emission of Chemicals - PVC Flooring Materials. Indoor Air 1999;9:202–8.

[19] Zhang LZ, Niu JL. Effects of substrate parameters on the emissions of volatile organic compounds from wet coating materials. Building and Environment 2003;38:939–46.

[20] Wilke O, Jann O, Brodner D. VOC and SVOC emissions from adhesives, floor coverings and complete floor structures. Indoor Air 2004;14:98–107.

[21] Jarnstrom H, Saarela K, Kalliokoski P, Pasanen A. Reference values for structure emissions measured on site in new residential buildings in Finland. Atmospheric Environment 2007;41:2290–302.

[22] Marć M, Zabiegała B, Namieśnik J. Testing and sampling devices for monitoring volatile and semi-volatile organic compounds in indoor air. Trends in Analytical Chemistry 2012;32:76–86.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

[23] Blondel A, Plaisance H. Validation of a passive flux sampler for on-site measurement of formaldehyde emission rates from building and furnishing materials. Analytical Methods 2010;2:2032.

[24] Shinohara N, Fujii M, Yamasaki A, Yanagisawa Y. Passive flux sampler for measurement of formaldehyde emission rates. Atmospheric Environment 2007;41:4018–28.

[25] Shinohara N, Kajiwara T, Ohnishi M, Kodama K, Yanagisawa Y. Passive emission colorimetric sensor (PECS) for measuring emission rates of formaldehyde based on an enzymatic reaction and reflectance photometry. Environmental Science & Technology 2008;42:4472–7.

[26] Shinohara N, Kai Y, Mizukoshi A, Fujii M, Kumagai K, Okuizumi Y, et al. On-site passive flux sampler measurement of emission rates of carbonyls and VOCs from multiple indoor sources. Building and Environment 2009;44:859–63.

[27] Fujii M, Shinohara N, Lim A, Otake T, Kumagai K, Yanagisawa Y. A study on emission of phthalate esters from plastic materials using a passive flux sampler. Atmospheric Environment 2003;37:5495–504.

[28] Blondel A, Plaisance H. Screening of formaldehyde indoor sources and quantification of their emission using a passive sampler. Building and Environment 2011;46:1284–91.

[29] Michelot N, Mandin C, Ramalho O, Riberon J, Marchand C, Malherbe L, et al. Campagne pilote de surveillance de la qualité de l’air dans les écoles et crèches en France : Résultats de la première phase. Pollution Atmosphérique 2011:267–79.

[30] Michelot N, Marchand C, Ramalho O, Delmas V, Carrega M. Monitoring indoor air quality in French schools and day-care centres. Results from the first phase of a pilot survey. Healthy Buildings 2012, 10th International Conference, hal-00747458, 2012.

[31] ASTM E 741-00. Standard test method for determining air change in a single zone by means of tracer gas dilution. 2006.

[32] ISO 16000-3. Indoor air - Determination of formaldehyde and other carbonyl compounds in indoor air and test chamber air - Part 3 : Active sampling method. 2002.

[33] Lewis CW, Zweidinger RB. Apportionment of residential indoor aerosol, voc and aldehyde species to indoor and outdoor sources, and their source strengths. Atmospheric Environment Part A General Topics 1992;26:2179–84.

[34] Liu W, Zhang J, Zhang L, Turpin B, Weisel C, Morandi M, et al. Estimating contributions of indoor and outdoor sources to indoor carbonyl concentrations in three urban areas of the United States. Atmospheric Environment 2006;40:2202–14.

[35] Dunky M. Urea–formaldehyde (UF) adhesive resins for wood. International Journal of Adhesion and Adhesives 1998;18:95–107.

[36] Brown SK. Chamber Assessment of Formaldehyde and VOC Emissions from Wood-Based Panels. Indoor Air 1999;9:209–15.

[37] Moriske HJ, Ebert G, Konieczny L, Menk G, Schöndube M. Concentrations and decay rates of ozone in indoor air in dependence on building and surface materials. Toxicology Letters 1998;96-97:319–23.

[38] Baughman A, Arens E. Indoor Humidity and Human Health--Part I: Literature Review of Health Effects of Humidity-Influenced Indoor Pollutants. ASHRAE Transactions 1996;102:193–211.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

[39] Traynor GW, Anthon DW, Hollowell CD. Technique for determining pollutant emissions from a gas-fired range. Atmospheric Environment 1982;16:2979–87.

[40] Blondel A. Identification des sources intérieures de Composés Organiques Volatils et estimation de leur contribution aux teneurs observées dans des logements français. Thèse de l’Université de Lille, 2010.

[41] Nazaroff WW, Cass GR. Mathematical modeling of chemically reactive pollutants in indoor air. Environmental Science & Technology 1986;20:924–34.

[42] Journal officiel de la république française n°0111. Arrêté du 19 avril 2011 relatif à l’étiquetage des produits de construction ou de revêtement de mur ou de sol et des peintures et vernis sur leurs émissions de polluants volatils 2011.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Figure captions Figure 1: Scatter plot of replicates. PFS #1 and #2 are two measurements performed on the same material. Open and solid symbols represent replicates performed in 2011 and 2012 respectively. The solid line is the 1:1 line. Dashed lines are upper and lower bounds of the 3σ precision expected from laboratory testing Figure 2: Breakdown of emission sources into building and furnishing materials – 2011 field campaign. Some values displayed in these figures may be different than contribution values reported in Table III. This difference is due to the contribution of several items to a specific class. For example, the ceiling for School 1 is the sum of ceiling 1, ceiling 2, and wooden beams from Table III. Figure 3: Breakdown of emission sources into building and furnishing materials – 2012 field campaign. Some values displayed in these figures may be different than contribution values reported in Table III. This difference is due to the contribution of several items to a specific class. Figure 4: Scatter plot of total emissions. The solid curve represents the 1:1 line. Solid diamonds are total emission rates calculated using Eq. 4 without an additional sink of formaldehyde. Open diamonds are total emission rates calculated with an additional sink of formaldehyde (see text). Figure 5: Disagreement observed between total emission rates of formaldehyde measured by PFS and calculated from measured concentrations of formaldehyde (Eq. 4). Empty and solid symbols are values calculated with and without an additional sink of formaldehyde respectively (see text). The red curve indicates a ratio of 1. Figure 6: Impact of formaldehyde reduction strategies - 2011 field campaign. Abatement values calculated by Eq. 5 are shown as bars while abatement values accounting for an additional loss of formaldehyde (see text) are shown using dashed lines. Light grey bars are abatement values calculated by increasing the air exchange rate to 1 h-1. White bars are abatement values calculated after the removal or the replacement of the main source of emission (see sections 2.2.4 and 3.4). Dark grey bars are abatement values calculated after the removal or the replacement of the main source of emission and by increasing the air exchange rate to 1h-1. Values displayed under the series name are formaldehyde concentration (µg/m3) measured during the 2009-2011 French national field campaign (Table I). Values displayed above each bar are concentrations (µg/m3) calculated after abatement. Figure 7: Impact of formaldehyde reduction strategies - 2012 field campaign. Abatement values calculated by Eq.5 are shown as bars while abatement values accounting for an additional loss of formaldehyde (see text) are shown using dashed lines. Light grey bars are abatement values calculated by increasing the air exchange rate to 1h-1. White bars are abatement values calculated after the removal or the replacement of the main source of emission (see sections 2.2.4 and 3.4). Dark grey bars are abatement values calculated after the removal or the replacement of the main source of emission and by increasing the air exchange rate to 1h-1. Values displayed under the series name are formaldehyde concentrations (µg/m3) measured

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

during the 2009-2011 French national field campaign (Table I). Values displayed above each bar are concentrations (µg/m3) calculated after abatement.

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table captions Table I: Description of the measurement sites. Table II: Measurements of environmental parameters. Each measurement was averaged over 6 hours. Table III: Formaldehyde emissions measured during the 2011 campaign. Table IV: Formaldehyde emissions measured during the 2012 campaign. Table V: Total emissions inferred from Eq. 4 and measured emission rates (µg/h)

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table I:

School # Sampling date

Type of school

Type of area

Formaldehyde concentration a

(µg/m3)

Number of identified sources

1 periurban 54.5 36 2 rural 59.4 39 3 periurban 52.7 45 4

June-July 2011

industrial 42.6 45 5 urban 46.4 39 6

primary

rural 52.4 42 7 day-care urban 46.6 29 8

May 2012

primary urban 43.3 31 a Concentrations of formaldehyde measured during the 2009-2011 national campaign

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table II :

School # Temperature (°C)

Relative Humidity

(%)

Air exchange rate * (h-1)

Indoor formaldehyde concentration *

(µg/m3)

Outdoor formaldehyde concentration

(µg/m3)

1 21.4 70.6 0.14 208 1.9 2 29.6 63.6 0.22 350 6.7 3 28.0 53.9 0.19 156 7.1 4 24.2 57.5 0.12 64 1.4 5 18.6 50.0 0.04 42 1.4 6 14.4 67.7 0.03 115 1.2 7 19.6 51.7 0.13 79 1.7 8 19.2 52.4 0.14 62 4.6

* These values should be interpreted with caution and do not represent individual exposure under normal use of the rooms (see text).

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table III:

Material Replicates Formaldehyde emission rate (Ti in µg/m2/h)

Si (m²)

Tot. emission (Si×Ti in µg/h)

Contribution to total emission

(CSi, Eq. 2)

School #1

Detection limit 3.0 Ceiling #1 (painted plaster tiles)* 1 114 36.9 4207 52 % Floor (linoleum) 2 14 49.8 697 9 % Heaters (painted metal) 1 12 31.5 378 5 % Ceiling #2 1 10 25.3 253 3 % Wall #1 1 23 10.8 248 3 % Wall #2 1 19 11.6 220 3 % Wall #3 1 28 7.1 199 2 % Rest of sources (36/43) N/A N/A 1847 23 %

School #2

Detection limit 1.2 Ceiling (chipboard)* 3 252 58.3 14692 69 % Furniture #1 (painted plywood) 1 45 18.7 842 4 % Floor 2 16 47.3 757 4 % Wall 4 13 53.1 690 3 % Chairs 1 66 6.6 436 2 % Furniture #2 1 38 8.9 338 2 % Bookshelf 1 43 7.1 305 1 % Rest of sources (33/40) N/A N/A 3189 15 %

School #3

Detection Limit 1.4 Ceiling (concrete + roughcast) 2 44 83.4 3670 32 % Deco items #1 (foam)** 2 116 22.0 2552 22 % Deco items #2 (foam)** 2 187 6.0 1122 10 % Floor 1 13 60.4 785 7 % Wall #1 2 14 34.0 476 4 % Wall #2 2 18 20.7 373 3 % Wood panels 1 82 3.4 279 2 % Rest of sources (31/38) N/A N/A 2264 20 %

School #4

Detection limit 7.0 Ceiling (mineral fiber tiles)* 2 25 56.8 1420 38 % Cabinet (melamine-chipboard) 1 40 8.5 340 9 % Rack #1 1 78 2.4 187 5 % Tables 1 16 11.0 176 5 % Mobile Wall 1 93 1.8 167 4 % Wall 1 11 14.2 156 4 % Rack #2 1 9 13.8 124 3 % Rest of sources (24/31) N/A N/A 1191 32 %

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Only 7 materials ranked by values of total emission are reported. An average emission rate was calculated when several replicates were available. *replaced by a less emissive material or **removed from the room to evaluate strategies of formaldehyde reduction (see section 3.4).

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Material Replicates Formaldehyde emission rate (Ti in µg/m2/h)

Si (m²)

Tot. emission (Si×Ti in µg/h)

Contribution to total emission

(CSi, Eq. 2)

School #5

Detection limit 1.2 Ceiling (mineral fiber tiles)* 2 14 80.0 1120 43 %

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table IV:

Rack (painted plywood) 3 13 15.8 205 8 % Wall #1 2 7 23.5 165 6 % Furniture #1 1 14 11.0 154 6 % Furniture #2 1 15 10.1 152 6 % Furniture #3 1 16 8.3 133 5 % Wall #2 2 3 31.1 93 4 % Rest of sources (27/34) N/A N/A 580 22%

School #6

Detection limit 2.3 Ceiling #1 (chipboard)* 2 70 22.8 1596 27 % Ceiling #2 (mineral fiber tiles) 2 51 21.2 1081 18 % Furniture #1 1 21 26.6 559 9 % Ceiling #3 2 18 28.6 515 9 % Wall 1 6 51.5 309 5 % Floor 2 6 51.2 307 5 % Wooden chairs 1 39 7.1 277 5 % Rest of sources (32/39) N/A N/A 1314 22 %

School #7

Detection limit 1.0 Melamine foam on ceiling** 2 63 33.6 2117 51 % Ceiling 2 8 48.3 386 9 % Wall 3 6 51.7 310 7 % Blanket 1 20 12.5 250 6 % Floor 2 5 49.4 247 6 % Wooden piece of furniture 1 48 4.2 202 5 % Partitions 2 6 31.3 188 4 % Rest of sources (20/27) N/A N/A 488 12 %

School #8

Detection limit 0.6 Ceiling (mineral fiber tiles)* 2 14.8 55.6 822.9 52 % wall #1 (wood) 1 4.6 23.0 105.8 7 % Wooden board #1 2 10.6 7.0 74.2 5 % Wooden board #2 2 4.5 15.6 70.2 4 % Floor 2 1.2 55.6 66.7 4 % Chair 1 15.2 3.5 53.2 3 % Foam board 1 2.6 16.4 42.6 3 % Rest of sources (7/14) N/A N/A 351 22 %

Only 7 materials ranked by values of total emission are reported. An average emission rate was calculated when several replicates were available. *replaced by a less emissive material or **removed from the room to evaluate strategies of formaldehyde reduction (see section 3.4).

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

Table V:

School # Total emission

from PFS (µg/h)

Total emission from Eq. 4

(µg/h)

k from Eq. 6 (m/h)

Total emission from Eq. 4, replacing a by (a+k)

(µg/h)

1 7972 3922 0.20 9744 2 21482 15304 0.18 27852 3 11864 5310 0.19 10842 4 4322 1107 0.22 3172 5 1916 480 0.20 3011 6 6501 646 0.34 7566 7 3922 2325 0.16 5251 8 1611 850 0.21 2234

k was calculated using v = 0.105 m/h (see text)

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

MANUSCRIP

T

ACCEPTED

ACCEPTED MANUSCRIPT

• Methodology developed to apportion formaldehyde emissions from building and furnishing materials in indoor environments.

• Identification of the main sources of emission in 8 public schools. • Evaluation of formaldehyde reduction strategies using a mass balance model.

• An unknown sink of formaldehyde may be operating in indoor environments.