hierarchical cluster analysis of carbonyl compounds emission profiles from building and furniture...

lable at ScienceDirect

Building and Environment 75 (2014) 40e45

Contents lists avai

Building and Environment

journal homepage: www.elsevier .com/locate/bui ldenv

Technical note

Hierarchical cluster analysis of carbonyl compounds emission profilesfrom building and furniture materials

H. Plaisance a,b,c,*, A. Blondel a,b, V. Desauziers c, P. Mocho d

aUniversité Lille Nord de France, F-59000 Lille, Franceb Ecole des Mines de Douai, Département Chimie et Environnement, 941 rue Charles Bourseul, BP 10838, 59508 Douai Cedex, Francec Pôle RIME C2MA, Ecole des Mines d’Alès, Hélioparc, 2 Avenue Pierre Angot, 64053 Pau Cedex 9, Franced Laboratoire Thermique Energétique et Procédés, Université de Pau et des Pays de l’Adour, BP, Pau 7511-64075, France

a r t i c l e i n f o

Article history:Received 13 November 2013Received in revised form21 January 2014Accepted 23 January 2014

Keywords:Indoor materialsVolatile organic compoundsFormaldehydeEmissionEmission test chamberMultivariate analysis

* Corresponding author. Pôle RIME C2MA, Ecole dAvenue Pierre Angot, 64053 Pau Cedex 9, France. Tel.

E-mail address: herveplaisance.dubois@sfr.fr (H. P

0360-1323/$ e see front matter � 2014 Elsevier Ltd.http://dx.doi.org/10.1016/j.buildenv.2014.01.014

a b s t r a c t

Emission profiles of carbonyl compounds from twenty-three indoor materials were assessed by chambertests and compared by means of a hierarchical cluster analysis. This multivariate method provides apartition into six clusters of materials having statistically similar chemical profiles. Formaldehyde is themost dominant component of emissions mainly related to two types of wood composite products(chipboards and medium-density fibreboards (MDF)) and one finishing plaster. The analysis of clustersreveals that the emission profiles of materials belonging to a same category can have various degrees ofvariability. Some common pressed-wood products as chipboards and medium-density fibreboards haverelatively uniform profiles characterized by its unique emission of formaldehyde. On the contrary, theprofiles of Oriented Strand Boards (OSB) and finishing plasters appear very heterogeneous and unspecificin terms of relative dominance between different carbonyl compounds. The finishing plasters areidentified as sources of carbonyl compounds (formaldehyde and acetaldehyde, especially). These fin-ishing products have not yet been listed as potential formaldehyde and acetaldehyde emitters. Accordingto these results, the wood composite products can also be ranked in the decreasing order of formalde-hyde emission as follows: Chipboards > MDFs > Plywoods > OSBs. In light of these results, more sys-tematic surveillance program on the emissions from materials should be set up by Public Health servicesto require or request product changes for building and furnishing applications.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Among indoor air pollutants, carbonyl compounds are ofparticular interest due to their abundance in indoor air and to theiradverse effects on health [1,2]. In France, a national survey wascarried out during the 2003e2005 period by the OQAI (French In-door Air Quality Observatory) in 554 dwellings designed to berepresentative of the 24 million French dwellings [3]. It provides afull overview of the state of indoor air quality at the national leveland shows that three aldehydes: formaldehyde (median concen-tration: 19.6 mg m�3), hexanal (median concentration: 13.6 mg m�3)and acetaldehyde (median concentration: 11.6 mg m�3) are amongthe most abundant identified volatile organic compounds. Forthese compounds, ratios between indoor and outdoor concentra-tions largely exceed 1 (10 for formaldehyde, 27 for hexanal and 9 for

es Mines d’Alès, Hélioparc, 2: þ33 6 80 13 42 42.laisance).

All rights reserved.

acetaldehyde) indicating the majority part of the contaminationcomes from indoor sources. Analysis of the highest indoor levels offormaldehyde found in previous studies [4,5] pointed out moreoften as sources the wood composite products, especially whenthese materials are new. Recent field studies combining emissionand air concentration on-site measurements confirmed that ma-terial emissions are themajor contributors to indoor contaminationof formaldehyde [6e8].

Moreover, some extensive programs of chamber tests allowed toidentify a set of sources of carbonyl compounds among the mate-rials and household products present indoors like the wood-pressed products for formaldehyde, acetaldehyde and hexanal [9],paints for benzaldehyde and hexanal [10] and cleaning products foracetone [11]. However, the diversity of possible sources and addi-tion of their contribution to indoor concentrations make complexidentifying the main sources effectively responsible for contami-nation in a real indoor environment. Analysis of chemical profiles ofsources including the emission factors of multiple volatile organiccompounds can help to identify the signature of specific sources.Knowledge of these chemical profiles is useful to apply and

H. Plaisance et al. / Building and Environment 75 (2014) 40e45 41

interpret the results of multivariate receptor models like FactorAnalysis [12], Positive Matrix Factorization [13] and Chemical MassBalance [14] which have been used intensively for source appor-tionment in ambient air. In the field of indoor air, the application ofthese receptor models is muchmore limited probably due to lack ofspecific source profiles clearly identified and discriminant.

Moreover, the associations of VOC exposure to ill health havebeen controversially discussed [15,16]. However, some studiesshowed that exposure to certain VOC including formaldehyde atlow concentration levels may already have adverse effects onhealth [17,18].

The compound concentrations as well as the proportion of thesecompounds compared to the total exposure could influence, interms of positive or negative trends, the health relevance of themixture [19]. The proportion of health relevant chemicals in a VOCexposure is a critical parameter to consider for the assessment ofindoor environments and indoor sources.

In this paper, we present a detailed statistical analysis of emis-sions of six carbonyl compounds from a series of building andfurniturematerials. The aim of the paper is threefold: (1) to identifymaterials not yet listed as indoor sources of carbonyl compounds;(2) to compare the chemical profiles of emissions between themaccording to the nature and use of materials; and (3) to look for thespecific characteristics of these profiles of emissions.

2. Materials and methods

2.1. Material samples

Twenty-three materials selected were purchased from a DIYstore and represent a diverse set of product categories available forsale in France that are commonly found in homes. They were notintended to be a statistically representative sampling of thosecategories. The selection process provided that when a materialwas identified as a carbonyl source, additional materials belongingto the same category were also tested. Through this, we try toappreciate the uniformity degree of emission profiles within the

Fig. 1. Emission test

classes of source materials. Sample selection includes both wetproducts: one glue for wallpaper, four furnishing plasters, one sil-icone and one expanding foam, and dry products: two Orientedstrand boards (OSBs), four Medium-density fibreboards (MDFs),two chipboards, two plywoods, one composite board, linoleum, oneceiling tile, one gypsum board and two raw woods (beech andpine).

Drymaterials were cut into sample of 10 cm by 15 cm in order tohave an exposed area of the material of 0.015 m2. For testing, onlyone main face of material (i.e. 0.015 m2) is exposed, the edges andopposite face of sample are covered with aluminium foil.

Wet materials were placed into a petri dish of 12 cm diameterwith a product thickness of 0.5 cm. Thus, the exposed area of wetmaterials was 0.011 m2 under test conditions close to that of drymaterials. After filling of the petri dish, the sample is dried underairflow at ambient temperature until its weight stabilizes. Thedrying time takes several days for the finishing plasters especially.

2.2. Emission test chamber method

Each selected material was tested by the emission test chamberstandard method to assess emission rates of six carbonyl com-pounds (formaldehyde, acetaldehyde, acetone, propanal, benzal-dehyde and hexanal) that constitute the chemical profile ofmaterial. The experimental device set up for these emission tests ispresented in Fig. 1. The emission test chamber is consisted of a glasscylinder with a capacity of 35 L (length: 50 cm and diameter:30 cm) and contains a glass plate separating the lower part devotedto the generation of air movements by means of three axial fansregulated by a potentiometer from the upper part where the ma-terial sample is placed on the glass plate. The cylindrical form of thechamber favours the air recirculation giving a thorough mixing ofair in the emission chamber. The air opening in the emission testchamber is produced by a compressor and is dried and chemicallyfiltered in an air purifier (AZ 2020 manufactured by Claind). A firstair flow is produced directly by the air purifier and a second air flowcomes from a humidifier system consisting of a bubbler filled up

chamber device.

H. Plaisance et al. / Building and Environment 75 (2014) 40e4542

with demineralized water. These two airflows are regulated bymass flow controllers. They are mixed at the chamber inlet,generating various test atmospheres. Ambient conditions in theemission test chamber are continuously monitored and recordedby means of temperature, relative humidity and wind velocitymultifunction probes (Datalogger Testo term 400 and temperature,humidity, wind velocity sensor 0635.1540, Testo, France). Theemission test chamber is placed in a thermostatic enclosuremaintained at constant temperature. Material sample are intro-duced or removed thanks to the opening of a removable cover. Thetemperature was 23 � 2 �C, the relative humidity was 50 � 5% andthe air velocity was maintained below 0.2 m s�1. The ventilationrate was chosen to be close to that found in indoor environments,0.5 h�1. The area of the material ranges from 0.011 to 0.015 m2

which corresponds to a loading factor of 0.32e0.42 m2 m�3

depending on the sample. To avoid contamination and to achieve asteady state of formaldehyde emissions, the material samples wasplaced in a closed environmental chamber and incubated for atleast 12 h before sampling. Then, the concentration of formalde-hyde in the chamber was determined by active sampling usingDNPH-Silica Sep-Pak cartridges (purchased from Waters, Guyan-court, France). The sampling rate was 200 mL min�1 and thesampling duration was 2 h. The used cartridges were sealed with acolour cap and stored in the dark at �20 �C until analysis. Thesecartridges were extracted and analysed by HPLC according to theprotocol described in the following section.

The formaldehyde emission rate was calculated by means of thefollowing equation:

F ¼ C � VS� t

(1)

where F (mgm�2 h�1) is the formaldehyde emission rate, C (mgm�3)the formaldehyde concentration measured in chamber, V (m3) thechamber volume, S (m2) the area of material sample and t (h) thecollection time. In the experiments, V ¼ 0.036 m3, S ¼ 0.011e0.015 m2 and t ¼ 2 h.

The experiments were carry out basing on normalized methods(ISO 16000-9 [20] and ISO 16000-3 [21]) relative to emission testchamber method and active sampling method for the determina-tion of carbonyl compounds.

The chamber is considered as an ideal continuously-stirredreactor operating at quasi-steady state conditions. Losses of com-pounds due to factors other than ventilation rate (i.e., sink effects)are ignored; consequently, the calculated values were net ratesassessed under conditions of temperature, humidity, air velocityand ventilation rate which are intended to represent thoseencountered indoors.

2.3. Analysis

For analysis, the DNPH cartridges are eluted with 3 mL ofacetonitrile (HPLC grade, Waters). Then, the extracted DNPH-formaldehyde derivates are analysed using a high performanceliquid chromatography system (HPLC, Waters 2695, Waters) with aphotodiode array detector at a wavelength of 365 nm. 20 mL ofextraction solution were injected onto a Restek ULTRA C18 column(Restek, France, length 25 cm, inner diameter 4.6 mm, particle size5 mm). The mobile phase is initially a 30% acetonitrile, 20% tetra-hydrofurane and 50% water mixture which progressively evolvefor 15 min to achieve a 65% acetonitrile, 5% tetrahydrofuran and30% water mixture maintained for a further 12 min. The flow rateis maintained at a constant 1.5 mL min�1. The gas-phase concen-tration of FA was quantified from an external calibration curvemade before sampler analysis. The solid standard DNPH derivative

of FA was purchased from Chem-Service. Known concentrations ofhydrazones were then prepared in acetonitrile and injected intothe HPLC in order to set up the calibration curves (HPLC peak areasversus formaldehyde concentrations). The calibration curve waslinear for the hydrazone concentration range between 1 and1000 mg L�1 and the correlation coefficients were larger than 0.99.Before and after each set of analyses, a single point of this cali-bration curve with a known concentration was injected to checkthe accuracy of the HPLC response coefficient (deviation < 5%).

Detection limits of emission rates for the six carbonyl com-pounds were estimated considering the background concentra-tions present in the emission chamber without material sample(blank test) and detection limits of the active sampling method inthe case where the compound is not detected in the blank test.These detection limits were estimated as follows: 3.9 mg m�2 h�1

for formaldehyde; 1.3 mg m�2 h�1 for acetaldehyde; 1 mg m�2 h�1

for acetone; 0.9 mg m�2 h�1 for propanal; 0.65 mg m�2 h�1 forbenzaldehyde and 0.9 mg m�2 h�1 for hexanal.

2.4. Hierarchical clustering analysis

Hierarchical clustering analysis was used to group in classessites having similar chemical VOC profiles. Before the hierarchicalclustering analysis, the concentrations were standardized so thatthe variation extent of each emission rate has an equivalentcontribution to the formation of the classification. To transform thereal emission rate of compound j for material i (Fij) to standardizedvalue (F0ij), the following equation was applied:

F 0ij ¼Fij � FjmaxþFjmin

2Fjmax�Fjmin

2

(2)

where Fjmax and Fjmin are the maximum and minimum of emissionrate of compound j. The standardized emission rates (F0 ij) varybetween �1 and 1.

Every material is described by a series of quantitative variables(standardized emission rates of six compounds) defining a chemi-cal profile. The hierarchical cluster analysis uses an iterative algo-rithm which aggregates at each iteration the two most similarchemical profiles. The similarity between two chemical profiles ismeasured by Euclidean distance. The aggregationmode used in thiscluster analysis is the Ward Algorithm [22].

A tree, also called a dendrogram, is built by iterations,beginning with the initial partition P0 consisting of the 23emission profiles until the last partition P22 where all profiles areaggregated into a single node. The distance between the twoaggregated profiles is attached to each Pi partition. This aggre-gation distance indicates the heterogeneity of Pi. The sequence(P0,...,Pn,..., P22) is then naturally in ascending order from thisaggregation distance. A cut-off of dendrogram is then chosen atone aggregation distance level in order to form several homo-geneous classes with common characteristics clearly marked andinterpretable.

In order to help in the interpretation of the tree obtained byhierarchical clustering analysis, a graph of multivariate data calledthe radar chart is used to represent the mean profile of eachmaterial cluster (group of materials with homogeneous emissionprofiles). Each radar chart consists of six rays drawn from a centrepoint where each ray represents the standardized emission rate ofone compound for one material cluster. The radar chart allows thevisual comparison of the different mean compositions ofemissions.

The PC of Statistica (version 10) is used for hierarchical clus-tering analysis and radar chart.

H. Plaisance et al. / Building and Environment 75 (2014) 40e45 43

3. Results and discussion

Table 1 presents the whole of emissions of six carbonyl com-pounds from 23 materials assessed by the emission chambermethod. The emission rates of target compounds vary greatly be-tween the materials and between the samples of same category.The results indicate that formaldehyde is the most dominantcomponent with a median of 26.7 mgm�2 h�1. The emission rates ofcarbonyl compounds can also be arranged in the following order:formaldehyde > acetaldehyde > acetone > hexanal > benzalde-hyde > propanal. The emission range has a very high magnitudefrom 1 to 1000 mg m�2 h�1 for acetone and from 1 to 400e500 mg m�2 h�1 for formaldehyde and acetaldehyde.

Note that this ranking is consistent with the order of abundanceof carbonyl compounds found in private houses in Europe [23] andin United States [24] suggesting that indoor materials have a majorcontributor to indoor contamination of carbonyl compounds.

In order to compare the emission profiles of 23 indoor materialsand aggregate those having similar profiles, the hierarchical clus-teringmethodwas applied to emission rate data reported in Table 1.Results are presented as a dendrogram in Fig. 2. A cut-off at a dis-tance level of 1.3 gives a distribution of materials in six homoge-neous classes with common characteristics clearly marked andinterpretable.

Cluster 1 includes the emission profiles of various materials(two MDF, one glue, one ceiling tile and one finishing plaster).These profiles are characterized by low to medium levels of form-aldehyde and acetone. Gall el al [25]. also found that emissions ofcarbonyls emitted from ceiling tiles are generally low (total car-bonyls below 50 mg m�2 h�1).

Cluster 2 gathers ten materials for which profiles are marked bythe lowest emission rates for all target compounds. These materialsare diverse in nature and form a class with low impact on indoor airquality if only carbonyl compounds are considered.

A previous test program performed on materials sold in Cali-fornia showed that gypsum boards emit formaldehyde and acetonewith emission rates of 15e19 mg m�2 h�1 and 35e110 mg m�2 h�1,respectively [26], that is in agreement with our results for form-aldehyde and somewhat differs for acetone.

Table 1Emission rates (mg m�2 h�1) of six carbonyl compounds from 23 materials assessed by e

Indoor materials Emission rate (mg m�2 h�1)

Formaldehyde Acetaldehyde Acetone

Glue for wallpaper 5.0 <1.3 92Finishing plaster_1 413 <1.3 <1Finishing plaster_2 36 3.6 241Finishing plaster_3 40 500 135Finishing plaster_4 26.7 291 38OSB_1 21.3 14.7 1071OSB_2 31 6.5 26.8Chipboard_1 203 11.1 <1Chipboard_2 245 3.4 <1Plywood_1 <3.9 32 29.8Plywood_2 8.3 4.8 <1MDF_1 136 3.1 <1MDF_2 255 1.6 3.1MDF_3 250 4.0 28.6MDF_4 92 12.5 382Composite board 4.0 <1.3 3.9Linoleum <3.9 <1.3 <1Silicone <3.9 14.4 12.2Expanding foam <3.9 1.6 <1Pine wood 14.0 57 <1Ceiling tile 99 <1.3 <1Gypsum board 15.5 1.9 14.1Beech wood <3.9 5.4 <1Median (minemax) 26.7 (<3.9e413) 4.0 (<1.3e500) 3.9 (<1e

Cluster 3 includes the emission profile of one finishing plaster.This material is characterized by its unique emission ofacetaldehyde.

Cluster 4 gathers two types of wood composite products(Chipboards and MDFs) and one finishing plaster. Their profilesare marked by the highest formaldehyde emission rates of allmaterials tested. He et al. [27] found that formaldehyde is thedominant carbonyl compound emitted from fresh wood-basedpanels with a range of emission rates from 50 to300 mg m�2 h�1 comparable with the values found in our study forChipboards andMDFs. Que et al. [28] also showed that Chipboardsand MDFs made with urea-formaldehyde adhesives were themost emitters of formaldehyde among 13 common wood com-posite products with emission rates comprised between 120 and470 mg m�2 h�1. Formaldehyde was the majority part of VOCemissions from these materials.

Cluster 5 identifies the profile of one finishing plaster domi-nated by the high emission levels of acetaldehyde andbenzaldehyde.

One OSB forms Cluster 6. Its profile is marked by the highestemission rates of acetone, propanal and hexanal.

This classification reveals that the emission profiles of materialswithin a same category can have various degrees of variability.Chipboards and in a lesser degree MDFs have relatively uniformprofiles dominated by formaldehyde emission. On the contrary, theprofiles of OSBs and finishing plasters are very heterogeneous. Theycan belong to the classes of low emission profiles (Clusters 1 and 2)like OSB_2 and Finishing plaster_2 or have a profile with one, twoor three carbonyl compounds strongly represented like Finishingplaster_4, Finishing plaster_3 and OSB_1.

The formaldehyde emission dominates profiles of wood com-posite products, except OSBs. They can be ranked in the decreasingorder of formaldehyde emission:Chipboards > MDFs > Plywoods > OSBs. The most commonpressed-wood products made with urea-formaldehyde adhesives(e.g. particleboard, hardwood plywood, MDF, and paneling) andthose less prevalent made with phenol-formaldehyde adhesives(e.g. softwood plywood and OSB) are known to release formalde-hyde [29]. Due to the reversibility of the urea-formaldehyde

mission chamber tests.

Propanal Benzaldehyde Hexanal

<0.9 4.6 1.6<0.9 0.84 <0.9<0.9 7.1 1.5<0.9 29.2 1.7<0.9 <0.65 <0.97.1 2.1 95<0.9 <0.65 9.2<0.9 3.1 12.9<0.9 4.4 8.31.6 0.8 4.1<0.9 <0.65 <0.9<0.9 <0.65 1.4<0.9 1.0 1.01.5 0.7 6.8<0.9 0.7 10.1<0.9 1.3 <0.9<0.9 2.1 2.1<0.9 1.9 1.7<0.9 0.7 <0.9<0.9 <0.65 <0.9<0.9 <0.65 <0.9<0.9 <0.65 19.3<0.9 0.8 <0.9

1071) <0.9 (<0.9e7.1) 0.8 (<0.65e29.2) 1.6 (<0.9e95)

Fig. 2. Dendrogram obtained by hierarchical clustering analysis of 23 emission profiles and radar charts representing the mean and maximum profiles of 7 material clusters.

H. Plaisance et al. / Building and Environment 75 (2014) 40e4544

reaction, formaldehyde emission from pressed-wood products maydecrease very slowly and continue for months or even years [30].

Another aspect of these results concerns the finishing plastersthat were so far not listed as major indoor sources of carbonylcompounds (formaldehyde and acetaldehyde, especially). Formal-dehyde was widely used as biocides in water-based paints andfungicidal products but it has now been replaced by other com-pounds such as isothiazolinones [30,31]. In the case of furnishingplasters, carbonyl compounds enter into the composition offurnishing plasters. Released amounts are high, thus, it appearsunlikely that carbonyl compounds only play the role of fungicidesor biocides. They could perhaps be used as solvents to improve theproperties of material: smoothing, drying and/or bleaching offurnishing plasters when they are applied on the walls.

4. Conclusion

Hierarchical cluster analysis appears as an appropriate tool toclassify materials according to their emission profile. This multi-variate method allows to compare the emission profiles of sixcarbonyl compounds associated with twenty-three indoor mate-rials. This multivariate method gives a partition into six clusters ofmaterials having statistically similar chemical profiles. The resultsshow that formaldehyde is the most dominant component ofemissions. The emission rates can also be arranged in thedecreasing order as follows: formaldehyde > acetaldehyde >

acetone> hexanal> benzaldehyde> propanal that is in agreementwith the order of abundance of carbonyl compounds found indoors.The analysis of clusters reveals that the emission profiles of mate-rials belonging to a same category can have various degrees ofvariability. Some common pressed-wood products as chipboardsand MDFs have relatively uniform profiles characterized by itsunique emission of formaldehyde. On the contrary, the profiles ofOSBs and finishing plasters appear very heterogeneous and un-specific. The finishing plasters are identified as sources of carbonylcompounds (formaldehyde and acetaldehyde, especially) whichthat can potentially contribute to the contamination of indoor air.

To our knowledge, the finishing plasters have not yet been reportedas potential formaldehyde and acetaldehyde emitters.

According to these results, the wood composite products canalso be ranked in the decreasing order of formaldehyde emission:Chipboards > MDFs > Plywoods > OSBs.

Note that the scope of this application is limited to samplestested and it would be useful to extend to other materials in orderto deepen and complete the observed trends.

In light of these results, more systematic surveillance programon the emissions from materials and their changes, as investigatedin a limited scale in this study, should be set up by Public Healthservices to limit human health risks associated with carbonylcompounds pollution in indoor environments. This could lead torequire or request product changes for building and furnishingapplications.

References

[1] Wieslander G, Norback D, Bjornsson E, Janson C, Boman G. Asthma and theindoor environment: the significance of emission of formaldehyde and vola-tile organic compounds from newly painted indoor surfaces. Int Arch OccupEnviron Health 1997;69:115e24.

[2] International Agency for Research on Cancer (IARC). Overall evaluations ofcarcinogenicity to humans. In: IARC monographs, vol. 1e96; 2007.

[3] Observatoire de la qualité de l’air intérieur (OQAI). Campagne nationalelogements e etat de la qualité de l’air dans les logements français; 2006. Finalreport.

[4] Yang W, Sohn J, Kim J, Son B, Park J. Indoor air quality investigation accordingto age of the school buildings in Korea. J Environ Manag 2009;90:348e54.

[5] Hodgson AT, Beal McIlvaine D. Sources of formaldehyde, other aldehydes andterpenes in a new manufactured house. Indoor Air 2002;12:235e42.

[6] Kim J, Kim S, Lee K, Yoon D, Lee J, Ju DY. Indoor aldehydes concentration andemission rate of formaldehyde in libraries and private reading rooms. AtmosEnviron 2013;71:1e6.

[7] Blondel A, Plaisance H. Screening of formaldehyde indoor sources and quan-tification of their emission using a passive sampler. Build Environ 2011;46:1284e91.

[8] Shinohara N, Kai Y, Mizukoshi A, Fujii M, Kumagai K, Okuizumi Y, et al. On-sitepassive flux sampler measurement of emission rates of carbonyls and VOCsfrom multiple indoor sources. Build Environ 2009;44:859e63.

[9] [Chapter 15] Salthammer T. Indoor air pollution by release of VOCs fromwood-based furniture. In: Salthammer T, editor. Organic indoor air pollutants:occurrence e measurement e evaluation. Weinheim, Germany: Wiley-VCH

H. Plaisance et al. / Building and Environment 75 (2014) 40e45 45

Verlag GmbH; 2007. pp. 204e18. http://dx.doi.org/10.1002/9783527613663.ch15.

[10] Marchand C, Le Calve S, Mirabel P, Glasser N, Casset A, Schneider N, et al.Concentrations and determinants of gaseous aldehydes in 162 homes inStrasbourg (France). Atmos Environ 2008;42:505e16.

[11] Wolkoff P, Schneider T, Kildesø J, Degerth R, Jaroszewski M, Schunk H. Risk incleaning: chemical and physical exposure. Sci Total Environ 1998;215:135e56.

[12] Hopke PK. Recent developments in receptor modeling. J Chemom 2003;17:255e65.

[13] Sauvage S, Plaisance H, Locoge N, Wroblewski A, Coddeville P, Galloo JC. Longterm measurement and source apportionment of non-methane hydrocarbonsin three French rural areas. Atmos Environ 2009;43:2430e41.

[14] Song Y, Dai W, Shao M, Liu Y, Lu S, Kuster W, et al. Comparison of receptormodels for source apportionment of volatile organic compounds in Beijing,China. Environ Pollut 2008;156:174e83.

[15] Nielsen GD, Larsen ST, Olsen O, Løvik M, Poulsen LK, Glue C, et al. Do indoorchemicals promote development of airway allergy? Indoor Air 2007;17:236e55.

[16] Wolkoff P. Indoor air pollutants in office environments: assessment of com-fort, health, and performance. Int J Hyg Environ Health 2013;216:371e94.

[17] Casset A, Marchand C, Purohit A, Le Calve S, Uring-Lambert B, Donnay C, et al.Inhaled formaldehyde exposure: effect on bronchial response to mite allergenin sensitized asthma patients. Allergy Eur J Allergy Clin Immunol 2006;61:1344e50.

[18] Weschler CJ, Wells JR, Poppendieck D, Hubbard H, Pearce TA. Workgroupreport: indoor chemistry and health. Environ Health Perspect 2006;114:442e6.

[19] Herbarth O, Matysik S. Long-term trend of indoor volatile organic compoundse a 15-year follow-up considering real living conditions. Indoor Built Environ2013;22:669e77.

[20] International Organization for Standardization (ISO). Indoor air, part 9:determination of the emission of volatile organic compounds from building

products and furnishing e emission test chamber method; 2006. ISO 16000e9.

[21] International Organization for Standardization (ISO). Indoor air, part 3:determination of formaldehyde and other carbonyl compounds e activesampling method. ISO 16000e3; 2002.

[22] Saporta G. Probabilités, analyse des données et statistique. Paris: TechnipEditions; 1990.

[23] Geiss O, Giannopoulos G, Tirendi S, Barrero-Moreno J, Larsen BR, Kotsias D.The AIRMEX study e VOC measurements in public buildings and schools/kindergartens in eleven European cities: statistical analysis of the data. AtmosEnviron 2011;45:3676e84.

[24] Liu W, Zhang J, Zhang L, Turpin BJ, Weisel CP, Morandi MT, et al. Estimatingcontributions of indoor and outdoor sources to indoor carbonyl concentra-tions in three urban areas of the United States. Atmos Environ 2006;40:2202e14.

[25] Gall E, Darling E, Siegel JA, Morrison GC, Corsi RL. Evaluation of three commongreen building materials for ozone removal, and primary and secondaryemissions of aldehydes. Atmos Environ 2013;77:910e8.

[26] Publication n� 433-2003-0015 Building material emissions study. CaliforniaIntegrated Waste Management, Board Public Affairs Office, PublicationsClearinghouse; 2003., www.ciwmb.ca.gov/Publications/.

[27] He Z, Zhang Y, Wei W. Formaldehyde and VOC emissions at differentmanufacturing stages of wood-based panels. Build Environ 2012;47:197e204.

[28] Que Z-L, Wang F-B, Li J-Z, Furuno T. Assessment on emission of volatileorganic compound and formaldehyde from building materials. Compos Part B2013;49:36e42.

[29] Kelly TJ, Smith DL, Satola J. Emission rates of formaldehyde frommaterials andconsumer products found in California homes. Environ Sci Technol 1999;33:81e8.

[30] Salthammer T, Mentese S, Marutzky R. Formaldehyde in the indoor envi-ronment. Chem Rev 2010;110:2536e72.

[31] Chang JCS. Evaluation of low-VOC latex paints. Indoor Air 1999;9:253e8.