Data on formaldehyde sources, formaldehyde concentrations and air exchange rates in European housings

Formaldehyde has been discussed as a typical indoor pollutant for decades. To evaluate the current state-of-the-art in formaldehyde research and to identify the plethora of regulated and unregulated formaldehyde sources in indoor and outdoor spaces, an extensive literature search was carried out. The acquired data were analyzed with the aid of Monte-Carlo methods to calculate realistic formaldehyde concentration profiles and exposure scenarios under consideration of aging, source/sink behavior and diffusion effects. Average concentrations of formaldehyde are within 20–30 µg/m³ for European households under residential-typical conditions. The assumption of an average air exchange rate of 0.5 h−1 is also plausible. Formaldehyde emission rates of materials and products for indoor use are widely spread and range from non-detectable to > 1000 µg/h. However, processes like combustion, cleaning activities, operation of air purifiers and indoor chemistry were identified as temporary but relevant formaldehyde sources, which might cause high peak concentrations.


Subject area
Environmental Sciences More specific subject area Indoor Air Type of data Indoor air concentrations and material emission rates How data was acquired Survey and evaluation of the current literature Data format As taken from the cited references Experimental factors If necessary, data were converted from ppb to mg/m³ and vice versa. Chamber concentrations were converted into area specific and unit specific emission rates. Experimental features A literature survey was performed to collect published data about formaldehyde emissions from building materials and consumer products for indoor use in different databases. Data source location The data were taken from different sources (see cited references) Data accessibility All data can be assessed via the cited references.

Related research article
This article provides the scientific basis for the research paper: T. Salthammer (2019) Formaldehyde sources, formaldehyde concentrations and air exchange rates in European housings, Building and Environment, accepted for publication.

Value of the data
This work was carried out to gather representative data in order to calculate realistic distributions of indoor related formaldehyde emission rates and formaldehyde concentrations in Europe.
Data concerning formaldehyde concentrations in indoor and outdoor air, temporary and permanent sources, as well as data on air exchange, were collected for the European region.
Material aging, source/sink behavior and diffusion effects were also considered. The data can be used to estimate human exposure to formaldehyde in the indoor environment under real-life conditions.

Data
An evaluation of potential formaldehyde sources, formaldehyde concentrations and air exchange rates is provided. A multitude of different permanent and temporary formaldehyde emission sources were identified. In addition to the typical building products, these also include chemical reactions occurring in indoor spaces, infiltrated outdoor air, combustion processes of all kinds, the operation of equipment such as air purifiers and emissions from human activities such as sauna, cooking and cleaning. The data represent the living behavior and indoor conditions in European housings. This means that all evaluated and presented formaldehyde emission rates of building and consumer products refer to their availability on the European market. Indoor and outdoor formaldehyde concentrations outside of Europe are not discussed.
Keywords were chosen in a way that the number of hits was reduced sequentially until all papers published from 1990 on were extracted which contained data on emission rates from products used in indoor environments. Papers containing chamber concentrations from which emission rates could be calculated were considered as well. For this procedure, keywords describing products of relevance were identified in advance, such as textile, wood, particleboard, fibreboard/fibreboard, OSB, laminate, carpet, flooring, paper, adhesive, ceiling, foil, gypsum, insulation, sealant, furniture, paint, varnish, lacquer, film, tile, wallpaper, building material, construction material.
The sequential extraction of papers from the databases was performed as follows: At the beginning, all entries with the keyword "formaldehyde" were compiled. The SF database delivered approximately 267,500 hits, the WoS database approximately 43,750. The number of hits then was reduced by specification with the keyword emission, by excluding patents and setting a time limit for the year of publication from 1990 on. Further, only English, German, French, Italian or Spanish written papers were chosen. This procedure gave app. 5250 hits for SF and app. 3100 hits for WoS. From these, papers were taken with the keywords emission rate combined with the product specifications listed above. As a result, app. 570 papers from SF and app. 300 from WoS were identified. As a next step, all publications which could be excluded to be relevant because the title did not comply with the subject were removed. The same procedure was done with the remaining ones by checking the abstracts. Moreover, all reports and publications representing biased data were removed. Consequently, the report and database by Hofmann and Plieninger [1] was not considered. Finally, together with some studies from the WKI fundus including entries in the WKI owned sample database ERAD, 165 papers were collected in an EndNote database and closer investigated for relevant data. In case where concentrations (C) are given together with air exchange rates (ACH) and loading factors (L) (in case of area specific sources), area specific emission rates (SER A ) and unit specific emission rates (SER u ) were calculated. Reports and journals not being covered by scientific databases (e.g. HK Holz-und Möbelindustrie, Holztechnologie, Holz-Zentralblatt, etc…) were searched separately.
For each product or scenario the available data were summarized and, if possible, percentiles (10 À P, 25 À P, 50 À P, 75 À P, 90 À P) were calculated. Then an appropriate function (normal, lognormal or a combination of both), which represents these percentiles best, was determined by use of a least-squares algorithm [2]. Finally, a stochastic Monte-Carlo approach was applied to calculate probability distributions from pseudo-random numbers with 100,000 runs per calculation. Ranges (uniform) are provided if the derivation of a statistical function was not possible.

Units
Many different units can be found in the international literature for the concentration of formaldehyde in air. In the following, only mass-related units will be used for the comparison of concentrations and emission rates. For the conversion of volume-related units (ppb and ppm) into massrelated units (mg/m³ and mg/m³) according to Eq. (1), a pressure of 1013 mbar (101,300 Pa), a temperature of 23°C (293 K) and M(HCHO) ¼ 30.03 g/mol will be assumed.
In comparison to the thermodynamic standard chamber temperature of 25°C (298 K) and a temperature of 20°C (293 K) there is a marginal difference in the conversion factor (1.24 vs. 1.25 and 1.23), which will be neglected in the discussion.

Statistical software
The scientific software OriginPro 2016G (OriginLab Corporation, Northhampton, USA) was applied. The LabTalk script was used to calculate probability distributions.
N is the number of calculated random variables, σ is the arithmetic standard deviation with σ g ¼ exp(σ), m is the arithmetic mean with GM ¼ exp(m). GM is the geometric mean and σ g is the geometric standard deviation.

Air exchange rates
Tables 1 and 2 and Figs. 1 and 2. Table 2 Influence of window opening on the average air exchange rates in housings [9].

Formaldehyde concentrations in indoor air under living conditions
Tables 4 and 5 and Fig. 4.   6. Formaldehyde concentrations in indoor air under steady-state conditions   Table 7.   Table 7 Formaldehyde concentrations in sauna cabins according to Wegscheider et al. [26] (see this reference for experimental details). The synonyms should be interpreted as follows: "cold": before operation; "hot": during operation; "Eucalyptus, Birch, Citrus, Mint, Herbs, Menthol": type of essence.

Formaldehyde from the burning of candles
In a so far unpublished WKI study by Wensing a formaldehyde emission rate of 96 mg/g was measured. With a mass loss of 4 g/h this can be converted to a time related value of 384 mg/h. Derudi et al. [39] measured formaldehyde emission rates between 2 mg/g and 3 mg/g from scented candles but did not determine the mass loss.
Petry et al. [40] also studied formaldehyde emission rates from fragranced and unfragranced candles. The results are as follows: 137.

Formaldehyde from incense burning
Lee and Wang [42] studied the release of formaldehyde from 10 types of incense sticks in an 18.26 m³ stainless-steel chamber at T ¼ 23°C, RH ¼ 5¼ % and ACH ¼ 0.5 h À 1 . The average burn time was between 25 min and 51 min. The chamber concentrations ranged from approx. 20 mg/m³ to 300 mg/m³. Mass related formaldehyde emission rates ranged from approx. 400 mg/g to 1700 mg/g (Table 10).
Maupetit and Squinazi [41] studied the release of formaldehyde from incense sticks and incense cones in a 32.3 m³ test house at T ¼ 20°C and ACH ¼ 0.6 h À 1 . The burnt mass of the sticks was between 0.16 g and 1.25 g with a 50 ÀP value (median) of 0.32 g. The duration of combustion was between 15 min and 64 min with a 50 À P value (median) of 29 min. The burnt mass of the cones was between 0.39 g and 0.90 g with a 50 À P value (median) of 0.49 g. The duration of combustion was between 10 min and 25 min with a 50 À P value (median) of 17 min. Table 10 Formaldehyde concentrations in a test house from the burning of incense sticks and cones. The data are taken from Maupetit and Squinazi [41]. Abbreviations P1-P4 refer to the nomenclature in the paper.

Formaldehyde from the consumption of conventional and electronic cigarettes
As in the case of other combustion sources, the emission rate is often presented in the unit μg/cig (mass emitted per cigarette burnt) (Tables 11 and 12). A summary of formaldehyde emissions from conventional cigarettes can be found in the review by Salthammer et al. [113].

Table 11
Indoor air concentrations (mg/m³) of formaldehyde measured during a 2 hour use of e-cigarettes containing different liquids with (þ ) or without ( À ) nicotine in a 45 m³ room at ACH ¼ 0.56 h À 1 . The data were taken from Schober et al. [43].

Compound
No   Peng et al. [47] studied effects of cooking method, cooking oil, and food type on aldehyde emissions in cooking oil fumes. The formaldehyde concentrations in the oil fumes were between 4 mg/m³ and 27 mg/m³, depending on the cooking oil (palm rapeseed, sunflower, soybean) and the cooking method (pan-frying, deep-frying, stir-frying). The formation and emission of formaldehyde and other organic compounds from the heating of fatty acids and fatty acid esters was reviewed by Abdullahi et al. [48]. Formaldehyde is also formed by Strecker degradation in Maillard systems [49].
Bednarek et al. [50] performed a study on human exposure to air pollutants during a dinner. Seven adults volunteered in a 55 m³ room at ACH ¼ 0.29 h À 1 . During the cooking phase (indoor barbecue) the formaldehyde concentration increased from 23 mg/m³ to 58 mg/m³ within two hours. The consumption of 33 cigarettes led to a further increase of the formaldehyde concentration to 154 mg/m³.

Formaldehyde from ethanol fireplaces
Guillaume et al. [52] also measured high formaldehyde concentrations between 0.4 mg/m³ and 0.9 mg/m³ in the exhaust gas of four decorative ethanol fireplaces (Table 14). Höllbacher et al. [53] studied a single device and measured 62 mg/m³ formaldehyde in a model room. Formaldehyde sources from combustion (candles, ethanol fireplaces, mosquito coils, etc.) were reviewed by Szulejko and Kim [54].  Table 13 refer to oven cleaning rather than the cooking process itself. When taking into account the available data, a normal distribution was calculated with a mean value m ¼ 700 mg/h and a standard deviation σ ¼ 100 mg/h. It should, however, be mentioned that the assumed normal distribution only provides a very rough estimation of a realistic cooking scenario. ORIGIN LabTalk: normal(100,000)*100þ 700.  Tables 15 and 16.

Formaldehyde from wood combustion
Tables 17 and 18.  Table 16 Data from the study by Lefebvre et al. [58]. Subject blanks (bathroom with study subject), range of maximum air concentrations of formaldehyde in the bathroom after product application and mean bathroom concentrations. The conditions were as follows: Lévesque et al. [64] investigated 31 Canadian homes and found no difference in the HCHO concentrations in relation to the sampling location nor in relation to whether a combustion appliance was present or not.

Formaldehyde from air cleaning devices and paints
Sidheswaran et al. [67] demonstrated that at room temperature and 80% RH the indoor formaldehyde concentrations increased from 9-12 μg/m³ to 12-20 μg/m³ when synthetic filters were replaced with fiberglass filtration media in the HVAC units (Tables 19 and 20).

Formaldehyde from carpet
Hodgson et al. [76] determined the area specific emission rates from at least four samples. In one case the emission rate could be quantified with SER A (24 h) ¼ 57.2 mg/(m² h) and SER A (168 h) ¼ 18.2 mg/(m² h). In all other experiments, the maximum formaldehyde concentrations in the chamber were 5 ppb or less (Fig. 11).
Morrison and Nazaroff [31] studied carpet for area specific emission rates of formaldehyde in test chambers at T ¼ 23°C and RH ¼ 50%. In three cases the emission rates were between 9 mg/(m² h) and 15 mg/(m² h). In the other five cases, the emission rates were below 4 mg/(m² h).

Table 21
Formaldehyde steady-state concentrations and emission rates from chamber experiments (T ¼ 23°C, RH ¼ 45%, ACH ¼ 1.0 h À 1 ) and results from extraction analysis. The data were taken from Aldag et al. [75]  In the work by Katsoyiannis et al. [77], the 72 h chamber concentrations obtained from three carpets in three different chambers were between 2.8 mg/m³ and 14 mg/m³. Under assumption of steady-state conditions the calculated area specific emission rates are between 3.5 mg/(m² h) and 17.5 mg/(m² h).

Formaldehyde from surface coatings
Reiss et al. [29] measured emission rates between 0.05 mg/h and 3.45 mg/h with a median of 0.21 mg/h of 11 types of latex paint in a flow reactor. Chang et al. [81] studied the drying process of latex paint in a chamber at T ¼ 23°C, RH ¼ 50%, ACH ¼ 0.5 h À 1 and L ¼ 0.48 m²/m³. Within 50 h after application the formaldehyde chamber concentration of one paint was in the range of 0.5 mg/m³, while the chamber concentration of a different paint was 0.01 mg/m³. In a second study under identical chamber conditions, Chang et al. [82] followed the drying process of a freshly applied latex paint and measured a chamber concentration of about 0.1 mg/m³ after 300 h (Tables 24-26 and Fig. 13).

Table 22
Calculated area specific emission rates for the release of formaldehyde from different types of wallcoverings (paper, textile, vinyl, acrylic) in the steady-state at T ¼ 23°C and RH ¼ 45%. The experiments were performed in test chambers and by use of the WKI flask method between 1990 and 1992 (see Salthammer et al. [78] [88] published results of a European inter-laboratory comparison on raw particleboard. The mean of area specific emission rates from six independent laboratory results was 58.5 mg/(m² h) with a relative standard deviation of 9.6%. Horn et al. [80] measured the formaldehyde emission from seven OSB and found a range from 7 mg/(m² h) to 88 mg/(m² h) with a 50 À P value of 33 mg/(m² h).   The available data did not allow the calculation of log-normally distributed emission rates. The Shapiro-Wilk test did not reject the hypothesis of normally distributed data on a 95% confidence level. When considering the small number of data, a conservative approach was applied to calculate a normal distribution with a mean value of 4 mg/(m² h) and a standard deviation of 1 mg/(m² h). ORIGIN LabTalk: normal(100,000)*1.0 þ4.0.

Table 29
Test chamber conditions and unit specific emission rates in the study by Galinkina et al. [89].

Object
T

Formaldehyde from laminate
An et al. [92] studied the release of formaldehyde in a 20 l chamber and in the Field and Laboratory Emission Cell (FLEC) at T ¼ 25°C and RH ¼ 50%, respectively. After 7 days testing time the calculated formaldehyde emission rates were between 7 mg/(m² h) and 15 mg/(m² h) (Tables 31 and 32 and Fig. 19). Table 31 Chamber testing of laminate, formaldehyde steady-state concentrations after 28 days testing time (see Marutzky [91] for details).
Carrier  Pierce et al. [93] investigated the impact of laminate flooring manufactured in China on formaldehyde concentrations in a model room. In complementary chamber tests with two selected products and under so-called non-destructive test conditions the chamber concentrations after seven days were 0.018 ppm (product 1) and 0.012 ppm (product 2). The test conditions were T ¼ 25°C, (77°F), RH ¼ 50%, ACH ¼ 0.5 h À 1 , L¼ 0.43 m²/m³.
The Centers for Disease Control and Prevention (CDC) [94] released a report on formaldehyde emission from Chinese À produced laminate. Increased emission rates with a geometric mean of 41.7 m g/(m² h), a geometric standard deviation of 2.3 mg/(m² h) and maximum value of 350 mg/(m² h) at T ¼ 24.5 -25.7°C and RH ¼ 46.0 -51.5% are reported.
Wiglusz et al. [95] studied the effect of temperature on the emission of formaldehyde from laminate flooring. The tested materials did not show formaldehyde emissions at temperatures of 23°C and 29°C. At 50°C one of the materials showed a formaldehyde emission rate of approx. 40 mg/(m² h) after 20 days testing time. Tables 33 and 34.   Table 34 Types of measured door leafs and door frames, chamber conditions and area specific emission rates. The data were taken from the study by Wensing and Bliemetsrieder [97]. See also Wensing et al. [98].

Formaldehyde from mineral wool
An inter laboratory comparison experiment on the determination of formaldehyde emitted from mineral wool board using small test chambers has been described by Wiglusz et al. [99]. Eleven laboratories took part and the most reliable testing round yielded a range between 44 mg/(m² h) and 210 mg/(m² h) with a 50 À P value of 57 mg/(m² h). So far unpublished WKI data from eight different samples of mineral wool (four glass wool, four stone wool) are shown in Fig. 20. The concentrations after 96 h were between 10 mg/m³ and 66 mg/m³ with a geometric mean of 31.0 mg/m³.

Aging effect
Few studies deal with the long-term emission behavior of materials and products. Most available data refer to test chamber results of freshly produced materials measured after 28 days. Colombo et al. [100] applied an empirical potential function to extrapolate the formaldehyde emission rate of particleboard, fiberboard and plywood in large environmental chambers. For plywood, taking the 28 days value as a starting point, reductions of 33% after 1 year and 42% after 2 years can be calculated from the fit parameters. For particleboard, taking again the 28 days value as a starting point, reductions of 45% after 1 year and 66% after 2 years are obtained from the fit parameters. Brown [101] studied the release of formaldehyde from particleboard and MDF in test chambers and found that formaldehyde emission factors for all products assessed were approximately 300-400 mg/(m² h) in the first few weeks after product manufacture and 80-240 mg/(m² h) after 6-10 months. Liang et al. [102] studied the long-term formaldehyde emissions from MDF in a full-scale experimental room and found that concentrations decreased by 20-65% in the corresponding months of the second year. Under the assumption that the lifetime of wood-based materials in housing is ten years or more, a weighting factor of 0.4 can be estimated. Fig. 21 shows a Monte-Carlo simulation under assumption of a normal distribution.

Source/sink behavior and diffusion effects
Many other studies have shown that materials like textiles, wool, zeolites, etc. act as strong but partly reversible sources for formaldehyde [104][105][106][107] (Figs. 22-24).   22. Formaldehyde sorption/desorption experiment with a ceiling tile (mineral wool covered with glass fleece and paint) in a 1 m³ glass chamber. The figure was taken from Gunschera et al. [103]. Phase I: formaldehyde was doped from a gas bottle into an empty 1 m³ glass chamber for 4-6 h to achieve a steady-state concentration of approx. 150-160 ppb. Phase II: the chamber was loaded with the test specimen, loading factor 0.5 m²/m³, T ¼ 23°C, RH ¼ 50%, ACH ¼ 0.4 h À 1 and the formaldehyde concentration was continuously monitored for 70-75 h. Phase III: the formaldehyde supply was stopped and formaldehyde monitoring was continued for 24 h. Phase IV: the chamber was emptied and the decay of the formaldehyde concentration was measured for 12 h.
The barrier effect was also investigated by Yrieix [110][111][112] for different types of wood-based materials. One study [112] focused on formaldehyde emissions from different coated particleboards (melamine faced board with two paper basis weights, laminate board, wood veneer with two porosities, not varnished finish foil). In a follow-up study Yrieix compared the barrier effect of melamine impregnated decorative papers to formaldehyde emissions according to their paper basis weight (low and high basis weight) and to paper printing (surface printing or in the mass of the paper, mineral content) [110] (Tables 35-37). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 [103]. A tray made from stainless steel was completely filled with commercially available glass wool or stone wool several kinds of mineral wool and capped with a pre-conditioned 0.01 m gypsum board. The gap between tray and board was sealed and fixed in a metal frame. This construction was set up in a 1 m³ glass chamber at T ¼ 23°C, RH ¼ 50% and ACH ¼ 0.5 h À 1 . If the mineral wool is only covered with gypsum board (A), a diffusion of formaldehyde into the chamber air is clearly visible, leading to a steady-state concentration of about 30 ppb. If the surface of the gypsum board is covered with foil (B) the decaying concentration curve proves that the diffusion effect is negligible. In case of gypsum board being treated with primer and wallcovering (C) a very slight increase of the formaldehyde concentration could be observed (3 ppb within 600 h testing time).

Acknowledgments
The data were collected due to information requirements on formaldehyde given in the ECHA decision letter "DECISION ON SUBSTANCE EVALUATION PURSUANT TO ARTICLE 46(1) OF REGULA-TION (EC) NO 1907/2006, for formaldehyde, CAS No 50-00-0 (EC No 200-001-8)". The study was funded by ReachCentrum on behalf of the REACH Consortium for Formaldehyde, Brussels, Belgium, contract no. P-I216/CT01/Fraunhofer20160325.

Transparency document. Supporting information
Transparency data associated with this article can be found in the online version at https://doi.org/ 10.1016/j.dib.2018.11.096.

Table 36
Reduction of the area specific formaldehyde emission rate from particleboard by different types of covering (1 m³ stainless steel chamber, T ¼ 23°C, RH ¼ 45%, ACH ¼ 0.5 h À 1 and L ¼ 0.5 m²/m³). WKI, unpublished data. With primer 30 70 3 With primer and dispersion paint 24 76 4 With primer and plaster 22 78 5 With primer and wallpaper (fleece) 6 94 6 With primer and latex paint 2 98 Table 37 Formaldehyde emission rates of raw wood-based materials and covered wood based materials after 28 d. The effect of formaldehye reduction is also presented. The data are taken from Yrieix [110].