Robustness Validation of Methods developed by CEN/TC 351/WG 2. (Draft TS / WI351006) Summary report issued by project consortium

Transcription

1 Robustness Validation of Methods developed by CEN/TC 351/WG 2 (Draft TS / WI351006) Summary report issued by project consortium 29. November 2012 Issued after approval and revision by CEN TC351 WG2 by Reinhard Oppl 1), Dr. Matthias Richter 2), Dr. Olaf Wilke 2), Dr. Frank Kuebart 3) 1) Eurofins Product Testing A/S, Galten, Denmark 2) BAM Bundesanstalt für Materialforschung und -prüfung, Berlin, Germany 3) eco-institutgmbh, Köln, Germany Project ordered and funded by European Commission via NEN Netherlands Standardization Institute for CEN / TC 351 / WG2 Validation of CEN/TC 351/WG 2 draft CEN/TS Page 1 / 109

2 Table of Contents 1. Introduction Background and objectives Scope This document Structure of work and report Verification of Available Comparative Data for Robustness Validation Sample History (see cl. 5 in draft TS 16516) Preparation of the Test Specimen (see cl. 6.2 in draft TS 16516) Film thickness (liquid products) Sample sealing (solid products) Test Specimen Dimension and Sample Homogeneity Test Chamber Size (see cl. 7 in draft TS 16516) Test Chamber Climate (see cl. 7 in draft TS 16516) Ventilation, Product Loading and Area Specific Air Flow Rate (see cl. 7 in draft TS 16516) Ventilation rate (see cl. 7 in draft TS 16516) Loading factor (see cl. 7 in draft TS 16516) Area specific air flow rate (see cl. 7 in draft TS 16516) Air velocity above test specimen (see cl. 7 in draft TS 16516) Conclusions Test Chamber Intermediate Storage (see cl. 6.2.g) in draft TS 16516) Capillary GC column for analysis (see cl in draft TS 16516) Tube Conditioning and Laboratory Blank Tubes (see cl in draft TS 16516) Sampling Test Chamber Air (see cl in draft TS 16516) Calibration and Analysis (see cl in draft TS 16516) Reference material for validation of the procedure (see cl in draft TS 16516) Determination of Total Volatile Organic Compounds (see cl in draft TS 16516) Determination of Formaldehyde / Volatile Carbonyls (see cl. 8.3 in draft TS 16516) Conclusions Robustness validation of draft horizontal VOC emissions testing standard Project design and selection of test samples Homogeneity testing Testing program and test results Work package 1: Temperature and humidity Work package 2.1: Chamber sizes Work package : Loading factor and ventilation Work package : Age of sample at start of test Work package 1, and : Determination of hexanal Work package 1, and : Determination of the VVOC n-pentane Work package 3: Techniques for sealing back and edges Work package 4: Reference material for method validation Work package 5: Tenax TA tubes and benzene artefact generation Validation of CEN/TC 351/WG 2 draft CEN/TS Page 2 / 109

3 4. Repeatability of testing within one laboratory Study design Findings Conclusions on repeatability within one laboratory Detailed findings per tested product Project results summary and interpretation Conclusions on robustness validation of draft CEN/TS Conclusions on comparability with other testing standards Further improvements Bibliography ANNEX: Data obtained by laboratory testing A.1 Introduction A.2 Testing Samples A.3 Homogeneity testing A.3.1 Testing plan A.3.2 Results A.4 Testing program and data A.4.1 Work package 1: Temperature and humidity A WP 1 Results: Wooden flooring type product A WP 1 Results: Wood-based panel type product A WP 1 Results: Mineral wool type product A WP 1 Results: Flooring type product A WP 1 Results: Liquid type product A WP 1 Results: Foam type product A.4.2 Work package 2.1: Chamber sizes A WP 2.1 Results: Foam type product A WP 2.1 Results: Liquid type product A WP 2.1 Results: Solid reference material A.4.3 Work package : Loading factor and ventilation A WP Results: Flooring type product A WP Results: Foam type product A.4.4 Work package 3: Sealing technique for back and edges A Results Wooden flooring type product A Results Solid product with high emissions from back A.4.5 Work package 4: Evaluate reference material for method validation A.4.6 Work package 5: Tenax TA tubes and benzene artifact generation Validation of CEN/TC 351/WG 2 draft CEN/TS Page 3 / 109

4 1. Introduction 1.1 Background and objectives The draft CEN/TS "Construction products Assessment of emissions of regulated dangerous substances from construction products" (Work Item WI of CEN/TC 351) is a draft standard for the determination of emissions from construction products into indoor air. The determination is done by the use of emission test chambers [1] in combination with appropriate sampling and analysis methods [2-5]. Robustness validation of this draft emission testing standard should show any impact on test result if relevant testing parameters are modified. This is a purely descriptive task. There is no criterion available for assigning robustness or no robustness to a testing method. Only the degree of robustness can be documented. Interpretation then can lead to confirmation or modification of the testing parameters and their accepted ranges of variation during testing. Based on this study, CEN/TC 351/WG 2 changed some details of draft CEN/TS compared to earlier working drafts. 1.2 Scope Draft CEN/TS as horizontal testing standard contains only very general specifications on taking samples for testing and preparation of test specimens, as these shall be specified later in product specific standards for CE marking. Therefore sampling, and specific issues of making a test piece, were not part of the robustness validation of the horizontal testing standard. The following internal documents of CEN/TC 351/WG 2 formed the basis of this study: N129 draft testing standard, now as final version in draft CEN/TS N146 List of testing parameters that needed to be evaluated, identifying the relevance for robustness validation for each parameter with a ranking from "0 = not relevant", via "1 = relevant", to "2 = very relevant". N157 detailed working plan for this project (repeated here in the respective clauses). 1.3 This document This document combines the internal documents N154, N173 and N174 of CEN/TC 351/WG 2. CEN/TC 351 supported the publication of this report under the provision that Publisher is the consortium, not TC 351, All test samples were characterized such that it is not possible to conclude on any specific products, Suppliers of the tested samples agreed with the publication in this form, It was made clear in the text that the tested samples were selected such that the expected emissions were higher than average; this does not allow at all to use the test data for generalized conclusions or data base entries presuming that the tested products were typical for emissions of the involved product groups, and any such misuse must not take place, Any recommendations to CEN/TC 351/WG 2 given by the consortium were deleted from the original texts. These requirements were fulfilled by the consortium with this present report. The combination of the three documents into one document required a number of editorial changes and refinements, and some references are updated in this document e.g. the references to specific chapters in the draft TC 351 testing standard have been updated to the latest version, draft CEN/TS 16516, as in the version of 7 th November 2012 (internal document N189 of CEN/TC 351/WG 2). Validation of CEN/TC 351/WG 2 draft CEN/TS Page 4 / 109

5 1.4 Structure of work and report In a first step available comparative data (both published data, and available but unpublished data) from emission chamber tests were evaluated for their relevance and consequences for robustness validation, with focus on those parameters that had been rated as category "2=very relevant" in CEN/TC 351/WG 2 document N 146. While reviewing the available literature it was found that there exists only little published work dealing with the systematic validation of emission chamber measurements. Most of the relevant data is based on results and conclusions of round robin tests, in which only partially comparison of chamber measurements under different testing conditions had been done or could be read from the reports. Document N154 of CEN/TC 351/WG 2 "Collection and Verification of Comparative Data available for Robustness Validation of Methods developed by CEN/TC 351/WG 2" summarized existing information on robustness of the involved testing methodology, see chapter 2 of this report, and resulted in a rough testing program for filling remaining knowledge gaps. This was discussed and approved by CEN/TC 351/WG 2 in its meeting on 18 th May 2011 (see the minutes in document N156 of CEN/TC 351/WG 2). Document N157 of CEN/TC 351/WG 2 "Interim report on the details of testing plan for robustness validation of methods developed by CEN/TC 351/WG 2 Version 2 (31 August 2011)" detailed the project plan in terms of testing and was confirmed by CEN/TC 351/WG 2. The following testing laboratories participated in the experimental part of the study: Eurofins Product Testing A/S, Denmark BAM Bundesanstalt für Materialforschung und -prüfung, Germany eco-institut GmbH, Germany (eco) Fraunhofer WKI Institut, Germany (WKI) IDMEC, Portugal Mapei s.p.a., Italy Saint-Gobain Isover, France Teknologisk, Denmark (DTI) VTT, Espoo, Finland This list included a large part of the leading VOC emissions testing laboratories in Europe. Document N173 of CEN/TC 351/WG 2 "Final presentation of test data for Robustness Validation of Methods developed by CEN/TC 351/WG 2" contains the final data of the robustness testing project, see the annex of this report. Document N174 of CEN/TC 351/WG 2 "Presentation and interpretation of the data obtained, and proposals to amend accordingly the draft horizontal standard N129" evaluates and interprets the test data and their implications for the draft testing standard, see chapter 3 of this report. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 5 / 109

6 2. Verification of Available Comparative Data for Robustness Validation This chapter is divided into sub-chapters following the structure of document N 146. All considerations are based on the requirements marked with relevance = 2, meaning very relevant for robustness validation. A reference to the clauses or sub-clauses of the draft standard is also given. This chapter was taken from CEN/TC 351/WG 2 internal document N154 and updated editorially. 2.1 Sample History (see cl. 5 in draft TS 16516) Representative sampling of product for testing is of essential importance for the significance of test results. This is explored in detail in a technical report, CEN/TR [6]. It is common practice for US American low VOC specifications that testing is performed in the frame of a certification process where the selection of samples for testing is in the hand of a third-party certifier and performed by a third-party auditor. All steps of sample taking, dispatch and receipt at the testing laboratory are documented in a form sheet, the so-called chain-of-custody form. All this shall ensure that experts are involved in selection of samples who are experienced in making such choices. There was identified no further information or publication on this issue. Unpublished experience indicates that both products exist where emissions may vary strongly (with more than ± 50%), but also products with very low variation of emissions within and between production batches. Regarding the accepted time between sampling date and receipt at laboratory, and the time between receipt at laboratory and start of test, there were not found any data on the effect of both delays on test results. When ageing of the sample may influence emissions, then this is an important issue. The requirements for taking the sample in different test methods range from immediately up to 8 weeks after sample is ready for distribution or up to three months for containerized products. The requirements for starting the test may range from max 5 weeks up to max 8 weeks after sample is ready for distribution or up to max 4 months for containerized products, see e. g. [7-9]. All these considerations have been drafted when writing clause 5 of TC 351 draft standard. Sample age and history need to be specified having in mind the product specific characteristics and intended uses. Further specifications and possibly investigations should therefore be initiated by the product specific TCs when writing or expanding product performance standards for CE marking. At the moment no published data are available on the influence of sample age on emission test results. 2.2 Preparation of the Test Specimen (see cl. 6.2 in draft TS 16516) There is evidence that the preparation of the test specimen has an influence on the emission rates of VOCs and formaldehyde. The main aspects are sealing of edges (particle boards), backside sealing (flooring materials) or weight and thickness (liquids or adhesives). For this reason most round robin tests have been conducted using material with no or only a simple preparation of test specimens. Therefore not much useful data is available. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 6 / 109

7 2.2.1 Film thickness (liquid products) The amount of material in test chamber (expressed as mass or volume) should correlate with the emission source strength and thus with the emission rate. On the other hand, when a liquid product starts drying, it will form a film on the surface. From then on, emissions are no longer determined mainly by evaporation from the surface, but by diffusion of volatiles through this surface film. Several unpublished studies showed that there is some correlation between film thickness and emission rate after 3, 7 and 28 days, but this was not linear in all cases. If two layers are applied then the waiting time before application of the second layer impacts the emissions test results. Sometimes there was even observed no change of emission rate with change of thickness at all. The total surface seems to influence the emission rate as is relevant for e.g. those adhesives where not the whole surface is covered, or where the adhesive is applied as structured surface. Time until formation of this film is decisive for how many volatiles have evaporated, meaning how many volatiles still are available for being emitted before this change of emission mechanism occurred, respectively. This may influence the duration during which emission can occur [10]. In a study carried out by debortoli et al. [11], the possible influence of paint film thickness on the specific emission rate values has been investigated, attempting to correlate values of Texanol specific emission rate after 48 h with the reported coating density of the paint specimens. However, no correlation could be identified. But the study showed large deviations of test results, which might have covered any effect of film thickness Sample sealing (solid products) Some cases have been seen (but not in published studies) where front and back of a product showed extremely different emissions especially when recycled material was used which cannot be controlled easily for their potential emissions. If only the upper surface is relevant for emissions, then in those cases the tightness of sealing technique for edges and back may impair test results [10]. If samples are cut and the fresh edges are showing different emissions than normally treated surfaces also then the tightness of sealing technique for edges and back may impair test results [10, 12]. Available sealing techniques are: Mechanical tightening of surfaces: o Pressing a flexible flooring into a tray Japanese seal box (e. g. JIS A 1901, Annex 2, cl. 2.2) Coverage (complete, or partial coverage if some open edges are typical for real application, e.g. for tiles) o with aluminum foil o with aluminum tape o by placing two test pieces back-to-back in test chamber Several studies have been performed by different test laboratories but no data were published. Results were used just to identify a tape that seals efficiently, but with very low emissions, not impairing the chamber test results. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 7 / 109

8 2.2.3 Test Specimen Dimension and Sample Homogeneity According to TC 351 draft standard the size of the test specimen has to be adjusted in accordance with clauses "Dimensions and loading factors in the reference room" and 7 "Test chamber conditions" of the testing standard. The intended use shall be taken into consideration. German DIBt testing protocol prescribes a minimum test chamber size of 0,225 m³ for testing solid floorings such as laminates and parquets larger than for other products, to ensure a sufficient size of the test piece, for ruling out material inhomogeneity. The same testing protocol also specifies the fraction of joints between flooring tiles as well as a preparation of test specimens that is close to the intended use [9]. For the validation study, material homogeneity is of high importance to ensure comparability of the different sub-samples. Inhomogeneity may influence the evaluation of emissions from a product. This is also supported by recent inter-laboratory comparisons [13-17]. Main problem in these tests was a variability of chamber testing results that might have been caused either by sample inhomogeneity causing different emissions of the test specimens, or by differences of the applied analytical procedures[18]. Solid products made from natural based raw materials, such as wood, cork or vegetable oils, may show considerable variation of emissions over surface due to inhomogeneous raw material. All solid materials may show variation of emissions over surface due to variations in production process (e.g. drying time, temperature, or drying duration). Any documentation on material inhomogeneity has to consider the uncertainty of the testing method itself. Only variation of emissions that are significantly higher than method uncertainty can be assigned to the tested material. In an inter-laboratory comparison organized by FCBA [19] a particle board glued with UF resin was conditioned unwrapped during 4 weeks in a climatic chamber at 23 ± 2 C and 50 ± 5 % RH at 1 air change per hour. So treated samples showed relative standard deviations of emission of TVOC: 13 %, alpha-pinene: 21 %, and formaldehyde: 10 %. It should be noted that fresh samples might show much higher variation of emissions. In the GUT round robin test 14 sub-samples of PA6 carpet tiles with bitumen backing were tested for homogeneity of emissions of 32 VOCs by direct thermal desorption [20]. Variation was between 10% and 60%, with higher relative variability if the emissions were very low. Caprolactam showed even higher variability of emissions this was later correlated with specific analytical challenges. 2.3 Test Chamber Size (see cl. 7 in draft TS 16516) The question was whether test chamber size has an impact on emissions test result even with equivalent area specific air flow rate. Some less volatile products were suspected to re-condense on surfaces and the higher available surface in larger test chambers could reduce air chamber concentrations compared to smaller test chambers. For any test chamber comparison only test results in terms of specific emission rate per surface area and time should be used. If different loading factors or ventilation rates were applied in different test chambers, then this fact alone will give different chamber air concentrations, even when the area specific rate remains constant: In theory, double the emission source should give 100% higher air concentration, and double ventilation rate should give 50% lower air concentration. This would be consistent with stating no impact of test chamber size on the test result. Therefore emission rates, rather than air concentrations, must be compared if different loading factors or different ventilation rates are applied. GEV round robin tests did not show any impact of test chamber size on test result [13]. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 8 / 109

9 Only little data originating from within laboratory validation work is available. Wilke et al. [21] conducted emission tests with rubber flooring in test chambers and cells of different volume (0,035 l (FLEC; 1 l; 20 l; 1 m³). The area specific air flow rate q was kept constant at 1,25 m³/m²h in all these cases. The comparative measurements of the emission from the rubber flooring showed an overall maximum standard deviation of 16 % for TVOC. A larger deviation was found between the FLEC test cell and the 1 m³ test chamber early during the test. Although a good comparability resulted for TVOC, single compounds showed some differences. For example styrene emission was much higher in the 1 liter cell, whereas highest emissions of sesquiterpenes were determined in the 20 liters chamber. But in contrast to this finding the other VOCs showed a very good correlation in all four test chambers. The authors saw a certain inhomogeneity of the material as a possible reason for the differences, even though a good homogeneity of the rubber flooring had been determined by headspace analysis. It also has to be taken into account that a proper sealing of the backside of the flooring material is very important for achieving good comparability between different chambers and cells (compare to chapters and of this reporterror! Reference source not found.). BAM performed a comparison of measurements in two 1 m³ and one 20 liters test chambers where equivalent results were obtained [22]. Here, a good comparability between the emission test results in the two test chambers of different size was found. The BAM round robin test with a lacquer had to be performed by the participants with an area specific air flow rate q = 1 m³/m² h [15]. When comparing two classes of test chamber sizes (1 m³, and liters) the test results from both test chamber types were statistically not equivalent for six out of seven compounds. Only for styrene equivalence could be demonstrated, probably because it showed the lowest standard deviation of all compounds in the test. Another comparison was made between the test results of measurements in large ( liters) and small ( liters) test chambers used in the BAM round robin test with a sealant as test material and an area specific air flow rate q = 44 m³/m² h [16]. Both groups of test chambers showed the same variation of test results. This did not allow clear conclusions on the influence of the test chamber size on test results. Investigations on the emission of diisobutyl phthalate (as an example for a SVOC) in test chambers with different volume showed an impact of the test chamber size on the test result [21]. The emission rate in a 1 m³ test chamber was three times higher than in a 20 m³ test chamber. When comparing test results of TVOC with those of TSVOC in small test chambers (0, liters), it turned out that the standard deviation for TSVOC was higher, probably due to sink effects. Sollinger et al. [23] investigated the influence of test chamber volume on the initial emissions of a carpet under dynamic (ventilated) conditions. Different pieces of the same carpet were tested in parallel in 30 l, 1 m³ and 38 m³ test chambers. The product loading factor used in the 30 l and in the 1 m³ test chambers was 0,8 m²/m³, the product loading factor in the 38 m³ test chamber was 0,4 m²/m³. The authors explained this difference with the geometry of the 38 m³ test chamber. The authors reported that after multiplying the measured concentrations in the 38 m³ test chamber by a factor of two (for the difference between the product loading factors in fact then comparing the area specific emission rates), the comparability between the three test chambers was confirmed. They concluded that the maximum concentrations in the gas phase of a compound emitted from the tested textile flooring depend on the chamber loading. Furthermore it was found that the test chamber volume has only little influence on the determined maximum concentration. This was supported by debortoli et al. [18]. Based on test results obtained in an inter-laboratory comparison they concluded that the wide range of test chamber volumes (0,035 l to 1.5 m³) does not introduce any systematic difference of test results. This conclusion was supported by the fact that no significant correlation was found between test results and test chamber volume (with a probability P > 0,95) for n-dodecane (when expressed as relative difference between expected and observed emission rate) and PVC tile (as emission factors). Validation of CEN/TC 351/WG 2 draft CEN/TS Page 9 / 109

10 However, larger test chambers showed a tendency to give less scattered test results than smaller test chambers for a PVC tile. This may be explained by the heterogeneity of the material, which is more important for test results with a smaller test specimen in a smaller test chamber. With respect to the test chamber material they found that glass and stainless steel appear to be equally suitable for test chamber measurements. Katsoyiannis et al. came to different results as reported above [24]. Emissions of VOCs and carbonyls from carpets of different type (wool, synthetic) over a time period of three days at T = 23 C, RH = 45 %, n = 0,5 h -1 and L = 0,4 m²/m³ were measured and kept constant. The measurements were carried out in test chambers with different volume (0,02; 0,28; 0,45 and 30 m³). All carpet samples were tested in a 0,02, 0,28 and 0,45 m³ test chamber, two samples additionally in a 30 m³ test chamber. Large differences between the emission behaviour of the carpet samples in the different test chambers were observed, but it should be considered that the concentrations for most of the measured substances were far below 10 µg/m³ or even could not be detected because falling below the detection limit. The highest concentrations were measured for 4-phenylcyclohexene (4-PCH, c = 170 µg/m³) and 2,2-butoxyethoxy-ethanol (2,2-BEE, c = 320 µg/m³) in the 30 m³ test chamber. This is also valid for the area specific emission rates, as a constant area specific air flow rate q had been applied (0,8 m³/m² h). Information about emission tests performed with the same area specific flow rate q in different emission test chambers can also be read from reports on round robin tests. However, these tests were normally not performed with the aim to validate robustness and are not systematic in this sense. Thus, the comparability of emission chamber testing in test chambers of different size is given only exceptionally. The most frequent explanation for this incomparability read from literature is inhomogeneity of the test specimens as well as huge variation of the analytical methods between different laboratories [14, 17, 19, 25-27]. Therefore, these data were not taken into consideration here. The examples mentioned above indicate that comparability between test results obtained with test chambers of different size sometimes could be demonstrated and sometimes not. But no publication of a systematic validation work could be found. Therefore, comparative tests were performed, see chapter of this report. 2.4 Test Chamber Climate (see cl. 7 in draft TS 16516) Temperature and relative air humidity in the test chamber are the most relevant climate parameters during emission test chamber measurements. The draft TC 351 standard starts from the specifications in ISO [1], but with further specifications, resulting in Temperature: 23 C ± 1 C (compare to ± 2 C in ISO ) Relative humidity: 50 % ± 5 %. Both parameters may have a significant impact on emissions, so they cannot be varied for any type of product or test equipment without impairing the comparability of emission test data. The draft TC 351 standard incorporates the specifications of the testing standards ISO , -6, -9 and -11. During the robustness validation process the comparability between measurement results was evaluated. This also included investigating the comparability between ISO standard series and the specifications for the measurement of the emission of formaldehyde given in EN [18], which slightly differs from the ISO standard series: Temperature: 23 C ± 0,5 C Relative humidity: 45 % ± 3 %. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 10 / 109

11 EN is the reference standard for formaldehyde emission testing for a number of products with urea-formaldehyde binder. But that test method was created for determining an equilibrium concentration for the special case where formaldehyde is created continuously by chemical reaction of the binder with water from air humidity (hydrolysis). This reaction and the resulting emissions can be very sensitive to variation of temperature and relative humidity. Wilke et al. investigated the influence of temperature, relative humidity and air change rate on VOC emissions [29]. They loaded beech veneered particleboard coated with an UV-curable acrylate lacquer containing benzophenone as photo initiator into 1 m³ emission test chambers. The test series started after a conditioning time of 10 weeks in the test chambers under standard conditions, so that stable concentrations and emission rates could be expected. The compounds of interest were the higher boiling benzophenone and the lower boiling butyl acetate. During the 31 days lasting test series parameters were set as follows: Temperature [ C]: 18; 23; 28 Relative humidity [%]: 33; 45; 65 Air change rate [h -1 ]: 0,3; 0,5; 1,3; 2,0 Only one parameter each was changed per experiment, the others were kept at standard conditions (in this study: 23 C, 50 % RH, 1,3 h -1 ). A change of the climate parameters gave different results when comparing butyl acetate and benzophenone concentration: When increasing the temperature from 23 C to 28 C, the benzophenone concentration and emission rate showed a disproportional increase by a factor of 2,7. This is different from the more volatile butyl acetate where the concentration and emission rate increased by a factor of only 1,6. Changing the relative humidity had about the same impact on both substances. Both concentrations and emission rates increased with higher humidity. Under worst case conditions (n = 0 h -1, RH = 50 %, T = 28 C), butyl acetate emission increased extremely by a factor of nearly 50 whereas benzophenone emission increased only by a factor of 6. For the results regarding ventilation rate, please see chapter Sollinger et al. came to similar conclusions [30]. They performed experiments under static conditions with ten types of floor coverings using two chambers under static (meaning not ventilated) conditions: A 33 liters glass chamber and a 1 m³ stainless steel chamber. The purpose of the static conditions was to identify initially emitted compounds and to determine their dependence on temperature and relative humidity. The temperature dependence of the emissions was studied by determining the static equilibrium concentrations at 23, 30, 40, 50, 61 and 71 C. Based on their results the authors stated that with lower volatility of a compound its temperature dependence increases. The influence of the relative humidity on the emission process was also investigated. Samples of textile floorings were exposed to relative humidity of 0 % and 45 %. Explicit results were not published, but it was reported that the equilibrium concentrations of these samples do not correlate with relative humidity. The authors concluded that for textile floorings it will not be necessary to maintain a constant and well-defined humidity of the test chamber air. In contrast to that, Gene Tucker [31] reported that test results may be affected by relative humidity in correlation with the molecular polarity of the compounds. He concluded that products containing highly polar compounds require good control of relative humidity. Non-polar compounds were only slightly affected by changing relative humidity of test chamber air. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 11 / 109

12 DeBortoli et al. [11] also reported temperature effects on the emission rate in the context of an interlaboratory comparison. In this case they studied the emission of Texanol from a paint. One participating laboratory ran its emission test chamber in a first test at 24,9 C, deviating from the instructions of the test protocol. In a second test they repeated the measurements at 23,4 C. The temperature increase by 1,5 C caused an average increase of emission rate by 11,7 %, which was statistically significant (p < 0,05). This result confirms that temperature is a very important factor when determining the emission rate of Texanol from that paint. Wolkoff reported that both temperature and relative humidity affect emission rates, but with a strong dependence on the type of building product and type of VOC [32]. In this study the emissions of ten VOCs of concern from five building products were tested with the field and laboratory emission cell (FLEC) during up to 250 days. The following climate conditions were applied: Three different temperatures (23, 35 and 60 C) and two different relative humidities (0 % and 50 % RH). Some of the VOC emissions were clearly influenced by different relative humidity and by different temperature, but the correlations were different per different type of emitted VOC. For example, while the emission / time profile was unchanged for Texanol with different relative humidity, the emission / time profile of propanediol reached a zero concentration at 0 % RH in less than one day. The author assumed that the low relative humidity may either have resulted in a different film structure due to a faster drying process, or alternatively the water vapor carried polar substances away from the surface when humidity was present in test chamber air. For most of the emitted VOCs, either a modest or a negligible effect was seen when increasing temperature from 23 C to 35 C, while large effects were observed if increasing temperature to 60 C. An increase up to 35 C did not appear to have a clear influence on the (primary) source emission, but the temperature effect was higher for VOC emissions of a liquid building product than of a vinyl flooring. The author came to the conclusion that control of temperature and relative humidity for the studied VOCs could be maintained within an interval of ± 2 C and ± 5 % RH without impairing validity of the test results. The test results given in the above cited literature showed that temperature and humidity may have a significant impact on the emissions from materials, depending on their physical and chemical properties. All test results show the same basic effects, although with different characteristics that probably are related to different matrices and different involved VOCs. Therefore, systematic test series were performed with different matrices to obtain information about the influence of climate parameters on products with different composition and in different matrices (solid, porous and liquid), see chapter of this report. 2.5 Ventilation, Product Loading and Area Specific Air Flow Rate (see cl. 7 in draft TS 16516) Any identified documentation was based on change of several parameters (e.g. loading factor and ventilation) at the same time and most often simultaneously. A comparison of the emission rate where these parameters were changed separately and one by one (e.g. the loading factor, but not the ventilation rate), had not yet been published. Such an investigation is essential for making tests more affordable, especially if tests have been performed for other purposes (e.g. for EN 717-1, or for US specifications) with different loading factors or with different ventilation rates than specified in the draft testing standard of TC 351. In these cases, it could be an advantage if available test results obtained with one testing standard could be interpreted for other testing standards without new testing, just by recalculation to the deviating specifications. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 12 / 109

13 2.5.1 Ventilation rate (see cl. 7 in draft TS 16516) Only little exploitable published work could be identified. Sollinger et al. [23] investigated the influence of the air change rate on the emission of 2,2,6,6-tetramethyl-4-methylideneheptane from a rubber flooring with a test chamber loading factor of 0,4 m²/m³ in a 1 m³ test chamber with three different air exchange rates (0,45; 0,9 and 1,9 h - 1 ). An increase of the air change rate from 0,45 to 1,9 h -1, i. e. by a factor of 4,2, reduced the concentration only by a factor of 2,5. It was assumed that the material transfer from the rubber phase into the gas phase at the highest air change rate is enhanced, due to the increased concentration gradient with increasing air exchange. A comparison of emission rates had not been made and could not be calculated for this study because no detailed data were published, but only diagrams. In the study of Wilke et al. [29] mentioned above in chapter 2.4 with beech veneered particleboard coated with an UV-curable acrylate lacquer containing benzophenone as photo initiator in 1 m³ emission test chambers, the test series started after a conditioning time of 10 weeks in the test chambers under standard conditions, so that stable concentrations and emission rates could be expected: Air change rate [h -1 ]: 0,3; 0,5; 1,3; 2,0 Climate: 23 C; 50 % RH A modification of the air change rate gave different results when comparing butyl acetate and benzophenone concentration: Decreasing the air change rate increased the butyl acetate concentration more than the concentration of benzophenone - the increase of the butyl acetate concentration was almost proportional to the decrease of the air change rate (as expected), which was not the case for benzophenone. In parallel, it could be found that any change of the air change rate had no significant influence on the emission rate of butyl acetate, whereas the emission rate of benzophenone showed considerable impact of the ventilation rate Loading factor (see cl. 7 in draft TS 16516) In the context of an inter-laboratory comparison organized by de Bortoli et al. [18] it was found that the loading factor did not seem to have a significant impact on the test results. Three participating laboratories with the smallest test chambers were using test specimen sizes which resulted in a loading factor higher than prescribed. Even in the case of a 4 liters test chamber, where the loading factor was more than ten-fold higher than specified, the test results were comparable to those of the other laboratories who worked in compliance with the testing protocol Area specific air flow rate (see cl. 7 in draft TS 16516) Jann et al. tested a PU lacquer on solid alder wood in three different test chambers (20 m³, 1 m³, FLEC) and a UV lacquer on beech veneered particle board in four different test chambers (20 m³, 1 m³, 20 l, FLEC) with different loading factors and different air change rates but always with the same area specific air flow rate q of 1 m³/m² h [33]. TVOC results for the PU lacquer showed a good comparability. TVOC results of a UV lacquer showed larger differences, especially for the 20 m³ test chamber. The reason was interpreted as a sink effect impairing the main compound benzophenone. The evaluation of recent round robin tests organized by BAM where a constant area specific air flow rate was prescribed also showed good comparability for the tests in different test chambers, partly with different absolute loading factors. The concentrations of the main compounds showed standard deviations in the range between 20 and 30 % [15, 16]. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 13 / 109

14 2.5.4 Air velocity above test specimen (see cl. 7 in draft TS 16516) In the context of a round robin test with a paint, debortoli et al. [34] addressed the issue of surface air velocity above the surface of the test specimen by requesting its measurement, prescribing a measurement position and urging the participants to arrange air circulation in such a way that the average air speed, at the point of measurement were as close as possible to 10 cm/s. The reported values of surface air velocity did not show any correlation with the emission test results. Therefore, apparently, the surface air velocity did not have a significant impact on test results. However, the measured air velocities showed large fluctuations, could not be controlled well in many test chambers, and it is uncertain how representative the values measured in the prescribed position are for the average air velocity. Moreover, the measurements were started when a solid film had already been formed at the paint surface in order to reduce the influence of surface air velocity this limits the validity of the conclusions significantly, as the draft testing standard of TC 351 requires immediate transfer of liquid applied test specimens into the test chamber. Therefore, the impact of air velocity above the test specimen on the test result was further investigated, see chapter of this report Conclusions No study could be identified investigating the impact on the emission rate of the single parameters (ventilation, loading, area specific air flow rate) separately. Therefore specific tests were needed to obtain such information. The work plan specified investigations in test chambers of four different volumes, 0,02 m³ 0,1 m³ - 0,2 m³ 1 m³ much larger than 1 m³; with the following test plan, all with the same product in test chambers of the same volume: Application of 3 different loading factors L at constant air change rate n Application of 3 different air change rates n at constant loading factors L Application of a constant area specific air flow rate q with simultaneously changing air change rates n and loading factors L on three different levels 2.6 Test Chamber Intermediate Storage (see cl. 6.2.g) in draft TS 16516) Draft standard of TC 351 requires storage of test specimen in test chamber during the whole testing period. One of the reasons were data reported by BAM showing an increase of SVOC concentrations of up to ten days after loading the test chambers. An intermediate storage of the test specimen outside the test chamber (but with re-loading into the test chamber three days before air sampling) could therefore lead to lower test results [35]. The other reason is the risk of cross-contamination during storage together with other test materials. An example is given by Wilke et al. [29]. A beech veneered particleboard (covered with an UV cured acrylate lacquer) was stored together with untreated wood over 27 days and was tested on the 28 th day under standard conditions in a 1 m³ test chamber. The consequence was a contamination with pinene, carene and phenol, when compared with a test without storage of the test specimen outside the test chamber. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 14 / 109

15 2.7 Capillary GC column for analysis (see cl in draft TS 16516) Different gas chromatographic capillary columns for analysis of Tenax tubes after thermal desorption will give different separation of complex mixtures into its constituents, resulting in different identification and then different quantification. While ISO proposes a non-polar column (made of 100% poly dimethyl siloxane), the draft standard of TC 351 specifies a slightly polar column (made of 5% phenyl / 95% methyl polysiloxane) for achieving a better separation and analysis for more polar substances than it is possible on the non-polar column. Today the involved testing laboratories make different selections of the column. TC 351 WG2 considered it to be essential that this practice is harmonized for reducing variability of test results between laboratories. While ISO allows selection of another column than the specified one, the draft TC 351 standard requires use of the one specified column only. The use of a slightly polar column is rational with respect to today s analytical requirements. An increasing number of polar compounds (e.g. alcohols, aldehydes, carbonic acids, esters and glycols) appear as emissions from interior construction products in the last few years, which can only be poorly chromatographed using a non-polar column. In a round robin test reported by Wilke et al. [16], detailed information on the analysis method that each participant used was collected. This included data on the capillary separation column (length, polarity), so that a comparison of the chromatography could be made. Figure 1 compares the test results using of butyldiglycol on a non-polar column and on a moderately polar column. It showed that the variation of the test results obtained with a non-polar column is significantly larger. The mean value of the results of the slightly polar column (red line) is in better agreement with the target value (54,5 ng). Figure 1: Results of the participants for butyldiglycol depending on the GC column used (DB1: nonpolar type column, DB5: moderately polar type column). Red line: general mean value, dashed and solid lines: 1 and 2 sigma (standard deviation of the mean value and expanded uncertainty). Also less polar components such as styrene (Figure 2) show a slight tendency to reach the target value easier when using the slightly polar column. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 15 / 109

16 Figure 2: Results of the participants for styrene depending on the GC column used (DB1: non-polar column, DB5: slightly polar column). Red line: general mean value, dashed and solid lines: 1 and 2 sigma (standard deviation of the mean value and expanded uncertainty). The advantage of using a moderately polar column can easily be seen from Figure 3, containing the standardized values (content normalized to the respective target value) for the two main column types. The mean values of the results of the slightly polar columns marked by pink squares show a good agreement with the respective target values and nearly all are close to 1. The values of the non-polar column are on average significantly lower (circa 0,85) than the target value. Figure 3: Overview of the results of the participants depending on the GC column used (DB1: nonpolar column, DB5: slightly polar column); the values are normalized to the target value. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 16 / 109

17 A comparable picture has also been obtained for the relative standard deviations: the results provided by the non-polar columns showed higher fluctuations than those obtained by the slightly polar columns (Figure 4). Figure 4: Standard deviation of the participant results depending on the GC column used (DB1: nonpolar column, DB5: slightly polar column). Since 9 participants used non-polar columns and 18 participants used polar columns, a fairly high confidence level in the statement can be assumed, although there were laboratories that used non-polar columns and nevertheless provided results well near the target value. A remaining challenge is that the definition of the VOC range is related to retention times on a nonpolar column. Retention times may be different on a slightly polar column, which can be relevant for considering a substance as VOC or not with a retention time close to the borders of the VOC range. In other words, on one column type a substance may be within the VOC range and on the other column type outside of this range In real testing this happens only in very rare cases and with few substances. TC 351 WG 2 therefore decided to rate this as a non-significant problem and to accept assignment of a substance as VOC or not on basis of analyses on a slightly polar column: "Any change in component elution order is minimal and can be ignored." (cl of draft CEN/TS 16516). The existing knowledge obtained from the round robin test reported by Wilke et al. [16] allowed such a decision, and no further tests were necessary. 2.8 Tube Conditioning and Laboratory Blank Tubes (see cl in draft TS 16516) The specified air sampling tubes filled with Tenax TA can generate blank values during air sampling. This can affect the detection limits of benzene, octanal and nonanal [36]. Other information from users of Tenax TA tubes with thermal desorption indicated decomposition of the Tenax TA material giving acetophenone, benzaldehyde and benzoic acid. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 17 / 109

18 BAM made the experience that nonanal is likely to be a reaction product of Tenax TA in the presence of ozone (which should not be present in test chamber air due to cleaning of the supply air). The main challenge is benzene. Benzene as artifact can be close to the limit value if this is required to be 1 µg/m³ due to its carcinogenic properties. The provided internal information from several laboratories did not give a clear picture [10]. Therefore the benzene artifact level on exposed Tenax TA sampling tubes was documented by further investigations, see chapter of this report. 2.9 Sampling Test Chamber Air (see cl in draft TS 16516) ISO specifies maximum safe sampling volumes for air sampling sorbents, among that Tenax TA. Within these limits theory would expect no correlation between both air sampling volume and velocity, and completeness of air sampling, and thus the emissions test result. The BAM round robin test in 2008 investigated any possible influence of air sampling volume and air sampling flow rate on test results [16]. Questionnaires supplied by the participants were evaluated for the parameters air sampling flow rate and air sampling volume. With most participants air sampling flow rate was l00 ml/min, its smallest value being 40 ml/min and the highest value 200 ml/min. No impact of air sampling flow rate on test results could be established from Figure 5 within this range. No influence of air sampling volume on test results could be established in the range between 1 and 9 liters. Figure 6 did not provide any tendency or dependence of the measured value on the air sampling volume. It can nevertheless happen that very volatile components break through and show false low test results if the sampling volume is too large. Figure 5: Measured value as a function of air sample flow rate Validation of CEN/TC 351/WG 2 draft CEN/TS Page 18 / 109

19 Figure 6: Measured value as a function of air sample volume As shown in Figure 5 and Figure 6, no significant differences could be seen with different air sample flow rate and air sample volume Calibration and Analysis (see cl in draft TS 16516) Horn et al. investigated the limit of quantification (LOQ) of all compounds listed on German LCI list and of 38 carcinogenic compounds [37]. An LOQ of 1 µg/m³ was confirmed for 78 % of the VOC with LCI limit value, assuming a sample volume of five liters. 18 % of the VOC had an LOQ of 5 µg/m³ and 4 % had an LOQ above 5 µg/m³. The carcinogenic substances were analyzed using the selected ion mode and an LOQ of less than 1 µg/m³ was confirmed for 34 out of 38 substances (89 %). Ethylene glycol showed to be a major component of VOC emissions from a special test adhesive for GEV round robin test 2003 [13]. The result for both ethylene glycol and TVOC depended heavily on whether this VOC was identified correctly and then quantified with its specific response factor. The latter was important because the analytical response factor of ethylene glycol is very different from that of toluene (which is used for quantification of non-identified compounds). 14 out of 20 participants identified ethylene glycol, but only 6 out of 20 used the correct response factor for quantification. This means that correct analysis can be decisive for the test result. In this case this made the difference whether the test result was above or below the acceptable TVOC level (at that time 500 µg/m³). A further step towards better reproducibility of emission testing could be the application of a uniform temperature program for gas chromatography. This would also standardize the evaluation of chromatograms since the separation of substances in the chromatogram would be more harmonized throughout Europe, increasing reproducibility of test results between different testing laboratories. For quality control reasons, draft CEN/TS recommends that testing laboratories regularly participate in round robin tests and/or use certified standard materials for calibration. Another possibility for a better and comparable analysis would be the use of the same standard solution for calibration by all test laboratories. Such a solution could be offered and distributed by an independent institute, e.g. by HSL within the WASP program. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 19 / 109

20 Further it would be an advantage to use a gas mixing and exposure system for sampling compounds on Tenax TA from a gas atmosphere with known and traceable concentration, see e. g. [38, 39]. The most efficient way of checking the complete test chamber method would be the use of a reference material for emission testing, see next chapter Reference material for validation of the procedure (see cl in draft TS 16516) NIST and Virginia Tech University now have available the know-how to produce a reference material for emission test chambers with toluene in a polymethylpentene (PMP) film. This was presented during a workshop at NIST in Gaithersburg (Maryland, USA) in April 2011 [40, 41]. Emission rate of that film is not constant over time after unpacking; but emission rate during the first days follows astonishingly well a mathematical model that relies on film properties (loading, diffusion properties) and chamber dimensions and ventilation. Some variation occurs with testing after 24h, but reliable results were obtained with different laboratories involved after 48h and 72h. This reference material is suitable to validate performance of the whole chamber testing procedure, including ventilation, mixing and sink effects of test chamber, air sampling, analytical determination of toluene in air samples. Until now these elements of emission chamber testing could be determined only separately; now the whole procedure can be checked against an independent standard that can be traced back to a primary unit the weight increase of the PMP film during loading. This makes this new material unique and relevant. This reference material until now is NOT suitable to validate test chamber testing performance with respect to simulation of re-adsorption/re-desorption of once emitted VOCs on surface of test specimen made from real product samples (e.g. porous products with high surface), simulation of the impact of drying / curing / ageing of real samples on emission rate, interaction of less volatile and of reactive VOCs with test chamber walls and sinks, and their impact on emission rate. Therefore use of the reference material cannot substitute a validation of the whole test method with specific types of real samples. Nevertheless it can test the performance of the testing procedure as such, with the non-polar VOC toluene, without taking into account the performance of the procedure for VOCs that are more polar and/or less volatile than toluene; o this could be overcome with the development of similar reference materials made of some 2 4 additional VOCs representing classes of such VOC (e.g. n-butanol, dodecane, glycols, formaldehyde), which was reported to be planned by Virginia Tech University. without taking into account product specific aspects; o this is no problem for a method validation, as the method should be generic, and then it should be adapted later to specific product groups if necessary. CEN TC 351 WG2 tested this reference material by including it in some comparisons, e.g. for comparing any impact of test chamber size on emission rate once for real samples, and once for reference material. And CEN TC 351 WG2 made use of this new development by including a test run with this reference material for toluene including all involved laboratories. In conclusion, this was included as an option in clause on quality control of draft TS Validation of CEN/TC 351/WG 2 draft CEN/TS Page 20 / 109

21 2.12 Determination of Total Volatile Organic Compounds (see cl in draft TS 16516) TVOC is the total of all volatile organic compounds. This can be calculated in different ways: as sum of all individual VOCs above a certain threshold (e. g. 5 µg/m³), each calibrated with its own response factor, with all non-identified VOCs calibrated as toluene equivalent (with the response factor of toluene). as sum of o all individual VOCs on a target list (e. g. German LCI list), each calibrated with its own response factor; o plus all identified VOCs NOT on that target list, calibrated as toluene equivalent (with the response factor of toluene); o plus all non-identified VOCs, calibrated as toluene equivalent (with the response factor of toluene); o in any case VOCs are included only if above a certain threshold (e.g. 5 µg/m³); o this is the procedure in use for German DIBt testing. as sum of all individual VOCs above a certain threshold (e.g. 5 µg/m³), each calibrated as toluene equivalent (with the response factor of toluene) - this is the procedure specified in draft CEN/TS as total area of the chromatogram calibrated as toluene equivalent (with the response factor of toluene) - this is the procedure in use for ISO The threshold of 5 µg/m³ was selected for including only VOCs that can be determined with a reasonable accuracy. Traces of VOC generate very weak signals during analysis which is much more vulnerable to analytical errors than higher VOC amounts. Table 1 shows how TVOC values, obtained with MS or with FID. 27 pairs of TVOC test results were compared, taken from unpublished studies on floorings, coatings and glues in different laboratories. TVOC test results were available either as toluene equivalent, or calculated as for German DIBt regulation. Table 1: Correlation of TVOC calculated as for DIBt regulation or obtained as toluene equivalent. Measurements were done with MS or with FID. Range of ratio TVOC (DIBt) / TVOC (TE) Number of values Total <0,9 0,9 - <1 = 1 >1-1,5 >1,5-2 > In Table 1 a ratio of 1 means that both TVOC values are identical. A ratio x larger than 1 means that TVOC as in above DIBt option is x times larger than a TVOC, expressed as toluene equivalent. In 5 cases TVOC results obtained with the DIBt method were lower than those obtained as TVOC equivalent. In 2 cases TVOC the results were identical. In the majority of cases TVOC results obtained with the DIBt method was higher than those obtained as TVOC equivalent. This shows that each of these calculation methods gives different results. It is necessary to select one of these options to make sure that results are comparable. Further testing was not needed. The same applies to total SVOC determinations. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 21 / 109

22 2.13 Determination of Formaldehyde / Volatile Carbonyls (see cl. 8.3 in draft TS 16516) ISO standard is the state-of-the-art for the determination of formaldehyde and other volatile carbonyl compounds. Data on precision and uncertainty is summarized in Annex A of that standard. Another exercise for validation was performed during development of a reference material within a European project [40]. This method is validated regularly in ring trials organized by Health and Safety Laboratory (UK) within its program WASP [41]. Yrieix et al. and Salthammer et al. reported an underestimation of pentanal and hexanal when using DNPH cartridges compared to Tenax thermal desorption [19, 42] Conclusions From the review of literature about the impact of different test chamber parameters (chamber size, loading, air change rate, climate, and details of air sampling and analysis) it can be seen that only little systematic research has been performed for validation of the test chamber method (ISO , -3, - 6). Although the first inter laboratory comparison on test chamber measurements for VOC was performed in 1997 [11] no validation of this test method was established until today. The following conclusions for the experimental validation work were drawn from the collected available data. Preparation of test specimen No data was available comparing the effect of different sealing techniques for the test of flooring and other materials having cutting edges. Therefore a systematic study of the impact of sealing techniques on test result was performed within the experimental part of the robustness validation program, see chapter Test chamber conditions (chamber size) Only little data originating from tests in different chambers within one laboratory are available. Most of these data are only from round robin tests. Therefore a systematic study of the impact of chamber size on test result was performed within the experimental part of the robustness validation program, see chapter Test chamber conditions (climate conditions) Some studies had been done on the impact of temperature and relative test chamber air humidity with contradictory results depending on volatility and polarity of compounds. Therefore a systematic study of the impact of test chamber climate parameters on test result was performed within the experimental part of the robustness validation program with a solid, a porous and a liquid product, see chapter Test chamber conditions (ventilation, loading and area specific air flow rate) No study had been done investigating separately the impact of ventilation, loading, and area specific air flow rate on the emission rate. Therefore a systematic study was performed within the experimental part of the robustness validation program, see chapter Test chamber conditions (intermediate storage) There is data available showing that intermediate storage can cause contamination if different samples are stored together. There is also an influence on SVOC concentration when samples are placed in the test chambers only three days before testing and else stored outside the test chamber. No further tests are needed, see chapter 2.6. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 22 / 109

23 Tube conditioning and laboratory blank tubes Some artefacts are known to be generated from Tenax TA tubes which affect the tube blank value. The most problematic one is benzene. To improve the detection limit for benzene a systematic study on significance and control of benzene artifact generation was performed within the robustness validation program, see chapter Sampling test chamber air Available data did not show any influence of air sampling volume or air sampling flow within the specified range. No further tests are needed, see chapter 2.9. Capillary GC column The available data can be regarded as sufficient. Use of a slightly polar column is rated as beneficial. No further tests are needed, see chapter 2.7. Calibration and analysis The analytical procedures as specified in document N 129 and ISO , -9 and -11 allow a number of free selections regarding details of analysis procedure for VOCs sampled on Tenax TA and analyzed by thermal desorption, gas chromatography and mass spectroscopy. A harmonized analytical procedure as prescribed in clause 8 of draft CEN/TS is expected to help achieving smaller differences between test results obtained by different testing laboratories. Nevertheless, no further tests are needed. Increased use of reference material for validating elements of the testing procedure is recommended in draft CEN/TS Furthermore, establishment of applicability of a new reference material for validating the whole emissions testing procedure with known emissions of toluene is recommended, see chapter It was also deemed necessary to harmonize the interpretation/integration of chromatograms. Determining the emissions of total volatile organic compounds There are different definitions for the determination of TVOC. One of these had to be selected for achieving comparable results. No further tests are needed, see chapter Determination of formaldehyde and some other carbonyl compounds For formaldehyde, butanal, propanal, acetone and crotonaldehyde ISO can be used. For some other carbonyl compounds, such as pentanal and hexanal, underestimation is reported, when compared to ISO No further tests are needed, see chapters 2.13 and Validation of CEN/TC 351/WG 2 draft CEN/TS Page 23 / 109

24 3. Robustness validation of draft horizontal VOC emissions testing standard 3.1 Project design and selection of test samples Robustness validation of the draft emission testing standard, now draft CEN/TS 16516, had been performed by investigation of the impact on test result if selected testing parameters were modified one by one. Test samples were acquired from industry. The samples were selected in such a manner that they would represent different product types and different emissions mechanisms, and that they were expected to show measurable emissions even after 28 days. One sample even was spiked with some VOCs for better use within this validation project. This means that the test results do not represent the whole group of products at all. The test results reported here only represent the especially selected testing samples. 3.2 Homogeneity testing The samples were acquired from industry and then tested by BAM for homogeneity. Any effect of testing parameters observed in this report needs to show larger differences than the reported inhomogeneity for being rated as relevant. Homogeneity tests were run earlier and thus with fresher samples than all other tests. They had been performed before the test samples were dispatched to the laboratories for characterizing the samples sent. Findings Homogeneity testing showed variation of emissions within the test samples between less than 10% and some 20% relative standard deviation of 4 or 6 test results for most products. Only one product (wooden flooring type products, "D") showed higher variation and this only because one out of 6 test results was twice the other test results. 50% 40% 30% 20% 10% 0% Product inhomogeneity A1 A2 A3 A4 A5 B1 C1 C2 C3 C4 C5 D1 E1 E1 E3 F1 G1 Fig. 7 Inhomogeneity of products (see Annex A.3), given as relative standard deviation of test results A = Flooring type product, B = Mineral wool type product, C = Liquid type product, D = Wooden flooring type product, E = Foam type product, F = Solid product with high emissions from back, G = Wood-based panel type products. Numbers after letter = individual VOCs, see results in table in Annex A.3. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 24 / 109

25 3.3 Testing program and test results The test program was designed in a manner that test chamber parameters were varied one by one for detecting their impact on VOC emission. In theory, the area specific emission rate (the emitted mass of substance per hour and per square meter) should remain constant during comparative testing. But it cannot be expected that this area specific emission rate would be exactly the same in all individual tests due to influences such as ageing of samples, product inhomogeneity, and analytical differences between the involved laboratories Work package 1: Temperature and humidity Goal: Study the impact on test result of changed temperature in test chamber in the range 19 C 27 C with constant absolute humidity of chamber supply air; changed RH of chamber supply air in the range 45% 55% with constant temperature in test chamber for three materials emitting formaldehyde based on different release mechanisms, and three materials emitting VOC with different emissions characteristics. Temperature range was larger than the range accepted in draft CEN/TC 351 standard for better being able to detect any correlations with test results larger than material inhomogeneity and analytical uncertainty. Detailed findings Temperature in test chamber: The tested wooden flooring type product showed some increase of emissions with higher temperature but only for formaldehyde emissions (with a doubling of emission rate when increasing temperature from 19 C to 27 C). There was no clear trend for the other VOCs. The tested wood-based panel type product showed no clear trend, even for formaldehyde. It should be noted that other but much older investigations [e.g. 43] though under different testing setup had shown a stronger impact on the area specific emission rate of formaldehyde. The tested mineral wool type product showed an increase only of formaldehyde emissions with higher temperature, but only after three days, not later on. The same product showed a slight decrease of ethylhexanol after three days. But these differences were not much larger than inhomogeneity of the material and analytical uncertainty. And it cannot be precluded that this VOC was emitted from the aluminium tape used for sealing back and edges, and not from the product itself. The tested flooring type product showed some increase of emissions with higher temperature (with a doubling of emission rate when increasing temperature from 19 C to 27 C for some VOCs, especially for hexanal and for hexanoic acid, but not for acetic acid). The tested liquid type product showed high variation of results. Experience from equivalent tests allows explaining this with very high initial concentrations of VOC and of water in test chamber. No observable trend could be detected that would have been stronger than this variation. The tested foam type product showed large decrease of emissions of the VVOC n-pentane over storage time before start of test. No observable trend of emissions with increased temperature could be detected that would have been stronger than this decrease. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 25 / 109

26 150% 100% 50% 0% Figure 8 Examples of change of specific emission rate with temperature (in C); shown as % of test result obtained at 23 C after 4 weeks in test chamber (see Annex A.4.1). - Formaldehyde, emitted from wooden flooring type products, wood-based panel type products, Mineral wool type product. 200% 150% 100% 50% 0% Emission rate and temperature Formaldehyde Wooden flooring, wood based panel, and mineral wool type product Emission rate and temperature Acetic acid (3 x), hexanal, DEP Figure 9 Examples of change of specific emission rate with temperature (in C); shown as % of test result obtained at 23 C after 4 weeks in test chamber (see Annex A.4.1). - Acetic acid, emitted from wooden flooring type products, wood-based panel type products, flooring type product. - Hexanal, emitted from flooring type product. - DEP, emitted from liquid type product. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 26 / 109

27 Relative humidity of test chamber inlet air: The tested wooden flooring type product showed a slight increase of formaldehyde emissions of some 20% (in some cases even some 50%) with higher relative humidity in the range 45% 55%. This variation was slightly above analytical uncertainty and material inhomogeneity. There was no clear trend for other VOCs. The tested wood-based panel type product showed an increase of emissions (+50% 100%) of hexanal with higher relative humidity in the range 45% 55%. There was no clear trend for other VOCs, even formaldehyde. It should be noted that other but much older investigations [e.g. 43] though under different testing setup had shown a stronger impact on area specific emission rate of formaldehyde. 150% 100% 50% 0% The tested mineral wool type product showed an only slight increase of formaldehyde emissions (+20%) with higher relative humidity in the range 45% 55%. This variation was close to analytical uncertainty and material inhomogeneity. There was no clear trend for the other VOCs. The tested flooring type product showed no clear trend of VOC emissions with increased relative humidity. The tested liquid type product showed high variation of specific emission rates because of very high initial concentrations of VOC and of water in test chamber. No observable trend of emissions with increased relative humidity could be detected that would have been stronger than this variation. The tested foam type product showed large decrease of emissions over time. No observable trend of emissions with increased relative humidity could be detected that would have been stronger than this decrease. Emission rate and relative humidity Formaldehyde Wooden flooring, wood based panel, and mineral wool type product Figure 10 Examples of change of specific emission rate with relative humidity (in %); shown as % of result obtained at 50%RH after 4 weeks in test chamber (see Annex A.4.1). - Formaldehyde, emitted from wooden flooring type products, wood-based panel type products, Mineral wool type product. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 27 / 109

28 200% Emission rate and relative humidity Acetic acid (3 x), hexanal, DEP 150% 100% 50% 0% Figure 11 Examples of change of specific emission rate with relative humidity (in %); shown as % of result obtained at 50%RH after 4 weeks in test chamber (see Annex A.4.1). - Acetic acid, emitted from wooden flooring type products, wood-based panel type products, flooring type product. - Hexanal, emitted from flooring type product. - DEP, emitted from liquid type product. Wet applied products: The tested liquid type product that has been wet applied on an inert substrate showed large variation of test results. This can be explained by the very high concentrations in test chamber in the early phase of emissions testing: High initial VOC concentrations can be partly adsorbed at, and later re-desorbed from test chamber walls, leading to contribution of "early" emissions to "later" emissions, even to those tested after e.g. 14 or 28 days. This is only partly predictable because these sink effects depend on test chamber geometry and air flow through test chamber, and this can easily differ between different types of test chambers. It should be noted that the tested product was especially prepared for showing high emissions. High initial water concentrations can occur in case of water-based wet-applied products with high water content. This can lead to temporary condensation of water on test chamber walls. It cannot be excluded that some VOCs then can be dissolved temporarily in such condensed water, thus impairing the test result at certain points of time. During periods with high humidity in the test chamber even some condensation of water in air sampling tubes cannot be excluded. This water can contain dissolved VOCs and then induce high VOC test results if such condensation occurs. Kramberger et al. [44] reported relative humidity around 100% in test chamber air when they placed 4 test specimens of plaster or putty products into different test chambers. Application amount was g/m², and loading factor was 1,4 m²/m³ (scenario = walls + ceiling). It took between 5 and 8 days before humidity in test chamber air decreased down to 50%. This underlines the relevance of high initial water concentrations in test chamber air. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 28 / 109

29 Summary of findings, conclusions Some increase of emission rate with higher temperature was observed for some VOCs and some of the tested products, but in most cases to a low extent, and not for all VOCs and all products. Both some increase and some decrease of emission rate with higher relative humidity was observed for some VOCs and some of the products, but to a lower extent than for temperature, and not for all VOCs and all products. In some cases contradictory trends were observed for substances with different chemical and/or physical properties, within one set of comparative experiments. CEN/TC 351/WG 2 decided to maintain the acceptable range of temperature and relative humidity within ± 1 C and ± 5% RH, but with further specifications allowing short-term higher deviations, as long as the overall stability of temperature and relative humidity is given. A narrower interval cannot be justified by the obtained test data. A broader interval was rejected with the intention to avoid any impact of temperature and relative humidity on emissions, even for substances and products with highest climate sensitivity of emission rates. Wet-applied products High initial concentrations of water and of VOC in test chamber disturbed reliability of VOC determination after 3 days, but also impacted testing after 14 days and 28 days. It is important to understand that this will not happen in reality because these initial concentrations will be adsorbed on walls made of wood, gypsum or concrete, and these adsorbed VOCs will not be desorbed from these surfaces as fast as from stainless steel test chamber walls. A possible solution is a separate pre-conditioning period outside the testing chamber before start of actual test that can avoid such effects. This would reflect the fact that rooms with freshly coated surfaces normally will not be used immediately after coating, or at least with increased ventilation. Test specimens then are placed each in a separate environment (e.g. another chamber) during e.g. 3 days. Any water and VOCs initially adsorbed on walls then will stay in the pre-conditioning device and no longer impact emission testing. In that case only a dry or pre-dried product would be tested in actual test chamber this would increase reliability of testing significantly. If a pre-conditioning period is applied for certain wet-applied products then it is essential to ensure that no cross-contamination between different test specimens can occur. Separate pre-conditioning chambers or similar solutions are specified in draft CEN/TS 16516, with ventilation and climate parameters being similar to actual test chamber, in any case within the range accepted by draft horizontal standard. It is important that the time of transfer of test specimen into actual test chamber is regarded as starting time of test, and no emission test shall be performed at all immediately after that transfer. Stable air concentrations in the test chamber require efficient mixing during several air changes before a reliable air measurement is possible. Formaldehyde Formaldehyde showed smaller influence of changes in temperature and humidity than expected. Observable effects were seen mainly for mineral wool type products and for wooden flooring type products, but not for wood-based panel type products. Therefore results obtained with the testing parameters specified in EN ( (23 ± 0,5) C and (45 ± 3)% RH) showed to be comparable with VOC testing if relative humidity remains within the overlapping interval of (45-48)% RH and vice versa. It was demonstrated in chapter that a recalculation between different loading factors and ventilation rates is possible within the relevant range. This is in contradiction with theoretical calculations of formaldehyde emissions that have been reported earlier [43] for a limited number of wood-based panels showing higher ranges of formaldehyde emissions than investigated in this study. A hypothesis is that the "Andersen formula" type theoretical calculations may not give precise predictions of experimentally determined formaldehyde emissions from low emitting wood-based type products as they are manufactured today, especially if the final product is coated. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 29 / 109

30 3.3.2 Work package 2.1: Chamber sizes Goal: Study the impact on test result (in terms of area specific emission rate) when using 4 different test chamber sizes (20 l, 119 l, 1 m³, 3 m³) with constant loading factor and ventilation rate with products of different susceptibility to this parameter. According to theory, the area specific emission rate (µg/m²h) should remain constant in the selected range. Detailed findings The tested foam type product showed a slight tendency to higher specific emission rate in a smaller test chamber for some VOCs, namely styrene and ethyl benzene. This could be explained by the different geometry of such samples in different chamber volumes. Even if all edges and the back have been covered by aluminium foil, it cannot be precluded that some small gaps between surface and foil lead to unexpected additional emissions. As height of test specimen had to be kept constant, the surface of edges and back relative to top surface is larger, and therefore the impact of these additional emissions on test result from edges is stronger in smaller test chambers than in larger ones. The VVOC n-pentane showed the opposite trend. This could be explained by the fact that n-pentane can be released already during the preparation of the test specimen. The smaller the test chamber, the larger is the total surface of the test specimen relative to its volume, and the larger is the potential loss of n-pentane before the test specimen is prepared and placed in the test chamber. Further to that, air sampling on Tenax TA is not a good method for VVOC monitoring due to low adsorption capacity of very volatile substances. This fact may impair the reliability of this finding. The tested liquid type product showed some differences between test results in test chambers of different size, but these did not show a systematic trend. The reference material showed a satisfactory correlation between all test chambers except the 3 m³ test chamber which had the lowest loading factor. It could not be resolved whether the model for calculating expected emissions does not fit well under these conditions, or whether the 3 m³ test chamber (where also a lower loading factor had been applied) showed lower recovery (55% after 2 days and 62% after 3 days) than the involved smaller test chambers. 200% 150% 100% 50% Emission rate and test chamber size L=3,2 m³, M=1 m³, S=0,12 m³, VS=0,02 m³ Styrene (3d, 14d, 28d), DEP (28 d), n pentane (3d) 0% L M S VS L M S VS L M S VS L M S VS L M S VS Figure 12 Examples of specific emission rate in test chambers of different size; shown as % of result obtained in 1 m³ chamber after 4 weeks in test chamber (see Annex cl. A and A.4.2.2). - Styrene and n-pentane emitted from foam type product - DEP emitted from liquid type product Validation of CEN/TC 351/WG 2 draft CEN/TS Page 30 / 109

31 200% Emission rate and test chamber size L=3,2 m³, M=1 m³, S=0,12 m³, VS=0,02 m³ Toluene, reference material (1d, 2d, 3d) 150% 100% 50% 0% L M S VS L M S VS L M S VS Figure 13 Examples of specific emission rate in test chambers of different size; shown as recovery as % of predicted emission rate - Toluene emitted from spiked films (reference material)(see Annex cl. A.4.2.3). Note: The test in large test chamber (3,2 m³) was performed at lower loading factor than in the other test chambers. Summary of findings, conclusions All in all, the involved test chambers with volumes between 20 liters and 3 m³ were comparable because no general trend was observable. It could not be seen that larger or smaller chamber size always and systematically would induce different area specific emission rate. Any limitations of largest and smallest acceptable test chamber size only can be founded on inhomogeneity of sampling material: Larger test specimens, as required for larger test chambers, will rule out inhomogeneity of tested products better than smaller test specimens for smaller test chambers. The achieved data confirmed the present specifications in the draft horizontal testing standard. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 31 / 109

32 3.3.3 Work package : Loading factor and ventilation Goal: Study the impact on test result (in terms of area specific emission rate) of different air change rates with constant loading factor (0,4 m²/m³), different loading factors with constant air change rate, and with air change rate and loading factor changed simultaneously (with constant area specific air flow rate q at three levels). According to theory, the area specific emission rate (µg/m²h) should remain constant in the selected range. Detailed findings Flooring type product: A clear trend of VOC emissions when changing testing parameters could not be observed. 200% 150% 100% 50% 0% Figure 14 Examples of specific emission rate in test chambers at different loading factors and ventilation rates: Left: 0,3 / 0,7 / 1,5 m²/m³ with ventilation 0,7 /h; shown as % of 0,7 m²/m³ Center: 0,3 / 0,7 / 1,5 /h with loading factor 0,7 m²/m³; shown as % of 0,7 /h Right: 0,3 / 0,7 / 1,5 m²/m³ with ventilation 0,3 / 0,7 / 1,5 /h; shown as % of 0,7 m²/m³ with 0,7 /h Blue = 1 m³ test chamber, red = 0,125 m³ test chamber; all after 4 weeks in test chamber (see Annex cl. A.4.3.1). 200% 150% 100% 50% Emission rate, loading, ventilation Flooring type product Acetic acid (28 d) 0,3 0,7 1,5 0,3 0,7 1,5 0,3 0,7 1,5 Emission rate, loading, ventilation Flooring type product Hexanal (28 d) 0% 0,3 0,7 1,5 0,3 0,7 1,5 0,3 0,7 1,5 Figure 15 Examples of specific emission rate in test chambers at different loading factors and ventilation rates: details as in figure 14 (see Annex cl. A.4.3.1). Validation of CEN/TC 351/WG 2 draft CEN/TS Page 32 / 109

33 Foam type product: No clear trend of emissions of styrene and ethyl benzene when changing testing parameters could be observed. Results for the VVOC n-pentane were strongly influenced by age of sample. This dominated any possible other effect. The tested foam type product showed large decrease of emissions of n-pentane over time. No observable trend could be detected that would have been stronger than this decrease. 200% 150% 100% 50% 0% Figure 16 Examples of specific emission rate in test chambers at different loading factors and ventilation rates: Left: 0,3 / 0,7 / 1,5 m²/m³ with ventilation 0,7 /h; shown as % of 0,7 m²/m³ Center: 0,3 / 0,7 / 1,5 /h with loading factor 0,7 m²/m³; shown as % of 0,7 /h Right: 0,3 / 0,7 / 1,5 m²/m³ with ventilation 0,3 / 0,7 / 1,5 /h; shown as % of 0,7 m²/m³ with 0,7 /h Blue = 1 m³ test chamber, red = 0,255 m³ test chamber; all after 4 weeks in test chamber (see Annex cl. A.4.3.2). 200% 150% 100% 50% 0% Emission rate, loading, ventilation Foam type product styrene (14 d) 0,3 0,7 1,5 0,3 0,5 0,7 1,5 0,3 0,7 1,5 Emission rate, loading, ventilation Foam type product n pentane (14 d) 0,3 0,7 1,5 0,3 0,5 0,7 1,5 0,3 0,7 1,5 Figure 17 Examples of specific emission rate in test chambers at different loading factors and ventilation rates: - details as in figure 16 (see Annex cl. A.4.3.2). Validation of CEN/TC 351/WG 2 draft CEN/TS Page 33 / 109

34 Liquid type product: There was a large variation of test results, probably due to very high concentration of both water and VOCs in test chamber in the early phase of emission testing, as outlined earlier in more detail (see discussion in chapter on work package 1). No clear trend for VOC emissions was observed when changing testing parameters. Air velocity over test specimen did not show any effect within the range 0,1 0,5 m/s. 500% 400% 300% 200% 100% 0% Figure 18 Examples of specific emission rate in test chambers at different loading factors and ventilation rates: Please note that the scale of the Y axis (% values) is different from the other tables in this chapter. Left: 0,3 / 0,7 / 1,5 m²/m³ with ventilation 0,7 /h; shown as % of 0,7 m²/m³ Center: 0,3 / 0,7 / 1,5 /h with loading factor 0,7 m²/m³; shown as % of 0,7 /h Right: 0,3 / 0,7 / 1,5 m²/m³ with ventilation 0,3 / 0,7 / 1,5 /h; shown as % of 0,7 m²/m³ with 0,7 /h Blue = 1 m³ test chamber, red = 0,115 m³ test chamber, green = 0,024 m³ test chamber; all after 4 weeks in test chamber (see Annex cl. A.4.3.3). 200% 150% 100% 50% 0% Emission rate, loading, ventilation Liquid type product butyldiglycol (28 d) 0,3 0,7 1,5 0,3 0,7 1,5 0,3 0,7 1,5 Emission rate, loading, ventilation Liquid type product diethylphthalate (28 d) 0,3 0,7 1,5 0,3 0,7 1,5 0,3 0,7 1,5 Figure 19 Examples of specific emission rate in test chambers at different loading factors and ventilation rates: - details as in figure 18 (see Annex cl. A.4.3.3). Validation of CEN/TC 351/WG 2 draft CEN/TS Page 34 / 109

35 200% Emission rate, air velocity Liquid type product 0,1 / 0,3 / 0,5 m/s Butyldiglycol, diethylphthalate (28 d) 150% 100% 50% 0% 0,1 0,3 0,5 0,1 0,3 0,5 Figure 20 Examples of specific emission rate in test chambers with different air velocity above test specimen; shown as % of result with 0,3 m/s after 4 weeks in test chamber (see Annex cl. A ). - Butyldiglycol emitted from liquid type product - Diethylphthalate emitted from liquid type product Summary of findings, conclusions The solid products showed the expected constancy of area specific emission rates for the involved VOC. Only the VVOC n-pentane did not show the same trend in case of foam type product because any possible trend was overruled by a strong decrease of n-pentane emissions over storage time before start of test. The tested liquid type product showed: High variation of area specific emission rate resulted in the fact that for some VOCs (e.g. butyldiglycol, TXIB) there was no clear trend observable that were larger than the variation of the analytical results. For some other VOCs (e.g. diethylphthalate) there were indications that higher ventilation rate resulted in lower area specific emission rates, and higher loading factor gave higher area specific emission rates. Changing the air velocity over test specimen did not show any effect within the range 0,1 0,5 m/s. Analysis of the achieved data confirmed the present specifications of accepted ranges of loading factor and ventilation rate in the draft horizontal testing standard. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 35 / 109

36 3.3.4 Work package : Age of sample at start of test The studies in work packages revealed unintended but valuable information on the impact of sample age on test result. Findings, conclusions The freshly produced foam type product showed strong decrease of emissions of n-pentane between start of testing from August to December in both involved laboratories, giving lower test results the later the test was started. The linoleum flooring did not show that effect. This can be explained by the fact that linoleum has to cure a sufficient time before dispatch which will avoid elevated emissions initially after receipt at either customer or testing laboratory. The liquid type product did not show that behavior because each test specimen is produced freshly from the original product which was stored in a well tightened container. The other products did not show any observable ageing effects on emissions during the investigation period of 5 months as long as the sample was packaged properly and air tight. The earlier specification ("start of test maximum 8 weeks after taking the sample") can be too long for products that are very sensitive to loss of emissions, but it can be too restrictive for products with more stable emissions. In future product specific standards, a specification on maximum age of sample at start of test could be differentiated per type of product when knowing the respective behavior of emissions over storage time. 400% 300% 200% 100% 0% Aug Emission rate, age of sample Foam type product n pentane (28 d), 3 labs Lab 1 Lab 2 Lab 3 Sept Oct Nov Sept Oct Figure 21 Examples of specific emission rate depending on age of a sample showing strong decrease of emissions; shown as % of result of the earliest test performed by the respective lab in September; - n-pentane emitted from foam type product; 3 different labs; all results after 4 weeks in test chamber (see Annex cl. A.4.1.6, A.4.2.1, A 4.3.2). It is worth noting that other samples did not show the same decrease of specific emission rate, see text. Nov Aug Sept Oct Validation of CEN/TC 351/WG 2 draft CEN/TS Page 36 / 109

37 3.3.5 Work package 1, and : Determination of hexanal The studies in work packages 1 and revealed unintended but valuable information on the impact of selection of Tenax TA based (ISO ) or DNPH based (ISO ) air sampling and analysis on test result. Some analyses were conducted in parallel with both above mentioned methods. Earlier findings [42] led to the conclusion that DNPH method gives lower values for aldehydes with more than 4 carbon atoms (e.g. butanal) than Tenax TA method. Findings, conclusions Earlier findings on higher results obtained with Tenax TA method have not been confirmed. DNPH method gave higher results than Tenax TA results in all cases. Difference was small in many cases, but strong in several other cases. The horizontal testing standard (draft CEN/TS 16516) specifies only one testing method for hexanal and other carbonyl compounds for reducing systematic differences of test results between laboratories. 100% 80% 60% 40% 1 5 Determination of hexanal Tenax TA result as % of DNPH result Range: µg/m²h Median 87%; 100% would mean identical results Figure 22 Determination of n-hexanal with Tenax TA (ISO ) or DNPH (ISO ) - Tenax TA result (ISO ) shown as % of DNPH result (ISO ), plotted against number of test (X axis)( see Annex cl. A.4.1.4, A , A , A ) Validation of CEN/TC 351/WG 2 draft CEN/TS Page 37 / 109

38 3.3.6 Work package 1, and : Determination of the VVOC n-pentane The studies in work packages 1 and 2.2 revealed unintended but valuable information on challenges when determining high emissions of the very volatile organic compound (VVOC) n-pentane. Findings, conclusions Parallel sampling of test chamber air with different air sampling volumes gave highly different results as long as the sample was not too old for determining any n-pentane emissions. But sampling on Tenax TA is not an appropriate method for determining n-pentane because the adsorption capacity of Tenax TA is too small for an appropriate air sampling of this very volatile compound. Use of another adsorbent will help solving this challenge. This finding is not applicable to all VVOC. ISO and Annex D of ISO (2011 version) give guidance on selection of an appropriate testing method. 100% 80% 60% 40% 20% 0% Determination of VVOC n pentane: 5,4 l air sample % of 2,7 l air sample Figure 23 Air sampling volume and test result when tubes are overloaded with n-pentane: Result obtained with 5,4 l air sampling volume, shown as % of the result obtained with 2,7 l air sampling volume; plotted against area specific emission rate (µg/m²h) obtained with 2,7 l air sampling volume (X axis) (see Annex cl. A.4.1.6) Validation of CEN/TC 351/WG 2 draft CEN/TS Page 38 / 109

39 3.3.7 Work package 3: Techniques for sealing back and edges Goal: Study the impact on test result (in terms of area specific emission rate) of different techniques for sealing back and edges of a flat solid product if emissions from back or edges are much higher than emissions from top surface and if only the top surface is in contact with indoor air under intended use conditions. Findings, conclusions An investigation of the impact of techniques used for sealing back and edges on VOC area specific emission rate showed the effectiveness of different standard sealing techniques, but with minor ranking in effectiveness. The isolation of back and edges resulted in a complete or almost complete suppression of emissions from the back. Most efficient techniques were back to back storage of plates, with edges covered with aluminium tape, tight coverage of edges and back with aluminium foil (almost identical result), seal box as specified in JIS A Inclusion of a joint in wooden flooring type products test specimen did not cause any difference in area specific emission rate when compared with test specimens not including a joint. Less efficient sealing techniques were pressing into a tray, fixing on a plate with tape (both specified in California CDPH Section 1350). Any purchased aluminium tape should be checked for its emissions in á test chamber after being applied on a glass or metal plate. Aluminium tapes are available that show only very low emissions of 2-3 glue specific VOCs, and only during the first 3 days. If a tape is covering rough surfaces (e.g. wood) then slightly higher emissions from the tape can be observed in the low µg/m³ range even later than after 3 days. The achieved data confirmed the present specifications in the draft horizontal testing standard (draft CEN/TS 16516). It is rated important to specify the appropriate sealing techniques for specific products in product specific standards. 100% 90% 80% 70% 60% back-to-back, WFTP * Sealing techniques for sealing back and edges, after 3 days back-to-back, SPHEB fixed on glass, SPHEB seal box, SPHEB CDPH frame, SPHEB 50% Figure 24 Efficiency of sealing technique for 3 different main VOCs emitted from back side after 4 weeks in test chamber (see Annex cl. A.4.4.1, A.4.4.2); red = complete retention. *: Almost identical result for tight coverage of back and edges with aluminium foil and aluminium tape, including a joint. WFTP: Wooden flooring type product; SPHEB: Solid product with high emissions from back Validation of CEN/TC 351/WG 2 draft CEN/TS Page 39 / 109

40 3.3.8 Work package 4: Reference material for method validation Goal: Study the significance of a solid reference material for toluene for validation of the whole method, with participation of all involved laboratories. The reference material was a thin film spiked by Virginia Tech University with definite amounts of toluene and placed into sample holders and then in the test chambers [45]. Previous investigations showed good stability of emission rates after 2 and after 3 days if stored cool all the time before test [46]. Transport of these reference samples occurred in containers with dry ice. Dry freezing was required for storage in laboratory. Findings, conclusions Most test results were between 80% and 120% recovery, especially after 2 days. This could serve as benchmark for both test chambers and for the prediction model. The use of such reference materials allows a quality check of the whole procedure including all steps from test chamber operation to VOC analysis. The draft standard recommends using this or equivalent reference material within quality assurance programs, especially for checking newly purchased test chambers. 120% 100% 80% 60% 40% 20% 0% 3,2 1,0 1,0 Reference material Films spiked with toluene 1,0 1,0 0,225 0,225 Figure 25 Recovery of toluene from spiked film after 48 hours, plotted against test chamber volume (as m³) (X axis) (see Annex cl. A.4.5). Red lines: Proposed benchmark for test chambers. It should be noted that loading factor in 3,2 m³ test chamber was much lower than in the other test chambers. 0,19 0,125 0,119 0,11 0,02 Validation of CEN/TC 351/WG 2 draft CEN/TS Page 40 / 109

41 3.3.9 Work package 5: Tenax TA tubes and benzene artefact generation Goal: Investigation of the generation of benzene during air sampling and analysis in different labs with different experimental settings, for identifying options to control and minimize benzene artefact formation. Findings, conclusions The investigation showed that some laboratories reported unexpected increase of benzene levels on Tenax TA tubes after sampling from an atmosphere known to be free of benzene. In these cases, this benzene level was higher than blank level determined from the same Tenax TA tube before air sampling. In most (but not all) of the cases benzene increased during air sampling in the low nanogram range (on the sampling tube), resulting in false chamber air concentrations most times below 0,2 µg/m³, but up to 2 µg/m³ in other cases. This can impair accuracy of benzene determination in low levels in the µg/m³ range. Lab Sample volume Increase benzene Thermal desorption Main VOCs in sampled atmosphere no. Liters ng C Minutes , DEP, texanol 2 2,8 5, n-butanol, n-hexanal, butylglycol , D3-carene, α-/β-pinene, aldehydes, n-butanol, 4 9,6 0,5 1, ,2,4,6,6-pentamethylheptane 5 0,9 2,5 0 1, cyclohexane, octane, decane, hexadecane decane, dodecane, and traces of more VOCs 7 5,4 0 1, propylene glycol, C10-C16 alkanes 8 4,3 4,7 1,6 2, organic acids and aldehydes 9 6 0,4 1, α-pinene, acetone Table 2 Formation of benzene artefacts during air sampling of atmosphere free of benzene (see Annex cl. A.4.6). If the column "increase benzene - ng" contains a 0 value then the benzene level detected in a sampling tube after sampling air from an atmosphere free of benzene was not higher than the blank value of the same sampling tube before sampling. One possible assumption is that benzene is generated on Tenax TA tubes during air sampling under certain conditions. The mechanism has not been identified. It cannot be excluded that certain VOC mixtures after adsorption undergo chemical reaction while sampling air continues to pass over the surface of Tenax TA. As benzene test results in the low µg/m³ range can be falsified by artefact generation to a significant extent, the draft standard recommends to verify low-level test results with an independent second testing method before assessing a test result showing small benzene levels against any low limit value of e.g. 1 µg/m³. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 41 / 109

42 4. Repeatability of testing within one laboratory 4.1 Study design The studies within work packages 1 and 2 delivered unintended but valuable information on the repeatability of VOC emissions testing within one laboratory. All test results analyzed for repeatability were obtained by testing two or three test specimens from the same sample under identical conditions (in terms of temperature, relative humidity, loading factor, and ventilation rate) in the same laboratory during this study. All of the testing conditions were within the accepted ranges as specified in the draft TS Each result was analyzed for each main VOC. The following tables in the Annex of this document have been analyzed for repeatability of tests that have been conducted under identical conditions (in terms of temperature, relative humidity, loading factor, and ventilation rate): Ch. A.4.1: Tests at 23 C and 50% RH (2 tables per product, 12 tables in total) Ch. A , A , A : Tests with 0,7 m²/m³ and 0,7 /h (3 tables per product and test chamber size, 6 tables in total) Ch. A , A , A : Tests with 0,7 m²/m³ and 0,7 /h (3 tables per product and test chamber size, 6 tables in total) Ch. A , A , A : Tests with 0,7 m²/m³ and 0,7 /h (3 tables per product and test chamber size, 9 tables in total) Ch. A : Tests with 0,7 m²/m³ and 0,7 /h (3 tables) This resulted in: Number of chamber tests: 36. Number of involved products: 6. Number of involved testing laboratories: 9. Number of repeatability data (= number of individual VOCs analyses in these tests): 178. o 57 data for individual VOCs from duplicate determination o 121 data for individual VOCs from tests with triplicate determination. 4.2 Findings The deviation of individual emission chamber test results from their mean value was calculated as % of their mean value. Repeatability resulted as follows: 50% of all test data showed a deviation of individual test results from their mean value below 14% (the median of all findings). 75% of all test data showed a deviation of individual test results from their mean value below 27% (75 percentile of all findings). 95% of all test data showed a deviation of individual test results from their mean value below 54% (95 percentile of all findings). Standard deviation (1 ) of all test results was 17%. The expanded uncertainty (2 ) of all test results, representing the 95% confidence interval, was 35%. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 42 / 109

43 Repeatability of VOC emissions testing depended strongly on: Inhomogeneity of the tested product. Chemical characteristics of the identified VOCs. Height of emissions when testing; small traces of VOC emissions and very high emissions are difficult to analyze. Emissions mechanisms, and emissions decay over time of the tested product. Storage duration of test sample before testing, especially in the case of the foam type product. The selected VOCs for analysis; in the case of the liquid type product not all laboratories selected the same list of VOCs for testing. This gives less data for the evaluation and statistics. 4.3 Conclusions on repeatability within one laboratory The repeatability of VOC emissions testing within one laboratory needs to be specified per product group when establishing product specific standards because it can be influenced by the above mentioned factors, and some of these are specific to certain types of products. In general, it has been found that repeatability can vary between very low (= below ± 10%) and high or even very high (± 50%, or even more in the case of the VVOC n-pentane). It has to be considered that material inhomogeneity between test specimens taken from the same received test sample contributes to this variation. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 43 / 109

44 4.4 Detailed findings per tested product 25% 20% 15% 10% 5% 0% Figure 26 Repeatability within one laboratory Wood flooring type product 30% 25% 20% 15% 10% 5% 0% Repeatability within one lab Deviation from mean value: Wooden flooring type product Repeatability within one lab Deviation from mean value: Wood based panel type product 0% 3d 10d 28d Figure 27 Repeatability within one laboratory Wood-based panel type product 0% 3d 10d 28d Formaldehyde Acetaldehyde Acetic acid Formaldehyde Acetaldehyde Hexanal Validation of CEN/TC 351/WG 2 draft CEN/TS Page 44 / 109

45 60% 50% 40% 30% 20% 10% 0% Repeatability within one lab Deviation from mean value: Flooring type product, 3 days Figure 28 Repeatability within one laboratory Flooring type product, 3 days 60% 50% 40% 30% 20% 10% 0% Repeatability within one lab Deviation from mean value: Flooring type product, 10 days Figure 29 Repeatability within one laboratory Flooring type product, 10 days 60% 50% 40% 30% 20% 10% 0% Repeatability within one lab Deviation from mean value: Flooring type product, 28 days Lab A Lab B Lab A Lab B Lab A Lab B Figure 30 Repeatability within one laboratory Flooring type product, 28 days Validation of CEN/TC 351/WG 2 draft CEN/TS Page 45 / 109

46 100% Repeatability within one lab Deviation from mean value: Foam type product, 3 days 80% 60% 40% 20% 0% n Pentane Styrene Ethylbenzene Figure 31 Repeatability within one laboratory Foam type product, 3 days 100% 80% 60% 40% 20% 0% Repeatability within one lab Deviation from mean value: Foam type product, 10 days n Pentane Styrene Ethylbenzene Figure 32 Repeatability within one laboratory Foam type product, 10 days 100% 80% 60% 40% Repeatability within one lab Deviation from mean value: Foam type product, 28 days Lab A Lab B Lab C Lab A Lab B Lab C Lab A Lab B Lab C 20% 0% n Pentane Styrene Ethylbenzene Figure 33 Repeatability within one laboratory Foam type product, 28 days Validation of CEN/TC 351/WG 2 draft CEN/TS Page 46 / 109

47 70% 60% 50% 40% 30% 20% 10% 0% Figure 34 Repeatability within one laboratory Liquid type product, 3 days 70% 60% 50% 40% 30% 20% 10% 0% Repeatability within one lab Deviation from mean value: Liquid type product, 3 days Repeatability within one lab Deviation from mean value: Liquid type product, 10 days Figure 35 Repeatability within one laboratory Liquid type product, 10 days 70% 60% 50% 40% 30% 20% 10% 0% Repeatability within one lab Deviation from mean value: Liquid type product, 28 days Lab A Lab B, 1 m³ Lab B, 24 l Lab A Lab B, 1 m³ Lab B, 24 l Lab A Lab B, 1 m³ Lab B, 24 l Figure 36 Repeatability within one laboratory Liquid type product, 28 days Validation of CEN/TC 351/WG 2 draft CEN/TS Page 47 / 109

48 5. Project results summary and interpretation This project had the goal to deliver scientific data and information for robustness validation of VOC emission chamber testing as specified in the draft horizontal testing standard CEN/TS The work was performed on basis of the earlier draft version as in the document CEN/TC 351/WG 2/N129. The experimental part of the project concentrated on delivering data for filling those information gaps that had been identified in chapter Conclusions on robustness validation of draft CEN/TS The validity of most parts of the draft testing standard could be confirmed. Only minor changes had been recommended to CEN/TC 351/WG 2, many of these of editorial nature and most of these had been confirmed by CEN/TC 351/WG 2. There are no criteria available for assigning "robustness" to a testing standard. It is only possible to show the degree of robustness of the test method against modifications of testing parameters with potential impact on the test result. It could be shown that the specifications of draft CEN/TS allow determination of emissions of volatile organic compounds with a degree of reliability that is not impaired by systematic errors from any specifications of testing parameters. Remaining uncertainty of test results is caused partly by inhomogeneity of the tested products, and partly by differences in the analytical practice in the involved testing laboratories that cannot be improved without unacceptable impact on testing costs. 5.2 Conclusions on comparability with other testing standards It has been demonstrated in above documentation that a test under slightly different testing parameters (temperature, RH, ventilation, loading) gave comparable test results as when using the testing parameters of draft CEN/TS 16516, as long as the test results either are expressed as specific emission rate, or are recalculated to the specifications of the European Reference Room (see cl. 4 of draft CEN/TS 16516). This allows some flexibility (even though within narrow limitations) of the testing laboratories. It also allows performing a test under the specifications of other testing standards (e.g. US American standards, or EN 717-1, applying other ventilation rate and different loading factors), and then obtaining the test results for CEN/TS just by recalculation, without the need of another test, see the specifications in clause 7 and the calculation formulas in clause 9 of draft CEN/TS This applies to all products except wood-based panels with formaldehyde-urea binder. There still is an on-going dispute whether determination of formaldehyde generation from those specific products requires use of the testing parameters as specified in EN without any deviations. 5.3 Further improvements Some further improvement of reliability of VOC emission testing is possible if the involved testing laboratories compare their performance in round robin tests and by testing of reference material, as specified in cl. 8.4 of draft CEN/TS This allows early determination and remediation of any non-appropriate working routines in case of bad performance. The project consortium would welcome if standardization committees inside and outside CEN would use the findings of this study, together with final text of CEN/TS 16516, for further optimization and for global harmonization of VOC emission testing standards. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 48 / 109

49 6. Bibliography 1. International Organization for Standardization (ISO), Indoor Air - Part 9: Determination of the emission of volatile organic compounds from building products and furnishing - Emission test chamber method (ISO :2006), Beuth, Berlin 2. International Organization for Standardization (ISO), Indoor Air - Part 3: Determination of formaldehyde and other carbonyl compounds - Active sampling method (ISO :2001), Beuth, Berlin 3. International Organization for Standardization (ISO), Indoor air - Part 6: Determination of volatile organic compounds in indoor and test chamber air by active sampling on Tenax TA sorbent, thermal desorption and gas chromatography using MS/FID (ISO :2004), Beuth, Berlin 4. International Organization for Standardization (ISO), Indoor air - Part 11: Determination of the emission of volatile organic compounds from building products and furnishing - Sampling, storage of samples and preparation of test specimens, Beuth, Berlin 5. International Organization for Standardization (ISO), Indoor, ambient and workplace air - Sampling and analysis of volatile organic compounds by sorbent tube/thermal desorption/capillary gas chromatography - Part 1: Pumped sampling (ISO :2000), Beuth, Berlin 6. CEN/TR Construction products - Assessment of release of dangerous substances - Complement to sampling, California Department of Public Health and California Health and Human Services Agency, Standard method for the testing and evaluation of volatile organic chemical emissions from indoor sources using environmental chambers - Version 1.1, Gemeinschaft Emissionskontrollierte Verlegewerkstoffe, Klebstoffe und Bauprodukte e.v. (GEV), Bestimmung flüchtiger organischer Verbindungen zur Charakterisierung emissionskontrollierter Verlegewerkstoffe, Klebstoffe, Bauprodukte und Parkettlacke (in German; Determination of volatile organic compounds for the characterization of emission controlled flooring materials, adhesives, building materials and parquet lacqeurs), Deutsches Institut für Bautechnik (DIBt), Grundsätze zur gesundheitlichen Bewertung von Bauprodukten in Innenräumen (in German; Principles of health evaluation of building products), in: DIBt-Mitteilungen 5/2010, 2010: p Confidential data in unpublished work that was allowed to be used in this study 11. European Commission - Joint Research Centre - Environment Institute, ECA Report No 21 - European Inter-laboratory Comparison on VOC emitted from building materials and products, Office for Official Publications of the European Communities, Luxembourg 12. Kim, S., Kim, J.-A., Kim, H.-J., Hyoung Lee, H. and Yoon, D.-W., The effects of edge sealing treatment applied to wood-based composites on formaldehyde emission by desiccator test method. Polymer Testing, (7): p Windhövel, U. and Oppl, R., Praktische Überprüfung des Konzepts zur gesundheitlichen Bewertung von Bauprodukten (in German; Practicable evaluation of the approach for health evaluation of building products). Gefahrstoffe - Reinhaltung der Luft, (3): p Kirchner, D., Emissionsmessungen auf dem Prüfstand (in German; Emissions tests put to test). DIBt Mitteilungen, 2007: p Validation of CEN/TC 351/WG 2 draft CEN/TS Page 49 / 109

50 15. Horn, W. Wiegner, K., Wilke, O., Jann, O., Richter, M., Kalus, S., Brödner, D., Juritsch, E., Till, C., Entwicklung eines allgemeinen, externen Qualitätsmanagementsystems für den Nachweis von relevanten chemischen Schadstoffen der produktemission und in Innenräumen (in German; General, external quality management system for the determination of relevant chemical compounds from the emission of products or in indoor air), BAM Bundesanstalt für Materialforschung und -prüfung, Final Report, Berlin 16. Wilke, O., Horn, W., Wiegner, K., Jann, O., Bremser, W., Brödner, D., Kalus, S., Juritsch, E., Till, C., Investigations for the improvement of the measurement of volatile organic compounds from floor coverings within the health-related evaluation of building products, BAM Bundesanstalt für Materialforschung und -prüfung, Final Report, Berlin Hansen, V., Larsen, A. and Wolkoff, P. Nordic round-robin emission testing of a lacquer: Consequences of product in-homogeneity. in Proceedings of Healthy Buildings Espoo, Finland. 18. European Commission - Joint Research Centre - Environment Institute, ECA Report No 13 - Determination of VOCs emitted from indoor materials and products - Inter-laboratory Comparison of small chamber measurement, Office for Official Publications of the European Communities, Luxembourg 19. Yrieix, C., Dulaurent, A., Laffargue, C., Maupetit, F., Pacary, T. and Uhde, E., Characterization of VOC and formaldehyde emissions from a wood based panel: Results from an inter-laboratory comparison. Chemosphere, (4): p GUT-Ringversuch Bestimmung flüchtiger organischer Komponenten aus einer Teppichfliese nach dem Prüfkammerverfahren (in German; Determination of volatile organic compounds of a carpet tile with test chamber methodology), Internal working document of GUT 21. Wilke, O., Jann, O., Brödner, D. and Rother, I. Comparison of different types of emission test chambers and cells regarding VOC- and SVOC-emission. in Proceedings of Indoor Air Conference Beijing, China. 22. Jann, O., Wilke, O., Brödner, D., Entwicklung eines Prüfverfahrens zur Ermittlung der Emission flüchtiger organischer Verbindungen aus beschichteten Holzwerkstoffen und Möbeln (in German; Development of a test method for the determination of volatile organic compounds from coated wood based materials and furniture), BAM Bundesanstalt für Materialforschung und -prüfung, Final Report, UBA-Texte 74/99, Berlin 23. Sollinger, S., Levsen, K. and Wünsch, G., Indoor air pollution by organic emissions from textile floor coverings. Climate chamber studies under dynamic conditions. Atmospheric Environment. Part B. Urban Atmosphere, (2): p Katsoyiannis, A., Leva, P. and Kotzias, D., VOC and carbonyl emissions from carpets: A comparative study using four types of environmental chambers. Journal of Hazardous Materials, (2): p De Bortoli, M., Kephalopoulos, S., Kirchner, S., Schauenburg, H. and Vissers, H., State-of-the-Art in the Measurement of Volatile Organic Compounds Emitted from Building Products: Results of European Inter laboratory Comparison. Indoor Air, (2): p Oppl, R. and Winkels, K. Uncertainty of VOC and SVOC measurement - how reliable are results of chamber emission testing? in Proceedings of Indoor Air Monterey, USA. 27. Oppl, R., Reliability of VOC emission chamber testing - progress and remaining challenges. Gefahrstoffe - Reinhaltung der Luft, (3): p Validation of CEN/TC 351/WG 2 draft CEN/TS Page 50 / 109

51 28. Deutsches Institut für Normung (DIN), Wood based panels - Determination of formaldehyde release - Part 1: Formaldehyde emission by the chamber method; German version EN 717-1:2004, Beuth, Berlin 29. Wilke, O., Jann, O. and Brödner, D. Effects of temperature, humidity, air exchange rate, loading factor and storage-conditions on VOC-emissions. in Proceedings of Indoor Air Conference Edinburgh, Scotland. 30. Sollinger, S., Levsen, K. and Wünsch, G., Indoor pollution by organic emissions from textile floor coverings: Climate test chamber studies under static conditions. Atmospheric Environment, (14): p Gene Tucker, W., Emission of organic substances from indoor surface materials. Environment International, (4): p Wolkoff, P., Impact of air velocity, temperature, humidity, and air on long-term voc emissions from building products. Atmospheric Environment, (14-15): p Jann, O., Wilke, O. and Brödner, D. VOC-emissions from furnitures and coated wood based products. in Proceedings of Healthy Buildings Conference Washington DC, USA. 34. European Commission - Joint Research Centre - Environment Institute, ECA Report No 16 - Determination of VOCs emitted from indoor materials and products - Second Inter-laboratory Comparison of small chamber measurement, Office for Official Publications of the European Communities, Luxembourg 35. Wilke, O., Jann, O., Brödner, D., Untersuchung und Ermittlung emissionsarmer Klebstoffe und Bodenbeläge (in German; Investigations on and Determination of low-emitting adhesives and flooring materials), BAM Bundesanstalt für Materialforschung und -prüfung, Final Report, UBA-Texte 27/03, Berlin 36. European Commission - Joint Research Centre - Environment Institute, ECA Report No 8 - Guideline for the Characterization of Volatile Organic Compounds emitted from Indoor Materials and Products using Small Test Chambers, Office for Official Publications of the European Communities, Luxembourg 37. Horn, W., Jann, O., Kasche, J., Bitter, F., Müller, D., Müller, B., Environmental and health provisions for building products - Identification and evaluation of VOC emissions and odour exposure, BAM Bundesanstalt für Materialforschung und -prüfung, Final Report, UBA-Texte 21/07, Berlin 38. Richter, M., Jann, O., Horn, W. and Pyza, L. Development and validation of a new gas mixing device for low concentrations. in Proceedings of Healthy Buildings Conference Syracuse, USA. 39. Richter, M., Horn, W., Jann, O., Brödner, D. and Till, C. Application of a new gas mixing device to test adsorptive wall materials. in Proceedings of Healthy Buildings Conference Syracuse, USA. 40. Levin, J.-O., Lindahl, R., Heeremans, C.E.M. and van Oosten, K., Certification of reference materials related to the monitoring of aldehydes in air by derivatization with 2,4- dinitrophenylhydrazine. ANALYST, (9): p WASP: Salthammer, T. and Mentese, S., Comparison of analytical techniques for the determination of aldehydes in test chambers. Chemosphere, (8): p Edmone Roffael: Die Formaldehyd-Abgabe von Spanplatten und anderen Werkstoffen. DRW- Verlag, Stuttgart, 1982, pp Validation of CEN/TC 351/WG 2 draft CEN/TS Page 51 / 109

52 44 Personal correspondence by H. Kramberger, J. Beilstein, Dr. Robert-Murjahn-Institute, Cox, S. S., Liu, Z., Little, J. C., Howard-Reed, C., Nabinger, S., and Persily, A. Diffusion-Controlled Reference Material for VOC Emissions Testing: Proof of Concept, Indoor Air, Vol. 20, No. 5, 2010, pp Howard-Reed, C., Liu, Z., Benning, J., Cox, S. S., Samarov, D., Leber, D., Hodgson, A., Mason, S., Won, D. and Little, J. C. Diffusion-Controlled Reference Material for Volatile Organic Compound Emissions Testing: Pilot Inter-laboratory Study, Building and Environment, Vol. 46, 2011, pp Literature that was taken into consideration but not included in this study Brown, V and Crump, D (2011). Optimization of analytical parameters for the determination of VOCs emitted by construction and consumer products. Proceedings of the 2011 Annual UK Review meeting on outdoor and indoor air pollution research, Cranfield University, May 2011 IEH web report (in press). ECA Report No. 2, 1989: Formaldehyde Emission from Wood-based Materials. Guideline for the Determination of Steady State Concentrations in Test Chambers. Lor, M., et. al. Horizontal Evaluation Method for the Implementation of the Construction Products Directive (HEMICPD), Final report, published in 2010 by the Belgian Science Policy, Marutzky, R., Schripp, T. Erarbeitung der Grundlagen zur Evaluierung und Aktualisierung der bauaufsichtlichen Bestimmungen für die Formaldehydabgabe aus Baustoffen: Holzwerkstoffe, Abschlussbericht zum Forschungsvorhaben für das DIBt, Yu,C. W. F. and Crump, D. R. Small Chamber Tests for Measurement of VOC Emissions from Flooring Adhesives. Indoor and Built Environment 2003; Vol. 12 (issue 5), pp Yu, C. and Crump, D. Methods for measuring VOC emission from interior paints. Surface Coating International, Vol. 83, No. 11, p , November Yu, C., Crump, D. R. Testing for Formaldehyde Emissions from Wood-based Products A Review. Indoor and Built Environment, Vol. 8, 1999, p Validation of CEN/TC 351/WG 2 draft CEN/TS Page 52 / 109

53 ANNEX: Data obtained by laboratory testing A.1 Introduction A.2 Testing Samples Testing samples were acquired from leading manufacturers. The project team expresses its gratitude to the suppliers of the test samples. The samples were selected in such a manner that they were expected to show measurable emissions even after 28 days, higher than normal products. One sample was even spiked with some VOCs for better use within this project. For these reasons, the test results cannot be linked to any typical emissions of the selected product groups; there are good reasons to assume that the measured emissions were rather higher than the emissions from normal products available in the market. Some of the suppliers asked the project team not to disclose more details because their samples were not regarded representative for any product group, or for an application area of any certain CEN TC dealing with specific product groups. The testing samples were selected for representing different types of products with respect to VOC emissions mechanisms: Flooring type product Mineral wool type product Liquid type product Wooden flooring type product Foam type product Solid product with high emissions from back Wood-based panel type product Validation of CEN/TC 351/WG 2 draft CEN/TS Page 53 / 109

54 A.3 Homogeneity testing A.3.1 Testing plan A Flooring type product 30 m² (2 m x 15 m) of the test material was delivered as a roll. Test specimens for homogeneity testing and for robustness validation exercise were cut out as shown in figure 37. X X X Figure 07: Sampling scheme for flooring type product The test specimens for homogeneity testing were cut into pieces of 20 cm x 20 cm in size. Four pieces (marked with X) were selected as depicted in Figure 37for testing with test cells of 1 l volume. The compounds with highest concentrations were monitored for the estimation of homogeneity. A Mineral wool type product X discarded sample for homogeneity testing sample for validation testing sample for homogeneity testing sample for validation testing sample for homogeneity testing discarded The test sample was delivered as a roll with dimensions 1,2 m x 7,5 m. Test specimens for homogeneity testing and for robustness validation exercise were cut out as shown in figure 38. The six samples marked BAM (10 cm x 10 cm) were selected for homogeneity testing and sealed individually in hermetic bags. Homogeneity testing took place in 24 l test chambers. Only formaldehyde was monitored for the estimation of homogeneity. Figure 38: Sampling scheme for mineral wool type product Validation of CEN/TC 351/WG 2 draft CEN/TS Page 54 / 109

55 A Liquid type product A liquid type product spiked with six VOC was delivered in a can of 20 liters. After its homogenization with a mixer the product was divided into 16 portions and filled into plastic bottles with a volume of 500 ml each. 6 out of these bottles were selected randomly for homogeneity testing in µ-chambers. Only the spiked target compounds were monitored for the estimation of homogeneity. A Wooden flooring type product Samples were taken out of two packages of wooden flooring. In each case three boards from the middle of the staple (boards 1 to 3 from package 1 and boards 4 to 6 from package 2) were cut as depicted in figure A 1 B 1 C 1 BAM 1 D 1 E 1 F 1 A: Sample 19 C C 2 A 2 B 2 BAM 2 F 2 D 2 E 2 B: Sample 23 C B 3 C 3 A 3 BAM 3 E 3 F 3 D 3 C: Sample 27 C D 4 E 4 F 4 BAM 4 A 4 B 4 C 4 D: Sample 45% r.h. 5 F 5 D 5 E 5 BAM 5 C 5 A 5 B 5 E: Sample 50% r.h. 6 E 6 F 6 D 6 BAM 6 B 6 C 6 A 6 F: Sample 55% r.h. Figure 39: Sampling scheme for wooden flooring type product The six BAM marked samples were selected for homogeneity testing and the other ones were shipped to the involved partner institutes for robustness validation testing. Homogeneity testing took place in 24 l test chambers. Edges were sealed with aluminum tape. Only formaldehyde was monitored for the estimation of homogeneity. A Foam type product Six randomly chosen test specimens of a foam type product with 17 cm x 17 cm x 10 cm in size were delivered by Eurofins. Homogeneity testing took place in 24 l test chambers. The edges were sealed with aluminum tape. The compounds with highest concentrations were monitored for the estimation of homogeneity. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 55 / 109

56 A Solid product with high emissions from back Six randomly chosen test specimens of a solid product with 20 cm x 20 cm in size were cut out of the delivered roll by Eurofins as shown in figure 40. Figure 40: Sampling scheme for solid product Four of these test specimens were randomly chosen and tested top-side with test cells of 1 l. Phenol was the only compound detected and, thus, monitored for the estimation of homogeneity. A Wood-based panel type product 12 samples of a wood-based panel type product with 19 cm x 19 cm in size were delivered as a staple by WKI. Six test specimens were randomly chosen from the middle of the staple and tested in test chambers of 24 l. The edges were sealed with aluminum tape. Only formaldehyde was monitored for the estimation of homogeneity. Validation of CEN/TC 351/WG 2 draft CEN/TS Page 56 / 109