Volatile organic compounds in 169 energy‐efficient dwellings in Switzerland

Abstract Exposure to elevated levels of certain volatile organic compounds (VOCs) in households has been linked to deleterious health effects. This study presents the first large‐scale investigation of VOC levels in 169 energy‐efficient dwellings in Switzerland. Through a combination of physical measurements and questionnaire surveys, we investigated the influence of diverse building characteristics on indoor VOCs. Among 74 detected compounds, carbonyls, alkanes, and alkenes were the most abundant. Median concentration levels of formaldehyde (14 μg/m3), TVOC (212 μg/m3), benzene (<0.1 μg/m3), and toluene (22 μg/m3) were below the upper exposure limits. Nonetheless, 90% and 50% of dwellings exceeded the chronic exposure limits for formaldehyde (9 μg/m3) and TVOC (200 μg/m3), respectively. There was a strong positive correlation among VOCs that likely originated from common sources. Dwellings built between 1950s and 1990s, and especially, those with attached garages had higher TVOC concentrations. Interior thermal retrofit of dwellings and absence of mechanical ventilation system were associated with elevated levels of formaldehyde, aromatics, and alkanes. Overall, energy‐renovated homes had higher levels of certain VOCs compared with newly built homes. The results suggest that energy efficiency measures in dwellings should be accompanied by actions to mitigate VOC exposures as to avoid adverse health outcomes.

include their off-gassing from building materials, 12 paints, 13 consumer and household products, 14 occupants, 15 secondary formation owing to indoor chemistry, 16,17 and intrusion of VOC-enriched outdoor air. 18 Owing to changes in residential materials, construction techniques and associated energy-saving measures, increased use of consumer products, and altered occupants' habits, the type and abundance of indoor VOCs have dramatically changed over the last decades. 19 Understanding the VOC levels in residences is therefore important to better interpret their influence on occupants and to develop adequate control interventions.
Several studies provide abundant information on VOC contamination status in a sample of residences via population-based investigation approach, as summarized by Logue et al, 20 some of which investigated associations between dwelling characteristics and measured VOCs.
Park and Ikeda measured 17 VOCs and 11 aldehydes in 1417 homes in Japan; they found that the VOC levels decrease with building age. 21,22 Raw et al 23 reported a nationwide survey of 876 residences in England, in which concentrations of formaldehyde and TVOC (total VOC) were higher in newer than in older homes. Correlation between indoor formaldehyde and building age was also found in 100 dwellings in Hong Kong, while no such trend was witnessed for 15 other measured VOCs. 24 The study in California investigated 24 VOCs in 108 newly built homes and found that formaldehyde concentrations were affected by ventilation type and geographical location of the dwellings. 25 Similar conclusions were drawn in the measurement campaign in 305 dwellings in Sweden. 26 The French Observatory for Indoor Air Quality (IOAQ) measured VOCs in 567 French homes, in which the influence of building characteristics and socioeconomic factors on VOCs was analyzed. 27,28 It was found that VOC levels are influenced by relative humidity, building age, garage type, and family wealth status.

Cheng et al 29 characterized indoor VOCs in 40 dwellings in Australia
and found an association between proximity of major roads and indoor concentrations of alkanes and aromatics. VOCs in energy-efficient buildings are another area of increased public interest. 30 Overwhelming focus on energy savings may often be in conflict with maintaining the recommended levels of indoor VOCs. The requirement for airtightness in energy-efficient buildings can lead to low air infiltration, and if not sufficiently compensated by intentional ventilation, it can lead to higher VOC concentrations than that of conventional buildings. 31 Thus, Langer et al 32 reported higher TVOC concentrations in passive low-energy houses compared with conventional ones in Sweden. Introduction of thermal retrofitting materials during residential energy renovation was found to contribute to elevated concentrations of indoor formaldehyde and TVOC. 33 A recent field investigation by Du et al 34 found that BTEX (benzene, toluene, ethylbenzene, and xylenes) concentrations significantly increased after energy renovation of multifamily dwellings in Finland. A field campaign in France showed that energy-efficient houses had higher levels of several VOCs and aldehydes (acetaldehyde, hexaldehyde, n-decane, n-undecane, o-xylene, and styrene) compared with the national average levels. 35 In contrary, several studies reported reduced VOC levels in low-energy houses compared with those conventionally built. 36 38 Recently, Switzerland introduced the "Energy Strategy 2050" to reduce the energy-related environmental impact. 41

Practical implications
• The results from this study provide a new large dataset on the individual VOC levels in energyefficient dwellings, which is valuable in relation to how exposure to VOCs influences human health.
• The levels of the most prevalent VOCs in Swiss dwellings are comparable to those in other European countries.
• Thermal retrofit of dwellings and absence of mechanical ventilation system are associated with elevated levels of formaldehyde, toluene and butane indoors.
• Energy-efficiency measures in dwellings should be accompanied with actions to mitigate VOC exposures.
• The results are of potential utility for improving the indoor air quality models, for enhancing the ventilation design in energy-efficient dwellings, and for improving the energy renovation processes.

| Study design
Study samples were collected within the framework of the largescale survey conducted in "Mesqualair" New Regional Policy collaborative project on IAQ evaluation in 650 energy-efficient dwellings from January 2013 to March 2016 in Western Switzerland. 43,44 200 participants, largely building owners, were invited to take part in a complementary analysis of VOCs and aldehydes in their dwellings.

| Study sample
The locations of the 169 investigated dwellings are graphically represented in Figure S1. Table 1

| VOC and aldehyde quantification
The sampling kit consisted of two passive devices for VOCs and aldehydes (TOXpro SA, Switzerland) in compliance with ISO 16017-2 45 and ISO 16000-4 46 standards, respectively. Following step-by-step instructions, the occupants placed one passive badge sampler for VOCs (carbon molecular sieve, Anasorb 747) and one passive sampler for aldehydes (2,4-dinitrophenylhydrazine impregnated silica gel) in the master bedroom of each sampled dwelling. The two samplers were placed between 1.0 m and 1.7 m above the ground, away from windows and any prominent VOC emission sources-including perfume, potpourri, and scented candles. The distance between the two samplers was larger than 0.3 m, to avoid cross-contamination of the samplers, and less than 1.0 m to ensure the measurement in the same area of the bedroom. The VOC and aldehyde measurement lasted for seven days. During the sampling period, the occupants were asked to keep their living habits as usual, without touching or moving the samplers. The occupants had an option to phone the project team in the event that any questions had arisen.
The collected samplers were sent to laboratory (Advanced Chemical Sensors Co. Ltd, Florida, USA) where they were analyzed under ISO 17025 47 accreditation scheme. Chemicals retained in the VOC passive samplers experienced solvent desorption with carbon disulfide as described by OSHA Method 7. 48 Then, the extracted TA B L E 1 A summary of the characteristics of the 169 dwellings sampled in this study. Responses of "I do not know" are excluded Dwelling characteristics  The laboratory experiments indicated that the relative humidity has no measurable effect on the analysis results. Increases in air temperature from 24°C up to 37°C were found to have less than a 10% effect, which was within the overall measurement error.

| Statistical analyses
The statistical analyses were performed using SPSS 21 software and customized coding in MATLAB R2014 software. The concentrations of formaldehyde and logarithmical transformed TVOC were normally distributed (seen in Figure S2). Therefore, the parametric t test (number of categories k = 2) and analysis of variance (ANOVA) test (k > 2) were performed to test the relationship of the two variables with the dwelling characteristics. We also performed the nonpara- comparisons. 51,52 The ES larger than 0.2, 0.5, and 0.8 indicated small, medium, and large effects of the variables, respectively. 51 VOCs were found in more than 50% of sampled homes (see Table 2).

| Descriptive VOC data
This proportion is lower than that observed in the Australia campaign 29 (77 out of 97 VOCs (79%)), but higher than that observed in the Swedish campaign 26 (11 out of 124 (9%)). In our study, formaldehyde, hexaldehyde, and toluene had the highest incidence-they were found in all the sampled dwellings, followed by ethanol, benzaldehyde, butane, and acrolein. Unlike our findings, Swedish dwellings had the highest reported incidence of benzene, 2-ethyl-1-hexanol, and 1-butanol. 26 The median concentrations of most frequently detected VOCs

| Summary comparisons of VOC data
To understand the status of indoor VOC contamination in Swiss dwellings, we benchmarked the measured VOC data against several  For formaldehyde, the concentrations in all sampled dwellings were below the maximum recommended limit from the WHO 53 and in France 54 of 100 μg/m 3 and from the FOPH of 125 μg/m 3 . 55 Compared to the 8-hour and chronic exposure limit value of 9 μg/m 3 proposed by the OEHHA, 56 the formaldehyde concentrations in as many as 90% of the sampled dwellings exceeded the threshold, which can lead to nasal obstruction and discomfort, lower airway discomfort, and eye irritation. 57 For TVOCs, 8% of dwellings failed to stay below the Swiss upper exposure limit of 1000 μg/m 3 . 55 The proportion of homes exceeding the TVOC limits was 53% and 40% when compared to the lower and upper 8-hour exposure limit values from Germany (200 and 300 μg/m 3 , respectively). 58 Benzene was detected in only 37% of sampled dwellings, among which 10% exceeded the French long-term exposure guideline value of 10 μg/m 3 . 59 Compared with OEHHA's chronic exposure limit of 3 μg/m 3 , 60 benzene concentrations were exceeded in over 25% homes-levels known to be associated with decreased peripheral blood cells. 61 Considering the carcinogenicity, WHO recommends no safe level for exposure to benzene. 53 Most of the sampled dwellings (>95%) were below the toluene concentration recommended by WHO (260 μg/m 3 ), 53 OEHHA, 62 and Germany 58 (300 μg/m 3 ). Propionaldehyde may mainly originate from outdoors, while other compounds predominantly came from indoors, as suggested by the previously reported indoor/outdoor ratios in other studies. 29 When the dwellings were ventilated via mechanical systems or opening the windows, the outdoor air introduced propionaldehyde to indoor environment but removed other indoor VOCs, leading to the negative correlations between the concentrations of propionaldehyde and other compounds. For aromatics (Table S3)

| Influence of dwelling characteristics on VOC concentrations
Relative to naturally ventilated residences, dwellings with installed mechanical ventilation systems had significantly lower median con- Thus, it can be assumed that in winter, the disparity in VOC levels between mechanically and naturally ventilated dwellings would be even larger, owing to reduced dilution of airborne contaminants in naturally ventilated homes.  (Table S5) and low air infiltration rate. 26 We also probed the effects of dwelling material structure and garage type on the level of individual VOCs. Wood homes had significantly higher concentrations of several VOCs, that is, acrolein, toluene, glutaraldehyde, ethyl acetate, and 1-butyl alcohol, compared to dwellings with masonry and mixed structures (Table S7), as similarly reported in the literature. 27 Glutaraldehyde is widely used as the modification chemical to control wool moisture, 66 which contributes to the higher concentrations in buildings of wooden structure.
However, given the low sample size of wooden buildings (17), the results should be interpreted with care. Houses with attached garages had higher concentrations of formaldehyde, aromatics, and alkanes compared to those with detached garages (Table S8). The aromatics and alkanes are commonly associated with emissions from vehicles in garages. 67 The infiltration of the VOCs from the attached garages to the living spaces contributed to the higher concentrations in the sampled bedrooms.

| Study limitations
In interpreting the study results, a few limitations should be ac-