- Email: [email protected]
Modelling the impact of room temperature on concentrations of polychlorinated biphenyls (PCBs) in indoor air
Chemosphere 144 (2016) 2127–2133
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Modelling the impact of room temperature on concentrations of polychlorinated biphenyls (PCBs) in indoor air Nadja Lynge Lyng a,∗, Per Axel Clausen b, Claus Lundsgaard c, Helle Vibeke Andersen a a
Danish Building Research Institute, Aalborg University Copenhagen, A.C. Meyers Vænge 15, DK-2450, Copenhagen SV, Denmark National Research Centre for the Working Environment, Lersø Parkallé 105, DK-2100, Copenhagen Ø, Denmark c Scandinavian Bio-Medical Institute (SBMI), Rungstedvej 21, DK-2970, Hørsholm, Denmark b
h i g h l i g h t s • • • •
The effect of temperature on PCB air concentrations indoors. Change in concentration due to temperature follows Clausius–Clapeyron equation. Model predictions of concentrations match field intervention measurements, R2 > 0.94. Additional field data substantiate the model.
a r t i c l e
i n f o
Article history: Received 17 July 2015 Received in revised form 13 October 2015 Accepted 26 October 2015 Available online 14 November 2015 Handling editor: Andreas Sjodin Keywords: PCBs Polychlorinated biphenyls Indoor air Enthalpy of evaporation Temperature dependency SVOCs
a b s t r a c t Buildings contaminated with polychlorinated biphenyls (PCBs) are a health concern for the building occupants. Inhalation exposure is linked to indoor air concentrations of PCBs, which are known to be affected by indoor temperatures. In this study, a highly PCB contaminated room was heated to six temperature levels between 20 and 30 C, i.e. within the normal fluctuation of indoor temperatures, while the air exchange rate was constant. The steady-state air concentrations of seven PCBs were determined at each temperature level. A model based on Clausius–Clapeyron equation, ln(P) = −H/RT + a0 , where changes in steady-state air concentrations in relation to temperature, was tested. The model was valid for PCB28, PCB-52 and PCB-101; the four other congeners were sporadic or non-detected. For each congener, the model described a large proportion (R2 >94%) of the variation in indoor air PCB levels. The results showed that one measured concentration of PCB at a known steady-state temperature can be used to predict the steady-state concentrations at other temperatures under circumstances where e.g. direct sunlight does not influence temperatures and the air exchange rate is constant. The model was also tested on field data from a PCB remediation case in an apartment in another contaminated building complex where PCB concentrations and temperature were measured simultaneously and regularly throughout one year. The model fitted relatively well with the regression of measured PCB air concentrations, ln(P) vs. 1/T, at varying temperature between 16.3 and 28.2 °C, even though the measurements were carried out under uncontrolled environmental condition. © 2015 Elsevier Ltd. All rights reserved.
1. Introduction Polychlorinated biphenyls (PCBs) were first reported synthesised in 1864 (Griess, 1864) and were widely used in industry in the 1950–1980s. Until the late 1970s, it was widely used as an additive in building materials, e.g. caulks (Danish Energy Agency report, 2013; Kohler et al., 2005; Harrad et al., 2006). PCBs have gradually been recognised as a risk for both health and the environment. Among health risks associated with exposure to PCBs are increased risk of obesity and type-2 diabetes and effects on the cardiac system, and moreover adverse immunological, ∗
Corresponding author. E-mail address: [email protected] (N.L. Lyng).
http://dx.doi.org/10.1016/j.chemosphere.2015.10.112 0045-6535/© 2015 Elsevier Ltd. All rights reserved.
reproductive, and dermatological effects (Carpenter, 1998; Faroon et al., 2003; Jensen, 2013; Carpenter, 2015). Effects on the neuropsychological functions and poorer cognitive development were found by Walkowiak et al. (2001) and Ribas-Fitó et al. (2001). In 2013, PCBs were categorised as carcinogenic by the International Agency for Research on Cancer (Lauby-Secretan et al., 2013). Blood levels of PCBs were found to be elevated in occupants of PCBcontaminated buildings (Gabrio et al., 2000; Herrick et al., 2011; Knobeloch et al., 2012; Meyer et al., 2013). These elevated blood levels indicate buildings as a significant contributor to the overall PCB load in those exposed. Several environmental factors are known to affect the exposure level in PCB-contaminated buildings, e.g. ventilation rate (Lyng et al., 2014, 2015), and indoor temperatures
2128
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133
(Kohler et al., 2005; Bent et al., 2000). Few studies have investigated the influence of temperature on PCB air concentration indoors. In Bent et al. (2000), PCB air concentrations, indoor temperature and outdoor temperature were regularly measured in a control room. Temperature fluctuations were found to cause a multiple increase in the PCB air concentrations. In Kohler et al. (2005), indoor temperature was acknowledged as a driving parameter for increased PCB air concentrations (n = 112). In an apartment used for pilot testing of PCB remediation methods, PCB air concentrations were measured on a regular basis throughout one year and found to be exponentially related to indoor temperatures. These measurements will be presented as additional field data in this article and used for model evaluation of predictions based on Clausius–Clapeyron relation. Finally, Benthe et al. (1992) and Balfanz et al. (1993) found that indoor air concentrations of PCBs correlated positively with outdoor temperatures. Outdoor air concentrations of PCBs were found strongly dependent on temperature and at air temperatures above 0 °C this dependence could be modelled by the Clausius–Clapeyron equation (Halsall et al., 1999; Carlson and Hites, 2005). Halsall et al. (1999) suggest that equilibrium between atmosphere and surface is achieved as a function of either liquid vapour pressure or the octanol–air partitioning coefficient. The purpose of the present study was to investigate the effect of temperature on PCB air concentrations in indoor settings, in order to parametrise the change and thereby the possibility to predict PCB concentrations within a range of normal indoor temperature. 2. Method 2.1. Design of the intervention study The intervention study was conducted in one PCB contaminated bedroom in an apartment building. The bedroom was heated incrementally to six different temperature levels: 20, 22, 25 and 28 °C (Round 1) and one year later 30 and 24 °C (Round 2). Temperature levels were achieved in the mentioned order, i.e. Round 2 in the opposite order of Round 1. Four measurements of concentrations during one week showed that steady-state concentration was reached within 2.5 days after a changed temperature setting (Fig. S1). In the following intervention study the temperature setting was held for three days prior to air sampling in order to ensure steady-state temperature and concentration in the room. At each temperature level, the PCB air concentrations were measured twice; once during night time and once the following day, resulting in a total of 12 sampling periods. Round 1 was carried out from the beginning of April to the middle of May in 2014, Round 2 one year later in March and April 2015. 2.2. Intervention room characteristics In the apartment used for this study, the air concentration (PCBtotal ) was previously measured to 4750 ng/m3 (Golder Associates, 2013). PCBtotal is calculated as the sum of (PCB-28, PCB-52, PCB-101, PCB-118, PCB-138, PCB-153 and PCB-180) multiplied by a factor of 5 (VDI 2464 part 1, 2009). The intervention room was a bedroom measuring approximately 30 m3 . The room had one external wall with one window (facing west). The building’s mechanical ventilation consisted of exhausts in the kitchen and bathrooms. The only known primary source of PCBs in the room was the exterior caulk around the window. Pieces of caulk might be present beneath the wooden floors as it has been the case in other apartments. Interior surfaces have additionally been contaminated over the years due to elevated air concentrations of PCBs and have be-
come secondary sources (Guo et al., 2011). See supplementary material for further details on the building’s characteristics. 2.3. Environmental variables 2.3.1. Indoor PCB levels and chemical analysis Sampling of air was done for 12 sampling periods (day and night respectively for the six temperature levels). Concentrations were measured in duplicate at two locations (n = 24) and a few samples of caulk and wall paint were collected. Chemical analysis of 7 PCBs in air and material samples was done by Dansk Miljøanalyse Aps (Danish Environmental Analysis LLC). Description of sampling and analytical procedures can found in supplementary material. 2.3.2. Temperature, ventilation and humidity Air temperature and humidity were monitored continuously, whereas surface temperatures were measured in the beginning and end of each sampling period. Air exchange rate/infiltration was measured once during each sampling period by CO2 decay (ASTM E741). An oscillating fan ensured an evenly distributed temperature in the room. A description of temperature, humidity and ventilation measurements can be found in the supplementary material and measured values in Table S1. 2.4. Additional field data To further substantiate the study on impact of room temperature on PCB concentration, data from a comprehensive monitoring program of PCB indoor concentrations in Farum Midtpunkt, Denmark were included. Farum Midtpunkt housing estate is described in Frederiksen et al. (2012). During a one-year pilot testing of PCB remediation methods in a three bedroom apartment, PCB concentration and temperature were measured on 38 occasions. Measurements were performed in 1–6 different locations simultaneously (n = 98 measurements). The apartment was heated during the period, however room temperature varied between 16.3 and 28.2 °C depending on primarily season and weather conditions. Mechanical ventilation was constant with exhaust vents in kitchen and bathrooms and slit valves in windows for inlet. All interior doors between rooms were kept open. Measurements were performed by similar procedure as in the intervention study described above, except for differences described in supplementary material. Prior to the investigation period the apartment was vacated and removal of the following was completed: furniture, interior wooden doorframes, some curtain walls, PCB containing caulks (10–28% Clophen A40 or Aroclor 1248) around window elements, curtain walls, interior doors and the accessible adjacent concrete (to 3–5 cm distance from caulks). Paints on walls and ceilings were contaminated with 100–300 mg PCBtotal /kg, and this source was not removed. The measuring period was between 22nd of May 2010 and 22nd of May 2011. Minor additional remedial measures were carried out incrementally throughout the year involving additional removal of adjacent concrete, wooden floor sanding and refinishing, testing of silicate based barrier on one bedroom façade, and finally removal of kitchen units and wooden floors. 2.5. Theoretical framework The hypothesis is that the concentration, expressed as partial pressures pi (Pa) in a defined volume indoors, follows Clausius– Clapeyron equation at different temperatures T (K):
ln(Pi ) =
a1 + a0 T
(1)
where a1 = −H/R, H is the enthalpy of evaporation of liquid PCB which can be assumed constant in the studied temperature
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133
range (i.e. well below the boiling point) and R = 8.314 J/(K mol) is the gas constant (Halsall et al., 1999). The constant a0 is determined by the general concentration level and will vary from case to case. The enthalpy of evaporation of sub-cooled liquid PCB congeners H is 78.0, 80.8 and 86.4°KJ/(K mol), respectively, for PCB28, PCB-52 and PCB-101 calculated from the slopes (ml ) in Table S2 from Falconer and Bidleman (1994). A further assumption for the hypothesis is that source concentration is constant during the intervention period, i.e. the amount of PCB emitted is relatively low compared with the content in the source. To predict the concentration at different temperatures by Eq. (1), an estimate of a0 is required. a0 can be estimated if at least one steady-state concentration measurement, P1 (Pa), at a given temperature, T1 (K) is known. In this study, there were multiple measurements and the average temperature of the sampling periods was chosen as T1 . The estimated corresponding concentration P1 was calculated by the regression function (Table S3) of measured concentration, ln(P) versus 1/T, thus all measurements in the study are used to estimate the general contamination level, a0 . Concentration predictions at different temperatures were tested on data from the intervention study as well as the additional field data from Farum Midtpunkt.
2.6. Statistical analyses One-way ANOVA and Tukey’s HSD were used for assessing the difference in mean concentrations measured at the different temperature levels. The analyses were performed using R software version 3.0.0 (2013-04-03, R Foundation for Statistical Computing, Vienna, Austria). The significance threshold was p < 0.05. Simple linear regression was done in Microsoft Excel as well as the calculation of the coefficient of determination, r2 . The coefficient of determination between measured concentrations and the model (Eq. (1)), here distinguished as R2 , was calculated by the equation:
R2 = 1 −
SSres SStot
(2)
where SSres is the residual sum of squares, and SStot is the total sum of squares.
2129
3. Results 3.1. Intervention study Air temperatures were measured day and night, respectively, at 1-min intervals throughout the sampling periods and are shown in Fig. S2. Few sampling periods were left out for reasons explained in supplementary material. No significant differences in concentrations between night and day at each temperature level were found. Of the 7 analysed congeners, only PCB-28, PCB-52 and PCB-101 were detected at all temperature levels. In Figs. S3, S4 and S5 the PCB concentrations measured in the 12 sampling periods (including the omitted concentrations) are displayed as a function of the mean temperature in the respective sampling periods (standard deviation < 0.8 °C). Concentrations increased almost exponentially with the temperature. PCB air levels in Round 2 are around 20% lower than Round 1, but considered within the uncertainty of measurements and chemical analysis. One-way ANOVA showed that measured concentrations of the specific congeners are significantly different between different temperature levels (P-value < 0.01 for PCB-28 and < 0.001 for PCB-51 and PCB-101). In order to use Eq. (1) in predicting concentrations, an estimate of the general contamination, a0 is required. Thus, at least one reference temperature, T1 and the corresponding concentration, P1 is needed. The overall average temperature of the measuring periods was chosen as T1 (24.4 °C) for the intervention study. The corresponding concentrations were calculated by the simple regression function of the measured concentrations, ln(P), versus inverse temperature. Regression slopes and intercepts of measurements can be found in Table S3. In Fig. 1, the model predictions are shown by the thick grey line and the simple regressions by the black dashed lines. In Table 1, coefficients of determination, R2 , for the measured concentrations and the model are shown for the three detected congeners PCB-28, PCB-52 and PCB-101. Surface temperatures as well as air temperatures were measured in order to assess whether the room had reached steady-state temperatures. The measured mean values are shown in Table S1. The window caulk had a congener composition similar to Aroclor 1248 (<1% w/w). Painted wallpaper in the room was determined to have a PCBtotal concentration of 166 mg/kg.
Fig. 1. Intervention study: The model (thick grey lines), the measured concentrations [Pa] (dots) and their regression lines (thin black dashed lines) of PCB-28, PCB-52 and PCB-101 at different temperatures [K].
2130
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133
Table 1 Coefficients of determination for fitting simple regression (r2 ) and comparing models (R2 ) to measured concentrations of PCB. The slopes, intercept and R2 of the regression for the correlation between modelled and measured concentrations. r2 of simple regressiona ln(P) vs. 1/T
Intervention study
Field data
a b c d e
PCB-28 PCB-52 PCB-101 PCB-28 PCB-52 PCB-101
0.94 0.94 0.95 0.49 0.44 0.45
R2 by Eq. (2) when fitting model predictionsb to measuring data
0.94 0.94 0.95 0.47 0.44 0.45
Correlation between modelled and measured concentrationsc Linear regression equation terms Slope (95% CI)d
Intercepts (95% CI)e
1.00 (0.88; 1.14) 0.99 (0.87; 1.11) 1.05 (0.93; 1.16) 1.23 (0.98; 1.48) 1.14 (0.88; 1.40) 1.09 (0.84; 1.34)
0.08 (−1.58; 1.74) −0.10 (−1.64; 1.43) 0.72 (−1.13; 2.57) 3.22 (−0.30; 6.74) 1.91 (−1.70; 5.52) 1.50 (−2.80; 5.80)
Regression is based on the measuring data. Model predictions are based on Eq. (1). Correlation plots with regression lines is found in supplementary material (Fig. S6). 95% Confidence interval of the slopes. 95% Confidence interval of the intercepts.
3.2. Additional field data In order to further investigate the ability of the model to predict observed data, the model was applied on data from an apartment in Farum Midtpunkt, Denmark. In Fig. 2 concentrations of PCB-28, PCB-52, PCB-101 are shown with the regression lines (thin black lines), the 95% confidence interval (dashed lines) and the model (thick gray lines). Model predictions are based on overall average temperature of measurements, T1 (20.5 °C), and the corresponding concentration calculated from regression of measured data. The coefficient of determination, r2 , for the simple regression of the measured data is 0.44–0.49, whereas the coefficient of determination is R2 = 0.44–0.47 for the model and the measured data. PCB-138, PCB-153 and PCB-180 were not detected. PCB-118 was detected in less than 10% of the cases and only in low concentrations and was left out for that reason. 3.3. Model validation The correlation between measured concentration and concentrations modelled by the Clausius–Clapeyron equation was investigated in order to evaluate the ability of the model to predict indoor air concentrations at any temperature in the studied range. For each measured concentration at a given temperature, the corresponding modelled concentration was calculated. Plots of measured concentration vs. their corresponding modelled concentrations are given in the supplementary material (Fig. S6) From linear regression the slopes, intercepts and the 95% CI of the correlation as well as the coefficient of determination can be found in Table 1. Slopes are close to the value 1 and confidence intervals of intercepts all includes the value of 0. In addition coefficients of determination, r2 , are given between fit of simple regression of the measured partial pressures. Also the coefficient of determination, R2 , between the models and the actual measured data. From the slopes, a1 , of the simple regression lines (measured data) in Figs. 1 and 2, enthalpy of evaporation can be calculated. In Fig. 3, these values are plotted with 95% confidence intervals alongside the values of enthalpy of evaporation as reported in literature (Falconer and Bidleman, 1994). The values from literature are within the confidence interval of the measured values in the intervention study as well as the additional field data. 4. Discussion The general analytic uncertainty stated by the analysis laboratory was a maximum of 20% for the intervention study. Deviation
between the duplicate measurements from the intervention study was below 14%, covering uncertainties of both analysis and sampling. In spite of the relative high uncertainty of the PCB measurements the intervention study showed a surprisingly good fit between actual measured concentrations and predictions based on Clausius–Clapeyron equation. This indicates that steady-state air concentration is achieved rapidly (within a few days) with temperature changes. The measurements in the additional field data from Farum Midtpunkt, Denmark, were originally aimed at investigating the effect of the remediation measures. The measurements have large variation in the concentration of the detected congeners PCB-28, PCB-52 and PCB-101 even at similar indoor temperatures. Only 43–48% of the variation in data is explained by the temperature. Since data were obtained from 6 different rooms at different stages of remediation and throughout a whole year with different seasons and fluctuating weather conditions and without mixing of air in the apartment, large variation was expected. Nevertheless the model is found to be within the confidence interval of the regression line for measured PCB partial pressures and inverse temperature as shown in Fig. 2. The most influencing factor on the general PCB levels in indoor air is the source characteristics including physico-chemical properties of source, concentration of PCB in source, amount, size and location. This enhances the variation in air concentration from building to building as well as room to room. Therefore the model needs the contamination level of the specific room. This is achieved through one or more measurements of steady-state air concentration and temperature; these values are used to determine the background level, a0 in Eq. (1). In this study we had multiple measurements and chose the overall average temperature and calculated the corresponding concentration by the simple regression equation; therefore the model is not fully theoretically derived. However the essential part of the model is that the steadystate concentrations at different temperatures measured in this intervention study can be predicted by Clausius–Clapeyron equation, and measurements could be used to calculate enthalpy of evaporation (Fig. 3). These facts suggest that PCB sources in the room behave as liquid PCBs, though steady-state air concentrations are far from saturated. Change in saturated vapour pressure and probably octanol–air partitioning coefficient (Koa ) due to temperature can also predict temperature dependence of equilibrium partial air pressures (Halsall et al., 1999). PCBs are categorised as semi-volatile organic compounds (SVOCs), thus PCB emissions are likely to be controlled by “external” emission mechanisms like solid (source) to gas-phase partitioning and mass transport limitation by diffusion in the boundary layer rather than mass transport controlled by “internal” diffu-
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133
2131
Fig. 2. Field Data: Measured partial pressures [Pa] of A) PCB-28, B) PCB-52 and PCB-101 (dots), their regression lines (thin black lines) with 95% confidence intervals and the models (thick grey lines) against inverse temperatures [K].
sion in the source matrix as is generally the case for VOCs (volatile organic compounds) (Xu and Little, 2006). For this reason, emissions are expected to be affected by air temperature in the boundary layer rather than source surface temperature. However, the air temperature immediately adjacent to the source surfaces will be correlated to the surface temperature. Temperature in the boundary layer will gradually approximate the air temperature measured in the middle of the well-mixed room air with the increasing distance to the surface. In the intervention study surface temperatures measured before and after each PCB sampling indicated steadystate surface temperatures close to the air temperature (Table S1). This suggests that room air temperature measured in the middle of the room is a good estimate for the emission temperature. In reality indoor air temperature and source surface temperature is rarely maintained in steady-state, homogeneously distributed or without temperature gradient as provided in the intervention study. Direct sun and other local sources of heat might heat up sections of indoor surfaces (secondary sources) and the thermal capacity of air is much lower than building materials and
provides a faster change in indoor air temperature compared with material temperatures which counteract constant temperature in the air. Assumption of steady-state temperature is required for model predictions, and might be difficult to achieve in regular indoor environments. Model predictions in the field data were not significantly different than regression of concentration, ln(P) vs. inverse temperature, still model prediction and temperature regression explain less than 50% of the variation in concentration. Failure to meet the assumption of steady-state temperature and concentration as well as changes in background level a0 due to the renovation can explain the remaining variation in concentration measurements. In Clausen et al. (2012), the steady-state concentration of the SVOC bis(2-ethylhexyl) phthalate (DEHP) was measured at different temperatures (23, 35, 47, 55 and 61 °C) and was also found to be exponentially increasing with increasing temperature. Liang and Xu (2014a; 2014b) found that the equilibrium gas-phase concentration of DEHP immediately adjacent to the source surface, which is a key parameter that control emission, is proportional to the con-
2132
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133
Fig. 3. Enthalpies of evaporation and 95% confidence intervals calculated from measured data (black dots and bars) and reported values (grey dots) from Falconer and Bidleman (1994).
tent of DEHP in the emitting materials. This intervention study had air concentrations of PCBs below 0.026% of the saturated vapour pressure. The relatively low air concentration is probably due to the low content of PCB in the emitting materials compared to liquid PCBs. PCB contaminated buildings might react differently to influencing factors with regard to the steady-state air concentrations, and it is unknown whether this model can be used for prediction of concentration and temperature relations in other cases and more cases needs to be investigated. The presented model depends on estimations/measurements of PCB enthalpy of evaporation by Falconer and Bidleman 1994 and if other estimations were used the model might deviate from the model presented in this study. 5. Conclusion A significant and exponential increase in the concentration of PCB in indoor air was observed with increasing temperature, even at temperature changes of only a few degrees. In the temperature intervention study a high correlation (R2 > 0.94) was found between the actual measured concentrations of PCB in air and the predicted concentrations based on the Clausius–Clapeyron equation. This was found for three out of the seven measured congeners; PCB-28, PCB-52 and PCB-101. The remaining congeners were only sporadic or non-detected. In this specific case, the predictions can be used to estimate the indoor concentration of PCB at a given indoor steady-state temperature. However factors like air exchange rate, outdoor temperature, direct sun etc. are likely to affect the steady-state conditions and the concentration independently, which can complicate the estimation of concentration levels. In the additional field data from a case study in Farum Midtpunkt, Denmark, with uncontrolled environmental conditions but known temperatures, temperature explained between 44 and 49% of the variation in air concentrations (n = 98). When the Clausius– Clapeyron equation was used on these data, the model predictions occur within the 95% confidence interval of the simple regression. The model fitted the data reasonably well (R2 = 0.44–0.47) and the model appear to be robust, even though the measurements were carried out under completely un-controlled environmental condition. Acknowledgement We would like to thank the Danish Social Housing Association (DAB) for giving us access to a highly contaminated apartment and
to Copenhagen Social Housing Association (KAB) for letting us use data from Farum Midtpunkt. The study was supported by grants from Realdania and measurements in Farum Midtpunkt were supported by LBF (Landsbyggefonden, Denmark). Appendix A. Supplementary data Supplementary data related to this article can be found at http: //dx.doi.org/10.1016/j.chemosphere.2015.10.112. References ASTM E741 11 Standard Test Method for Determining Air Change in a Single Zone by Means of a Tracer Gas Dilution. DOI: 10.1520/E0741-11 Balfanz, E., Fuchs, J., Kieper, H., 1993. Sampling and analysis of polychlorinated biphenyls (PCB) in indoor air due to permanently elastic sealants. Chemosphere 26, 871–880. Bent, S., Rachor-Ebbinghaus, R., Schmidt, C., 2000. Remediation of highly polychlorinated biphenyls (PCBs) contaminated rooms by complete removal of primary and secondary sources. In German: sanierung von hochgradig mit polychlorierten biphenylen (PCB) belasteten Innenräumen durch komplettes Entfernen der Primärund Sekundärquellen. Gesundheitswesen 62, 86–92. Benthe, C., Heinzow, B., Jessen, H., Rotard, S.M.W., 1992. Polychlorinated biphenyls. Indoor air contamination due to Thiokol-rubber sealants in an office building. Chemosphere 25, 1481–1486. Carlson, D.L., Hites, R.A., 2005. Temperature dependence of atmospheric PCB concentration. Environ. Sci. Technol. 39, 740–747. Carpenter, D.O., 1998. Polychlorinated biphenyls and human health. Int. J. Occup. Med. Environ. Health 11, 291–303. Carpenter, D.O., 2015. Exposure to and health effects of volatile PCBs. Rev. Environ. Health 30, 81–92. Clausen, P., Liu, Z., Kofoed-Sørensen, V., Little, J., Wolkoff, P., 2012. Influence of temperature on the emission of Di-(2-ethylhexyl)phthalate (DEHP) from PVC flooring in the emission cell FLEC. Environ. Sci. Technol. 46, 909–915. Danish Energy Agency report, 2013. PCB Screening in Material and Indoor Air. In Danish: Kortlægning Af PCB I Materialer Og Indeluft. Project A030835 accessed 12 12 2014 http://www.ens.dk/sites/ens.dk/files/byggeri/sikre-sunde-bygninger/ pcb_kortlaegning_dec13.pdf. Falconer, R.L., Bidleman, T.F., 1994. Vapor pressures and predicted particle/gas distributions of Polychlorinated Biphenyl congeners as functions of temperature and ortho-chlorine substitution. Atmos. Environ. 28, 547–554. Faroon, O.M., Keith, L.S., Smith-Simon, C., De Rosa, C.T., 2003. Polychlorinated biphenyls: human health aspects concise. International Chemical Assessment. Document 55. WHO accessed 7 10 2014 http://www.inchem.org/documents/ cicads/cicads/cicad55.htm . Frederiksen, M., Meyer, H.W., Ebbehøj, N.E., Gunnarsen, L., 2012. Polychlorinated biphenyls (PCBs) in indoor air originating from sealants in contaminated and uncontaminated apartments within the same housing estate. Chemosphere 89, 473–479. Gabrio, T., Piechotowski, I., Wallenhorst, T., Klett, M., Cott, L., Friebel, P., Link, B., Schwenk, M., 2000. PCB-blood levels in teachers, working in PCB-contaminated schools. Chemosphere 40, 1055–1062. Golder Associates A/S, 2013. Complete Summary of Measured Indoor Concentrations of PCBs. In Danish: Resultatoversigter over PCB-målinger I Indeluften. Project No. 12501130119, 12501130141, 12501130142, 12501130143. Griess, P., 1864. On a new class of compounds in which nitrogen is substituted for hydrogen. Proc. Royal Soc. Lond. 13, 375–384.
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133 Guo, Z., Liu, X., Krebs, K., 2011. Laboratory study of polychlorinated biphenyl (PCB) contamination and mitigation in buildings. Part 1. Emiss. Sel. Prim. Sources EPA/600/R-11/156) U.S. Environmental Protection Agency (EPA). Halsall, C.J., Gevao, B., Howsam, M., Lee, R.G.M., Ockenden, W.A., Jones, K.C., 1999. Temperature dependence of PCBs in the UK atmosphere. Atmos. Environ. 33, 541–552. Harrad, S., Hazrati, S., Ibarra, C., 2006. Concentrations of polybrominated diphenyl ethers in indoor air and dust and polychlorinated biphenyls in indoor air in Birmingham, United Kingdom: Implications for human exposure. Environ. Sci. Technol. 40, 4633–4638. Herrick, R.F., Meeker, J.D., Altshul, L., 2011. Serum PCB levels and congener profiles among teachers in PCB-containing schools: a pilot study. Environ. Health 10, 56. Jensen, A.A., 2013. Health risks of PCB in the indoor climate in Denmark – background for setting recommended action levels. Dan. Health Med. Auth. ISBN: 978-87-7104-004-3. Kohler, M., Tremp, J., Zennegg, M., Seiler, C., Minder-Kohler, S., Beck, M., Lienemann, P., Wegmann, L., Schmid, P., 2005. Joint sealants: an overlooked diffuse source of polychlorinated biphenyls in buildings. Environ. Sci. Technol. 39, 1967–1973. Knobeloch, L., Turyk, M., Imm, P., Anderson, H., 2012. Polychlorinated biphenyls in vacuum dust and blood of residents in 20 Wisconsin households. Chemosphere 86, 735–740. Lauby-Secretan, B., Loomis, D., Grosse, Y., Ghissassi, F.E., Bouvard, V., BenbrahimTallaa, L., Guha, N., Baan, R., Mattock, H., Straif, K.on behalf of the International Agency for Research on Cancer Monograph Working Group IARC, 2013. Carcinogenicity of polychlorinated biphenyls and polybrominated biphenyls. Lancet Oncol. 14, 287–288.
2133
Liang, Y., Xu, Y., 2014a. Emission of phthalates and phthalate alternatives from vinyl flooring and crib mattress covers: the influence of temperature. Environ. Sci. Technol. 48, 14228–14237. Liang, Y., Xu, Y., 2014b. Improved method for measuring and characterizing phthalate emissions from building materials and its application to exposure assessment. Environ. Sci. Technol. 48, 4475–4484. Lyng, N.L., Trap, N., Andersen, H.V., Gunnarsen, L., 2014. Ventilation as mitigation of PCB contaminated air in buildings: review of nine cases in Denmark. Proceedings of the 13th International Conference: Indoor Air 2014. Lyng, N.L., Gunnarsen, L., Andersen, H.V., 2015. The effect of ventilation on the indoor air concentration of PCB: an intervention study. Build. Environ. 95, 305– 312. Meyer, H.W., Frederiksen, M., Göen, T., Ebbehøj, N.E., Gunnarsen, L., Brauer, C., Kolarik, B., Müller, J., Jacobsen, P., 2013. Plasma polychlorinated biphenyls in residents of 91 PCB-contaminated and 108 non-contaminated dwellings—An exposure study. Int. J. Hyg. Environ. Health 216, 755–762. Ribas-Fitó, N., Sala, M., Kogevinas, M., Sunyer, J., 2001. Polychlorinated biphenyls (PCBs) and neurological development in children: a systematic review. J. Epidemiol. Community Health 55, 537–546. Verein Deutscher Ingenieure, 2009. Ambient Air Measurement – Indoor Air Measurement – Measurement of Polychlorinated Biphenyls (PCBs) – GC-MS Method for PCB 25, 52, 101, 138, 153, 180 (VDI 2464, part 1) Issue German/English. Walkowiak, J., Wienerb, J., Fastabend, A., 2001. Environmental exposure to polychlorinated biphenyls and quality of the home environment: effects on psychodevelopment in early childhood. Lancet 358, 1602–1607. Xu, Y., little, J., 2006. Predicting emissions of SVOCs from polymeric materials and their interaction with airborne particles. Environ. Sci. Technol. 40, 456–461.
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Modelling the impact of room temperature on concentrations of polychlorinated biphenyls (PCBs) in indoor air Nadja Lynge Lyng a,∗, Per Axel Clausen b, Claus Lundsgaard c, Helle Vibeke Andersen a a
Danish Building Research Institute, Aalborg University Copenhagen, A.C. Meyers Vænge 15, DK-2450, Copenhagen SV, Denmark National Research Centre for the Working Environment, Lersø Parkallé 105, DK-2100, Copenhagen Ø, Denmark c Scandinavian Bio-Medical Institute (SBMI), Rungstedvej 21, DK-2970, Hørsholm, Denmark b
h i g h l i g h t s • • • •
The effect of temperature on PCB air concentrations indoors. Change in concentration due to temperature follows Clausius–Clapeyron equation. Model predictions of concentrations match field intervention measurements, R2 > 0.94. Additional field data substantiate the model.
a r t i c l e
i n f o
Article history: Received 17 July 2015 Received in revised form 13 October 2015 Accepted 26 October 2015 Available online 14 November 2015 Handling editor: Andreas Sjodin Keywords: PCBs Polychlorinated biphenyls Indoor air Enthalpy of evaporation Temperature dependency SVOCs
a b s t r a c t Buildings contaminated with polychlorinated biphenyls (PCBs) are a health concern for the building occupants. Inhalation exposure is linked to indoor air concentrations of PCBs, which are known to be affected by indoor temperatures. In this study, a highly PCB contaminated room was heated to six temperature levels between 20 and 30 C, i.e. within the normal fluctuation of indoor temperatures, while the air exchange rate was constant. The steady-state air concentrations of seven PCBs were determined at each temperature level. A model based on Clausius–Clapeyron equation, ln(P) = −H/RT + a0 , where changes in steady-state air concentrations in relation to temperature, was tested. The model was valid for PCB28, PCB-52 and PCB-101; the four other congeners were sporadic or non-detected. For each congener, the model described a large proportion (R2 >94%) of the variation in indoor air PCB levels. The results showed that one measured concentration of PCB at a known steady-state temperature can be used to predict the steady-state concentrations at other temperatures under circumstances where e.g. direct sunlight does not influence temperatures and the air exchange rate is constant. The model was also tested on field data from a PCB remediation case in an apartment in another contaminated building complex where PCB concentrations and temperature were measured simultaneously and regularly throughout one year. The model fitted relatively well with the regression of measured PCB air concentrations, ln(P) vs. 1/T, at varying temperature between 16.3 and 28.2 °C, even though the measurements were carried out under uncontrolled environmental condition. © 2015 Elsevier Ltd. All rights reserved.
1. Introduction Polychlorinated biphenyls (PCBs) were first reported synthesised in 1864 (Griess, 1864) and were widely used in industry in the 1950–1980s. Until the late 1970s, it was widely used as an additive in building materials, e.g. caulks (Danish Energy Agency report, 2013; Kohler et al., 2005; Harrad et al., 2006). PCBs have gradually been recognised as a risk for both health and the environment. Among health risks associated with exposure to PCBs are increased risk of obesity and type-2 diabetes and effects on the cardiac system, and moreover adverse immunological, ∗
Corresponding author. E-mail address: [email protected] (N.L. Lyng).
http://dx.doi.org/10.1016/j.chemosphere.2015.10.112 0045-6535/© 2015 Elsevier Ltd. All rights reserved.
reproductive, and dermatological effects (Carpenter, 1998; Faroon et al., 2003; Jensen, 2013; Carpenter, 2015). Effects on the neuropsychological functions and poorer cognitive development were found by Walkowiak et al. (2001) and Ribas-Fitó et al. (2001). In 2013, PCBs were categorised as carcinogenic by the International Agency for Research on Cancer (Lauby-Secretan et al., 2013). Blood levels of PCBs were found to be elevated in occupants of PCBcontaminated buildings (Gabrio et al., 2000; Herrick et al., 2011; Knobeloch et al., 2012; Meyer et al., 2013). These elevated blood levels indicate buildings as a significant contributor to the overall PCB load in those exposed. Several environmental factors are known to affect the exposure level in PCB-contaminated buildings, e.g. ventilation rate (Lyng et al., 2014, 2015), and indoor temperatures
2128
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133
(Kohler et al., 2005; Bent et al., 2000). Few studies have investigated the influence of temperature on PCB air concentration indoors. In Bent et al. (2000), PCB air concentrations, indoor temperature and outdoor temperature were regularly measured in a control room. Temperature fluctuations were found to cause a multiple increase in the PCB air concentrations. In Kohler et al. (2005), indoor temperature was acknowledged as a driving parameter for increased PCB air concentrations (n = 112). In an apartment used for pilot testing of PCB remediation methods, PCB air concentrations were measured on a regular basis throughout one year and found to be exponentially related to indoor temperatures. These measurements will be presented as additional field data in this article and used for model evaluation of predictions based on Clausius–Clapeyron relation. Finally, Benthe et al. (1992) and Balfanz et al. (1993) found that indoor air concentrations of PCBs correlated positively with outdoor temperatures. Outdoor air concentrations of PCBs were found strongly dependent on temperature and at air temperatures above 0 °C this dependence could be modelled by the Clausius–Clapeyron equation (Halsall et al., 1999; Carlson and Hites, 2005). Halsall et al. (1999) suggest that equilibrium between atmosphere and surface is achieved as a function of either liquid vapour pressure or the octanol–air partitioning coefficient. The purpose of the present study was to investigate the effect of temperature on PCB air concentrations in indoor settings, in order to parametrise the change and thereby the possibility to predict PCB concentrations within a range of normal indoor temperature. 2. Method 2.1. Design of the intervention study The intervention study was conducted in one PCB contaminated bedroom in an apartment building. The bedroom was heated incrementally to six different temperature levels: 20, 22, 25 and 28 °C (Round 1) and one year later 30 and 24 °C (Round 2). Temperature levels were achieved in the mentioned order, i.e. Round 2 in the opposite order of Round 1. Four measurements of concentrations during one week showed that steady-state concentration was reached within 2.5 days after a changed temperature setting (Fig. S1). In the following intervention study the temperature setting was held for three days prior to air sampling in order to ensure steady-state temperature and concentration in the room. At each temperature level, the PCB air concentrations were measured twice; once during night time and once the following day, resulting in a total of 12 sampling periods. Round 1 was carried out from the beginning of April to the middle of May in 2014, Round 2 one year later in March and April 2015. 2.2. Intervention room characteristics In the apartment used for this study, the air concentration (PCBtotal ) was previously measured to 4750 ng/m3 (Golder Associates, 2013). PCBtotal is calculated as the sum of (PCB-28, PCB-52, PCB-101, PCB-118, PCB-138, PCB-153 and PCB-180) multiplied by a factor of 5 (VDI 2464 part 1, 2009). The intervention room was a bedroom measuring approximately 30 m3 . The room had one external wall with one window (facing west). The building’s mechanical ventilation consisted of exhausts in the kitchen and bathrooms. The only known primary source of PCBs in the room was the exterior caulk around the window. Pieces of caulk might be present beneath the wooden floors as it has been the case in other apartments. Interior surfaces have additionally been contaminated over the years due to elevated air concentrations of PCBs and have be-
come secondary sources (Guo et al., 2011). See supplementary material for further details on the building’s characteristics. 2.3. Environmental variables 2.3.1. Indoor PCB levels and chemical analysis Sampling of air was done for 12 sampling periods (day and night respectively for the six temperature levels). Concentrations were measured in duplicate at two locations (n = 24) and a few samples of caulk and wall paint were collected. Chemical analysis of 7 PCBs in air and material samples was done by Dansk Miljøanalyse Aps (Danish Environmental Analysis LLC). Description of sampling and analytical procedures can found in supplementary material. 2.3.2. Temperature, ventilation and humidity Air temperature and humidity were monitored continuously, whereas surface temperatures were measured in the beginning and end of each sampling period. Air exchange rate/infiltration was measured once during each sampling period by CO2 decay (ASTM E741). An oscillating fan ensured an evenly distributed temperature in the room. A description of temperature, humidity and ventilation measurements can be found in the supplementary material and measured values in Table S1. 2.4. Additional field data To further substantiate the study on impact of room temperature on PCB concentration, data from a comprehensive monitoring program of PCB indoor concentrations in Farum Midtpunkt, Denmark were included. Farum Midtpunkt housing estate is described in Frederiksen et al. (2012). During a one-year pilot testing of PCB remediation methods in a three bedroom apartment, PCB concentration and temperature were measured on 38 occasions. Measurements were performed in 1–6 different locations simultaneously (n = 98 measurements). The apartment was heated during the period, however room temperature varied between 16.3 and 28.2 °C depending on primarily season and weather conditions. Mechanical ventilation was constant with exhaust vents in kitchen and bathrooms and slit valves in windows for inlet. All interior doors between rooms were kept open. Measurements were performed by similar procedure as in the intervention study described above, except for differences described in supplementary material. Prior to the investigation period the apartment was vacated and removal of the following was completed: furniture, interior wooden doorframes, some curtain walls, PCB containing caulks (10–28% Clophen A40 or Aroclor 1248) around window elements, curtain walls, interior doors and the accessible adjacent concrete (to 3–5 cm distance from caulks). Paints on walls and ceilings were contaminated with 100–300 mg PCBtotal /kg, and this source was not removed. The measuring period was between 22nd of May 2010 and 22nd of May 2011. Minor additional remedial measures were carried out incrementally throughout the year involving additional removal of adjacent concrete, wooden floor sanding and refinishing, testing of silicate based barrier on one bedroom façade, and finally removal of kitchen units and wooden floors. 2.5. Theoretical framework The hypothesis is that the concentration, expressed as partial pressures pi (Pa) in a defined volume indoors, follows Clausius– Clapeyron equation at different temperatures T (K):
ln(Pi ) =
a1 + a0 T
(1)
where a1 = −H/R, H is the enthalpy of evaporation of liquid PCB which can be assumed constant in the studied temperature
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133
range (i.e. well below the boiling point) and R = 8.314 J/(K mol) is the gas constant (Halsall et al., 1999). The constant a0 is determined by the general concentration level and will vary from case to case. The enthalpy of evaporation of sub-cooled liquid PCB congeners H is 78.0, 80.8 and 86.4°KJ/(K mol), respectively, for PCB28, PCB-52 and PCB-101 calculated from the slopes (ml ) in Table S2 from Falconer and Bidleman (1994). A further assumption for the hypothesis is that source concentration is constant during the intervention period, i.e. the amount of PCB emitted is relatively low compared with the content in the source. To predict the concentration at different temperatures by Eq. (1), an estimate of a0 is required. a0 can be estimated if at least one steady-state concentration measurement, P1 (Pa), at a given temperature, T1 (K) is known. In this study, there were multiple measurements and the average temperature of the sampling periods was chosen as T1 . The estimated corresponding concentration P1 was calculated by the regression function (Table S3) of measured concentration, ln(P) versus 1/T, thus all measurements in the study are used to estimate the general contamination level, a0 . Concentration predictions at different temperatures were tested on data from the intervention study as well as the additional field data from Farum Midtpunkt.
2.6. Statistical analyses One-way ANOVA and Tukey’s HSD were used for assessing the difference in mean concentrations measured at the different temperature levels. The analyses were performed using R software version 3.0.0 (2013-04-03, R Foundation for Statistical Computing, Vienna, Austria). The significance threshold was p < 0.05. Simple linear regression was done in Microsoft Excel as well as the calculation of the coefficient of determination, r2 . The coefficient of determination between measured concentrations and the model (Eq. (1)), here distinguished as R2 , was calculated by the equation:
R2 = 1 −
SSres SStot
(2)
where SSres is the residual sum of squares, and SStot is the total sum of squares.
2129
3. Results 3.1. Intervention study Air temperatures were measured day and night, respectively, at 1-min intervals throughout the sampling periods and are shown in Fig. S2. Few sampling periods were left out for reasons explained in supplementary material. No significant differences in concentrations between night and day at each temperature level were found. Of the 7 analysed congeners, only PCB-28, PCB-52 and PCB-101 were detected at all temperature levels. In Figs. S3, S4 and S5 the PCB concentrations measured in the 12 sampling periods (including the omitted concentrations) are displayed as a function of the mean temperature in the respective sampling periods (standard deviation < 0.8 °C). Concentrations increased almost exponentially with the temperature. PCB air levels in Round 2 are around 20% lower than Round 1, but considered within the uncertainty of measurements and chemical analysis. One-way ANOVA showed that measured concentrations of the specific congeners are significantly different between different temperature levels (P-value < 0.01 for PCB-28 and < 0.001 for PCB-51 and PCB-101). In order to use Eq. (1) in predicting concentrations, an estimate of the general contamination, a0 is required. Thus, at least one reference temperature, T1 and the corresponding concentration, P1 is needed. The overall average temperature of the measuring periods was chosen as T1 (24.4 °C) for the intervention study. The corresponding concentrations were calculated by the simple regression function of the measured concentrations, ln(P), versus inverse temperature. Regression slopes and intercepts of measurements can be found in Table S3. In Fig. 1, the model predictions are shown by the thick grey line and the simple regressions by the black dashed lines. In Table 1, coefficients of determination, R2 , for the measured concentrations and the model are shown for the three detected congeners PCB-28, PCB-52 and PCB-101. Surface temperatures as well as air temperatures were measured in order to assess whether the room had reached steady-state temperatures. The measured mean values are shown in Table S1. The window caulk had a congener composition similar to Aroclor 1248 (<1% w/w). Painted wallpaper in the room was determined to have a PCBtotal concentration of 166 mg/kg.
Fig. 1. Intervention study: The model (thick grey lines), the measured concentrations [Pa] (dots) and their regression lines (thin black dashed lines) of PCB-28, PCB-52 and PCB-101 at different temperatures [K].
2130
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133
Table 1 Coefficients of determination for fitting simple regression (r2 ) and comparing models (R2 ) to measured concentrations of PCB. The slopes, intercept and R2 of the regression for the correlation between modelled and measured concentrations. r2 of simple regressiona ln(P) vs. 1/T
Intervention study
Field data
a b c d e
PCB-28 PCB-52 PCB-101 PCB-28 PCB-52 PCB-101
0.94 0.94 0.95 0.49 0.44 0.45
R2 by Eq. (2) when fitting model predictionsb to measuring data
0.94 0.94 0.95 0.47 0.44 0.45
Correlation between modelled and measured concentrationsc Linear regression equation terms Slope (95% CI)d
Intercepts (95% CI)e
1.00 (0.88; 1.14) 0.99 (0.87; 1.11) 1.05 (0.93; 1.16) 1.23 (0.98; 1.48) 1.14 (0.88; 1.40) 1.09 (0.84; 1.34)
0.08 (−1.58; 1.74) −0.10 (−1.64; 1.43) 0.72 (−1.13; 2.57) 3.22 (−0.30; 6.74) 1.91 (−1.70; 5.52) 1.50 (−2.80; 5.80)
Regression is based on the measuring data. Model predictions are based on Eq. (1). Correlation plots with regression lines is found in supplementary material (Fig. S6). 95% Confidence interval of the slopes. 95% Confidence interval of the intercepts.
3.2. Additional field data In order to further investigate the ability of the model to predict observed data, the model was applied on data from an apartment in Farum Midtpunkt, Denmark. In Fig. 2 concentrations of PCB-28, PCB-52, PCB-101 are shown with the regression lines (thin black lines), the 95% confidence interval (dashed lines) and the model (thick gray lines). Model predictions are based on overall average temperature of measurements, T1 (20.5 °C), and the corresponding concentration calculated from regression of measured data. The coefficient of determination, r2 , for the simple regression of the measured data is 0.44–0.49, whereas the coefficient of determination is R2 = 0.44–0.47 for the model and the measured data. PCB-138, PCB-153 and PCB-180 were not detected. PCB-118 was detected in less than 10% of the cases and only in low concentrations and was left out for that reason. 3.3. Model validation The correlation between measured concentration and concentrations modelled by the Clausius–Clapeyron equation was investigated in order to evaluate the ability of the model to predict indoor air concentrations at any temperature in the studied range. For each measured concentration at a given temperature, the corresponding modelled concentration was calculated. Plots of measured concentration vs. their corresponding modelled concentrations are given in the supplementary material (Fig. S6) From linear regression the slopes, intercepts and the 95% CI of the correlation as well as the coefficient of determination can be found in Table 1. Slopes are close to the value 1 and confidence intervals of intercepts all includes the value of 0. In addition coefficients of determination, r2 , are given between fit of simple regression of the measured partial pressures. Also the coefficient of determination, R2 , between the models and the actual measured data. From the slopes, a1 , of the simple regression lines (measured data) in Figs. 1 and 2, enthalpy of evaporation can be calculated. In Fig. 3, these values are plotted with 95% confidence intervals alongside the values of enthalpy of evaporation as reported in literature (Falconer and Bidleman, 1994). The values from literature are within the confidence interval of the measured values in the intervention study as well as the additional field data. 4. Discussion The general analytic uncertainty stated by the analysis laboratory was a maximum of 20% for the intervention study. Deviation
between the duplicate measurements from the intervention study was below 14%, covering uncertainties of both analysis and sampling. In spite of the relative high uncertainty of the PCB measurements the intervention study showed a surprisingly good fit between actual measured concentrations and predictions based on Clausius–Clapeyron equation. This indicates that steady-state air concentration is achieved rapidly (within a few days) with temperature changes. The measurements in the additional field data from Farum Midtpunkt, Denmark, were originally aimed at investigating the effect of the remediation measures. The measurements have large variation in the concentration of the detected congeners PCB-28, PCB-52 and PCB-101 even at similar indoor temperatures. Only 43–48% of the variation in data is explained by the temperature. Since data were obtained from 6 different rooms at different stages of remediation and throughout a whole year with different seasons and fluctuating weather conditions and without mixing of air in the apartment, large variation was expected. Nevertheless the model is found to be within the confidence interval of the regression line for measured PCB partial pressures and inverse temperature as shown in Fig. 2. The most influencing factor on the general PCB levels in indoor air is the source characteristics including physico-chemical properties of source, concentration of PCB in source, amount, size and location. This enhances the variation in air concentration from building to building as well as room to room. Therefore the model needs the contamination level of the specific room. This is achieved through one or more measurements of steady-state air concentration and temperature; these values are used to determine the background level, a0 in Eq. (1). In this study we had multiple measurements and chose the overall average temperature and calculated the corresponding concentration by the simple regression equation; therefore the model is not fully theoretically derived. However the essential part of the model is that the steadystate concentrations at different temperatures measured in this intervention study can be predicted by Clausius–Clapeyron equation, and measurements could be used to calculate enthalpy of evaporation (Fig. 3). These facts suggest that PCB sources in the room behave as liquid PCBs, though steady-state air concentrations are far from saturated. Change in saturated vapour pressure and probably octanol–air partitioning coefficient (Koa ) due to temperature can also predict temperature dependence of equilibrium partial air pressures (Halsall et al., 1999). PCBs are categorised as semi-volatile organic compounds (SVOCs), thus PCB emissions are likely to be controlled by “external” emission mechanisms like solid (source) to gas-phase partitioning and mass transport limitation by diffusion in the boundary layer rather than mass transport controlled by “internal” diffu-
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133
2131
Fig. 2. Field Data: Measured partial pressures [Pa] of A) PCB-28, B) PCB-52 and PCB-101 (dots), their regression lines (thin black lines) with 95% confidence intervals and the models (thick grey lines) against inverse temperatures [K].
sion in the source matrix as is generally the case for VOCs (volatile organic compounds) (Xu and Little, 2006). For this reason, emissions are expected to be affected by air temperature in the boundary layer rather than source surface temperature. However, the air temperature immediately adjacent to the source surfaces will be correlated to the surface temperature. Temperature in the boundary layer will gradually approximate the air temperature measured in the middle of the well-mixed room air with the increasing distance to the surface. In the intervention study surface temperatures measured before and after each PCB sampling indicated steadystate surface temperatures close to the air temperature (Table S1). This suggests that room air temperature measured in the middle of the room is a good estimate for the emission temperature. In reality indoor air temperature and source surface temperature is rarely maintained in steady-state, homogeneously distributed or without temperature gradient as provided in the intervention study. Direct sun and other local sources of heat might heat up sections of indoor surfaces (secondary sources) and the thermal capacity of air is much lower than building materials and
provides a faster change in indoor air temperature compared with material temperatures which counteract constant temperature in the air. Assumption of steady-state temperature is required for model predictions, and might be difficult to achieve in regular indoor environments. Model predictions in the field data were not significantly different than regression of concentration, ln(P) vs. inverse temperature, still model prediction and temperature regression explain less than 50% of the variation in concentration. Failure to meet the assumption of steady-state temperature and concentration as well as changes in background level a0 due to the renovation can explain the remaining variation in concentration measurements. In Clausen et al. (2012), the steady-state concentration of the SVOC bis(2-ethylhexyl) phthalate (DEHP) was measured at different temperatures (23, 35, 47, 55 and 61 °C) and was also found to be exponentially increasing with increasing temperature. Liang and Xu (2014a; 2014b) found that the equilibrium gas-phase concentration of DEHP immediately adjacent to the source surface, which is a key parameter that control emission, is proportional to the con-
2132
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133
Fig. 3. Enthalpies of evaporation and 95% confidence intervals calculated from measured data (black dots and bars) and reported values (grey dots) from Falconer and Bidleman (1994).
tent of DEHP in the emitting materials. This intervention study had air concentrations of PCBs below 0.026% of the saturated vapour pressure. The relatively low air concentration is probably due to the low content of PCB in the emitting materials compared to liquid PCBs. PCB contaminated buildings might react differently to influencing factors with regard to the steady-state air concentrations, and it is unknown whether this model can be used for prediction of concentration and temperature relations in other cases and more cases needs to be investigated. The presented model depends on estimations/measurements of PCB enthalpy of evaporation by Falconer and Bidleman 1994 and if other estimations were used the model might deviate from the model presented in this study. 5. Conclusion A significant and exponential increase in the concentration of PCB in indoor air was observed with increasing temperature, even at temperature changes of only a few degrees. In the temperature intervention study a high correlation (R2 > 0.94) was found between the actual measured concentrations of PCB in air and the predicted concentrations based on the Clausius–Clapeyron equation. This was found for three out of the seven measured congeners; PCB-28, PCB-52 and PCB-101. The remaining congeners were only sporadic or non-detected. In this specific case, the predictions can be used to estimate the indoor concentration of PCB at a given indoor steady-state temperature. However factors like air exchange rate, outdoor temperature, direct sun etc. are likely to affect the steady-state conditions and the concentration independently, which can complicate the estimation of concentration levels. In the additional field data from a case study in Farum Midtpunkt, Denmark, with uncontrolled environmental conditions but known temperatures, temperature explained between 44 and 49% of the variation in air concentrations (n = 98). When the Clausius– Clapeyron equation was used on these data, the model predictions occur within the 95% confidence interval of the simple regression. The model fitted the data reasonably well (R2 = 0.44–0.47) and the model appear to be robust, even though the measurements were carried out under completely un-controlled environmental condition. Acknowledgement We would like to thank the Danish Social Housing Association (DAB) for giving us access to a highly contaminated apartment and
to Copenhagen Social Housing Association (KAB) for letting us use data from Farum Midtpunkt. The study was supported by grants from Realdania and measurements in Farum Midtpunkt were supported by LBF (Landsbyggefonden, Denmark). Appendix A. Supplementary data Supplementary data related to this article can be found at http: //dx.doi.org/10.1016/j.chemosphere.2015.10.112. References ASTM E741 11 Standard Test Method for Determining Air Change in a Single Zone by Means of a Tracer Gas Dilution. DOI: 10.1520/E0741-11 Balfanz, E., Fuchs, J., Kieper, H., 1993. Sampling and analysis of polychlorinated biphenyls (PCB) in indoor air due to permanently elastic sealants. Chemosphere 26, 871–880. Bent, S., Rachor-Ebbinghaus, R., Schmidt, C., 2000. Remediation of highly polychlorinated biphenyls (PCBs) contaminated rooms by complete removal of primary and secondary sources. In German: sanierung von hochgradig mit polychlorierten biphenylen (PCB) belasteten Innenräumen durch komplettes Entfernen der Primärund Sekundärquellen. Gesundheitswesen 62, 86–92. Benthe, C., Heinzow, B., Jessen, H., Rotard, S.M.W., 1992. Polychlorinated biphenyls. Indoor air contamination due to Thiokol-rubber sealants in an office building. Chemosphere 25, 1481–1486. Carlson, D.L., Hites, R.A., 2005. Temperature dependence of atmospheric PCB concentration. Environ. Sci. Technol. 39, 740–747. Carpenter, D.O., 1998. Polychlorinated biphenyls and human health. Int. J. Occup. Med. Environ. Health 11, 291–303. Carpenter, D.O., 2015. Exposure to and health effects of volatile PCBs. Rev. Environ. Health 30, 81–92. Clausen, P., Liu, Z., Kofoed-Sørensen, V., Little, J., Wolkoff, P., 2012. Influence of temperature on the emission of Di-(2-ethylhexyl)phthalate (DEHP) from PVC flooring in the emission cell FLEC. Environ. Sci. Technol. 46, 909–915. Danish Energy Agency report, 2013. PCB Screening in Material and Indoor Air. In Danish: Kortlægning Af PCB I Materialer Og Indeluft. Project A030835 accessed 12 12 2014 http://www.ens.dk/sites/ens.dk/files/byggeri/sikre-sunde-bygninger/ pcb_kortlaegning_dec13.pdf. Falconer, R.L., Bidleman, T.F., 1994. Vapor pressures and predicted particle/gas distributions of Polychlorinated Biphenyl congeners as functions of temperature and ortho-chlorine substitution. Atmos. Environ. 28, 547–554. Faroon, O.M., Keith, L.S., Smith-Simon, C., De Rosa, C.T., 2003. Polychlorinated biphenyls: human health aspects concise. International Chemical Assessment. Document 55. WHO accessed 7 10 2014 http://www.inchem.org/documents/ cicads/cicads/cicad55.htm . Frederiksen, M., Meyer, H.W., Ebbehøj, N.E., Gunnarsen, L., 2012. Polychlorinated biphenyls (PCBs) in indoor air originating from sealants in contaminated and uncontaminated apartments within the same housing estate. Chemosphere 89, 473–479. Gabrio, T., Piechotowski, I., Wallenhorst, T., Klett, M., Cott, L., Friebel, P., Link, B., Schwenk, M., 2000. PCB-blood levels in teachers, working in PCB-contaminated schools. Chemosphere 40, 1055–1062. Golder Associates A/S, 2013. Complete Summary of Measured Indoor Concentrations of PCBs. In Danish: Resultatoversigter over PCB-målinger I Indeluften. Project No. 12501130119, 12501130141, 12501130142, 12501130143. Griess, P., 1864. On a new class of compounds in which nitrogen is substituted for hydrogen. Proc. Royal Soc. Lond. 13, 375–384.
N.L. Lyng et al. / Chemosphere 144 (2016) 2127–2133 Guo, Z., Liu, X., Krebs, K., 2011. Laboratory study of polychlorinated biphenyl (PCB) contamination and mitigation in buildings. Part 1. Emiss. Sel. Prim. Sources EPA/600/R-11/156) U.S. Environmental Protection Agency (EPA). Halsall, C.J., Gevao, B., Howsam, M., Lee, R.G.M., Ockenden, W.A., Jones, K.C., 1999. Temperature dependence of PCBs in the UK atmosphere. Atmos. Environ. 33, 541–552. Harrad, S., Hazrati, S., Ibarra, C., 2006. Concentrations of polybrominated diphenyl ethers in indoor air and dust and polychlorinated biphenyls in indoor air in Birmingham, United Kingdom: Implications for human exposure. Environ. Sci. Technol. 40, 4633–4638. Herrick, R.F., Meeker, J.D., Altshul, L., 2011. Serum PCB levels and congener profiles among teachers in PCB-containing schools: a pilot study. Environ. Health 10, 56. Jensen, A.A., 2013. Health risks of PCB in the indoor climate in Denmark – background for setting recommended action levels. Dan. Health Med. Auth. ISBN: 978-87-7104-004-3. Kohler, M., Tremp, J., Zennegg, M., Seiler, C., Minder-Kohler, S., Beck, M., Lienemann, P., Wegmann, L., Schmid, P., 2005. Joint sealants: an overlooked diffuse source of polychlorinated biphenyls in buildings. Environ. Sci. Technol. 39, 1967–1973. Knobeloch, L., Turyk, M., Imm, P., Anderson, H., 2012. Polychlorinated biphenyls in vacuum dust and blood of residents in 20 Wisconsin households. Chemosphere 86, 735–740. Lauby-Secretan, B., Loomis, D., Grosse, Y., Ghissassi, F.E., Bouvard, V., BenbrahimTallaa, L., Guha, N., Baan, R., Mattock, H., Straif, K.on behalf of the International Agency for Research on Cancer Monograph Working Group IARC, 2013. Carcinogenicity of polychlorinated biphenyls and polybrominated biphenyls. Lancet Oncol. 14, 287–288.
2133
Liang, Y., Xu, Y., 2014a. Emission of phthalates and phthalate alternatives from vinyl flooring and crib mattress covers: the influence of temperature. Environ. Sci. Technol. 48, 14228–14237. Liang, Y., Xu, Y., 2014b. Improved method for measuring and characterizing phthalate emissions from building materials and its application to exposure assessment. Environ. Sci. Technol. 48, 4475–4484. Lyng, N.L., Trap, N., Andersen, H.V., Gunnarsen, L., 2014. Ventilation as mitigation of PCB contaminated air in buildings: review of nine cases in Denmark. Proceedings of the 13th International Conference: Indoor Air 2014. Lyng, N.L., Gunnarsen, L., Andersen, H.V., 2015. The effect of ventilation on the indoor air concentration of PCB: an intervention study. Build. Environ. 95, 305– 312. Meyer, H.W., Frederiksen, M., Göen, T., Ebbehøj, N.E., Gunnarsen, L., Brauer, C., Kolarik, B., Müller, J., Jacobsen, P., 2013. Plasma polychlorinated biphenyls in residents of 91 PCB-contaminated and 108 non-contaminated dwellings—An exposure study. Int. J. Hyg. Environ. Health 216, 755–762. Ribas-Fitó, N., Sala, M., Kogevinas, M., Sunyer, J., 2001. Polychlorinated biphenyls (PCBs) and neurological development in children: a systematic review. J. Epidemiol. Community Health 55, 537–546. Verein Deutscher Ingenieure, 2009. Ambient Air Measurement – Indoor Air Measurement – Measurement of Polychlorinated Biphenyls (PCBs) – GC-MS Method for PCB 25, 52, 101, 138, 153, 180 (VDI 2464, part 1) Issue German/English. Walkowiak, J., Wienerb, J., Fastabend, A., 2001. Environmental exposure to polychlorinated biphenyls and quality of the home environment: effects on psychodevelopment in early childhood. Lancet 358, 1602–1607. Xu, Y., little, J., 2006. Predicting emissions of SVOCs from polymeric materials and their interaction with airborne particles. Environ. Sci. Technol. 40, 456–461.