Measuring partition and diffusion coefficients for volatile organic compounds in vinyl flooring

Atmospheric Environment 35 (2001) 3823–3830 Measuring partition and diffusion coefficients for volatile organic compounds in vinyl flooring Steven S. Cox, Dongye Zhao1, John C. Little* Department of Civil & Environmental Engineering, Virginia Tech, Blacksburg, VA 24061, USA Received 15 December 2000; accepted 23 February 2001 Abstract Interactions between volatile organic compounds (VOCs) and vinyl flooring (VF), a relatively homogenous, diffusion-controlled building material, were characterized. The sorption/desorption behavior of VF was investigated using single-component and binary systems of seven common VOCs ranging in molecular weight from n-butanol to npentadecane. The simultaneous sorption of VOCs and water vapor by VF was also investigated. Rapid determination of the material/air partition coefficient (K) and the material-phase diffusion coefficient (D) for each VOC was achieved by placing thin VF slabs in a dynamic microbalance and subjecting them to controlled sorption/desorption cycles. K and D are shown to be independent of concentration for all of the VOCs and water vapor. For the four alkane VOCs studied, K correlates well with vapor pressure and D correlates well with molecular weight, providing a means to estimate these parameters for other alkane VOCs. While the simultaneous sorption of a binary mixture of VOCs is noncompetitive, the presence of water vapor increases the uptake of VOCs by VF. This approach can be applied to other diffusion-controlled materials and should facilitate the prediction of their source/sink behavior using physically-based models. # 2001 Elsevier Science Ltd. All rights reserved. Keywords: Building material; Emission; Indoor air; Microbalance; Sink; Sorption 1. Introduction A variety of building materials (e.g., adhesives, sealants, paints, stains, carpets, vinyl flooring, and engineered woods) can act as indoor sources of volatile organic compounds (VOCs). Following their installation or application, these materials typically contain residual quantities of VOCs that are then emitted over time. Once installed and depending upon their properties, these materials may also interact with airborne VOCs through alternating sorption and desorption cycles (Zhao et al., 1999b, 2001). Consequently, building materials can have a significant impact on indoor air *Corresponding author. Fax: +1-540-231-7916. E-mail address: jcl@vt.edu (J.C. Little). 1 Present address: Department of Soil and Water, Connecticut Agricultural Experiment Station, New Haven, CT 06504, USA. quality both as sources of and sinks for volatile compounds. Current methods for characterizing the source/sink behavior of building materials typically involve chamber studies. This approach can be time-consuming and costly, and is subject to several limitations (Little and Hodgson, 1996). For those indoor sources and sinks that are controlled by internal diffusion processes, physicallybased diffusion models hold considerable promise for predicting emission characteristics when compared to empirical methods (Cox et al., 2000b, 2001b). The key parameters for physically-based models are the material/air partition coefficient (K), the materialphase diffusion coefficient (D), and, in the case of a source, the initial concentration of VOC in the material (C0 ). Rapid and reliable determination of these key parameters by direct measurements or by estimations based on readily available VOC/building material properties should greatly facilitate the development 1352-2310/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII: S 1 3 5 2 - 2 3 1 0 ( 0 1 ) 0 0 1 7 5 - 3 3824 S.S. Cox et al. / Atmospheric Environment 35 (2001) 3823–3830 and use of mechanistic models for characterizing the source/sink behavior of diffusion-controlled materials (Zhao et al., 1999a; Cox et al., 2000a, 2001a). Several procedures have been used to measure D and K of volatile compounds in building materials. D and K have been inferred from experimental data obtained in chamber studies (Little et al., 1994). A procedure using a two-compartment chamber has also been used for D and K measurement. A specimen of building material is installed between the two compartments. A concentration of a particular compound is introduced into the gas-phase of one compartment while the gas-phase concentration in the other compartment is measured over time. D and K are then indirectly estimated from gas-phase concentration data (Bodalal et al., 2000; Meininghaus et al., 2000). A complicating feature of this method is that VOC transport between chambers may occur by gas-phase diffusion through pores in the building material in addition to solid-phase Fickian diffusion, confounding estimates of the mass transport characteristics of the solid material. A procedure based on a European Committee for Standardization (CEN) method has also been used to estimate D. A building material sample is tightly fastened to the open end of a cup containing a liquid VOC. As the VOC diffuses from the saturated gas-phase through the building material sample, cup weight over time is recorded. Weight change data can be used to estimate D. (Kirchner et al., 1999). A significant drawback of this method is that D has been shown to become concentration dependent in polymers at concentrations approaching saturation (Park et al., 1989). In accordance with a previously proposed strategy for characterizing homogeneous, diffusion-controlled, indoor sources and sinks (Little and Hodgson, 1996), the objectives of this study were to (1) develop a simple and rapid experimental method for directly measuring the key equilibrium and kinetic parameters, (2) examine the validity of several primary assumptions upon which the previously mentioned physically-based models are founded and (3) develop correlations between the O and K, and readily available properties of VOCs. removed from the thin slabs by conditioning in clean air at 708C for 24 h. A high-resolution (0.1–0.5 mg) dynamic microbalance (Model D200-02, Cahn) equipped with a PC-based dataacquisition system (DAQ) was used to measure and record changes in VF sample weight during sorption/ desorption tests. To minimize mechanical vibration, the microbalance was placed on a marble balance stand isolated from the floor by vibration dampening pads. An enclosure was erected around the microbalance and covered with foil-faced polyethylene insulation to minimize potential signal fluctuations due to thermal variation or electromagnetic radiation. The temperature in the microbalance enclosure was maintained at 25.6
0.38C using a constant temperature circulator (Isotemp, 1028D, Fisher Scientific) connected to a heat exchanger in the enclosure. The sample chamber temperature was monitored with a temperature transducer (RTD, Model 2Pt100G3050, Omega). A diagram of the system is shown in Fig. 1. Clean air was supplied from gas cylinders (Medical Air USP, UN1002, Air Products). The water vapor content in the air as delivered was 16 ppmv. The flow path was constructed of 3.2-mm O.D. 304 stainless steel and Teflon tubing with stainless steel fittings. The sample chamber was constructed of borosilicate glass. A glass frit was installed at the inlet end of the sample chamber to improve gas flow distribution. For sorption tests, a gas concentration of a specific VOC was generated using a constant temperature diffusion cell (Dynacalibrator Model 190, VICI Metronics, Inc.) modified by substituting a stainless steel/glass flow path. Mass flow controllers (MFC, Model FC280S, Tylan-General) were used to control the air flow rate. Gas-phase VOC concentration was determined by dividing the diffusion cell VOC emission rate by the air flow rate. To minimize errors induced by drag forces acting on the VF sample, gas-phase VOC concentrations were controlled by adjusting the diffusion cell temperature while the air flow rate was held constant. Diffusion cell VOC emission rates were determined gravimetrically. MFCs were calibrated using a soap bubble meter. The accuracy of gas-phase concentration measurements is a function of the variability associated with the use of these primary standards. 2. Materials 3. Methods A commercial vinyl flooring (VF) manufactured for use in offices, schools, and hospitals, was selected for study. This VF contains approximately 50% (by weight) limestone filler (calcium carbonate), as well as polyvinyl chloride (PVC), plasticizers, pigments, and stabilizers (Tshudy, 1998). A microtome (Model 820-II, ReichertJung) was used to cut the VF into thin slabs, ranging from 0.28 to 0.37 mm in thickness. Residual VOCs were A VF sample was placed on the microbalance in the sample chamber. The sample weight was first stabilized by passing clean air through the sample chamber until equilibrium was obtained. An air stream containing a constant and known VOC concentration was then passed through the sample chamber. VOC sample mass gain over time was monitored until equilibrium was S.S. Cox et al. / Atmospheric Environment 35 (2001) 3823–3830 3825 Fig. 1. Diagram of the microbalance test system. reached. Influent air was then switched to clean air and the desorption process was monitored until equilibrium was re-established. Equilibrium was assumed when a five-point moving average rate of mass change reached 1% of the maximum rate of change. 3.1. Determination of K and D Using the sorption and desorption data recorded by the microbalance, the equilibrium and kinetic parameters, K and D, can be determined. For a particular VOC, the sorption equilibrium is described using a partition coefficient, or K¼ C ; y ð1Þ where C is the equilibrium concentration in the materialphase (g-VOC m3), and y is the corresponding concentration of the species in the gas-phase (g-VOC m3). For a linear relationship, a higher K value represents a higher sorption capacity for a specific VOC. C is obtained from the difference between the initial and equilibrium weight of the specimen divided by specimen volume, whereas y is calculated from y¼ E ; Q ð2Þ where E is the constant emission rate of VOC generated by the diffusion cell, and Q is the air flow rate through the system. The diffusion coefficient, D, is determined by fitting a diffusion model to experimental sorption and desorption data. The VF sample conforms to the geometry of a thin slab. Under the experimental conditions, the rate of change in mass due to Fickian diffusion is given (Crank, 1975) by   1 X Mt 8 Dð2n þ 1Þ2 p2 t exp ; ð3Þ ¼1 2 2 4L2 M1 n¼0 ð2n þ 1Þ p where Mt is the total mass of a VOC that has entered or left the slab in time t, M1 is the corresponding quantity at saturation reached, and 2L is the thickness of the VF sample. 4. Results and discussion 4.1. Transient sorption and desorption of phenol Fig. 2 shows sorption and subsequent desorption profiles for phenol with VF at three different gas-phase concentrations. Equilibrium was reached in about 80 h for both sorption and desorption. The sorption and desorption profiles are highly symmetrical. It is also evident that the sorption of phenol is completely reversible. 3826 S.S. Cox et al. / Atmospheric Environment 35 (2001) 3823–3830 the D values determined for sorption are essentially the same as those determined for desorption. The experimental data in Fig. 2 were normalized by dividing Mt by M1 , as shown in Fig. 4. The coincidence of the normalized mass change curves supports the assumption that D is independent of concentration. Fig. 4 also shows the excellent fit of the diffusion model (Eq. (3)) to phenol sorption and desorption data using the average D value of 1.20 1013 m2 s1, further supporting the premise that diffusion in VF is described by Fick’s law. 4.2. K and D for a range of VOCs and water vapor Fig. 2. Transient mass gain/loss of a VF sample during sorption/desorption of phenol. Table 2 summarizes the K and D values measured for other compounds. Data were obtained by subjecting the VF sample to multiple sorption/desorption cycles at various gas-phase concentrations. Gas-phase concentrations ranged from 130,000 to 730,000 mg m3 for toluene and from 2000 to 3500 mg m3 for n-pentadecane. The gas-phase water vapor concentrations ranged from 6.0 106 to 14 106 mg m3 (26–61% RH). All data conformed to linear sorption isotherms and simple Fickian diffusion. The precision of the method for determining K varied from
14% for n-decane to
1.1% for n-tetradecane as measured by relative standard deviation. The precision of the method for determining D varied from
31% for water to
5.2% for n-dodecane as measured by relative standard deviation. 4.3. Correlations of K and D with VOC properties Fig. 3. Linear sorption isotherm for phenol in VF. Fig. 3 shows the equilibrium concentrations of phenol in VF as a function of the imposed gas-phase phenol concentration and confirms the linear relationship assumed in Eq. (1) over the range of concentrations studied. Table 1 shows the K and D values inferred from the sorption/desorption data given in Figs. 2 and 3. The sorption rates are independent of concentration, and It is known that VOC diffusion coefficients in polymeric materials often decrease as the molecular weight of the compound increases and that partition coefficients generally increase as the vapor pressure of the compound decreases (Little and Hodgson, 1996). For example, Berens and Hopfenberg (1982) correlated experimentally determined D values with van der Waals molar volume and mean diameter for various inorganic gases and organic vapors in PVC, polystyrene, and PMMA. Pankow (1989) pointed out that for certain types of VOC the sorption capacities of polyurethane foam may be correlated with vapor pressure. Table 1 Values of D and K obtained from phenol sorption/desorption data Gas-phase concentration (mg m3) Partition coefficient (-) Diffusion coefficient (Sorption) (m2 s1) 65,000 27,800 11,500 123,300 123,800 123,200 1.24 1.17 1.10 1013 1013 1013 Diffusion coefficient (Desorption) (m2 s1) 1.16 1.12 1.11 1013 1013 1013 3827 S.S. Cox et al. / Atmospheric Environment 35 (2001) 3823–3830 tion behavior is consistent with the linear partitioning mechanism. 4.5. Simultaneous sorption of VOCs and water vapor The influence of water vapor (50% RH) on the sorption of VOCs was evaluated by exposing a VF sample to gas streams containing phenol and water vapor and n-dodecane and water vapor. The sorption profiles of these binary systems were compared to the sums of the sorption profiles resulting from singlecompound sorption at identical concentrations as presented in Figs. 8 and 9. The data show that water vapor is rapidly absorbed, reaching equilibrium after about 1 h. In contrast, phenol requires 60 h to reach equilibrium and n-dodecane requires 25 h. When comparing the individual uptake curves for phenol and water vapor to the curve for the binary system, there is no observable difference in overall uptake until 3 h. After this, the VF specimen exposed to the multi-component gas stream takes up more mass than would be expected if the sorption processes were completely independent. Similar results were obtained from the n-dodecane and water vapor system. The data for both systems suggest that sorbed water molecules increase the total uptake of VOCs. The apparent increase in sorption capacity of VOCs in VF in the presence of water may be attributed to several causes. Firstly, water can exist in polymers in bound or bulk form (Sammon et al., 1998). Consequently, VOCs could dissolve into bulk water that may be present in the pores of the VF. However, the relatively small VF/water vapor partition coefficient suggests that little bulk water is present in VF. Even if all of the water in VF at equilibrium was in bulk form, calculations using Henry’s law constants for phenol and n-dodecane show that the VOC mass absorbed into the bulk water would be small compared to the apparent increase in sorption capacity Fig. 4. Fitting transient sorption/desorption data (symbols) to a diffusion model (lines) for determination of D. Figs. 5 and 6 show that for the alkane VOCs evaluated in this study, the logarithm of K correlates well with the logarithm of vapor pressure (R2 ¼ 0:998), and that D correlates well with molecular weight (R2 ¼ 0:983). 4.4. Simultaneous sorption of two VOCs The effect of simultaneous sorption of two VOCs was examined by exposing a VF sample to a gas stream containing phenol at 28,000 mg m3 and n-dodecane at 33,000 mg m3. The sorption profile of the binary gas stream was compared to the sum of the sorption profiles of the individual compounds conducted at the same concentrations as in the binary system. Fig. 7 shows that for the compounds and concentration levels studied, the sorption of one VOC is unaffected by the simultaneous sorption of another VOC. This non-competitive sorp- Table 2 Summarized values of D and K obtained from sorption/desorption experiments Compound MW Vapor pressure (mm Hg at 208C) K a (-) Da,b (m2 s1) Water n-Butanol Toluene Phenol n-Decane n-Dodecane n-Tetradecane n-Pentadecane 18 74 92 94 142 170 198 212 17 4.1 22 0.22 0.89 0.074 0.0071 0.0014 78
6.8 810
77 980
34 120,000
3000 3000
420 17,000
260 120,000
1300 420,000
38,000 3.6
1.1 6.7
0.4 6.9
1.2 1.2
0.1 4.5
1.1 3.4
0.2 1.2
0.1 6.7
1.1 a Mean
standard deviation. Obtained from both sorption and desorption rate measurements. c Number of experimental sorption–desorption cycles. b 1012 1013 1013 1013 1013 1013 1013 1014 Cyclesc 4 2 3 4 5 3 2 3 3828 S.S. Cox et al. / Atmospheric Environment 35 (2001) 3823–3830 Fig. 5. Correlation of log K vs. log vapor pressure for alkanes. Fig. 7. Transient mass gain of a VF sample during sorption of phenol and n-dodecane. Fig. 6. Correlation of D vs. MW for alkanes. of VF in the presence of water vapor. Therefore, dissolution in water alone cannot account for the observed increase in sorption capacity. Another possible mechanism is that bound water molecules could disrupt the dipole–dipole interactions between relatively polar PVC chains effectively further plasticizing the PVC in the VF (Tsukruk et al., 2000). Additional plastification would increase void volume within the PVC matrix, possibly increasing sorptive capacity. From a thermodynamic viewpoint, the free energy of the VF/solute system is lower for the mixture of solutes than for a single solute. System free energy is the sum of the chemical potential of each species present in the system. Molecular interactions between solute species sorbed to VF could lower the chemical potential of each sorbed species. Overall system equilibrium would then shift to minimize the total gas/solid system free energy. The equilibrium shift due to solute interactions would result in more molecules sorbed to the VF. The higher Fig. 8. Transient mass gain of a VF sample during sorption of phenol and water vapor. apparent sorptive capacity of the VF/phenol/water system compared to the VF/n-dodecane/water system could result from greater molecular affinity between phenol and water. Phenol and water have similar polarities, which may create a lower free energy state compared to the VF/water/dodecane system. 5. Conclusions The gravimetric method for directly measuring K and D in VF is simple and effective and can be applied to other indoor materials that can be accommodated in a microbalance. For the compounds and concentration ranges studied, K and D do not depend on concentration. This concentration independence should hold at the lower concentrations typically associated with S.S. Cox et al. / Atmospheric Environment 35 (2001) 3823–3830 3829 References Fig. 9. Transient mass gain of a VF sample during sorption of n-dodecane and water vapor. gas and material-phases in the indoor environment, confirming two of the key assumptions on which the previously developed source/sink diffusion models are based (Little et al., 1994; Little and Hodgson, 1996; Cox et al., 2000b, 2001b). The observed partition and diffusion coefficients for a series of alkane VOCs correlate well with vapor pressure and molecular weight, respectively, providing a convenient means for estimating K and D for other alkane VOCs in this type of VF without resorting to experimental measurements. Individual VOCs behaved independent of one another during binary sorption experiments, suggesting that the diffusion models may be applied to mixtures of VOCs that are either sorbing or desorbing simultaneously. In contrast, experiments with VOCs and water vapor showed that the presence of sorbed water molecules moderately increases the total uptake of VOCs. The relatively ideal behavior of the VF studied is somewhat surprising since it is not a perfectly homogeneous material (Cox et al., 2000a, 2001a). VOCs in pure polymers generally behave ideally if the concentration of VOCs in the material-phase is lower than 1% by weight (Schwope et al., 1989). The VF material studied here contained about 50% (by weight) calcium carbonate, which might be expected to alter the polymer’s behavior. These results are encouraging because they suggest that other relatively homogeneous building materials can be characterized in a similar fashion. Acknowledgements Financial support for this research was provided by the National Science Foundation (NSF) through an NSF CAREER Award (Grant No. 9624488). We thank Al Hodgson for his insightful review of the manuscript. Berens, A.R., Hopfenberg, H.B., 1982. Diffusion of organic vapors at low concentrations in glassy PVC, polystyrene, and PMMA. Journal of Membrane Science 10, 283–303. Bodalal, A., Zhang, J.S., Plett, E.G., 2000. A method for measuring internal diffusion and equilibrium partition coefficients of volatile organic compounds for building materials. Building and Environment 35, 101–110. Cox, S.S., Hodgson, A.T., Little, J.C., 2000a. A new method to measure the concentration of volatile organic compounds in vinyl flooring, Proceedings of the Engineering Solutions to Indoor Air Quality Problems Symposium, Raleigh, NC, 17–19 July. Air and Waste Management Association, Pittsburgh, PA, pp. 65–75. Cox, S.S., Little, J.C., Hodgson, A.T., 2000b. A new method to predict emission rates of volatile compounds from vinyl flooring, Proceedings of Healthy Buildings 2000, Espoo, Finland, 6–10 August, Vol. 4. SIY Indoor Air Information Oy, Helsinki, Finland, pp. 169–174. Cox, S.S., Hodgson, A.T., Little, J.C., 2001a. Measuring concentrations of volatile organic compounds in vinyl flooring, Accepted by Journal of the Air and Waste Management Association. Cox, S.S., Little, J.C., Hodgson, A.T., 2001b. Predicting the emission rates of volatile organic compounds from vinyl flooring, submitted to Environmental Science & Technology. Crank, J., 1975. The Mathematics of Diffusion. 2nd Edition. Oxford University Press, New York, pp. 44–49. Kirchner, S., Badey, J.R., Knudsen, H.N., Meininghaus, R., Quenard, D., Saarela, K., Sallee, H., Saarinen, A., 1999. Sorption capacities and diffusion coefficients of indoor surface materials exposed to VOCs: proposal of new test procedures, Proceedings of the eighth International Conference on Indoor Air Quality and Climate}Indoor Air ’99, Edinburgh, Scotland, 8–13 August, Vol. 1. Construction Research Communications Ltd., London, pp. 430–435. Little, J.C., Hodgson, A.T., 1996. A strategy for characterizing homogeneous, diffusion-controlled, indoor sources and sinks. Standard Technical Publication 1287, American Society for Testing and Materials, pp. 294–304. Little, J.C., Hodgson, A.T., Gadgil, A.J., 1994. Modeling emissions of volatile organic compounds from new carpets. Atmospheric Environment 28, 227–234. Meininghaus, R., Gunnarsen, L., Knudsen, N., 2000. Diffusion of sorption of volatile organic compounds in building materials-impact of indoor air quality. Environmental Science and Technology 34, 3101–3108. Pankow, J.F., 1989. Overview of the gas phase retention volume behavior of organic compounds on polyurethane foam. Atmospheric Environment 23, 1107–1111. Park, J.K., Holsen, T.M., Bontoux, L., Jenkins, D., Selleck, R.E., 1989. Permeation of plastic pipes by organic chemicals. Sanitary Engineering and Environmental Health Research Laboratory (SEERHL) Report, University of California, Berkeley, CA, January. Sammon, C., Mura, C., Yarwood, J., Everall, N., Swart, R., Hodge, D., 1998. FRIR-ART studies of the structure and dynamics of water molecules in polymeric matrixes, a comparison of PET and PVC. Journal of Physical Chemistry B 102, 3402–3411. 3830 S.S. Cox et al. / Atmospheric Environment 35 (2001) 3823–3830 Schwope, A.D., Lyman, W.J., Reid, R.C., 1989. Methods for assessing exposure to chemical substances, USEPA Report (EPA 560/5-85-015), Vol.11. Tshudy, J.A., 1998. Personal communication. Tsukruk, V.V., Gorbunov, V.V., Huang, Z., Chizhik, S.A., 2000. Dynamic microprobing of viscoelastic polymer properties. Polymer International 49, 441–444. Zhao, D.Y., Cox, S.S., Little, J.C., 1999a. Source/sink characterization of diffusion controlled building materials, Proceedings of the eighth International Conference on Indoor Air Quality and Climate}Indoor Air ’99, Edinburgh, Scotland, 8–13 August, Vol. 1. Construction Research Communications Ltd, London, pp. 408–413. Zhao, D.Y., Rouques, J., Little, J.C., Hodgson, A.T., 1999b. Effect of reversible, diffusion-controlled sinks on VOC concentrations in buildings, Proceedings of the eighth International Conference on Indoor Air Quality and Climate}Indoor Air ‘99, Edinburgh, Scotland, 8–13 August, Vol. 5. Construction Research Communications Ltd, London, pp. 264–269. Zhao, D.Y., Little, J.C., Hodgson, A.T., 2001. Dynamic modeling of the sink effect in response to transient contaminant sources, submitted to Indoor Air.