Article pubs.acs.org/est
Adsorption of Phthalates on Impervious Indoor Surfaces Yaoxing Wu,† Clara M. A. Eichler,† Weinan Leng,†,‡ Steven S. Cox,† Linsey C. Marr,† and John C. Little*,† †
Department of Civil and Environmental Engineering, Virginia Tech, Blacksburg, Virginia 24061, United States Virginia Tech Institute for Critical Technology and Applied Sciences (ICTAS), Blacksburg, Virginia 24061, United States
‡
S Supporting Information *
ABSTRACT: Sorption of semivolatile organic compounds (SVOCs) onto interior surfaces, often referred to as the “sink effect”, and their subsequent re-emission significantly affect the fate and transport of indoor SVOCs and the resulting human exposure. Unfortunately, experimental challenges and the large number of SVOC/surface combinations have impeded progress in understanding sorption of SVOCs on indoor surfaces. An experimental approach based on a diffusion model was thus developed to determine the surface/air partition coefficient K of di-2-ethylhexyl phthalate (DEHP) on typical impervious surfaces including aluminum, steel, glass, and acrylic. The results indicate that surface roughness plays an important role in the adsorption process. Although larger data sets are needed, the ability to predict K could be greatly improved by establishing the nature of the relationship between surface roughness and K for clean indoor surfaces. Furthermore, different surfaces exhibit nearly identical K values after being exposed to kitchen grime with values that are close to those reported for the octanol/air partition coefficient. This strongly supports the idea that interactions between gas-phase DEHP and soiled surfaces have been reduced to interactions with an organic film. Collectively, the results provide an improved understanding of equilibrium partitioning of SVOCs on impervious surfaces.
■
INTRODUCTION Tens of thousands of manufactured chemicals are currently used in consumer products in the United States and at least a thousand more are introduced each year.1,2 Semivolatile organic compounds (SVOCs), including many phthalates, flame retardants, and preservatives, are organic molecules with vapor pressures between 10−9 and 10 Pa.3,4 SVOCs are often present in consumer products as additives that are used to enhance product performance and may be present in the product at levels up to tens-of-percent by mass.5−8 Human exposure to SVOCs in consumer products may occur following emission from the source and migration to other media. The potential adverse health effects associated with some SVOCs include endocrine disruption, asthma, and allergies.9−11 For these reasons, indoor emissions, transport, and potential human exposure to these SVOCs have become a high priority research area that informs chemical safety and public health. Although there is mounting interest in human exposure to SVOCs in the indoor environment, substantial analytical challenges and extensive surface adsorption due to their low volatility have impeded progress in studying them.3,12 The need for indoor air quality and exposure assessment models is thus becoming more relevant.12 However, there is still significant uncertainty associated with the treatment of compound-specific physicochemical properties in those models. The surface/air partition coefficient K is a typical model input to predict the interaction of SVOCs with interior surfaces, but is a © 2017 American Chemical Society
characteristic that is not yet well understood. Interior surfaces play an important role in the occurrence and transport of SVOCs in the indoor environment, as interphase mass transfer among air, building and occupant surfaces, particles, and dust can redistribute SVOCs throughout rooms and buildings. SVOC partitioning onto indoor surfaces tends to result in higher mass on surfaces than in the gas phase,3,13 meaning that dermal contact with contaminated surfaces can be an important route of exposure, substantially increasing the potential dose of SVOCs. Therefore, characterization of the partitioning of SVOCs to interior surfaces is essential to better predict indoor SVOC exposure and develop reliable risk assessments. Among the considerable efforts made to investigate the mass transfer mechanisms of SVOCs from indoor sources, only limited research has been conducted to determine K experimentally for limited combinations of SVOCs and materials. The most common approach is to pass an air stream with a constant concentration of target compounds through a ventilated chamber initially containing clean test materials. The values of K are determined after adsorption equilibrium is established between the material and air. For example, in a previous study, a sandwich-like chamber has been used to Received: Revised: Accepted: Published: 2907
November 20, 2016 January 25, 2017 January 31, 2017 January 31, 2017 DOI: 10.1021/acs.est.6b05853 Environ. Sci. Technol. 2017, 51, 2907−2913
Article
Environmental Science & Technology measure K of di-2-ethylhexyl phthalate (DEHP) on a stainless steel surface.14 Liu et al. developed a modified small chamber test method to measure the adsorption of polychlorinated biphenyl (PCB) congeners on common indoor materials including glass and polyethylene.15 Morrison et al. measured the K values for methamphetamine on skin oil and cotton fabrics, and for diethyl phthalate (DEP) and di-n-butyl phthalate (DnBP) on cotton fabrics in small chamber experiments.16,17 Bi et al. measured K for benzyl butyl phthalate (BBzP) and DEHP on different interior surfaces (dust, dish plates, windows, mirrors, fabric cloth, and wood) using field measurements in a test house.18 Since equilibrium partitioning is a prerequisite for these measurements, a noticeable constraint is the long experimental duration, which is usually more than 10 days. Furthermore, the time required to reach equilibrium could be impractically long (>85 days) for SVOC/surface pairs with large K according to the calculation of Morrison et al.16 To alleviate this problem, Xiong et al. recently developed a sealed-chamber approach based on the early stage C-history method to measure K values of DEHP, DiBP, and DnBP partitioning onto cotton clothing.19 However, the sink of SVOCs on the chamber wall is neglected in this method, which could compromise its accuracy. In addition, the requisite time to make reliable measurements can still be long for compounds with large K values such as DEHP (more than 1680 h at 25 °C). We recently developed a simple method based on a passive sampling technique to measure the gas-phase SVOC concentration (y0) immediately adjacent to the material surface in a consumer product.20 The method employs standard sorbent tubes, with values of y0 and K, the tube surface/air partition coefficient inside the sorbent tube, obtained by fitting a diffusion model to the sampling data. However, given that the commercial sorbent tubes serving as diffusive samplers are only available in stainless steel and glass, the determination of SVOC surface/air partition coefficients using sorbent tubes was limited to these two types of materials. The adsorption (partitioning) process is affected by surface morphology, and most surfaces are to some extent rough. Unfortunately, surface roughness is usually ignored in the studies of surface/air partitioning of SVOCs. In addition, Weschler et al. have proposed that most indoor surfaces are covered with a thin film of “grime” that accumulates indoors.21 These films are mixtures of sorbed water, sorbed organics, inorganic molecules and ions, and deposited airborne particles. Several studies showed that such films could develop at the interface of most impervious indoor surfaces with room air.3,22−24 This may affect the surface-air exchange of SVOCs because the gas-phase SVOCs could partition into the film layer. However, this mechanism has received little attention, and the implications for the measurement of K are currently unknown. Therefore, the objective of this study is to improve the technique for measuring K for different types of surfaces, with an emphasis on impervious materials. A new passive sampler was designed and applied to characterize the adsorption of DEHP on several impervious indoor surfaces that are commonly used in cookware and furniture and are thus related to dietary and dermal exposure. The effect of surface roughness and the influence of a thin organic layer on the adsorption behavior of DEHP on those impervious surfaces were considered. The results of this study will provide an improved understanding of the factors that determine SVOC adsorption to indoor impervious surfaces, which is essential in
assessing human exposure to SVOCs and improving the quality of the indoor environment.
■
EXPERIMENTAL SECTION Chemicals. DEHP (certified reference material grade) was obtained from Sigma-Aldrich, St. Louis, MO, U.S.A. D4-dibutyl phthalate-3,4,5,6 (d4-DiBP, 98 atom % D) used as internal standard (IS) and d4-bis(2-ethylhexyl) phthalate-3,4,5,6 (d4DEHP, 98 atom % D) used as surrogate for the determination of recovery rates were also obtained from Sigma-Aldrich. Reagent-grade acetone and methanol used for cleaning and >99.9% methylene chloride (DCM) and >99.9% methyl tertbutyl ether (MTBE) used as solvents were all obtained from Fisher Scientific, Waltham, MA, U.S.A. Materials. Polyvinyl chloride (PVC) flooring material with the value of y0 previously characterized in chamber tests was chosen as the SVOC emission source material.20,25,26 The DEHP content in the vinyl flooring used for this study is 23 ± 3 wt %.20 Prior to the test, the flooring was wrapped in aluminum foil and stored at room temperature. Six impervious indoor materials including aluminum, stainless steel, steel, polished glass, ground glass, and acrylic were tested to characterize their respective K values for DEHP (Figure S1 of the Supporting Information). These impervious materials are of interest since they are commonly used in indoor environments for different applications including silverware, cookware, dishware, furniture, tools, and computer equipment. Therefore, these materials are closely related to dietary and dermal exposure. All materials were obtained in disk shape with a diameter of 25.4 mm and a thickness of 3 to 6.4 mm. The aluminum and stainless steel #1 disks were obtained from a local hardware store; stainless steel #2 and steel disks were purchased from SHARPE products, New Berlin, WI. The ground glass disks were cut from rods in a glass shop, and the polished glass disks were ordered from Technical Glass Products Inc., Painesville, OH. The acrylic disks were obtained from Delvies Plastics, Salt Lake City, UT. Measurement of Surface Roughness. Atomic-force microscopy (AFM) was used to characterize the surfaces of the test materials (WITec alpha500, Ulm, Germany). The AFM was operated in AC mode with a silicon nitrite probe having a force constant of 42 N/m. The images were captured with a scan rate of 1 s/line and a 5 μm scan size. A resolution of 1024 × 1024 was used with a reference area (area footprint) of 24.9 μm2. Three to four scan areas were randomly selected for each sample to measure the true surface area. The roughness factor of the surface is defined as follows:27 roughness factor =
true surface area area footprint
(1)
Diffusive Sampler. The design of the diffusive sampler is based on the model described in a subsequent section and shown schematically in Figure 1. The cylindrical diffusion chamber consisted of one aluminum shim (20.2 mm ID, 25.9
Figure 1. Schematic representation of the diffusive sampler. 2908
DOI: 10.1021/acs.est.6b05853 Environ. Sci. Technol. 2017, 51, 2907−2913
Article
Environmental Science & Technology
clean air. A set of diffusive samplers was placed in a temperature-controlled cabinet (Lunaire CEO-932) operated at 25 ± 0.5 °C. After the materials were exposed to DEHP for different sampling periods ranging from 1 to 14 days, the sample disks were removed using precleaned tweezers and immediately placed in 20 mL glass jars for extraction and analysis. Duplicate sampling was conducted for each time period. Sample Analysis and Quality Control. The DEHP from all sample disks was extracted using 10 mL of DCM except for the acrylic disks. For those, MTBE was used to prevent decomposition of the acrylic material. To monitor the recovery efficiency of solvent extraction, the solvent of each sample was spiked with 10 μL of 20 μg/mL d4-DEHP solution prior to sample extraction. All extracts were subsequently concentrated to 500−1000 μL using a gentle stream of high purity nitrogen. A total of 10 μL of IS (250 μg/mL d4-DiBP) was added prior to analysis for volume correction and signal intensity normalization. All samples were analyzed using gas chromatography−mass spectrometry (GC−MS, Finnigan TraceGC ultra with Finnigan Trace DSQ, Thermo Electron Corporation, Waltham, MA). Five-point calibration lines ranging from 20 to 500 ng/mL (lower range) or 100 to 2000 ng/mL (upper range) were used for quantification of the target analytes, depending on the sample concentration. All glassware was heated to 400 °C for 2 h before each experiment to minimize contamination. Two procedural blanks using unexposed disks were included in each batch of samples to detect background contamination during the experimental process. The blanks were subject to the same extraction and analysis as were the actual samples. Results showed that DEHP was detected at low levels in all blanks (40%) could be associated with random analytical errors or background contamination. The numerical model provides a reasonable fit to the sampling data (Figure 4). The coefficients of determination R2, which provide a measure of how well the sampling data are fitted by the model, are shown in Table 2. Among the materials tested, acrylic and polished glass present a lower R2 value since the experimental sampling data display greater variation, while the agreement between the experimental results and fitted curves appears to be acceptable (>0.84) for the other materials. The
■
RESULTS AND DISCUSSION Surface Roughness of Clean Surfaces. Table 1 lists the surface characteristics determined by AFM for all test materials. Steel has the highest surface roughness among the test materials, followed by ground glass and stainless steel #2. 2910
DOI: 10.1021/acs.est.6b05853 Environ. Sci. Technol. 2017, 51, 2907−2913
Article
Environmental Science & Technology
equilibrium mass. Figure S2 shows that tequ of phthalates increases linearly with increasing K, which is consistent with the findings of Cao et al.31 As aluminum, polished glass, and acrylic exhibit relatively lower adsorption capacity, the adsorption equilibrium is reached in less than 300 h. For ground glass and steel, the value of tequ is much longer (>2000 h). This agrees well with both experimental observation and model prediction (Figure S3). The findings suggest that tequ varies substantially for different impervious surfaces even for one SVOC, and the equilibrium time scale can be substantial if the partition coefficient is large. For indoor impervious surfaces that are cleaned daily or weekly, such as ceramic countertops, silverware, cookware, and dishware, the relatively slow adsorption of DEHP from air to those surfaces may result in limited daily dietary ingestion and dermal sorption. However, dermal uptake from impervious surfaces that are not frequently cleaned may be significant, for example, dermal contact with furniture and computer equipment. It is also worth noting that the tequ in this study is calculated based on a quiescent environment where diffusion is the only mass transfer mechanism; tequ in the actual indoor environment may be shorter due to turbulent mass transfer. Partition Coefficient K of Soiled Surfaces. Figure 6 shows the mass of DEHP measured on the surface of the soiled
Table 2. Approximate Values of Surface/Air Partition Coefficients K for DEHP on Impervious Materials at 298 K (25 °C)a K (m) aluminum stainless steel #1 stainless steel #2 steel polished glass ground glass acrylic
this study
R2
× × × × × × ×
0.84 0.88 0.94 0.97 0.68 0.97 0.62
6 1.2 1.3 1 6 4.2 5
2
10 103 103 104 102 103 102
literature 1700b, 1500c
758 (21 °C)d
a K is defined as the mass adsorbed per unit surface area divided by the concentration in the air. bWu et al., 2016.20 cLiang and Xu, 2014.25 dBi et al., 2015.18
fitted approximate values of the only unknown parameter K for clean impervious surfaces are also listed in Table 2, which are in the range of 5 × 102 to 1 × 104 m. Steel displays the highest K value, while acrylic shows the lowest. The value of K is expected to decrease with increasing temperature. Figure 5 plots the measured surface/air partition coefficients versus the roughness factor of the test materials, and K appears
Figure 5. Measured surface/air partition coefficient K (m) versus roughness factor of test materials.
Figure 6. Measured DEHP mass accumulated on the soiled surfaces as a function of time and best-fit model curve.
to linearly correlate with the roughness factor. Although larger data sets are clearly needed to support the association, the present data suggest that the surface roughness plays an important role in the adsorption process. If this association holds true, as increasing numbers of measurements for different clean surfaces are completed, then the data could be used to establish the relationship between K and surface roughness for specific SVOCs. The understanding of surface adsorption involving SVOCs would thus be greatly facilitated if the value of K could be approximately predicted from the surface roughness, instead of being measured each time when a new impervious material is investigated. Table 2 also shows experimentally determined partition coefficients reported in the literature for the same type of material, although remarkably few measured values are available for phthalates. The data determined in this study shows reasonable agreement with literature values. The discrepancies in K values could be related to the different surface roughness. Figure S2 also suggests that the value of K determines the time scale required for the chamber to reach equilibrium (tequ). On the basis of the measured values of K listed in Table 2, we calculated tequ for test materials, assuming the adsorption process reaches equilibrium when Mq reaches 95% of its
material disks over time for aluminum, polished glass, and ground glass. The RSD of the replicate samples range from 1 to 18%. Although the clean surfaces exhibit different adsorption properties as discussed in the previous section, the soiled surfaces display a completely different partitioning behavior. A nearly identical accumulation of DEHP on the surface of those soiled materials was observed, providing direct evidence that the interactions between gas-phase DEHP and different soiled surfaces have been reduced to interactions with grime films. This is consistent with our expectation as organic films have been found to develop rapidly on clean impervious surfaces when the material disks were placed in the kitchen environment. The film is derived from cooking activities, presumably condensation and deposition of cooking oil because of its low vapor pressure.24 As a result, the sampling data of all soiled disks was fitted with only one adsorption isotherm, and the fitted K (5 × 104 m) is orders of magnitude higher than the values of K for all clean materials. It is worth noting that the soiled surface with such a high K is more like an infinite and instantaneous sink, which keeps removing DEHP from the air. The concentration gradient across the sealed chamber remains at a maximum for a long period of time, thus the accumulation 2911
DOI: 10.1021/acs.est.6b05853 Environ. Sci. Technol. 2017, 51, 2907−2913
Article
Environmental Science & Technology
partitioning into an organic film layer. Further investigation of K for a range of SVOCs with different vapor pressures and surfaces of different roughness is needed. Moreover, considering that a large proportion of indoor surfaces is covered by pervious materials such as wood, laminate, and fabric, further research to determine the interaction between SVOCs and pervious materials including partition and diffusion is also of interest. Nevertheless, the new knowledge provides improved insight in estimating K values for impervious surfaces and a sound basis for further method development for pervious surfaces.
rate of SVOCs on the surface is not very sensitive to K values (Figure S4). Therefore, the fitted K for soiled surfaces might be subject to some uncertainty in this case. The surface/air partitioning coefficients K for soiled disks were then converted to the dimensionless organic matter/gas partitioning coefficient Kom, which is expressed as Csurf/C. Csurf (μg/m3) is the concentration of DEHP in the organic matter on the soiled surface that would be in equilibrium with gasphase DEHP concentration C.3 The relationship between K and Kom is as follows: Kom =
A ·ρom q A ·ρom A·q Csurf = = · = K C Vom·C Vom·ρom C Mom
■
(8)
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.6b05853. Figure S1: Photograph of surface materials. Figure S2: The correlation between the time to reach adsorption equilibrium (tequ) and the value of surface/air partition coefficient K for impervious surface. Figure S3: Predicted concentrations of gas-phase DEHP (μg/m3) in the sealed chamber a) aluminum at 350 h; b) ground glass at 300 h. The center of the chamber is the left edge in the diagram; the source is at the bottom. Figure S4: The total mass accumulated on the surface and mass flux to the surface at the time of 350 h as a function of surface/air partition coefficient K for the current sampling setup (PDF)
where A (m2) is the base area of the cylindrical chamber. Vom (m3) is the volume of organic matter accumulated on the surface of soiled disks. Mom (g) is the mass of organic matter accumulated on the surface of soiled disks. The density of the organic matter ρom (g/cm3) is assumed to be close to the density of cooking oil, as the material disks were mainly exposed to kitchen grease in this test. The average density of cooking oil, 0.92 g/cm3, was used in the calculation. On the basis of the mass of measured organic matter and average oil density, the average bulk film thickness was calculated to be 2 × 103 nm, which is higher than the film thickness reported in the literature (
