From Clothing to Laundry Water: Investigating the Fate of Phthalates

Aug 10, 2016 - Group 2 samples were then dried in a desiccator to remove water after which they were extracted (again using ASE with hexane, DCM and a...
0 downloads 9 Views 2MB Size
Article pubs.acs.org/est

From Clothing to Laundry Water: Investigating the Fate of Phthalates, Brominated Flame Retardants, and Organophosphate Esters Amandeep Saini,† Clara Thaysen,‡ Liisa Jantunen,§,‡ Rachel H. McQueen,∥ and Miriam L. Diamond*,‡,† †

Department of Physical and Environmental Sciences, University of Toronto Scarborough, 1265 Military Trail, Toronto, Ontario M1C 1A4 Canada ‡ Department of Earth Sciences, 22 Russell Street, University of Toronto, Toronto, Ontario M5S 3B1 Canada § Air Quality Processes Research Section, Environment and Climate Change Canada, 6248 Eighth Line, Egbert, Ontario L0L 1N0 Canada ∥ Department of Human Ecology, University of Alberta, Edmonton, Alberta T6G 2N1 Canada S Supporting Information *

ABSTRACT: The accumulation of phthalate esters, brominated flame retardants (BFRs) and organophosphate esters (OPEs) by clothing from indoor air and transfer via laundering to outdoors were investigated. Over 30 days cotton and polyester fabrics accumulated 3475 and 1950 ng/dm2 ∑5phthalates, 65 and 78 ng/dm2 ∑10BFRs, and 1200 and 310 ng/dm2 ∑8OPEs, respectively. Planar surface area concentrations of OPEs and low molecular weight phthalates were significantly greater in cotton than polyester and similar for BFRs and high molecular weight phthalates. This difference was significantly and inversely correlated with KOW, suggesting greater sorption of polar compounds to polar cotton. Chemical release from cotton and polyester to laundry water was >80% of aliphatic OPEs (log KOW < 4), < 50% of OPEs with an aromatic structure, 50−100% of low molecular weight phthalates (log KOW 4−6), and < detection−35% of higher molecular weight phthalates (log KOW > 8) and BFRs (log KOW > 6). These results support the hypothesis that clothing acts an efficient conveyer of soluble semivolatile organic compounds (SVOCs) from indoors to outdoors through accumulation from air and then release during laundering. Clothes drying could as well contribute to the release of chemicals emitted by electric dryers. The results also have implications for dermal exposure.



INTRODUCTION Clothing plays a central role in our society with global exports of textiles estimated at $294 billion USD in 2011.1 In Sweden, clothing, followed by household textiles (e.g., upholstery and curtains), was estimated to have the greatest surface area of all materials indoors.2 Clothing is unique in the indoor environment as it undergoes continual laundering. In terms of human exposure, clothing covers 85% of human skin3 and can act as a barrier to exposure to environmental and occupational airborne chemicals.4−7 For nonoccupational exposure, these chemicals are likely confined to indoors where many North Americans spend over 90% of the time.8 However, clothing can also be a source of exposure of intentionally and unintentionally added chemicals.9,10 For example, Morrison et al.11 conducted a series © XXXX American Chemical Society

of experiments showing that clothing can reduce or increase dermal uptake of phthalates at environmentally relevant concentrations, according to whether the clothing was clean (i.e., not contaminated) or exposed to phthalates. Textile fibers, from which fabrics are made, can be categorized according to origin into natural (e.g., cellulosebased cotton and protein-based wool), semisynthetic (e.g., rayon from regenerated cellulose), and synthetic fibers (e.g., polyester synthesized from terephthalic acid and ethylene glycol Received: April 24, 2016 Revised: July 11, 2016 Accepted: July 18, 2016

A

DOI: 10.1021/acs.est.6b02038 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

at −4 °C until extraction. Group 1 samples were extracted using ASE with hexane, dichloromethane (DCM) and acetone (2:1:1, v/v) (HPLC grade, Fisher Scientific) immediately after removal. Group 2 samples were laundered after removal, where the laundry water was collected for analysis. Group 2 samples were then dried in a desiccator to remove water after which they were extracted (again using ASE with hexane, DCM and acetone, 2:1:1). Group 3 samples were laundered, dried in an automatic dryer (LG electric dryer, DLEX3250R model) and the lint was collected. The machine dried fabrics and lint samples were extracted as per the other groups. Laundering, Drying, Extraction and Analysis. Laundering and Drying. Fabrics from groups 2 and 3 were laundered at ∼25 °C using Natural 2× concentrated liquid laundry detergent (Seventh Generation, Burlington, VT). Details of laundering and drying in an electric dryer are given in SI. Extraction. Fabrics and lint were extracted using ASE with hexane, DCM and acetone (2:1:1, v/v) at the operating conditions listed in SI. The in-cell extraction and clean up method of Saini et al.30 was modified by adding 5 g of precleaned silica gel (Fisher Scientific) to the ASE cells along with 10 g of anhydrous sodium sulfate (Fisher Scientific). The clean extract was volume reduced and solvent exchanged into iso-octane under a gentle stream of nitrogen (Zymark TurboVap II concentration workstation, Caliper Life Science, MA) to a final volume of 0.5 mL (HPLC grade, Fisher Scientific) Laundry water (∼1 L) was liquid−liquid extracted in a separatory funnel thrice using DCM (50 mL × 3) as described by Schreder and La Guardia.23 The DCM extract was dried over sodium sulfate after which the sample volume was reduced and exchanged into iso-octane using the TurboVap as above. Analysis. Five phthalates, 14 PBDEs, 11 “novel” brominated flame retardants (NFRs) and eight OPEs were analyzed (Supporting Information (SI) Table S1). The three isomers of TCPP were treated as distinct chemicals although their distinct physical-chemical properties have not been reported in the literature.31 We use the notation for these isomers recommended by Truong et al.31 See SI Table S1 for the notation used. Samples were analyzed using an Agilent 6890N gas chromatograph coupled with Agilent 5975 inert massselective detector (GC-MSD). Full details of the operating conditions are given in SI with instrumental detection limits and limits of quantitation listed in SI Table S2. QA/QC. Blanks and recoveries were monitored throughout the sampling and analytical measurements. Field blanks were collected during fabric deployment and were extracted and analyzed along with laboratory blanks and samples for each group of fabrics. Averages of field and laboratory blank levels of each group were used for blank correction (and discarding samples) according to the criteria explained by Saini et al.30 Any sample value below the limit of quantification (given in SI, obtained using a signal-to-noise ratio of 10:1) or discarded during blank correction was left out during data analysis. Surrogate standards were added prior to extracting fabrics, lint and laundry water samples to check recoveries. DEHP-d4 was used as the surrogate standard for phthalates; mPBBZ, mHBB (mass-labeled) and F-BDE-100, -154, and -208 (fluorinated BDEs) for BFRs; and dTnBP and mTPhP for OPEs (Accustandard, and Wellington Laboratories, Canada). Average recoveries for surrogates were 60% (DEHP-d4), 55−80% (BFRs), and 50−65% (OPEs). The data were recovery corrected. The extraction method was validated by extracting

monomers). Several studies have shown that sorption of polar semivolatile organic compounds (SVOCs) such as nicotine, is greater to fabrics of natural origin than nonpolar synthetics since natural fibers have polar functional groups, for instance hydroxyl (cellulose and protein) or amide groups and polar amino acid side chains (protein).12−14 The extent of sorption of polar compounds has been suggested to be a function of the hygroscopicity of a fabric,15,16 which is consistent with the importance of polar functional groups. Nonpolar compounds are expected to sorb preferentially to nonpolar fabrics and materials,17,18 where aromaticity is expected to play a role.19,20 Recently, Saini et al.21,22 conducted experiments under controlled and ambient indoor conditions that confirmed accumulation by cotton, rayon, and polyester of gas- and particle-phase phthalates and brominated flame retardants (BFRs). They reported similar accumulation and uptake rates regardless of chemical or fabric when normalized to planar surface area, which is consistent with air-side controlled uptake. Clothing has been hypothesized to transfer BFRs such as polybrominated diphenyl ethers (PBDEs) and organophosphate esters (OPEs) from indoors to outdoor surface waters via wastewater as a result of laundering.23 Based on the similarity in flame retardant profiles in dust and laundry water, Schreder and La Guardia23 hypothesized that this transfer occurs from indoors to outdoor waters due to the release of flame retardants (FRs) during laundering, where chemicals were accumulated by clothing via contaminated dust or air. Additionally, they showed that the transfer was greatest for water-soluble OPEs. Based on the findings and explanations in the literature, it stands to a reason that physical-chemical properties of SVOCs as well as fabrics play a role in their accumulation by clothing and release during laundering. Our goal was to investigate the role of clothing as a sorbent of indoor SVOCs and as a source to outdoors through laundering. We hypothesized that the physical−chemical properties of the SVOCs and fabrics control accumulation from air and release to laundry water. The study was designed to first investigate SVOC accumulation by fabrics followed by release to laundry water and/or retention after laundering, and the contribution of clothes drying. We focused on phthalates, BFRs and OPEs. Numerous studies have documented the prevalence of PBDEs, other BFRs and OPEs indoors.24−26 Phthalates and primarily nonhalogenated OPEs are used as plasticizers and a growing literature documents their elevated levels indoors.27−29



MATERIALS AND METHODS Test Fabrics. Fabrics purchased from Testfabrics, Inc. (Pennsylvania) were plain weave cotton (120 g/m2, bleached, unmercerized) and polyester (127 g/m2, poplin). The specific surface areas of cotton and polyester measured use the BET adsorption method (BET-SSA) were 1.17 and 0.14 m2/g, respectively. The method for measuring BET-SSA is described in Supporting Information (SI). Fabrics were cut into 35 × 35 cm2 squares and were precleaned by pressurized liquid extraction using an accelerated solvent extractor (ASE) (Dionex ASE 350, Thermo Scientific) with hexane (HPLC grade, Fisher Scientific). Cleaned fabric squares were dried in a desiccator and then fixed to a metal frame for deployment. Experimental Design. Three groups, 10 squares each of cotton and polyester fabrics, were hung vertically 1.5 m above the floor at ∼25 °C in an office located at University of Toronto. Fabrics were removed from the frames after 30 days and each piece was wrapped in clean aluminum foil and stored B

DOI: 10.1021/acs.est.6b02038 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology and analyzing spikes of native compounds added to the fabric samples. Average recoveries of spiked analytes ranged between 58−130% for phthalates, 79−110% for BDEs (except BDE209), 55−170% for NFRs (except OBTMPI and DBDPE), and 70−110% for OPEs. The data were quantified using internal standards, fluoranthene-d10, BDE-118 and mirex for phthalates, BFRs and OPEs, respectively. Microsoft Office Excel 2007 was used for descriptive data analyses. Nonparametric statistical tests (Mann−Whitney U test, MWU, and Kruskal−Wallis ANOVA, KWA) were performed using STATISTICA software version 8 (StatSoft Inc., OK).



RESULTS AND DISCUSSION For each type of fabric, the total masses extracted from fabric groups 1−3 (sum of masses in fabrics and water for groups 2 and 3) for the three chemical classes were not statistically different, even though they underwent different treatments (i.e., ASE extraction only, washing and drying in desiccator or dryer followed by ASE extraction). Chemical Accumulation by Fabrics Normalized to Planar Surface Area (Group 1). Out of 36 chemicals, 23 chemicals had ≥70% detection in the samples with exceptions noted below. The high detection limits of some chemicals such as BEH-TEBP, DBDPE and BDE-209 (SI Table S2) limited their detectability in the samples. When considering planar surface area normalized concentrations, ∑5phthalates were ∼50 and 25 times higher than ∑10BFRs, and three and six times higher than ∑8OPEs in cotton and polyester, respectively. Phthalates. Detection frequencies of the five target phthalates were ≥90% except for 70% for DiBP and DiNP in polyester (SI Table S3). The relative standard deviation (RSD) was 0.05) (Figure 1a, SI Table S3). DEHP had highest concentration of all chemicals reported. DEHP concentrations in cotton (1363 ± 492 ng/dm2 fabric) and polyester (1091 ± 460 ng/dm2 fabric) constituted 40 and 56% of ∑5phthalates, respectively. DiBP and DnBP were nine and five times higher in cotton (650 and 914 ng/dm2, respectively) than polyester (70 and 179 ng/dm2, respectively). Phthalates have a wide variety of usages: low molecular weight phthalates such as DiBP and DnBP are mainly used in adhesives, waxes, cosmetics, personal care, or cleaning products, whereas higher molecular weight BzBP, DEHP and DiNP are mainly used as plasticizers.32−34 Relatively high concentrations of phthalates in indoor environments compared to other SVOCs have been reported previously.28,30,35 BFRs. Ten BFRs had ≥90% detection frequency (SI Table S3) with RSDs of 0.05) except for BDE-99 which was significantly higher in polyester than cotton (MWU, p < 0.05) (Figure 1b, SI Table S3). BDE-47 had the highest concentration accumulated of the 10 BFRs reported, with an average of 46 ± 11 and 53 ± 18 ng/dm2 in cotton and polyester, respectively, constituting 70% of the ∑10BFR mass measured in both fabrics. BDE-99 contributed 15−20% of

Figure 1. Average concentrations of phthalates (a), BFRs (b), and OPEs (c) accumulated by cotton and polyester, expressed as ng/dm2 planar surface area of fabric. Error bars indicate standard deviation. Note: Y-axis is a log scale for BFRs and OPEs but is linear for phthalates. Asterisks represent a statistically significant difference between cotton and polyester (p < 0.05). Note: TCiPP is referred as TCPP-1.

∑10BFR with an average of 10 ± 2.9 and 15 ± 4.5 ng/dm2 in cotton and polyester, respectively. The abundance of BDE-47 and -99 among BFRs indoors has been widely reported.24,28,36 Among the NFRs, HBB, and EH-TBB were the main contributors with average levels of 0.4−1.0 ng/dm2 fabric. HBB is reported to be mainly used as an additive flame retardant in plastic, wood and textile goods.37,38 EH-TBB, along with BEHTEBP, are the main constituents of Firemaster 550 (FM 550) that is used as a major penta-BDE replacement in foam products.39,40 OPEs. The detection frequency of the eight (including three isomers of TCPP) OPEs analyzed was ≥70% except TnBP which was 50% for polyester (SI Table S3). Total average concentrations of ∑8OPEs were 1200 and 310 ng/dm2 for cotton and polyester, respectively. OPE concentrations (except TCEP) were up to seven times higher in cotton than polyester (MWU, p < 0.03) (Figure 1c, SI Table S3). ∑3TCPP (isomers 1, 2 and 3) dominated the OPEs with average total concentrations of the three isomers of 1050 ± 700 and 208 ± 150 ng/dm2, constituting 87 and 67% of ∑6OPEs in cotton and polyester, respectively. OPE concentrations were highly variable (SI Table S3), with RSDs ranging between 35 to 85%, some of which could be attributed to variable recoveries. Overall, ∑8OPEs in cotton and polyester were 18 and 4 times higher than ∑10BFRs, respectively. OPEs such as TCPP and TDCiPP are used as replacements for phased-out pentaBDE in flexible polyurethane foam products.26,41 TCPP is also added at levels of 2−25% to spray and blown polyurethane foam used as building insulation.42 High levels of OPEs have been reported indoors.23,27,43,44 TCEP, a contaminant in the commonly used commercial mixture Antiblaze V6, is reported to be used in foam in automotive and furniture applications45 and has also been measured in foam collected from baby products with concentrations of V6 as high as 4.6% of foam C

DOI: 10.1021/acs.est.6b02038 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

Figure 2. Percentage distribution of chemicals released to laundry water and remaining sorbed to cotton (top) and polyester (bottom). Percentages are based on concentrations in laundry water (ng/L.dm2) and remaining on fabric (ng/dm2).

cotton (16 ± 5 to 1684 ± 771 ng/L.dm2) and polyester (1.5 ± 1.5−168 ± 73 ng/L.dm2). Exceptions were TPhP and EHDPP for which 50% and 0.5, SI Table S4) with the notable exception of DBDPE. Group 3 fabrics showed high concentrations of DBDPE of 106 ± 45 and 211 ± 44 ng/dm2 in cotton and polyester, respectively, whereas this chemical was not detected in group 1 or 2 fabrics (in part because of high detection limits). DBDPE in group 3 fabrics appears to have originated from the electric dryer since field blanks (clean fabrics not deployed) that were also dried in the dryer for 20 min, had similarly high concentrations of DBDPE. We then compared concentrations of these fabrics with a single sample of dryer lint collected after drying 10 fabric squares of each of cotton and polyester. DBDPE concentrations were 17 000 and 8500 ng/g of lint from cotton and polyester fabrics, respectively, compared with 7 (cotton) and 18 (polyester) ng/g of fabric dried in the dryer (SI Figure S3). Stapleton et al.48 and Schecter et al.49 reported BDE-209 at levels up to 2890 and 2149 ng/g in dryer lint collected from homes, respectively. These results indicated that the dryer was the source of DBDPE, which is a major replacement for decaBDE, with reported use in electrical and electronic equipment, including its components.24,38 In this case, DBDPE was likely released from the dryer’s electrical components such as wiring and possibly plastic parts as well. Chemical Accumulation and Release As a Function of Physical-Chemical Properties. Factors related to the differences between chemicals accumulated from air by cotton and polyester were investigated by plotting the difference, (Ccotton − Cpolyester)/Ccotton, against physical-chemical properties. The physical-chemical properties were obtained from EPI Suite software tools (version 4.11).50

mass.46 Apart from their uses as flame retardants, OPEs such as TnBP, TPhP, and EHDPP are also used in hydraulic fluids, floor wax, adhesives, cosmetics, and/or as plasticizers.25,47 Chemical Accumulation by Fabrics Normalized to BETSSA. Concentrations of higher molecular weight phthalates (BzBP, DEHP, and DiNP) and BFRs that were similar on a per unit planar area of fabric were significantly lower (6−12 times) in cotton than polyester per unit BET-SSA (MWU, p < 0.05) (SI Figure S1). OPE concentrations that were significantly higher in cotton than polyester per unit planar area, were either similar (TnBP, TCPP-1, −2, and −3) or 3−9 times lower in cotton (TCEP, TDCiPP, TPhP, and EHDPP) than polyester (MWU, p < 0.05), when normalized to BET-SSA. These differences in concentration when expressed on a planar versus BET-SSA basis were consistent with the eight times higher BET-SSA of cotton than polyester (1.17 vs 0.14 m2/g, respectively). Chemical Release to Laundry Water (Group 2). The release of chemicals from fabric to laundry water was consistent between cotton and polyester with some notable exceptions discussed below (Figure 2, SI Figure S2; Table S4). As expected, the percentage of accumulated chemical released to laundry water was highest for chemicals with high water solubility. Specifically, the percentage of phthalates released to laundry water was up to 100% for DiBP with concentrations of 454 ± 79 and 59 ± 20 ng/L·dm2 of cotton and polyester, respectively. Eighty percent of DnBP and BzBP in cotton (565 ± 137 and 196 ± 52 ng/L.dm2, respectively) were released to laundry water, whereas 50 to 70% (118 ± 57 and 155 ± 31 ng/L.dm2, respectively) were released from polyester. For DEHP, ∼35% was released in laundry water from cotton and polyester (362 ± 296 and 339 ± 322 ng/L.dm2, respectively), whereas no release of DiNP to laundry water was detected for both fabrics. These losses of phthalates due to laundering were similar to those reported by Gong et al.7 Less than 10% of accumulated BFRs, in general, were released to laundry water (90% of the mass remained on laundered cotton (0.06 ± 0.02−33 ± 15 ng/dm2) and polyester (0.10 ± 0.04−61 ± 15 ng/dm2). OPEs showed very different behavior from BFRs with >80% release of chemicals accumulated by D

DOI: 10.1021/acs.est.6b02038 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology On a planar surface area basis, the difference (Ccotton − Cpolyester)/Ccotton showed a significant inverse relationship with the octanol−water partition coefficient (KOW) (r2 = 0.60, p < 0.001, Figure 3), a weak, inverse relationship with the Henry’s

inverse sigmoidal relationship with HLC (with TnBP, DnBP and DiBP as outliers) (SI Figures S7, S8). Further, a significant (p < 0.05) positive relationship was found between the percentage of accumulated chemical released to laundry water and the polarizability of eight of the 24 chemicals reported here, with the exception of DnBP (Figure 4b). Overall, the greatest release was measured for the most highly polarizable long-chain aliphatic OPEs, TDCiPP, and TCEP (and presumably TnBP as well), was less for aromatic TPhP, and was least for the PBDEs that have minimal polarizability. A relationship was not found between the percentage released to laundry water and gas-particle partitioning of phthalates and BFRs reported for the same location22,30 (SI Figure S9; gas-particle distributions were not available for OPEs). However, this graph clearly distinguished phthalates with higher release rates to laundry water, from BFRs with lower release rates, regardless of their gas-particle partitioning. Finally, we investigated the relationship between the difference in accumulation of cotton versus polyester and the percentage released to laundry water. We found that most aliphatic OPEs that sorbed more to cotton than polyester had >80% release to laundry water (SI Figures S10 and S11). BFRs, which showed no difference between accumulation by cotton versus polyester, had 80% release to laundry water, whereas chemicals (all BFRs) with log KOW > 6 showed 1900 fibers per wash from a single garment during laundering where microfibers are known to contain sorbed chemicals.55,56 We also did not consider the role

Figure 4. Percentage of accumulated chemical released to laundry water from cotton (squares) and polyester (stars) as a function of (a) octanol− water partition coefficient (KOW), and (b) polarizability of eight chemicals (data from Stenzel et al.51). Green, blue and pink ellipses enclose phthalates, BFRs and OPEs, respectively. E

DOI: 10.1021/acs.est.6b02038 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Article

Environmental Science & Technology

efficiently convey polar but not nonpolar SVOCs emitted indoors to outdoor surface waters via laundering.23 We calculated the chemical load transferred from indoor air to laundry water based on the 30 day deployment of fabrics in an office. If we assume that one load of cotton laundry weighs ∼4500 g and cotton has a density of 120 g/m2, one load of cotton laundry would contain about ∼40 m2 of fabric. The concentrations of ∑5phthalates, ∑5PBDEs, and ∑8OPEs in a single load of laundry would be ∼6000, 4, and 11 000 μg/L, respectively (calculated based on the average total laundry water concentrations measured here for phthalates, PBDEs and OPEs of 1580, 1.0, and 2965 ng/L·dm2, respectively). A typical laundry machine that uses about 50 L of water per load59 would thus release 300, 0.2, and 550 mg of phthalates, PBDEs and OPEs, respectively, per laundry load to wastewater. The estimates for PBDEs and OPEs are higher than those of Schreder and La Guardia who measured concentrations of BFRs and OPEs in residential laundry water.23 As such, our calculation could be viewed as a “worst case” scenario given 30 days of continuous uptake in an office which has higher flame retardant concentrations in comparison to homes.22,36,44,60 However, we assumed a fabric density of 120 g/m2 for this estimate whereas in reality, laundry contains clothing with higher density fabrics such as jeans. In wastewater treatment plants (WWTPs), OPEs, particularly chlorinated OPEs, have