ARTICLE pubs.acs.org/est
Predicting the Migration Rate of Dialkyl Organotins from PVC Pipe into Water William A. Adams,† Ying Xu,‡ John C. Little,§ Anthony F. Fristachi,|| Glenn E. Rice,*,^ and Christopher A. Impellitteri† †
U.S. EPA National Risk Management Research Laboratory, 26 West Martin Luther King Drive, Cincinnati, Ohio, United States Department of Civil, Architectural and Environmental Engineering, The University of Texas at Austin, Austin, Texas, United States § Department of Civil and Environmental Engineering, Virginia Tech, Blacksburg, Virginia, United States Fristachi Environmental Consulting, Los Alamos, New Mexico, United States ^ U.S. EPA National Center for Environmental Assessment, 26 West Martin Luther King Drive, Cincinnati, Ohio, United States
)
‡
bS Supporting Information ABSTRACT: Organotins (OTs) are additives widely used as thermal and light stabilizers in polyvinyl chloride (PVC) plastics. OTs can leach into water flowing through PVC pipes. This work examines the leaching rates of two potentially neurotoxic OTs, dimethyl tin (DMT) and dibutyl tin (DBT), from PVC pipe. Water was circulated in a closed loop laboratory PVC pipe system. Using a gas chromatograph-pulsed flame photometric detector (GC-PFPD), the change in concentrations of DMT and DBT in the water in the system was monitored over time and allowed to reach equilibrium. OT concentration as a function of time was analyzed using a mechanistic leaching rate model. The diffusion coefficient for OT in the PVC pipe material, the only unknown model parameter, was found to be 9 1018 m2/s. This value falls within with the range of values estimated from the literature (2 1018 to 2 1017 m2/s) thus increasing confidence in the leaching rate model.
’ INTRODUCTION Since the 1940s, organotins (OTs) have been used as additives in the plastics industry. By the year 2000, approximately 70% of the OTs produced in the United states were used as additives in the processing of polyvinyl chloride (PVC) and chlorinated polyvinyl chloride (CPVC).1 OT additives act as temperature and light stabilizers that prevent degradation of the polymer. When the pipe is exposed to high pressure and temperature or light, the polymer can undergo dehydrochlorination. During this process, the structural stability of the polymer can become compromised through the loss of HCl and subsequently become susceptible to chain scission through chemical attack from the HCl, which reduces polymer strength. OTs act as HCl scavengers and substitute ligands for chorine to prevent the degradation. Additionally, OTs can act as an antioxidant and prevent oxidation of the polymer.2 Although PVC and CPVC polymers have many applications, the introduction of OT additives during the manufacture of drinking water pipes and their potential release into potable water raises health concerns, because OTs are potential neurotoxicants. Most OTs used in industry have tin in the Sn(IV) oxidation state and follow the formula RnSnX4n where R = alkyl or aryl group, Sn = tin, and X = anionic species. The di- and monosubstituted forms, RSnX3 and R2SnX2, are typically used as r 2011 American Chemical Society
stabilizers,3 which would be added to PVC. Since these stabilizers are not chemically bound to the bulk material, OTs can migrate to the surface of the pipe and leach into the water. Residual OTs may also be present on the surface of the polymer, especially with new PVC pipe. These residual OTs result in initial contamination of the water flowing through the pipe as the residual OTs on the surface dissolve into the water. The leaching of organotins in both PVC and CPVC pipe has been previously studied.46 Wu et al.6 analyzed the organic tin extracted from water flowing through a closed loop PVC pipe system. Their analytical methods did not differentiate between tin species. Results showed that the tin concentration in the water reached equilibrium within about 12 h. Wu et al. also determined that the initial OT concentration could be reduced by rinsing the pipe several times prior to use, a process which removed the residual OTs initially present on the pipe surface. Forsyth et al.4 examined OT leachates from sections of CPVC pipe. Monobutyl tin and dibutyl tin were both shown to leach into tap water, and the concentration of extracted OTs increased with increasing Received: May 6, 2011 Accepted: July 5, 2011 Revised: July 4, 2011 Published: July 05, 2011 6902
dx.doi.org/10.1021/es201552x | Environ. Sci. Technol. 2011, 45, 6902–6907
Environmental Science & Technology temperature. However, the concentration of extractable OTs varied considerably among different CPVC pipes. Analytically, gas chromatography (GC) is typically preferred over liquid chromatography (LC) due to its resolution, ability to couple to different types of detectors, and its ability to separate many different OT compounds.7 However, a derivatization technique, such as ethylation8 or alkylation,9 is required for GC analysis because OTs are not volatile. Typically following derivatization, a liquidliquid extraction is performed, although solid phase microextraction (SPME) techniques have been used.10 Many types of GC detectors can be used for OT analysis;7 however, gas chromatographymass spectrometry (GC-MS),11 gas chromatographyinductively coupled plasma-mass spectrometry (GC-ICP-MS),12 and gas chromatography-pulsed flame photometric detection (GC-PFPD) 13 have been more frequently used due to their sensitivity and selectivity. Mass transfer of OTs from a PVC pipe into drinking water is predictable. Xu and Little14 developed a model that predicts emissions of semivolatile organic compounds (SVOCs) from polymers into air. While the model was developed to predict emissions of SVOCs such as phthalates, flame retardants, and biocides from polymer materials into air, the same set of mechanisms is expected to govern the leaching of OTs from PVC pipe into water. As will be described later, these mechanisms include diffusion through the PVC pipe material, equilibrium partitioning between the interior pipe surface and the water immediately adjacent to the surface, and finally convective masstransfer through a boundary layer into the bulk water. Previously, we used this leaching rate model coupled with an exposure model to simulate OT exposures in the U.S. population that drinks water transported in PVC pipes.15 Our results suggested that human OT exposures through tap water consumption are likely to be approximately 100-fold lower than the World Health Organization16 “safe” long-term concentration in drinking water (150 μg/L). However, uncertainty analyses showed that our exposure estimates were sensitive to the predicted OT leaching rates. Considering this sensitivity, the widespread residential use of PVC pipe to transport drinking water, and the potential neurotoxicity of OTs, we thought it prudent to acquire additional data and further evaluate our OT leaching model. This paper reports on the migration of the alkyl groups dimethyl tin (DMT) and dibutyl tin (DBT) from PVC pipe into water recirculating within a closed loop system. The DMT and DBT concentrations were measured over time using ethylation derivatization, liquidliquid extraction, and GC-PFPD analysis. The experimental data obtained chromatographically were analyzed using a model capable of predicting DMT and DBT leaching rates. The results lead to a determination of the diffusion coefficient (D), which quantitatively describes the migration of DMT and DBT from PVC pipe into water flowing through the pipe. Finally, we compare the new results to those obtained in our previous estimates of human exposure to OTs in PVC pipes, showing that our earlier estimates15 were reasonable.
’ EXPERIMENTAL METHODS Chemicals. Dimethyl tin dichloride, dibutyl tin dichloride, tetrabutyl tin, HPLC n-hexane (>97%), and sodium tetraethylborate (STEB) were purchased from Sigma-Aldrich (St. Louis, MO). Anhydrous sodium sulfate and sodium acetate trihydrate were purchased from Fisher (Fairlawn, NJ). Water used in the loop system was purified to 18.2 MΩ. A fresh
ARTICLE
Figure 1. PVC pipe loop system showing pump, flow meter, holding tank, and connecting tubes with partial view of PVC pipe.
STEB solution was prepared each day in which extractions would be performed. Between extractions, the STEB solution was stored at 4 C. Stock solutions were prepared for each OT at 100 mg/L. The stock solutions were diluted to 0.05, 0.50, 1.00, and 2.00 μg/L for each calibration curve. Tetrabutyl tin was used as an internal standard (IS) for GC analysis. A 17.5 mg/L tetrabutyl tin standard was prepared and added to samples prior to analysis. Concentrated nitric acid, concentrated sulfuric acid, and N,N-dimethylformamide (N,NDMF), used to determine the concentration of tin in the PVC material, were purchased from Fisher (Fairlawn, NJ). PVC Pipe Loop System. As shown in Figure 1, the PVC pipe loop consisted of 61 m FlowGuard Gold (Charlotte Pipe and Foundry Company, Charlotte, NC) 1 in. inner diameter CPVC pipe with an additional 0.6 m of fittings. The loop capacity was calculated to hold 31.0 L water. The pipe was connected to a stainless steel holding tank, and the pump was set at a flow rate of 7.6 L/minute. During experimental runs, the system was allowed to circulate continuously with samples taken from the holding tank. The rapid recirculation rate meant that the concentration of OTs in the water was essentially the same at any point in time, although it increased over the course of time until reaching equilibrium with the OTs in the pipe. Sampling. Initially, the pipe system was flushed with clean water in order to remove any interferences (i.e., residual OT) that may be present at the surface of the PVC. The holding tank was filled with 42 L of 18.2 MΩ water. The water in the system was allowed to circulate for 30 min, sampled (1 L), and then completely flushed. Six flushes were performed. Flushing was stopped when the percent deviation among replicates of OT responses were within 30%. Following the final flush, all water was removed from the system. Three separate circulations were performed. For each circulation, 50 L 18.2 MΩ water was added to the holding tank. The water was pumped through the system until the pipe was filled. The circulation time ranged from 20 min to in excess of 100 h. Extended runs to 1079 h were conducted in order to ensure equilibrium had been reached. Samples taken within the first hour of circulation were sampled in triplicate and analyzed. As will be shown later, the initial build up of OTs in the water is rapid, and we attempted to monitor this early increase in concentration as accurately as possible. After 1 h, triplicate samples were taken every other sample. During sampling, the 6903
dx.doi.org/10.1021/es201552x |Environ. Sci. Technol. 2011, 45, 6902–6907
Environmental Science & Technology
ARTICLE
volume that was removed was replaced with fresh 18.2 MΩ water (1 or 3 L, depending on the sample). The dilution caused by adding fresh water was accounted for in the final calculations. Prior to beginning the next circulation, the system was drained and fresh water was introduced. Pipe Water Sample Preparation. Samples were prepared by combining a 1 L water sample, 10 mL of 1 M sodium acetate, 4 mL of hexane, and 1 mL of 1% m/v STEB. The mixture was shaken mechanically for 20 min and allowed to rest for 15 min. The hexane layer was carefully removed and passed through a pipet containing sodium sulfate to remove any excess water. The sample extract was brought to a final volume of 5 mL using fresh hexane, and 10 μL of solution containing 17.5 mg/L tetrabutyl tin was added as an IS. Pipe Water Analysis. Samples were analyzed using a Varian CP-3800 gas chromatograph with pulsed flame photometric detector (GC-PFPD, Walnut Hills, CA). A sulfur filter was used with the pulsed flame photometric detector (PFPD). A splitless injection mode was used with an injection volume of 3 μL. The injector temperature was 220 C. The column was a VF-17 ms 30 m 0.25 mm i.d. 0.25 μm film thickness (Lake Forest, CA). The temperature program was 40 C for 3 min and then 8 C/minute to 250 C and held for 5 min with a flow rate of 2.0 mL/minute. The PFPD used a photomultiplier voltage of 550 V with a gate delay of 4.5 ms and a gate width of 5 ms. The trigger level was 200 mV. The gases were set as Air 1 (17.0 mL/minute), Air 2 (10.0 mL/minute), and H2 (13.0 mL/minute). A calibration curve that ranged between 0.05 and 2.00 μg/L was prepared for both DMT and DBT. All standards were extracted and concentrated using the same derivatization and liquidliquid extractions as the samples. All results were fit to the calibration curve and normalized to the IS during data processing. PVC Pipe Total Tin Analysis. Sections of PVC pipe were analyzed using a PerkinElmer Optima 2100 DV inductively coupled plasma-optical emission spectrometer (Shelton, CT), or ICP-OES, for the presence of tin. The FlowGuard Gold pipe was cut into 150.0200.0 mg pieces and dissolved in 20 mL of N, N-DMF under sonication. Three different pipe locations were used and analyzed in triplicate. The organic solvent was evaporated, and 10 mL of concentrated sulfuric acid and 2 mL of concentrated nitric acid were added. The sample was heated above 300 C on a hot plate and reduced to a volume of 4 mL. This process was repeated two more times in order to remove all carbon from the residual polymer such that the solution was clear. The samples were then diluted to 50 mL in 20% nitric acid and analyzed for total tin on the ICP-OES. Leaching Rate Model. The governing equation describing transient diffusion through the pipe material is ∂Cðx, tÞ ∂2 Cðx, tÞ ¼D ∂t ∂x2
ð1Þ
Figure 2. Schematic representation of mechanisms controlling migration of OT from PVC pipe into water. Key parameters include the concentration of OT in the PVC pipe material, C0, the diffusion coefficient for OT in the PVC pipe material, D, the equilibrium partition coefficient between the pipe surface and the water immediately adjacent to the surface, K, and the convective mass-transfer coefficient controlling transfer through the water boundary layer into the bulk water, h.
wall boundary condition is assumed to be ∂Cðx, tÞ ¼ 0 for t > 0, x ¼ 0 ∂x
indicating that the concentration of OT at the external pipe surface is not depleted. The second boundary condition imposed at the interior pipe surface is ∂Cðx, tÞ ¼ hðy0 ðtÞ yðtÞÞ for t > 0, x ¼ L D ∂x
Cðx, tÞ ¼ Ky0 ðtÞ for t > 0, x ¼ L
where L (m) is the thickness of the pipe wall, and C0 (mg/m ) is the initial material-phase OT concentration. The outside pipe
ð5Þ
where K (dimensionless) is the material/water partition coefficient. K is also assumed to be a constant. Similar to previous studies of gas-phase emissions of volatile organic compounds19 and SVOCs from vinyl flooring,14 the following analytical solution is obtained, based on eqs 15 Cðx, tÞ ¼ KyðtÞ þ
∞
∑ m¼1
sinðβm LÞ 2ðβ2m þ H 2 Þ cosðβm xÞ 3 βm Lðβ2m þ H 2 Þ þ H 3
3 ½ðC0 Kyð0ÞÞe
ð2Þ 3
ð4Þ
where h is the convective mass transfer coefficient (m/s), y0(t) is the dissolved concentration of the OTs in the water immediately adjacent to the inner pipe surface (mg/m3), and y(t) is the OT concentration in the bulk water (mg/m3). Equilibrium is assumed to exist between the OTs in the interior surface layer of the pipe and the water immediately adjacent to the surface, or
where C(x, t) is the material-phase concentration of OT (mg/m3), x is distance (m), and t is time (hours).17,18 The material-phase diffusion coefficient D (m2/s) is assumed to be a constant. The initial condition assumes that the OT is uniformly distributed in the pipe material, or Cðx, tÞ ¼ C0 for 0 e x e L
ð3Þ
Dβ2m t
Z
t
þ 0
eDβm ðt τÞ 3 KdyðτÞ 2
ð6Þ where H = ((h)/(KD)) and βm(m = 1,2,...) are the positive roots of βm tan(βmL) = H. Therefore, the migration rate per unit 6904
dx.doi.org/10.1021/es201552x |Environ. Sci. Technol. 2011, 45, 6902–6907
Environmental Science & Technology
ARTICLE
Figure 3. Observed DMT and DBT concentration versus time data.
surface area at time t is ∂Cðx, tÞ mðtÞ ¼ D 3 ∂x
x¼L
¼ D3
∞
Dβm t þ 3 ½ðC0 Kyð0ÞÞe 2
2ðβ2m þ H 2 Þ 2 m þ H Þ þ H
∑ sin2ðβm LÞ 3 Lðβ2 m¼1 Z
0
t
eDβm ðt τÞ 3 KdyðτÞ 2
ð7Þ Assuming that the bulk water within the pipe is rapidly recirculated, and therefore well-mixed, the accumulation of OTs in the water obeys the following mass balance (see Figure 2) dyðtÞ V ¼ A 3 mðtÞ dt 3
ð8Þ
where V is the volume of water within the pipe (m3), and A is the inner surface area of the pipe. The migration model that predicts the transfer of OTs from the PVC pipe into the water is obtained by combining eqs 7 and 8.
Figure 4. Comparison of “lumped” migration model with observed data. The material-phase diffusion coefficient, D, was the only parameter adjusted to obtain a good fit.
’ RESULTS AND DISCUSSION The data collected during the three experimental runs are shown in Figure 3 (Table S1 in the Supporting Information). The patterns of both DMT and DBT release from the PVC pipe are similar, with an initial rapid increase in concentration followed by a leveling-off as the OT concentration in the water approaches equilibrium with that in the pipe material. As will be shown later, C0, the initial material-phase concentration of OTs within the pipe material is fairly high and the migration rate is relatively low, and thus C0 remained essentially constant from one experiment to the next. Indeed, the concentration in the water approached the same level at the end of each run, or y0, the aqueous-phase concentration that is in equilibrium with C0. The ratio of these two concentrations can be used to calculate, K, the material/water partition coefficient. Because C0 is a critical parameter, we measured the total OT concentration in the pipe, as described earlier. The OTs in the PVC pipe are not chemically bound to the polymer and can be separated by dissolving the pipe material using a strong organic solvent. Once the OTs were free of the polymer matrix, they were treated with acids and analyzed for total tin. It should be noted that the total tin analysis cannot speciate all types of OTs that may be present in the pipe. This method allowed for a reasonable approximation of the target OT concentration in the PVC. The mean total tin concentration over three pipe locations was found to be 0.238 ( 0.016 wt % in the FlowGuard Gold pipe (Table S2
in the Supporting Information). This value was used together with the polymer density to estimate C0. The four model parameters that are needed to predict the migration rate of OTs out of the pipe and into the water are C0, D, K, and h. Given that flow within a pipe is well understood, an estimate of h was obtained using correlation equations based on flow rate as described in Fristachi et al.15 Preliminary calculations using the leaching rate model and a rough estimate of D showed that the mechanism controlling the migration rate is internal diffusion and not external convective mass transfer through the boundary layer. This means that the three key parameters that are required are C0, D, and K and that an accurate value of h is not important. In addition, if C0 is known, then K can be calculated19 from the value of y0 obtained directly from the experiments. The value of D was therefore the only unknown model parameter. Although the concentration of the individual OT (DMT and DBT) compounds in the pipe material could not be determined, it was possible to measure the total tin content in the PVC pipe. This “lumped” OT value was 0.238%, in good agreement with the nominal range of ∼0.1 to 0.5% obtained from the manufacturer. The measured value was used as an estimate of C0 and should be a reasonable approximation because DMT and DBT are the major stabilizers used in PVC pipes. Before applying the masstransfer model to the experimental data, we simply added the concentrations of DBT and DMT together and treated them as a single combined OT compound. Figure 4 shows the combined data, including the “lumped” value for y0, the equilibrium 6905
dx.doi.org/10.1021/es201552x |Environ. Sci. Technol. 2011, 45, 6902–6907
Environmental Science & Technology
ARTICLE
y0 (mg/m )
0.49
measured
’ ACKNOWLEDGMENT These materials have been reviewed and cleared for publication by the US Environmental Protection Agency. The views expressed in this manuscript do not necessarily reflect the views and policies of the U.S. Environmental Protection Agency. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. The authors would like to thank Vasudevan Namboodiri and Deborah Roose for their expertise in the total tin analysis of the PVC pipe using ICP-OES.
K () D (m2/s)
8.4 106 9 1018
calculated fitted
’ REFERENCES
Table 1. Model Parameter Summary parameter
value
comment
Q (m3/h)
0.45
measured
L (m)
0.003
measured
A (m2)
4.86
calculated
V (m3)
0.05
measured
h (m/s)
1.08 105
estimated
4.2 106
measured
C0 (mg/m3) 3
concentration, which was used together with the value of C0 to estimate K for combined OT. Table 1 provides a summary of the model parameters. As shown in Figure 4, and using eqs 7 and 8, a least-squares method was employed to determine the best-fit value of D. The value of D obtained is 9 1018 m2/s. The diffusion coefficient of organotin stabilizers in rigid PVC at 100 C is in the range of 3 1015 m2/s to 5 1014 m2/s,20 which corresponds to a range of 2 1018 m2/s to 2 1017 m2/s at 20 C based on corrections for temperature made using Piringer’s model.21 In the development of the migration model, the values of both D and K were assumed to be constant and independent of concentration. This is generally considered to be a reasonable assumption if the material-phase concentration is below 1 wt %. However, even if these parameters do depend on concentration, the fact that the concentration in the pipe material hardly changes over time means that the values of D and K also do not change over time. Therefore, the assumption that they are both constant is valid. The distribution of OT concentrations reported in this study was compared deterministically to our previous results.15 The Levene test for equal variance indicated that the variances were significantly different (t = 18.15, df = 3,166, p-value
