Environ. Sci. Technol. 2009 43, 9274–9280
Modeling the Global Fate and Transport of Perfluorooctane Sulfonate (PFOS) and Precursor Compounds in Relation to Temporal Trends in Wildlife Exposure JAMES M. ARMITAGE,† URS SCHENKER,‡ MARTIN SCHERINGER,‡ JONATHAN W. MARTIN,§ MATTHEW MACLEOD,‡ AND I A N T . C O U S I N S * ,† Department of Applied Environmental Science (ITM), Stockholm University, SE-10691 Stockholm, Sweden, Institute for Chemical and Bioengineering, Swiss Federal Institute of Technology, ETH Zürich, CH-8093 Zu ¨ rich, Switzerland, and Department of Laboratory Medicine & Pathology, Division of Analytical & Environmental Toxicology, University of Alberta, T6G 2G3, Edmonton AB, Canada
Received May 15, 2009. Revised manuscript received October 7, 2009. Accepted October 15, 2009.
A global-scale fate and transport model was applied to investigate the historic and future trends in ambient concentrations of perfluorooctane sulfonate (PFOS) and volatile perfluorooctane sulfonyl fluoride (POSF)-based precursor compounds in the environment. First, a global emission inventory for PFOS and its precursor compounds was estimated for the period 1957-2010. We used this inventory as input to a global-scale contaminant fate model and compared modeled concentrations with field data. The main focus of the simulations was to examine how modeled concentrations of PFOS and volatile precursor compounds respond to the major production phase-out that occurred in 2000-2002. Modeled concentrations of PFOS in surface ocean waters are generally within a factor of 5 of field data and are dominated by direct emissions of this substance. In contrast, modeled concentrations of the precursor compounds considered in this study are lower than measured concentrations both before and after the production phaseout. Modeled surface ocean water concentrations of PFOS in source regions decline slowly in response to the production phase-out while concentrations in remote regions continue to increase until 2030. In contrast, modeled concentrations of precursor compounds in both the atmosphere and surface ocean water compartment in all regions respond rapidly to the production phase-out (i.e., decline quickly to much lower levels). With respect to wildlife biomonitoring data, since precursor compounds are bioavailable and degrade to PFOS in vivo, it is at least plausible that declining trends in PFOS body burdens observed in some marine organisms are attributable to this exposure pathway. The continued increases in PFOS body burdens
* Corresponding author email:
[email protected]; phone: +46 (0)8 16 4012. † Stockholm University. ‡ Swiss Federal Institute of Technology, ETH Zurich. § University of Alberta. 9274
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 24, 2009
observed in marine organisms inhabiting other regions may reflect exposure primarily to PFOS itself, present in the environment due to production and use of this compound as well as degradation of precursor compounds.
Introduction Perfluorooctane sulfonate (PFOS) is widely distributed in the global environment and is detectable in biota in both source and remote regions (1-3). The presence of PFOS in the environment results from the manufacturing and use of this compound itself as a surfactant in various industrial applications as well as from the release and subsequent degradation of perfluorooctanesulfonyl fluoride (POSF)based precursor compounds (4-7). In this regard, there are obvious parallels with PFC(A)s (perfluorocarboxylic acids and perfluorocarboxylates), another class of perfluorinated compounds that are also released through emissions related to manufacturing and use (8) as well as degradation of their volatile precursor compounds (7, 9, 10). Recently, mass balance modeling approaches have been applied to investigate the potential contribution of different sources of PFC(A)s to the observed levels in the environment, particularly in the context of transport to the Arctic (11-15). These efforts have been facilitated by estimates of emission inventories (8, 15) and physical-chemical properties (16). PFOS is readily detectable in biological samples around the globe, often at higher wet weight concentrations than PFC(A)s (e.g., see ref 3), but global-scale mass balance models have not yet been applied to this substance. In particular, it remains unclear what fraction of PFOS present in the global environment is due to emissions of PFOS as opposed to the degradation of precursor compounds. One of the intriguing aspects of the distribution of PFOS in the global environment is the reported decline of measured concentrations in ringed seals inhabiting the Canadian Archipelago (17) and sea otters living along the Alaskan coast (18) following the production phase-out of POSF-based chemistry (including PFOS and precursors) by the major manufacturer in 2000-2002 (19). In other locations, however, a similar response to the phase-out is not evident. For example, measured PFOS concentrations in ringed seals (20) and polar bears (21) sampled in Greenland as well as guillemot eggs from the Baltic Sea (22) have not declined since 2000 and have rather been interpreted as evidence of ongoing or increasing exposure. Model simulations investigating how ambient concentrations of PFOS and its precursor compounds evolve following the production phase-out could provide useful insights into this apparent discrepancy. Therefore, the primary objective of this study is to simulate the long-term fate and transport of PFOS and known volatile precursors to compare and contrast historic and forecasted ambient concentrations. We derive historic emission estimates for both PFOS and selected precursor substances which then serve as input to a globalscale contaminant fate model. The plausibility of our emission estimates is assessed through comparisons of modeled concentrations to available monitoring data. The temporal trends observed in the marine environment after the phaseout of PFOS and its precursors are compared to model outputs, and we present hypotheses to explain the seemingly divergent temporal trends in biota in different regions of the globe. We also identify key data gaps and uncertainties that need to be addressed to improve our understanding of the abiotic and biological fate of PFOS and its precursors. 10.1021/es901448p CCC: $40.75
2009 American Chemical Society
Published on Web 11/12/2009
Methods The CliMoChem Model. CliMoChem is a contaminant fate model that divides the global environment into a series of latitudinal bands (23). For these simulations, the globe was divided into 5 zones per hemisphere, each spanning 18° of latitude (e.g., Zone 1 ) 72-90°N, Zone 2 ) 54-72°N, and so on). Each zone is subdivided into compartments representing bulk environmental media such as soil, vegetation, ocean water, and the overlying atmosphere. Additional modifications have recently been made (24) to allow the user to simulate multiple compounds simultaneously (e.g., parent compound and transformation products); this capability was recently exploited to model the potential contribution of two classes of fluorinated compounds known to degrade in the atmosphere under environmentally relevant conditions to yield perfluorooctanoic acid (PFOA) (13). In the current modeling exercise, the same basic model approach and parameterization was employed to simulate PFOS and its precursor compounds. Details of the emission estimates, treatment of PFOS precursor substances, and other key input parameters are presented in the following sections. Emission Estimation Methodology. Perfluorooctanesulfonyl fluoride (POSF) is the starting material for all subsequent production of a range of low (e.g., surfactants including PFOS and it salts) and high molecular weight materials (e.g., polymers) (4-6). For example, perfluorooctane sulfonic acid is manufactured in batch reactions by hydrolysis of POSF; PFOS salts are then produced by neutralizing the acid. POSF can also be reacted with methyl or ethyl amine to produce perfluorooctanesulfonamides (e.g., N-methyl and N-ethyl FOSA, collectively referred to here as xFOSAs). xFOSAs can be subsequently reacted with ethylene carbonate to form either N-methyl or N-ethyl sulfonamidoethanols (collectively xFOSE). The xFOSA and xFOSE intermediates (here collectively termed FOSx) are the principal building blocks of the higher molecular weight and polymeric fluorochemical (FC) products. As a consequence of the inefficiency of these various reactions however, PFOS, xFOSA, and xFOSE are known to be present as unbound residual impurities in final fluorochemical products (6). For the purposes of this study, it was necessary to consider emissions of PFOS and the precursors considered here (i.e., volatile FOSx released due to manufacturing and presence as residuals). Rather than estimate emissions of PFOS at each life stage of major POSF-derived products (i.e., manufacture, supply chain, product use, disposal) as was previously attempted (4, 5, 25), we adopt a simplified approach following the methodology proposed in ref 8 for PFC(A)s. PFOS emissions are divided into two categories: direct and indirect sources. Direct sources of PFOS refer to releases which occur during manufacturing and use of fluorochemical products that contain PFOS intentionally, e.g., aqueous film forming foams (AFFFs) and acid mist suppressants, while indirect sources refer to releases due to the presence of PFOS as a residual impurity (i.e., present unintentionally) and through degradation of precursor compounds (i.e., FOSx). A complete description of the methodology, assumptions and calculations used to derive historic (1957-2002) and contemporary (2003-2010) global emission estimates is provided in the Supporting Information (SI; Section S1). Mode of Entry and Global Distribution of Emissions. POSF-based precursor substances present as unbound residuals were assumed to be emitted 100% to the atmosphere and to be composed exclusively of xFOSE. This simplifying assumption is supported by the measurements of residuals in FC products analyzed in ref 26 and is consistent with the modeling approach in ref 13. All direct emissions of PFOS were assumed to occur to surface water, as were indirect emissions of residual PFOS present
in FC products. Emissions from POSF-based product manufacturing were assigned 50% to Zone 3 (36-54°N) and 50% to Zone 4 (18-36°N) of the model, based on the location and production volumes of the two main manufacturing facilities (Antwerp, Belgium and Decatur, AL). Emissions from finished products (direct, indirect) were distributed 90% to Zone 3 and 10% to Zone 4 based on population density in model zones where the majority of products were assumed to be marketed (e.g., Western Europe, North America, Japan). Atmospheric Fate of Precursor Substances. Schenker et al. (13) proposed a simplified atmospheric degradation scheme starting from fluorotelomer alcohols (FTOH) and xFOSE that includes potential reaction pathways leading to PFCAs and PFOS. The approach only explicitly considers chemical species that are stable for several hours and groups similar substances into “blocks” which are defined by an average set of physical-chemical properties. For the present study, we have retained the xFOSE, xFOSA, and INT blocks from ref 13 (see the SI, Section S2 and Figure S2); for further details see ref 13. Partition Coefficients and Degradation Rate Constants. All property values are summarized in the SI Section S3 and Table S9. We have used the same partition coefficients and degradation rate constants for the precursor blocks as selected by Schenker et al. (13). PFOS was assumed to have an organiccarbon-water partition coefficient (log KOC) of 2.5 (27), a negligible air-water partition coefficient (KAW), and to be stable in the environment (i.e., negligible degradation). We assumed a range (1-4.5%) for the overall atmospheric degradation yield of PFOS from xFOSE; the reasons for this range are discussed in the SI, Section S1. Degradation of precursors in surface compartments was assumed to be negligible (i.e., not competitive with other fate processes). Furthermore, no distinction has been made between branched and linear isomers of any substance in this study (i.e., we have assumed that differences in atmospheric reactivity and abiotic partitioning among isomers do not significantly influence overall environmental fate). This assumption is reasonable given the main purpose of this study, which is to contrast the general behavior of PFOS and FOSx. Model Application. Simulations were conducted from 1960 until the year 2050 based on the geometric mean emission estimates derived for xFOSE and PFOS. Modeled concentrations of PFOS in all environmental media attributable to direct and indirect sources were tracked separately, as were the concentrations of each precursor block. To assess the uncertainty of the model parametrization and the sensitivity of the model to key input parameters, model results based on estimated minimal and maximal parameter values for emissions and the PFOS degradation yield (see the SI, Section S2) were calculated.
Results and Discussion Historic POSF Production Volume. POSF production estimates were submitted to the Stockholm Convention Secretariat by the major manufacturer (3M) for the period 1985-2002 (28). Based on these estimates and similar assumptions as made by Paul et al. (25) for the years prior to 1985, we estimate total global historic POSF production to be approximately 65,160 t for the period 1957-2002. Prevedouros et al. (8) estimated total historic POSF production for the period 1960-2002 and arrived at a value of approximately 83,000 t, while Paul et al. (25) calculated a value of 122,500 t (96,000 t excluding unusable wastes) for the period 1970-2002. The discrepancy between our estimates and these values is mainly related to the higher POSF production by the major manufacturer assumed in refs 8 VOL. 43, NO. 24, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
9275
FIGURE 1. Measured and modeled concentrations of xFOSEs in the atmosphere (pg m-3) of the main source region (Zone 3, 36-54°N). The solid line represents model output assuming geometric mean emissions whereas the dotted line represents model output assuming maximum emissions. Vertical bars represent the range of reported concentrations of xFOSE while the midpoint of the range is indicated by a dash. The red-shaded area represents the period in which xFOSE was phased-out by the main manufacturer. Release lag (RL) refers to the length of time over which a given year’s emissions are distributed (see Supporting Information). and 25 for the period 1985-2002 in comparison to information in ref 28. Contemporary POSF-Based Production. There is limited information regarding continuing POSF-based manufacturing by other companies after the production phase-out by the major manufacturer (3M) between 2000 and 2002. Although some regulatory authorities have banned the use of PFOS, it has been granted exemptions for certain applications until 2011 (29) and production has continued. For example, PFOS was manufactured in Germany (20-60 t) and Italy (
