Environ. Sci. Technol. 2002, 36, 4761-4769 Predicted Distribution and Ecological Risk Assessment of a “Segregated” Hydrofluoroether in the Japanese Environment JOHN L. NEWSTED,† JUNKO NAKANISHI,‡ IAN COUSINS,§ KURT WERNER,| AND J O H N P . G I E S Y * ,⊥ ENTRIX, Inc., 4295 Okemos Road, Suite 101, Okemos, Michigan 48864; Graduate School of Environment and Information Services, Yokohama National University; 3M Center, DABT, Building 236-1B-10, St. Paul, Minnesota 55144; Trent University, Peterborough, Ontario K9J7B8; and National Food Safety and Toxicology Center, Institute of Environmental Toxicology, Department of Zoology, Michigan State University, Lansing Michigan 48824 An assessment of HFE-7500, a “segregated” hydrofluoroether, was conducted to evaluate the potential for exposure to and subsequent effects on humans and wildlife in Japan. The segregated hydrofluoroethers belong to a class of fluorochemicals currently being proposed as replacements for traditional fluorochemicals (CFCs and PFCs) that are currently being used in several industries, in particular, the semiconductor industry. These traditional compounds have been implicated as ozone-depleting or potent “greenhouse gases”. The segregated hydrofluoroethers have useful physical and chemical properties, but do not contribute to ozone depletion and have lower “global warming potential” (GWP) indices. Although the physical properties of these materials (low H2O solubility and high vapor pressure) suggest there would be a very low level of risk to aquatic systems, a thorough analysis had not been previously performed. Predicted environmental concentrations (PECs) of HFE-7500 in Japan were determined with the Higashino model, a Gausian puff and plume model that used an approximation of environmental releases to the atmosphere as input to the model. Allowable concentrations to protect aquatic life, wildlife, and humans from noncancer effects were determined as detailed in USEPA’s Final Water Quality Guidance for the Great Lakes Systems. Potential risk to ecological receptors and humans was determined by calculating hazard quotients and margins of safety. The results of the risk assessment indicate that HFE-7500 poses no significant risk to either aquatic or terrestrial wildlife species or humans living in the Japanese environment. The least margin of safety for any ecological receptor was 100 000, and a margin of safety greater than 100 000 000 for most receptors indicated that HFE-7500 poses no threat to human health. Because of a * Corresponding author phone: (517) 353-2000; fax: (517) 4321984; e-mail: JGIESY@AOL.COM. † ENTRIX, Inc. ‡ Yokohama National University. § Trent University. | 3M DABT. ⊥ Michigan State University. 10.1021/es0257321 CCC: $22.00 Published on Web 10/19/2002 2002 American Chemical Society scarcity of toxicity and exposure data, the risk assessment was based on very conservative assumptions. Therefore, the actual margins of safety for both humans and wildlife could have been 100- to 1000-fold greater if additional data were available such that less stringent uncertainty factors could be applied. These results suggest that the environmental impact of HFE-7500 should be inconsequential based on the marked improvement in its atmospheric properties relative to the traditional compounds currently in use. Given the short atmospheric lifetime and low global warming potential of this material, its replacement of CFCs and PFCs would result in a net improvement of environmental health and safety. Introduction Currently, perfluorocarbons (PFCs) are used in the semiconductor manufacturing industry as heat-transfer liquids to maintain process temperature. The industrial processes that use these liquids include ion implanters, dry etchers, deposition tools, steppers, automated test equipment (ATE), and other machines. In addition, PFCs are used to cool sensitive electronics, fuel cells, and lasers, etc. (1). However, these applications can result in losses of PFCs to the atmosphere, and as a result, PFCs have come under increased regulatory scrutiny due to their long atmospheric lifetimes and to their relatively great global warming potential (GWP). Therefore, under the U.S. Environmental Protection Agency’s (USEPA’s) significant new alternatives policy (SNAP), PFC liquids are allowed as heat transfer media only in those applications for which no alternatives are currently feasible. As gaseous emissions from plasma-aided processes are reduced, the percentage of the net PFC emissions attributable to PFC heat transfer will increase. National, industry, and individual company efforts to reduce emissions of compounds with significant GWP (greenhouse gases) have prompted development of replacements such as segregated hydrofluoroethers (HFEs). These compounds have shorter atmospheric lifetimes and less GWPs but share many of the desirable properties of PFCs needed for industrial application as heat-transfer liquids. Segregated hydrofluoroethers are compounds that contain hydrogen and fluorine atoms on a backbone consisting only of carbon and oxygen atoms. They are termed “segregated” because these chemicals possess a perfluorocarbon segment that is separated or “segregated” from the fully hydrogenated portion of the molecule by an ether linkage (Figure 1). Reliable, quantitative structure-activity relationships (QSARs) capable of predicting environmentally relevant ecotoxicological data or physical chemical properties of HFEs, based on their structure, do not yet exist. One HFE compound, currently registered for use in the United States, Europe, Korea, and most other nations is 3-ethoxyperfluoro (2methylhexane), which is marketed by 3M as Novec Engineered Fluid HFE-7500. Physical and chemical properties of HFE-7500 (Table 1) were measured in support of its registration for commercial applications in several countries. The atmospheric lifetimes of HFEs are approximately 100 times less than PFCs and, therefore, HFEs have GWPs that are much less than those of PFCs. For instance, the GWP of HFE-7500 is 210 (100-yr time horizon) vs 5000 to 11 000 for PFCs (19). Additionally, HFEs are inherently nonozone depleting (ODP) because neither the HFEs nor their degradation products catalyze ozone depletion (19). Currently, HFE-7500 is under consideration as a replacement for PFCs in several applicaVOL. 36, NO. 22, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4761 TABLE 2. Parameters Used in the Higashino Model for HFE-7500 parameter dissipation rate to upper mixing (1/sec) photochemical degradation ratea (1/sec) dry deposition soil (cm/sec) water (cm/sec) Henry’s law constant (unitless)b wet deposition ratec FIGURE 1. Structure of HFE-7500. TABLE 1. Physical and Chemical Properties of HFE-7500a chemical name structure CAS number molecular weight (g/mol) boiling point (°C) pour point (°C) vapor pressure (Pa) @ 20 °C water solubility (µg/L) @ 22 °C Henry’s law constant (Pa m3/mol)b Log Kow Log Koc BCF (laboratory value) HFE-7500 C7F15OC2H5 297730-93-9 414 128 -100 1000 13.3 3.1 × 107 4.9 4.9 8544 a HFE-7500 samples used for analyses were from 3M and were > 99% pure. b Calculated by dividing vapor pressure (Pa) by water solubility (mol/m3). tions in Japan. However, the environmental fate and potential risks of inadvertent release of HFE-7500 to the Japanese environment have not been evaluated. The objectives of this study were the following: (1) to determine whether there are sufficient data to conduct an environmental risk assessment for HFE-7500; (2) conduct an environmental risk assessment using USEPA methodology; and (3) to identify areas of uncertainty. Materials and Methods Approach. The methodology used to evaluate the toxicity and environmental fate of HFE-7500 was based on guidance outlined in the U.S. Environmental Protection Agency (EPA) document entitled “ Final Water Quality Guidance for the Great Lakes Systems” (2). This methodology provides the framework for conducting both human and environmental risk assessments of organic and inorganic chemicals. Specifically, the Great Lakes Water Quality Initiative (GLI) provides specific procedures and methodologies that use environmental fate and toxicity data to derive water quality values that are protective of aquatic organisms, upper trophic level wildlife, and humans. This approach was selected because it is one of the most comprehensive assessment strategies currently being used by regulatory agencies. Potential environmental concentrations (PECs) of HFE-7500 in Japanese surface freshwaters were determined on the basis of estimated environmental releases from industrial applications and a Gausian puff and plume model that has been validated in Japan (3). Toxicant reference values (TRVs) for wildlife (TRVw) and human health (TRVH) values were determined for each receptor from mammalian toxicity data by use of appropriate safety factors to account for data extrapolations. Species-specific bioaccumulation factors, water and diet consumption rates, and dietary compositions were based on parameters taken from the GLI guidance. This approach assumes that the primary route of exposure of wildlife and humans is from water through species-specific food chains. Because site-specific data were not available, the default values given in the GLI guidance were not adjusted for regional differences that may exist between the North 4762 9 value ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 22, 2002 reference 10-6 16 1.28 × 10-8 19 0.2 0.2 1.26 × 104 7.93 × 10-5 17 17 5.00 × a Converted from the reported atmospheric half-life (1.5 × 104 hours) The H value from Table 1 converted as H′ ) H/RT at 20 °C. c Calculated by dividing 1 by Henry’s law constant or H′ (unitless). b American Great Lakes and Japan. Potential risks posed by HFE-7500 to selected wildlife and humans were determined by calculating hazard quotients (HQ) and margins of safety (MOS) (Equations 1 and 2). HQ ) PEC PNEC (1) 1 HQ (2) MOS ) The HQ is the quotient of the predicted environmental concentration (PEC) divided by the TRV expressed as the predicted no-effect concentration (PNEC). Finally, an uncertainty analysis was conducted to evaluate data gaps and other questions that are pertinent to the risk assessment. Higashino Model. The Higashino model was developed to simulate long-term concentrations of chemicals in the atmosphere of Japan (3). The model is formulated for the entire Kanto plain, which includes metropolitan Tokyo and six surrounding prefectures. This area has more than 30 million inhabitants and is the most industrialized area in Japan. The analytical domain, with a scale of 276 × 224 km, has 40 × 60 grids, each equal to a 5-km square. This model was developed with the spatial resolution of 5 km2 and with the time resolution of one month. The atmospheric concentration of HFE-7500 was calculated in 4-h increments and averaged over one month. Data for wind velocity and direction, and precipitation from an automated meteorological data acquisition system (AMeDAS) collected in 1997 were used as meteorological input data in the simulation. The input parameters used in the Higashino model simulation are compiled in Table 2. The primary strength of the Higashino model in this analysis is that it has been calibrated to reflect the use patterns of other chemical products for which HFE-7500 has been proposed as a replacement compound. Specifically, this model has been validated for fate and distribution of trichloroethylene (TCE) and tetrachloroethylene (PCE). Thus, results from the Higashino model would reflect a “realistic” case study of the fate and distribution of HFE-7500 in the Kanto plain based on historic uses of chemicals that have similar use patterns and environmental behaviors. A copy of the model is available from the authors. Sources and Emissions. On the basis of its chemical properties, HFE-7500 is suited as a low- and mediumtemperature fluid for semiconductor applications such as etching, plasma vapor deposition (PVD), and ion implant and testing. Using the assumption that the sole uses of HFE7500 would be in the above-mentioned applications, a preliminary assessment suggested that release of HFE-7500 to the environment would occur primarily through volatilization during storage and use. Furthermore, since manu- FIGURE 2. Concentration map of HFE-7500 in the Kanto region as predicted by the Higashino Model. Each color represents a different atmospheric concentration of HFE-7500 in units of ng/m3. facturing and packaging of HFE-7500 would occur outside of Japan (4), these activities would not be a source of HFE7500 in Japan. Finally, releases of HFE-7500 “down the drain” to water treatment systems or directly to aquatic or terrestrial environments are not expected under normal use patterns, therefore, this route was not included in estimates of environmental releases. On the basis of the above assumptions and scenarios, and for purposes of this assessment, the estimated total releases of HFE-7500 to the Japanese environment would be 100 000 kg/yr in Japan (4). Based on the estimated total annual release of HFE-7500 to the Japanese environment, an allotment of the total emission to each prefecture was made by using the proportion of TCE as a surrogate for HFE-7500 emissions. These values reflect the historic emissions of TCE to the atmosphere in this region as reported in the Ministry of Economics, Trade and Industry (METI) extensive national emissions inventory of TCE (data not available to the public). Assuming that the pattern of emissions of HFE-7500 would mimic that of TCE, the Kanto plain accounted for 42% of total atmospheric emissions of HFE-7500 expected in Japan. Emissions of HFE7500 in the Kanto plain were allotted to each 5 km square grid in proportion to the monetary value of production in the electronic and electric industry for that grid. Results Higashino Model Simulations. The maximum annually averaged concentration of HFE-7500 in air for any grid in the TABLE 3. Maximum, Mean, and Minimum Atmospheric Levels of HFE-7500 for All Grids in the Kanto Plain, Japan (ng/m3) month maximum mean minimum January February March April May June July August September October November December 1.63 × 1.37 × 101 1.43 × 101 1.59 × 101 1.68 × 101 1.73 × 101 1.74 × 101 1.87 × 101 1.78 × 101 1.77 × 101 1.60 × 101 1.68 × 101 1.74 × 1.58 × 100 1.57 × 100 1.62 × 100 1.54 × 100 1.58 × 100 1.57 × 100 1.62 × 100 1.58 × 100 1.74 × 100 1.61 × 100 1.72 × 100 1.90 × 10-4 1.48 × 10-4 2.69 × 10-3 2.36 × 10-3 6.83 × 10-3 1.35 × 10-2 5.85 × 10-3 1.86 × 10-3 1.85 × 10-3 4.07 × 10-4 5.76 × 10-4 1.70 × 10-4 101 100 conceptualization of the Kanto plain was 16.4 ng/m3 (Figure 2). The estimated maximum, mean, and minimum monthly averaged concentrations for all grids in the model and the results indicated that the concentration of HFE-7500 varied slightly between summer and winter months (Table 3). The Higashino model was validated by comparing predicted atmospheric concentrations of TCE with measured atmospheric concentrations of the compounds in the Kanto plain (3). The predicted concentrations of TCE from the Higashino model (Figure 3) emphasize that the model is capable of estimating the long-term average distribution of chemicals VOL. 36, NO. 22, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4763 FIGURE 3. Predicted atmospheric trichoroethylene (TCE) concentrations by the Higashino model in the Kanto plain located in Japan. Each color represents atmospheric TCE emission rate in units of µg/m3. FIGURE 4. Comparison of observed and predicted values of atmospheric trichoroethylene concentrations in summer (subscript “s”) and in winter (subscript “w”) for various locations in the Kanto region. Each capital letter indicates an individual station measured or estimated in either summer or winter. over a wide flat area. The validation step of the model is important in that HFE-7500 is not currently not being used in Japan, and consequently, no measurable concentrations would be expected to be found in the Japanese environment at this time. Predicted atmospheric concentrations of TCE (Figure 4) were in agreement with atmospheric TCE concentrations measured in the Kanto plain (3). Thus, the Higashino model predicted the concentration of a highly 4764 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 22, 2002 volatile compound in the Japanese atmosphere with a reasonable degree of accuracy. Based on the model, the predicted atmospheric concentrations of HFE-7500 for the Kanto plain would be less than 5 ng/m3 (Figure 3). Furthermore, less than 1.7% of the entire Kanto plain would be expected to have atmospheric concentrations greater than 10 ng/m3. Surface water concentrations of HFE-7500 were estimated with a surface-water-to-air ratio derived from the results of simulations with the ChemCan model (5). Using an average air concentration of 16.4 ng/m3 and a water-to-air ratio of 3.13 × 10-4, the average surface water concentration of HFE7500 was estimated to be 5.13 × 10-6 ng/L. However, environmental models are subject to uncertainty that affects their ability to predict accurately the concentration of chemicals in environmental media. This is due, in part, to the simplification of complex processes in the model that can alter the spatial and temporal distribution of the chemical in an environment. Therefore, to account for this uncertainty and variability in the Higashino model predictions, uncertainty factors were applied to the results to ensure that actual environmental concentrations are not underestimated. Depending on how well a chemical’s properties and its environmental behavior are understood, safety factors can range from 10 to 1000. For this assessment, an uncertainty factor of 100 was applied to the predicted surface water concentration that resulted in a final water concentration of 5.13 × 10-4 ng/L. Mammalian Toxicity Evaluation. Results from acute toxicity tests conducted with HFE-7500 indicate that this compound exhibits relatively low toxicity. Rat LD50 values for dermal, inhalation, and oral (5 d) exposure were >2000 mg/kg, 10 000 mg/L, and >2 000 mg/kg, respectively. HFE- 7500 was negative in the Ames mutagenicity test and exhibited no clastogenic effects in the Chinese hamster chromosomal aberration bioassay. On the basis of these results, HFE-7500 has a toxicity profile that would not elicit a concern for carcinogenicity and is an unlikely candidate for investigation in a 2-y cancer assay. Therefore, noncancer endpoints were used to calculate TRVs. TRVs were based on a 28-d repeateddose study with rats dosed daily via gavage with 0, 40, 200, or 1000 mg/kg HFE-7500 (6). No changes in survival, clinical pathology, blood chemistry, body weight, food consumption, or behavior were observed in individuals at any dose group during the exposure or recovery phase of the study. Histological examination revealed that rats of both sexes had centrilobullar acidophilic changes in liver hepatocytes when dosed with HFE-7500 at 1000 mg/kg, but that these changes were not observed at the end of the recovery period. Therefore, the no-observed-effect level (NOEL) was estimated to be 200 mg/kg, and the lowest-observed-effect level (LOEL) was estimated as 1000 mg/kg. The selection of NOEL or LOEL values are a function of the experimental design and may not reflect the specific point in the dose response relationship where organisms are protected from adverse effects. The NOEL is likely to underestimate the maximum dose that could cause impairment, while the LOEL is likely to overestimate the minimum dose that will cause adverse effects in susceptible species. The minimum dose where adverse effects to species are likely to occur is therefore between the NOEL and the LOEL. For this analysis, both the NOEL and LOEL were used as toxic doses (TD) to provide a range of the potential risk of HFE-7500 to wildlife and humans in the Japanese environment. Ideally, a TD can be estimated from dietary toxicity studies where bioavailability and absorption of the chemical from food across the gastrointestinal tract is accounted for in the study design. Uptake of most chemicals is greater in oral gavage studies than that observed in dietary studies because of the effect of food proteins and fats that alter the absorption of chemicals from the GI tract (7). For this risk assessment, absorption of HFE-7500 from a dietary source was assumed to be 100 percent. Based on this conservative assumption, the NOEL and LOEL from the rat 28-d oral gavage study were used, instead of no-observedadverse-effect level (NOAEL) and lowest-observed-adverseeffect level (LOAEL) values, in the calculation of human and wildlife TRVs. Aquatic Toxicity Evaluation. Two studies with HFE-7500 have been conducted with fish (8). In a 48-h acute toxicity study with Japanese Medaka (Oryzias latipes), no mortality was observed for fish exposed to 25 mg/L or 50 mg/L. The lethal concentration to kill 50% of the population (LC50) was determined to be >50 mg/L. In a bioconcentration study conducted with carp (Cyprinus carpio), no mortality or adverse effects were observed in fish exposed to 0.5 mg/L for 6 wk. Based on the above results and the data quality criteria outlined in the GLI for deriving water quality criteria, an LC50 of 50 mg/L was assumed to be the threshold value. It is noteworthy that this value is approximately 3760 times that of the water solubility limit for this material. Aquatic Life Criteria. The GLI guidance provides a twotiered procedure to derive chemical-based water quality (WQC) for the protection of aquatic life. These WQCs are intended to protect aquatic organisms from continuous exposure to chemicals. These methods are very similar to those used in current guidelines for developing National WQC (section 304(a) of the U.S. Clean Water Act). Tier I values are derived using methodologies where there are sufficient data from acute and chronic studies with aquatic organisms which can be adopted as numeric criteria into a water quality standard. The Tier II methodology is used when there are fewer toxicity data than that needed to compute a Tier I value. The Tier II methodology generally produces more stringent values than the Tier I methodology to reflect the greater uncertainty in the absence of additional toxicity data. As more data become available, the derived Tier II values tend to become less conservative. That is, they more closely resemble Tier I numeric criteria. As there is a lack of acute and chronic toxicity data for aquatic species, a Tier II secondary acute value (SAV) was calculated. Because the SAV incorporates less information, safety factors are applied such that a conservative safe concentration can be determined. To calculate a SAV, the lowest acute toxicity result (LC50 or EC50) was divided by the secondary acute factor (SAF). The SAF is an adjustment factor specified in the water quality guidelines (2) that depends on the number of data requirements that have been satisfied for a Tier I criterion. For HFE-7500, the SAF was set at 21.9 because there was only one acute toxicity value available (Japanese Medaka LC50 ) 50 000 µg/L). A Tier II SAV of 2283 µg/L was calculated by dividing the LC50 by the SAF. Because no chronic aquatic studies have been conducted with HFE7500, a secondary chronic value (SCV) was calculated instead of a final chronic value (FCV). The GLI recommends an acuteto-chronic ratio (ACR) of 18. A SCV of 127 µg/L was calculated by dividing the SAV by the ACR. A Tier II water quality analysis results in WQC. The first is a secondary maximum concentration (SMC), which is equal to one-half the SAV, and the second is a secondary continuous concentration (SCC) which is equal to the SCV. For HFE-7500, the SMC and SCC are 1142 and 127 µg/L, respectively. Thus, aquatic organisms would not be unacceptably affected if the four-day average concentration of HFE-7500 does not exceed 1142 µg/L (approximately 86-times H2O solubility) more than once every three years on average. Nor would they be unacceptably affected if the 1-h average concentration does not exceed 127 µg/L more than once every three years on the average. Terrestrial Life Criteria. Japan has a flora and fauna that shares similar types of species with those found in the North American Great Lakes. Therefore, it was considered appropriate to conduct the exposure analysis with species used in the Great Lakes Water Quality Initiative that have been used to establish water quality criteria for persistent, lipophilic chemicals. Species included in the environmental risk analysis were mink, river otter, kingfisher, herring gull, and the bald eagle. These species represent organisms that inhabit upper trophic levels of aquatic food chains and are among the animals most sensitive to many classes of persistent, organic compounds. Furthermore, they can serve as surrogates for several Japanese threatened or endangered species including the Japanese river otter, Japanese ermine, the redcrown crane, the red-faced cormorant, and the Hodgson Hawk Eagle (18). HFE-7500 bioaccumulation factors (BAFs) for trophic level 3 and 4 organisms were derived using procedures outlined in the GLI guidance. The baseline BAF was calculated using a laboratory-measured bioconcentration factor (BCF) for carp. This study met the basic requirements outlined in the GLI guidance document and provided an experimental value that was more accurate than one derived from a structure activity model. Baseline BAFs for trophic level 3 (TL3) and trophic level 4 (TL4) fish were calculated (eq 3) as follows: Baseline BAF ) (FCM) [ ][ ] Measured BCFtT 1 ffd f1 (3) where BCFtT is the bioconcentration factor (BCF ) 8544) (8); fl is the fraction of the tissue that is lipid (fraction ) 0.041); ffd is the fraction of the total chemical in the water that is freely dissolved; and FCM is the food chain multiplier for trophic levels 3 and 4. The lipid fraction of fish was taken from experimental results from the carp bioconcentration study. The freely VOL. 36, NO. 22, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4765 TABLE 4. HFE-7500 Specific Bioaccumulation Factors (BAF) Derived for Human and Wildlife Species receptor trophic level standardized fraction lipid valuea food chain multiplierb baseline BAF trophic level BAF wildlife TL3 TL4 TL3 TL4 0.0648 0.1031 0.0182 0.0310 2.780 2.193 NA NA 613 612 483 802 613 612 483 802 37 421 47 112 10 543 14 166 human a Standard lipid values taken from GLI for human and wildlife. b FCM values were taken from Table B-1 of the GLI, based on log Kow of 4.9. TABLE 5. Assignment of Uncertainty Factors (UF) for the Calculation of Wildlife Values (WV) for HFE-7500 uncertainty factor *intertaxon extrapolation *exposure duration *toxicological endpoint *modifying factors threatened species relevance of endpoint lab to field extrapolation co-contaminants endpoint unclear study species sensitivity organ ratios intraspecies variability *overall UF notes The laboratory study that provided the threshold dose used rat as the test organism. All other species used in this report belong to the same class, but different orders, so Amammal ) 5. Because avian species belong to the same phylum, but different class, Aavian ) 10. The rat study used in this report was a 28-d oral gavage subacute study, so B ) 5. The 28-d rat study determined a NOAEL based on histological changes in the liver. The effects were reversible and classified as mild, so C ) 2. Japanese river otters are listed, so d1 ) 1.5. Liver histological changes in liver was the endpoint, so d2 ) 1.5. The oral gavage study duration was short, not dietary, and does not represent actual year-round intake rates of most mammals, so d3 ) 1.5. No co-contaminants were evaluated, so d4 ) 2. Liver lesions are not mechanistically clear relative ecological impact on natural populations, so d5 ) 1.5. No comparative data exist, so d6 )1. Tissue ratios were not used, so d7 ) 0. Only adult rats have been tested, no data have been collected on other life stages, so d8 ) 2. UFmammal ) (5 × 5 × 2 × (1.5 + 1.5 + 1.5 + 2 + 1.5 + 1.0 + 0 + 2) ) 550. UFavian ) (10 × 5 × 2 × (1.5 + 1.5 + 1.5 + 2 + 1.5 + 1 + 0 + 2) ) 1100. dissolved HFE-7500 fraction (ffd) in water as outlined in the GLI (eq 4) is as follows: ffd ) 1 (DOC)(Kow) + (POC)(Kow) 1+ 10 (4) where DOC is concentration of dissolved organic carbon (0.0000002 kg DOC/L); POC is concentration of particulate organic carbon (0.00000004 kg POC/L); and Kow ) octanolwater partitioning coefficient of HFE-7500 (Kow ) 79 432). The food chain multiplier (FCM) represents the ratio of a BAF to an appropriate BCF. The FCM is used in the calculation of trophic level specific baseline BAFs when there are no available field-measured BAFs or biota-sediment accumulation factors (BSAF). Based on the log Kow for HFE7500 (4.9), the FCM for trophic level 3 was 2.78, whereas the FCM for trophic level 4 was 2.193 (taken from Table B-1 of GLI). Human and wildlife TL3 and TL4 bioaccumulation factors for trophic levels 3 and 4 were calculated as follows (eq. 5) and are reported in Table 4: Human or Wildlife BAFTLx ) [(baseline BAFtlx)(Std lipid value) + 1](ffd) (5) Uncertainty related to the interpretation of toxicity test data among different species, use of different laboratory endpoints, and differences in experimental design are addressed by applying uncertainty factors (UFs). There are several methods that exist to estimate UFs based on different regulatory programs and their requirements (9). One method developed by USEPA Region 8 for the Rocky Mountain Arsenal (RMA; 10), uses rigorous study-selection criteria and a 4-step “balanced” uncertainty factor protocol to calculate TRVs. The RMA procedure uses four categories of uncertainty 4766 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 22, 2002 factors: (1) intertaxon variability extrapolation, (2) exposure duration extrapolation, (3) toxicologic endpoint extrapolation, and (4) modifying factors. The overall uncertainty factor is calculated as the product of all four categories. The selection and rationale for calculating mammalian and avian uncertainty using the RMA protocol are given in Table 5. Wildlife values (WV) for each ecological receptor were estimated with eq 6: TRVW ) TD or NOAEL × BW Oveall UF WC + ∑(FC TLi × BAFWL TLi) (6) where TRVw is the wildlife value (mg/L), TD is the test dose or threshold dose for the test species (mg/kg/day), BW is the average weight (kg) for the representative species, FCTLi is a species-specific average daily amount of food consumed (kg/ day), and WC is a species-specific average daily amount of water consumed (L/day). Ideally, concentrations for the protection of wildlife are derived from chronic toxicity studies in which an ecologically relevant endpoint has been assessed in the species of concern or in a closely related species. Although TRVws can be expressed or defined as NOAELs, the use of LOAELs is generally preferred because, by definition, NOAELs incorporate greater uncertainty than LOAELs. In the 28-d oral rat study, only a NOAEL was determined. In addition, the study duration was short, less than 10% of the lifetime of the test organism, and did not include reproductive or developmental endpoints. Because of the lack of a definitive chronic study, wildlife values were calculated using both NOEL and LOEL as determined in the 28-d rat study. Exposure parameters, including body weights (BW), feeding rates (FCTli), drinking rates (WC), and trophic level dietary composition (as food ingestion rate and food item percent in diet, for humans and TABLE 6. Exposure Parameters for Human and Five Surrogate Wildlife Species Evaluated for the Japanese Environment receptor adult BW (kg) human (adult) mammals mink otter birds kingfisher herring gull bald eagle food ingestion rate of each prey in each trophic level (kg/day)b water ingestion rate (L/day) trophic level of prey (% diet) 2.0a TL3: 0.0336; TL4: 0.0114 not applicable 0.80 7.4 0.081 0.600 TL3: 0.159; other: 0.0177 TL3: 0.977; TL4: 0.244 TL3: 90; other: 10 TL3: 80; TL4: 20 0.15 1.1 4.6 0.017 0.063 0.160 TL3: 0.0672 TL3: 0.192; TL4: 0.0480; other: 0.0267 TL3: 0.371; TL4: 0.0929 PB: 0.0283; other: 0.0121 TL3; 100 fish: TL3: 80; TL4: 20; other: 10 fish (92): TL3: 80; TL4: 20; birds (8): PB: 70; other: 30 70 a Mean water consumption including both drinking and incidental ingestion. b TL3 or TL4 ) trophic level 3 or 4 fish; PB) piscivorous birds; other ) nonaquatic birds and mammals. TABLE 7. Wildlife Values for Mammalian and Avian Species Based on NOEL and LOEL for HFE-7500 species mammalian species mink otter geometric mean avian species kingfisher herring gull bald eagle geometric mean NOEL-TRVw (ng/L) LOEL-TRVw (ng/L) 44 56 49 220 280 247 11 19 42 21 54 96 211 103 wildlife are given in Table 6. Wildlife values for both mammalian and avian species evaluated in this study are given in Table 7. Criteria Based on Noncancer Human Health. A human health criterion for HFE-7500 was derived to establish an ambient WQC that would be protective of humans who consumed fish that had accumulated HFE-7500. This criterion, if not exceeded, will protect individuals from adverse health impacts from that chemical due to consumption of contaminated fish and drinking water, or through the ingestion of water because of participation in water-oriented recreational activities. Noncancer endpoints were used to derive the maximum ambient water concentration of HFE7500 at which no adverse effects are likely to occur in human populations from lifetime exposures. Depending on the availability of suitable data, Tier I or Tier II human noncancer criterion (HNC) can be calculated. The minimum data requirements for calculating Tier I values include at least one well-conducted epidemiology study or a well-conducted chronic study with test animals. The animal studies should demonstrate a dose-response relationship involving one or more critical effects that are biologically relevant to humans. In addition, the duration of the study should span multiple generations of the exposed population or at least a major portion of its life span. For shorter studies, the study should be at least 90 d for rodents or 10% of the life span of other appropriate species. The minimum data requirements for Tier II values include at least one repeated dose study of 28 d (in rodents) with a dose-response relationship that includes endpoints that are relevant to humans. A review of the relevant toxicity data for HFE-7500 indicates that there is not sufficient data for the derivation of a Tier I TRCHNC value. Therefore, a TRVHNV was calculated as a conservative and interim level of protection for human life (eq 7): TRVHNV (mg/L) ) ADE × BW × RSC (7) HH WC + [(FCTL3 × BAFHH TL3) + (FC × BAFTL4)] where ADE is the acceptable daily exposure (mg/kg/day); BW is the weight of average human (kg); RSC is the relative source contribution factor for contaminant from aquatic sources (0.8); WC is total per capita water consumption including drinking water and incidental daily consumption (L/day); FCTLi is mean consumption of trophic level i fish by regional consumers of locally caught fish (kg/day); and HH BAFTli is the human bioaccumulation factor for edible portion of trophic level i fish. Bioaccumulation factors used in the human health criterion (eq 5) were calculated using the baseline BAF that was calculated to estimate wildlife values. The acceptable daily exposure (ADE) estimates were based on the NOEL and LOEL from the 28-d rat study. To account for the uncertainties in predicting acceptable dose levels for the general human population based on test animal data, uncertainty factors (UF) were selected on the basis of quantity and quality of the data. The selection of uncertainty factors was based on criteria outlined in the GLI document. Because the duration of the 28-d rat study was less than 10% of a rat’s normal life cycle, an uncertainty factor of 3000 was used in the analysis. This factor not only accounted for the uncertainty in the animal to human extrapolation, but also considered the uncertainty in subacute to chronic extrapolation. Furthermore, an additional uncertainty factor of 3.0 was included to account for the lack of reproductive or developmental endpoint evaluation during the rat study. This resulted in an overall uncertainty factor of 9000. The high Henry’s constant and apparent lack of biological activity also suggest the magnitude of uncertainty due to chronic toxicity or across species extrapolation has been overestimated. Dividing the NOEL or LOEL by the overall uncertainty factor resulted in an ADENOEL and ADELOEL of 0.022 and 0.111 mg/ kg/day, respectively. Noncancer human health values were estimated (eq 7) and are reported in Table 8. Risk Characterization. Hazard quotients (HQ) and margins of safety (MOS) were calculated to evaluate potential risks of HFE-7500 to humans and wildlife. Hazard quotients are ratios of predicted environmental concentrations to threshold concentration to which an organism can be exposed continuously for their entire life without suffering. If a predicted environmental concentration is equal to a TRV then the HQ would be 1.0. Thus, for HQ values greater than 1.0, some potential for injury can be inferred. HQ values less than 1.0 indicates that the potential for injury is low. Because very conservative (protective) assumptions are used in the calculations, HQ values less than 1.0 are associated with safe conditions. Margins of safety (MOS) are the inverse of the HQ and represent the magnitude of the difference between predicted environmental concentrations and no-effect concentrations. Aquatic Organisms. Based on the Tier II WQC, the release of HFE-7500 to the Japanese environment poses little risk to VOL. 36, NO. 22, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4767 TABLE 8. Hazard Quotients (HQ) and Margins of Safety (MOS) for Humans, Avian, and Mammalian Species Exposed to HFE-7500 organism avian NOEL LOEL mammal NOEL LOEL human NOEL LOEL predicted water conc. (ng/L) critical health value (ng/L) HQ MOS 5.13 × 10-4 5.13 × 10-4 21 103 2.5 × 10-5 5.0 × 10-6 4.0 × 104 2.0 × 105 5.13 × 10-4 5.13 × 10-4 49 247 1.0 × 10-5 2.1 × 10-6 9.7 × 104 4.8 × 105 5.13 × 10-4 5.13 × 10-4 6177 30 887 8.3 × 10-8 1.7 × 10-8 1.2 × 107 6.0 × 107 aquatic organisms. Under the assumptions used in this risk analysis, the HQ value was 4.0 × 10-9 with a MOS of 2.5 × 108. Mammal and Avian Organisms. The geometric mean of the mammalian and avian wildlife values were used as “safe values” and compared to predicted water concentrations taken from the Higashino model. The results (Table 8) show that HFE-7500 poses no risk to either mammalian or avian species. For either group, the margin of safety was greater than 50 000. The margin of safety is probably even greater because of the conservative application of 100 in safety factors to the predicted water concentration Noncancer Human Health Evaluation. Noncancer human health values were used in the risk assessment of HFE7500 rather than a cancer-based assessment because of the lack of whole organism carcinogenicity data. Furthermore, two in vitro tests, a mutagenicity and clastogenesis bioassay, indicated that HFE-7500 was not genotoxic. Results of the risk calculations are given in Table 8. Using a predicted water concentration of 5.13 × 10-4 ng/L, the noncancer HQ was 3.7 × 10-8 and the MOS was 2.7 × 107. Discussion On the basis of the results of the environmental fate modeling and toxicological evaluation of HFE-7500, it can be concluded that this chemical does not pose any risk to aquatic or terrestrial organisms in Japan. For the receptors evaluated in the study, all HQ values were significantly less than 1 with margins of safety greater than 200 000. In addition, these risk calculations were based on water concentrations that included a safety factor of 100 to account for the uncertainty in model predictions for surface water concentrations. Safety factors were also included in the calculation of wildlife values and human health values to take into account the lack of relevant toxicity data for both aquatic and terrestrial species. Thus, the predictions of risk were extremely conservative and may have overestimated the actual risk that HFE-7500 could pose to biota and humans in Japan. Evaluation of Uncertainty. The evaluation of uncertainty involves identifying sources of uncertainty associated with the ERA process that may potentially affect the conclusions of the assessment. According to the U.S. EPA (11), an uncertainty analysis increases the credibility of an environmental risk assessment by explicitly describing the magnitude and direction of uncertainties. This analysis also provides the basis for efficient data collection and application of refined methods to evaluate further the risk assessment’s assumptions and conclusions. Specifically, uncertainty associated with measurement endpoint results translates into uncertainty associated with the conclusions regarding HFE-7500 releases to the Japanese environment. To reduce the potential for uncertainty resulting in underestimates of actual risks of 4768 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 22, 2002 HFE-7500 to humans and wildlife in Japan, conservative methods and procedures were used throughout the assessment. One source of uncertainty in the analysis was the lack of subchronic and chronic toxicity data that included an evaluation of potential effects of HFE-7500 on reproductive and developmental processes in traditional laboratory species. Furthermore, no studies exist that evaluate the effect of HFE-7500 on wildlife, including semi-aquatic or terrestrial mammals, or avian species. Because of the lack of toxicity data, the calculation of wildlife values from test species data incorporated additional uncertainty factors to account for toxicity endpoint and interspecies extrapolations. A 100-fold safety factor was also used to account for the uncertainty in model predictions for surface water concentrations. Overall, because of the lack of toxicity studies to relevant species, the risk analysis used very conservative assumptions that resulted in risk quotients that were between 100- and 1000-fold larger than they might be if additional data were available. Again, the high Henry’s constant and apparent lack of biological activity also suggests the magnitude of uncertainty due to chronic toxicity or among-species variation has been overestimated. Another possible source of uncertainty was the lack of environmental fate properties for inclusion in the fate and distribution models. In particular, the lack of media-specific half-lives added significantly to the final uncertainty incorporated into this analysis. For instance, the only reactive loss value accounted for in the Higashino model was atmospheric degradation that proceeds through photochemical mechanisms. However, although the model did not directly address advective losses from the atmosphere, processes such as washout by precipitation would not be significant. This is because, due to the low water solubility of HFE-7500, partitioning of HFE-7500 from air into atmospheric water droplets would not be a significant loss process when compared to its reactive losses. The atmospheric lifetime of a compound is the time it takes for a compound to decrease to approximately 37% of its original concentration. To estimate the lifetime due to rainout is based on the assumption that rainout is the only mechanism contributing to atmospheric loss and that it is mostly dependent on Henry’s law constant (H). Thus, the concentration of a HFE-7500 in cloud-water droplets would be in equilibrium with the concentration of HFE-7500 in the surrounding air. The atmospheric lifetime due to rainout can then be estimated as a function of it’s Henry law constant (for compounds with H < 104 M/atm, eq 7 can be used)(20): tRainout ) 105/H (days) (7) The Henry’s law constant for HFE-7500 was 3.1 × 107 Pa m3/mol and is equivalent to an H of 3.27 × 10-6 M/atm. This results in an estimated atmospheric lifetime of 3.06 × 1010 days. Thus, rainout would be a minor process in the removal of HFE-7500 from the atmosphere, where expected to be removed from the atmosphere in rainwater, since H indicates it would partition almost entirely into the gas phase. Finally, there was no accounting for advective or reactive losses of HFE-7500 from surface waters. Again, due to its low water solubility and relatively great vapor pressure, it would not be expected to remain in water for any significant duration. However, the significance or uncertainty associated with these processes does not affect the overall conclusions of this study. This is because of the very large margins of safety used in our analyses. Thus, even if the total release of HFE7500 to the environment, the accumulation, and toxicity to receptors were underestimated, there would be essentially no chance of adverse effects. A third area of uncertainty in the risk assessment concerns the formation and accumulation of HFE-7500 breakdown products and their effect on environmental systems. Although the photodegradation pathway of HFE-7500 has not been fully characterized, an evaluation of this pathway provides valuable insight to potential breakdown products that could be found in ecological systems (12). A hydroxyl radical reaction rate measurement indicates that it would have an atmospheric lifetime of about 2.5 yrs (19). Based on analogous HFEs, as well as other related fluorochemicals, one can make reasonable predictions about the degradation products (2123). The predominant degradation intermediates of a related product, ethyl iso-perfluorobutyl ether (CF3CF(CF3)CF2OCH2CH3), are the acetate and the formate forms where the acetate predominates. For HFE-7500, the acetate and formate would have the structures C3F7CF(OC(dO)CH3)CF(CF3)2 and C3F7CF(OCH(dO))CF(CF3)2. By analogy to other HFEs, these two components are likely to be removed from the atmosphere by wet or dry deposition. Once out of the atmosphere, the acetate and formate are likely to react with water and lose an HF to form a perfluoroketone, C3F7C(O)CF(CF3)2. This hydrolytic reaction could also occur in water droplets within the atmosphere. If this ketone reenters the atmosphere, it would likely photodegrade further within days or weeks, breaking at the keto group to form two perfluororadicals, e.g., C3F7C(O) and CF(CF3)2. The straight chain perfluororadical would go through a series of photochemical reactions and fully degrade to HF and CO2. The branched chain perfluororadical would degrade to form HF, CO2, and trifluoroacetic acid. If deposition of the acetate and formate predominates, HFE-7500 would produce one mole of trifluoroacetic acid per mole of HFE-7500 and the remainder would degrade to HF and CO2. Perfluorobutyric acid and iso-perfluorobutyric acid are additional potential photochemical degradation products that could form in the atmosphere. Like trifluoroacetic acid, these water-soluble compounds would washout of the atmosphere and would be present in the environment as acetate salts. Hydrofluorocarbon breakdown products, such as trifluoroacetic acid (TFA), are not expected to cause adverse effects in ecological systems. Numerous studies have been conducted that evaluate the environmental properties and toxicity of trifluoroacetic acid (TFA) to algae, aquatic macrophytes, terrestrial plants, fish, animals, and humans (13). TFA is an organic acid with a pKa of 0.23. TFA exists in water in a completely dissociated state and is miscible in water (solubility over 1000 g/L). Its low octanol/water partition coefficient (Kow)0.008) indicates no potential to bioaccumulate. Toxicity tests on fish and aquatic invertebrates show no adverse effects at large concentrations (up to 1 g of TFA/ L). However, the NOEC for the algal species, Selenastrum capricornutum, was approximately 0.1 mg TFA/L. Toxicological data with mammals also show TFA hsd low toxicity (NOAEL ) 150 mg/kg/day). While TFA is a ubiquitous contaminant in the hydrosphere, concentrations of in rain and water are expected to be low (14, 15). Therefore, based on its low toxicity, the expectation that all its emissions will be to the atmosphere, and its low propensity to partition into water, the potential risk of HFE-7500 to wildlife and humans is very small. In addition, HFE-7500 has a relatively short environmental half-life, is not a contributor to pho- tochemical smog, and has a very low GWP. Therefore, HFE7500 would make a much smaller contribution to global warming relative to the materials it would replace and the substitution of HFE-7500 for currently used products would likely result in a net improvement in the environmental health and safety. Literature Cited (1) Tuma, P. E.; Tousignant, L. Solid State Technol. 2000, June, 175-182. (2) USEPA Final Water Quality Guidance for the Great Lakes System. Code of Federal Regulations, Title 40, Part 132, 1995. (3) Higashino, H.; Kitabayashi, K.; Yokoyama, O.; Takatsuki, M.; Yonezawa, Y. J. Jpn. Soc. Atmos. Environ. 2000, 35, 215-228. (4) Beach, S. A.; Gulbranson, D. J.; Purdy, R. E.; Reiner, E. A.; Lieder, P. H.; Werner, K. T. 3M Internal Report, 1999. (5) Cousins, I. T.; MacKay, D. 3M Internal Report, 2001. (6) Yamashita, K. Twenty-eight Day Repeated Oral Toxicity Study of T-7145 in Rats, Study No. 9L333. Sumitomo 3M Limited, 2000. (7) U.S. EPA. Wildlife Exposure Factors Handbook Volumes I and II; EPA/600/R-93/187b; Office of Research and Development: Washington, DC, 1993. (8) Shigeoka, T.; Saitoh, H. Bioconcentration Study of T-7145 with Carp, 9B213G; 2000. (9) Duke, L. D.; Taggart, M. Environ. Toxicol. Chem. 2000, 19, 16681680. (10) Henningsen, G.; Hoff, D. Uncertainty Factor Protocol for Ecological Risk Assessment: Toxicological Extrapolations to Wildlife Receptors. RMA-IEA/0056; 1997. (11) U.S. EPA. Guidelines for Ecological Risk Assessment: Final; EPA/ 630/R-95/002F; U.S. Government Printing Office: Washington, DC, 1998. (12) Cooper, D. L.; Cunningham, T. P.; Alan, N. L.; McCulloch, A. Atmos. Environ. 1993, 27A, 117-119. (13) Boutonnet, J. C.; Bingham, P.; Calamari, D.; De Rooij, C. G.; Franklin, J.; Kawano, T.; Libre, J. M.; McCulloch, A.; Malinverno, G.; Odom, J. M.; Rusch, G. M.; Smythe, K.; Sobolev, I.; Thompson, R.; Tiedje, J. M. Human Ecol. Risk Assess. 1999, 5, 59-124. (14) Wujcik, C. E.; Zehavi, D.; Seiber, J. N. J. Phys. Chem. A 1998, 101, 8264-8273. (15) Cahill, T. M.; Seiber, J. E. Environ. Sci. Technol. 2000, 34, 29022912. (16) MacKay, D.; Paterson, S.; Cheung, B. Chemosphere 1985, 14, 859-863. (17) Judekis, H. S.; Wren, A. G. Atmos. Environ. 1977, 11, 1221-1224. (18) Collar, N. J.; Andreev, A. V.; Chan, S.; Crosby, M. J.; Subramanya, S.; Tobias, J. A. Red Data Book. BirdLife International: Cambridge, U.K., 2001. (19) Guschin, A. G.; Molina, L. T.; Molina, M. J. Atmospheric Chemistry of L-15381, L-15566, and L-14703 and Integrated Band Strengths of L-14374, L-14375, L-14752, L-13453 and L-14703. 3M Internal Report, 1999. (20) Chang, D. T.; Dak, N. Chemicals in the Atmosphere. Atmospheric and Environmental Research, Inc.: San Ramon, CA, 1992. (21) Wallington, T. J.; Schneider, W. F.; Sehested, J.; Bilde, M.; Platz, J.; Nielsen, O. J.; Christensen, L. K.; Molina, M. J.; Molina, L. T.; Wooldridge, P. W. J. Phys. Chem. A 1997, 101, 8264. (22) Christensen, L. K.; Sehested, J.; Nielsen, O. J.; Bilde, M.; Wallington, T. J.; Guschin, A.; Molina, L. T.; Molina, M. J. J. Phys. Chem. 1998, 102, 4839. (23) Ninomiya, Y.; Kawasaki, M.; Guschin, A.; Molina, L. T.; Molina, M. J.; Wallington, T. J. Environ. Sci. Technol. 2000, 34, 29732978. Received for review April 19, 2002. Revised manuscript received August 21, 2002. Accepted August 26, 2002. ES0257321 VOL. 36, NO. 22, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4769