Chemosphere xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants Greta Stieger, Martin Scheringer ⇑, Carla A. Ng, Konrad Hungerbühler Institute for Chemical and Bioengineering, ETH Zürich, 8093 Zürich, Switzerland
a r t i c l e
i n f o
Article history: Received 14 October 2013 Received in revised form 11 January 2014 Accepted 15 January 2014 Available online xxxx Keywords: Brominated flame retardants PBT assessment Baseline toxicity Octanol–water partition coefficient
a b s t r a c t Polybrominated diphenylethers (PBDEs) and hexabromocyclododecane (HBCDD) are major brominated flame retardants (BFRs) that are now banned or under restrictions in many countries because of their persistence, bioaccumulation potential and toxicity (PBT properties). However, there is a wide range of alternative BFRs, such as decabromodiphenyl ethane and tribromophenol, that are increasingly used as replacements, but which may possess similar hazardous properties. This necessitates hazard and risk assessments of these compounds. For a set of 36 alternative BFRs, we searched 25 databases for chemical property data that are needed as input for a PBT assessment. These properties are degradation half-life, bioconcentration factor (BCF), octanol–water partition coefficient (Kow), and toxic effect concentrations in aquatic organisms. For 17 of the 36 substances, no data at all were found for these properties. Too few persistence data were available to even assess the quality of these data in a systematic way. The available data for Kow and toxicity show surprisingly high variability, which makes it difficult to identify the most reliable values. We propose methods for systematic evaluations of PBT-related chemical property data that should be performed before data are included in publicly available databases. Using these methods, we evaluated the data for Kow and toxicity in more detail and identified several inaccurate values. For most of the 36 alternative BFRs, the amount and the quality of the PBT-related property data need to be improved before reliable hazard and risk assessments of these substances can be performed. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Brominated flame retardants (BFRs) comprise a wide range of brominated aromatic and aliphatic compounds (Bergman et al. 2012). Two types of BFRs widely used in the last decades are polybrominated diphenylethers (PBDEs) and hexabromocyclododecane (HBCDD). Because of their hazardous properties regarding persistence, bioaccumulation potential and toxicity (PBT properties), the use of these substances is now restricted or prohibited under various national and international legislations (UNEP, 2013; ECHA, 2013a). However, PBDEs and HBCDD are often replaced by other BFRs such as decabromodiphenyl ethane, pentabromotoluene, pentabromoethylbenzene, and many more. Often, the chemical properties of these substitutes are only poorly known; thus concerns about their PBT characteristics remain (Covaci et al., 2011). PBT assessment is a component of chemicals assessment that focuses on three hazardous properties of organic chemicals. These ⇑ Corresponding author. Tel.: +41 44 632 3062; fax: +41 44 632 1189.
properties include persistence (P), potential for bioconcentration and bioaccumulation (B), and aquatic toxicity (T); Strempel et al. (2012) provide an overview of several PBT assessment schemes. Data required for a PBT assessment include degradation half-lives in water or soil for P, octanol–water partition coefficients, bioconcentration and, in some cases, also bioaccumulation factors for B, and toxic effect concentrations from chronic toxicity tests or, if no chronic toxicity data are available, acute tests for T (Strempel et al., 2012). Here we evaluate the availability and quality of chemical property data that are needed for a PBT assessment of alternative BFRs. We selected 36 BFRs that were listed by Bergman et al. (2012) and also found in the PBT database compiled by Strempel et al. (2012). To retrieve chemical property data for these chemicals, we extensively searched 25 publicly accessible databases. On this basis, we analyze the number of data points for each property and chemical, discuss the variability and scatter in the data, and evaluate the data for the octanol–water partition coefficient (Kow) and the aquatic toxicity in more detail.
E-mail address: [email protected] (M. Scheringer). http://dx.doi.org/10.1016/j.chemosphere.2014.01.083 0045-6535/Ó 2014 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
2
G. Stieger et al. / Chemosphere xxx (2014) xxx–xxx
2. Materials and methods The BFRs investigated are listed in Table 1, which is based on information from Bergman et al. (2012). For these 36 chemicals we retrieved chemical property data from 25 publicly accessible databases, including the database of the European Chemicals Agency (ECHA), http://www.echa.eu; and the OECD eChemPortal, which provides links to 24 participating data sources, http:// www.echemportal.org. The property data include biodegradation half-life in soil, t1/2,soil, for persistence, the octanol–water partition coefficient, Kow, and the bioconcentration factor, BCF, for bioaccumulation, and chronic and acute effect concentrations, Tc and Ta, for aquatic organisms (algae, daphnia, fish) for toxicity. The databases were searched by CAS number for each of the 36 BFRs and all available data entries for the PBT properties, including both experimental and estimated values, were collected and stored. The data were retrieved in November and December 2012. We determined the numbers of data points that were found for each chemical and property. For chemicals with several data points per property (Ta, BCF, Kow), we plotted the data in order to visualize the variability of the data. For the Kow and toxicity data in particular, we evaluated the plausibility of individual data points. The Kow of organic chemicals is largely determined by the chemicals’ solubility in water, Sw (Schwarzenbach et al., 2003). Sw is determined by the energy that is required to accommodate the solute in water, i.e. to break hydrogen bonds between water molecules and form a cavity that can take up the solute molecule. For neutral or weakly polar organic chemicals, this amount of energy is directly proportional to the molecular weight. In the case of polychlorinated biphenyls (PCBs), log Kow follows a highly significant positive linear relationship with increasing degree of chlorination and thus
increasing molecular weight (Schenker et al., 2005). Many brominated aromatic compounds in our set are also non-ionizing organic substances; for substances with phenolic groups, the Kow is measured at sufficiently low pH where ionization is suppressed and the substances are predominantly present in their neutral form (Sotomatsu et al., 1993; Kuramochi et al., 2004). Therefore, we expect that a similar relationship can be derived for the brominated aromatic substances in our set of 36 alternative BFRs. To establish this relationship, we looked for series of compounds with similar aromatic backbones, in analogy to the set of PCB congeners. From the set of 36 BFRs, two such subsets were available: (i) 14 monoaromatic BFRs and (ii) 9 di-aromatic structures, including TBBPA, 6 TBBPA derivatives, as well as DBDPE and BTBPE. Together, these two sets include 23 of the 36 BFRs. Because Kow values were available for only 11 of these 23 substances, we included the log Kow values collected by Strempel et al. (2012) and Bergman et al. (2012) in our analysis. Because the log Kow values for these 23 compounds showed a high degree of variability, we implemented a two-step process to find the best relationship between log Kow and molecular weight. First, we used all of the log Kow data to derive a linear relationship with molecular weight. We then used this linear relationship to determine, for each substance, which log Kow values seemed implausible because they were either too low or too high. In general, for chemicals where the variability of the Kow data was high, the lower log Kow values were (much) below the linear relationship whereas most of the higher ones were in line with it. In the second step, we excluded those log Kow values that were clearly below or above the linear relationship and calculated for each chemical the ‘‘best guess’’ log Kow as the average of the remaining log Kow values (these ‘‘best guess’’ estimates are the data points shown in Fig. 1, left).
Table 1 CAS numbers, common names, and practical abbreviations proposed by Bergman et al. (2012) of the 36 BFRs investigated in this work. Count
CAS number
Common name
Practical abbreviation
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
79-94-7 632-79-1 52434-90-9 66710-97-2 59447-55-1 34571-16-9 26040-51-7 39635-79-5 32588-76-4 38521-51-6 20566-35-2 87-82-1 3194-57-8 3278-89-5 25327-89-3 21850-44-2 87-83-2 3296-90-0 3322-93-8 85-22-3 25713-60-4 37853-59-1 84852-53-9 615-58-7 35109-60-5 37853-61-5 33798-02-6 3555-11-1 126-72-7 4162-45-2 42757-55-1 23488-38-2 608-71-9 51936-55-1 118-79-6 39569-21-6
Tetrabromobisphenol A 3,4,5,6-Tetrabromophthalic anhydride 1,3,5-Tris(2,3-dibromopropyl)-1,3,5-triazine-2,4,6-trione Tetrabromobisphenol A bis(2-hydroxyethyl)ether bisacrylate Pentabromobenzyl acrylate 1,2,3,4,7,7-Hexachloro-5-(2,3,4,5-tetrabromophenyl)-bicyclo[2.2.1]hept-2-ene Bis(2-ethylhexyl) tetrabromophthalate Tetrabromobisphenol S N,N0 -Ethylenebis (tetrabromophthalimide) Pentabromobenzyl bromide 2-(2-Hydroxyethoxy)ethyl 2-hydroxypropyl 3,4,5,6-tetrabromophthalate Hexabromobenzene 1,2,5,6-Tetrabromo-cyclooctane 2,4,6-Tribromophenyl allyl ether Tetrabromobisphenol A bis(allyl) ether Tetrabromobisphenol A bis(2,3-dibromopropyl) ether Pentabromotoluene Dibromoneopentyl glycol 4-(1,2-Dibromoethyl)-1,2-dibromocyclohexane Pentabromoethylbenzene Tris(2,4,6-tribromophenoxy)-s-triazine 1,2-Bis(2,4,6-tribromophenoxy)ethane Decabromodiphenyl ethane 2,4-Dibromophenol 2,4,6-Tribromophenyl 2,3-dibromopropyl ether Tetrabromobisphenol A bismethyl ether 3,30 ,5,50 -Tetrabromobisphenol A bisacetate Pentabromophenol allyl ether Tris(2,3-dibromopropyl) phosphate Tetrabromobisphenol A bis(2-hydroxyethyl) ether Tetrabromobisphenol S bis(2,3-dibromopropyl ether) 1,2,4,5-Tetrabromo-3,6-dimethylbenzene Pentabromophenol 5,6-Dibromo-1,10,11,12,13,13-hexachloro-11-tricyclo[8.2.1.02,9]tridecene 2,4,6-Tribromophenol 2,3,4,5-Tetrabromo-6-chlorotoluene
TBBPA TEBP-Anh TDBP-TAZTO TBBPA-BHEEBA PBB-Acr HCTBPH BEH-TEBP TBBPS EBTEBPI PBBB HEEHP-TEBP HBB TBCO TBP-AE TBBPA-BAE TBBPA-BDBPE PBT DBNPB DBE-DBCH PBEB TTBP-TAZ BTBPE DBDPE DBP TBP-DBPE TBBPA-BME TBBPA-BOAc PBP-AE TDBPP TBBPA-BHEE TBBPS-BDBPE TBX PBP DBHCTD TBP TBCT
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
3
G. Stieger et al. / Chemosphere xxx (2014) xxx–xxx
Excluded were 7 estimated values (4 too low, 3 too high), 7 measured values (all too low), and 4 values without any information about the method used (all 4 too low). Finally, we made a linear regression through the 23 best-guess log Kow values (see Fig. 1, left) and found the following equation for the correlation: log Kow = 0.0125MW + 0.3034, R2 = 0.88. For the 23 brominated aromatic substances in our set, we determined the deviation of individual data points from the value predicted by the relationship between log Kow and MW. Data points that deviate by several orders of magnitude from the value derived from the molecular weight were considered unreliable. To evaluate the data for aquatic toxicity, we used established relationships between LC50 for baseline toxicity and log Kow. Baseline toxicity or narcosis is caused by the partitioning of a chemical into the cell membrane lipids; it is an unspecific effect and represents the minimum toxicity that every chemical exerts (Escher and Hermens, 2002). Because it is determined by a chemical’s tendency to partition from water into lipids, the baseline toxicity of chemicals increases with increasing log Kow. The LC50 for baseline toxicity (denoted by LC50,bl in the following) can, therefore, be derived from the log Kow. If a chemical acts according to a specific mode of toxic action, this is an additional effect on top of the chemical’s baseline toxicity. In such a case, the actual LC50 of the chemical is lower than the LC50 for baseline toxicity derived from the log Kow. For a plot of LC50 data vs. log Kow, this implies that measured LC50 data can be expected to be close to the line defined by the relationship between log Kow and LC50,bl or, if the chemical acts according to a specific mode of toxic action, below this line. In contrast, LC50 data that are much higher than the LC50,bl derived from the log Kow are likely to be unreliable or incorrect. Fig. 1, right, shows six relationships between LC50 for baseline toxicity in fish and log Kow from the literature (Könemann, 1981; Veith et al., 1983; McCarty et al., 1992; Verhaar et al., 1995; Maeder et al., 2004). As pointed out by Mayer and Reichenberg (2006), the relationship between LC50,bl and log Kow also applies to high-Kow chemicals. Also shown in Fig. 1, right, are three domains of LC50 data: more toxic than baseline toxicity (more than 1 order of magnitude below the lowest baseline toxicity relationship); baseline toxicity (less than 1 order of magnitude below the lowest or above the highest baseline toxicity relationship); and inconclusive/incorrect (more than 1 order of magnitude above the highest baseline toxicity relationship).
values are persistence data (see Fig. 2, left). Values for Kow and BCF were found for all 19 of the 36 BFRs for which any data were available; values for t1/2,soil were found for 11 of the 36 BFRs, and values for Tc and Ta for 14 and 18 of the 36 BFRs, respectively. For all of these PBT properties, there are a few BFRs for which multiple values for the same property can be found, whereas for the other BFRs only one or no values are available for any of the properties (see Fig. 2, right). Among the few BFRs for which multiple entries are available for all PBT properties are TBBPA, CAS 79-94-7, and TBP, CAS 118-79-6. TBBPA is one of the most commonly used BFRs today (Bergman et al., 2012) and TBP is produced in high production volumes (>1000 t yr1) (WHO, 2005). In terms of data availability, we conclude that property data are scarce for persistence. Data on bioaccumulation potential and toxicity are more abundant; however, the data are unequally distributed: for some BFRs, in particular ones with high production volumes and extensive applications, many data are available whereas for most alternative BFRs only very little information is available. 3.2. Data variability Where multiple database entries were found for the same BFR and PBT property, the variability of the values is often substantial. As examples, the Kow values found for TBBPA range over five orders of magnitude (see Fig. 3, left, CAS 79-94-7) and the Ta values found for pentabromophenol (PBP) range over almost five orders of magnitude (see Fig. 3, right, CAS 608-71-9). This high variability and thus inconsistency of the available data raises questions about the quality of the data. 3.3. Data plausibility Plotting log Kow values against the BFRs’ molecular weight makes it possible to judge the plausibility of the available data, especially in cases where multiple and widely different data points are available for the same substance. The log Kow values retrieved from the various databases for the set of 23 aromatic BFRs are largely consistent with the linear relationship between log Kow and molecular weight. However, there are a few exceptions, see Fig. 4. One exception is the log Kow value of 3.55 provided by the ECHA database for DBDPE (CAS 84852-539, MW = 971 g mol1), which is too low by more than 8 log units (i.e. 8 orders of magnitude for the Kow) and was removed from the calculation of the best-guess log Kow for DBDPE, which is 12.3. For DBDPE, a high bioaccumulation factor, BAF, with log BAF values between 6.1 and 7.1 has been reported (He et al., 2012), which is consistent with a high log Kow. It is well known that high
3. Results 3.1. Data quantity Overall, data were available for 19 of the 36 BFRs. 47% of all collected data points are for Ta, whereas only 5% of the available 2
domain 3: effect concentrations above baseline toxicity: suspect
log K ow
log LC50,bl (mmol/L)
0
Verhaar et al. (1995) Könemann (1981) Verhaar et al. (1995) McCarthy et al. (1992) Veith et al. (1983) Maeder et al. (2004)
–2 –4 –6 domain 1: effect concentrations below baseline toxicity: specific toxicity
–8
domain 2: baseline toxicity
– 10
0 molecular weight, MW
2
4
6 8 log Kow
10
12
14
Fig. 1. Left: log Kow vs. molecular weight for 23 aromatic BFRs; symbols indicate best-guess values. Red line: relationship obtained by linear regression, log Kow = 0.0125MW + 0.3034. Right: six relationships between LC50 for baseline toxicity (in mmol L1) and log Kow as well as three domains of LC50 data in the plot of LC50 vs. log Kow.
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
G. Stieger et al. / Chemosphere xxx (2014) xxx–xxx
total number of datapoints found
4
46.7% 13.6%
19.9% 15.5%
4.3%
Kow
B
P
Tc
Ta CAS number
log Kow
effect concentration for acute aquatic toxicity (mg/L)
Fig. 2. Left: distribution of the total number of database entries found among the different PBT properties. Right: total number of database entries found for each of the 36 BFRs.
CAS number
TBBPA
CAS number
PBP
Fig. 3. Left: variability of database entries found for log Kow per chemical (chemicals indicated by CAS number). Right: variability of database entries found for acute aquatic toxicity (in mg L1) per chemical (chemicals indicated by CAS number). TBBPA (left) and PBP (right) are highlighted in gray as examples (see text).
14
Br
Br
DBDPE
log K ow
Br
Br
Br Br Br
Br Br
Br
ECHA: database of substances registered under REACH
ECHA H3C
Br
O
Br
CH3
Br
Br
HO
OH
HPVIS
Br
Br
TEBP-Anh
Br
Br
TBBPA
O
ECHA
Br
O Br
ECHA
Br
O Br
O
HPVIS: High Production Volume Information System (US EPA)
Br
BTBPE Br
molecular weight Fig. 4. log Kow of brominated aromatic compounds as a function of molecular weight. Red dots indicate values where the deviation from the linear relationship from Fig. 1, left, is well over 2 log units (sources: ECHA (2013b), EPA (2013)).
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
G. Stieger et al. / Chemosphere xxx (2014) xxx–xxx
log Kow values are difficult to measure because the water solubility of these compounds is very low (Pontolillo and Eganhouse, 2001). What is more surprising is that in the ECHA database the value of 3.55 for the log Kow of DBDPE is assigned the highest reliability score (‘‘reliable without restriction’’) (ECHA, 2013b). Three other cases are very low log Kow values that were found for tetrabromophthalic anhydride (TEBP-Anh), TBBPA, and BTBPE, see Fig. 4. Again, experimental difficulties in measuring the low concentrations in the water phase are the most likely explanation of these data. The data on aquatic toxicity can be tested for plausibility by relating them to effect concentrations for baseline toxicity as indicated by the relationships between LC50,bl and log Kow in Fig. 1, right. In order to test the Ta data that we were able to find for 18 of the 36 BFRs and to judge their plausibility, we plotted the log of the effect concentration for acute aquatic toxicity in mmol L1 against our best-guess log Kow values of these 18 BFRs (Fig. 5). Also indicated in this plot are the three domains defined in Fig. 1, right (below baseline relationship by more than one log unit; consistent with baseline relationship; above baseline relationship by more than one log unit). Baseline toxicants can be expected to be located in the middle domain, whereas specifically acting chemicals exhibit so-called excess toxicity in addition to their baseline toxicity and are therefore expected to lie below the baseline toxicity relationship. The majority of the BFR Ta data follows these trends, i.e. lie below the baseline toxicity relationship or in the domain defined by the baseline toxicity relationships in Fig. 1, right. However, there are several chemicals, in particular substances with a log Kow above 7, whose Ta data are located much above the baseline toxicity relationships. From a conceptual point of view, this is not possible (every chemical exerts at least its baseline toxicity). Closer inspection of several of these cases shows different possible sources of error. In the case of PBP, there are two effect concentrations reported in different databases that are different exactly by a factor of 1000 (blue dots connected by vertical line in Fig. 5) and the higher one clearly exceeds all other effect concentrations found for this substance. The higher effect concentration is 93 mg L1, the lower one is 0.093 mg L1. It is well possible that the higher value is a typing error where the decimal point has been lost.
2
BTBPE
EBTEBPI
PBP
DBDPE
log LC50 (fish) in mmol/L
0
–2
TBBPA-BAE
–4
domain 3
–6
rainbow trout fathead minnow Japanese killfish bluegill sunfish common carp zebra fish not specified
–8
– 10
0
2
domain 1
4
6
domain 2
8
10
12
14
log K ow Fig. 5. LC50 for acute toxicity to fish of 18 alternative BFRs vs. log Kow. The three domains separated by the dashed lines are from Fig. 1, right. Black line: relationship log LC50,bl = – log Kow + 2 (Maeder et al., 2004). Colored symbols indicate different fish species tested.
5
Another case is TBBPA allyl ether (TBBPA-BAE): here the effect concentration reported for TBBPA was used without any adjustment although the Kow of TBBPA-BAE exceeds that of TBBPA by three orders of magnitude (dashed horizontal line in Fig. 5). In principle, it is an established technique to extrapolate from the properties of known substances to the properties of structurally related substances. However, in this case the structural difference (two allyl ether groups instead of two hydroxyl groups) was completely disregarded although the difference between the log Kow values of these two substances clearly indicates that this structural difference is substantial. Finally, for two (EBTEBPI and DBDPE) of the three substances with very high effect concentrations in Fig. 5, BTBPE, EBTEBPI, and DBDPE, it is stated in the databases that the values represent nominal concentrations of solid substances instead of actual concentrations of dissolved substances. For the third substance, BTBPE, this is not stated in the database but is also a likely explanation of the very high values.
4. Discussion Conceptually, the Kow is a well-defined physicochemical quantity that has a single true value. However, very low solubilities in water are difficult to measure and therefore some scatter in experimentally derived log Kow values is to be expected. However, discrepancies of several log units are too big to be accepted as scatter caused by measurement uncertainties and it is important to identify and remove inaccurate values. A key point here is to evaluate Kow values reported for a chemical in the context of values for structurally similar chemicals. For neutral organic compounds, this can be done by establishing a relationship between log Kow and molecular weight. However, these relationships are not universal but need to be derived specifically for different groups of chemicals. Another important insight from our analysis is that log Kow values are not necessarily more reliable because they were determined experimentally and according to, for example, GLP standards. Especially for (very) high log Kow values, it is advisable to use estimates derived from quantitative structure–property relationships such as KOWWIN (Meylan and Howard, 1995). KOWWIN was trained on a set of 2351 chemicals and validated on a set of 6055 chemicals (Meylan and Howard, 1995). For non-ionizing organic compounds, log Kow values estimated with KOWWIN should always be included as a reference point. The toxicity data shown in Fig. 5 also show more scatter than could be attributed to normal measurement uncertainties. To some extent, this finding may be caused by interspecies variability (the Ta data shown in Fig. 3, right, represent a range of different fish species) and by different test conditions. However, the fact that the Ta data deviate most strongly from the baseline toxicity relationship for chemicals with high Kow, i.e. low water solubility, points to some systematic problems associated with the toxicity testing of these chemicals. Effect concentrations much above the baseline toxicity do not represent the actual toxicity of these chemicals. This has been pointed out by Mayer and Reichenberg (2006), who investigated the toxicity of super-hydrophobic chemicals. They concluded that the observation of a ‘‘solubility cut-off’’ for toxicity, i.e. a threshold below which chemicals may not be soluble enough to exert any appreciable toxicity, is an artifact that occurs when the solubility of the solid chemical is considered instead of the solubility of the sub-cooled liquid. It is the sub-cooled liquid solubility that needs to be considered in toxicity assessments, and this solubility is always, even for chemicals with very high Kow, high enough for the chemicals to reach concentrations where toxic effects are caused. In the light of this analysis by Mayer and
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
6
G. Stieger et al. / Chemosphere xxx (2014) xxx–xxx
Reichenberg (2006), our Ta data compilation indicates a need for improved toxicity testing schemes for highly hydrophobic chemicals. For these chemicals, it is not appropriate to test the solid substance; a promising technique is to use passive dosing systems (Smith et al., 2010) from which the chemicals can directly diffuse into the water or the tissue of the test organism without dissolution of the solid. In any case, nominal concentrations of solid high-Kow substances do not provide any information about the toxicity of these substances and should not be provided as test results. Another reason why Ta data of high-Kow chemicals may not reflect the actual toxicity of these substances is that high-Kow chemicals have long times-to-steady-state in aquatic organisms (Escher and Hermens, 2002). The time-to-steady-state increases with increasing Kow and increasing size of the organism and may well reach several weeks (Mayer and Reichenberg, 2006). In acute toxicity tests, which last for 48–96 h, these chemicals reach only a small fraction of their steady-state concentration in the organism. Accordingly, there is a strong need for more chronic toxicity tests for high-Kow chemicals. Currently, many chemical property data are collected in large databases and made publicly available, in particular in the context of REACH. It is highly desirable that these chemical property data be critically evaluated before they are incorporated into any database. For this purpose, relationships as we used them in the current work can be used. This may require more financial resources than initially anticipated, but without this kind of evaluation, there clearly is a danger that the value of new databases is seriously reduced because of an unknown and unidentified fraction of inaccurate chemical property data. Regarding the alternative BFRs, our analysis leads to several observations: First, the availability of chemical property data for these substances is still limited; for 17 of the 36 substances, no data at all were found in the 25 databases. This implies that for these chemicals there is no sufficient basis for their use as replacements of PBDEs or HBCDD. Also for 17 of the 19 chemicals for which there were some data, the number of data is too low for a reliable hazard and risk assessment. In particular, data on degradability are scarce; half-life estimates were found for only 11 chemicals, and the number of datapoints was very low. Second, in several cases where many data points were found, the data exhibt very wide scatter. This makes identification of the most reliable values difficult and, thereby, impedes hazard and risk assessment of the alternative BFRs. Third, for Kow and aquatic toxicity in particular, there are indications that Kow data that are too low by several orders of magnitude are reported and that effect concentrations exceeding the baseline toxicity of the chemicals by several orders of magnitude are listed. These data are inaccurate and will lead to erroneous conclusions if they are used in hazard and risk assessments. Such values should be removed from the databases with high priority. Finally, before regulatory decisions can be made that alternative BFRs are safe replacements of PBDEs and HBCDD, both the amount and the quality of chemical property data for these substances need to be substantially improved.
References Bergman, Å., Rydén, A., Law, R.J., de Boer, J., Covaci, A., Alaee, M., et al., 2012. A novel abbreviation standard for organobromine, organochlorine and organophosphorus flame retardants and some characteristics of the chemicals. Environ. Int. 49, 57–82. Covaci, A., Harrad, S., Abdallah, M.A.E., Ali, N., Law, R.J., Herzke, D., de Wit, C.A., 2011. Novel brominated flame retardants: a review of their analysis, environmental fate and behaviour. Environ. Int. 37, 532–556. ECHA, 2013a. Candidate list of Substances of Very High Concern for Authorisation. European Chemicals Agency, Helsinki, . ECHA, 2013b. Registered Substances. European Chemicals Agency, Helsinki, . EPA, 2013. High Production Volume Information System. US Environmenal Protection Agency, Washington DC, . Escher, B.I., Hermens, J.L.M., 2002. Modes of action in ecotoxicology: their role in body burdens, species sensitivity, QSARs, and mixture effects. Environ. Sci. Technol. 36, 4201–4217. He, M.-J., Luo, X.-J., Chen, M.-Y., Sun, Y.-X., Chen, S.-J., Mai, B.-X., 2012. Bioaccumulation of polybrominated diphenyl ethers and decabromodiphenyl ethane in fish from a river system in a highly industrialized area, South China. Sci. Total Environ. 419, 109–115. Könemann, H., 1981. Quantitative structure-activity relationships in fish toxicity studies. Toxicol 19, 209–221. Kuramochi, H., Maeda, K., Kawamoto, K., 2004. Water solubility and partitioning behavior of brominated phenols. Environ. Toxicol. Chem. 23, 1386–1393. Maeder, V., Escher, B.I., Scheringer, M., Hungerbühler, K., 2004. Toxic ratio as an indicator of the intrinsic toxicity in the assessment of persistent, bioaccumulative, and toxic chemicals. Environ. Sci. Technol. 38, 3659–3666. Mayer, P., Reichenberg, F., 2006. Can highly hydrophobic organic substances cause aquatic baseline toxicity and can they contribute to mixture toxicity? Environ. Toxicol. Chem. 25, 2639–2644. McCarty, L.S., Mackay, D., Smith, A.D., Ozburn, A.D., Dixon, D.G., 1992. Residuebased interpretation of toxicity and bioconcentration QSARs from aquatic bioassays: neutral narcotic organics. Environ. Toxicol. Chem. 11, 917–930. Meylan, W.M., Howard, P.H., 1995. Atom/fragment contribution method for estimating octanol–water partition coefficients. J. Pharm. Sci. 84, 83–92. Pontolillo, J., Eganhouse, R.P., 2001. The search for reliable aqueous solubility (Sw) and octanol–water partition coefficient (Kow) data for hydrophobic organic compounds: DDT and DDE as a case study. U.S. Geological Survey, Washington, DC. Schenker, U., MacLeod, M., Scheringer, M., Hungerbühler, K., 2005. Improving data quality for environmental fate models: a least-squares adjustment procedure for harmonizing physicochemical properties of organic compounds. Environ. Sci. Technol. 39, 8434–8441. Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Environmental Organic Chemistry. John Wiley, New York. Smith, K.E.C., Dom, N., Blust, R., Mayer, P., 2010. Controlling and maintaining exposure of hydrophobic organic compounds in aquatic toxicity tests by passive dosing. Aquat. Toxicol. 98, 15–24. Sotomatsu, T., Shigemura, M., Murata, Y., Fujita, T., 1993. Octanol–water partition coefficient of ortho-substituted aromatic solutes. J. Pharm. Sci. 82, 776–781. Strempel, S., Scheringer, M., Ng, C.A., Hungerbühler, K., 2012. Screening for PBT chemicals among the ‘‘existing’’ and ‘‘new’’ chemicals of the EU. Environ. Sci. Technol. 46, 5680–5687. UNEP (2013) Stockholm convention on persistent organic pollutants. . Veith, G.D., Call, D.J., Brooke, L.T., 1983. Structure-toxicity relationships for the fathead minnow, Pimephales promelas: narcotic industrial chemicals. Can. J. Fish. Aquat. Sci. 40, 743–748. Verhaar, H.J.M., Mulder, W., Hermens, J.L.M., Rorije, E., Langenberg, J.H., Peijnenburg, W.J.G.M., et al. 1995. Overview of structure-activity relationships for environmental endpoints, Part 1. Report of the EU-DG-XII Project QSAR for predicting fate and effects of chemicals in the environment. WHO, 2005, 2005. Concise International Chemical Assessment Document 66. 2,4,6Tribromophenol and other Simple Brominated Phenols. World Health Organization, Geneva.
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants Greta Stieger, Martin Scheringer ⇑, Carla A. Ng, Konrad Hungerbühler Institute for Chemical and Bioengineering, ETH Zürich, 8093 Zürich, Switzerland
a r t i c l e
i n f o
Article history: Received 14 October 2013 Received in revised form 11 January 2014 Accepted 15 January 2014 Available online xxxx Keywords: Brominated flame retardants PBT assessment Baseline toxicity Octanol–water partition coefficient
a b s t r a c t Polybrominated diphenylethers (PBDEs) and hexabromocyclododecane (HBCDD) are major brominated flame retardants (BFRs) that are now banned or under restrictions in many countries because of their persistence, bioaccumulation potential and toxicity (PBT properties). However, there is a wide range of alternative BFRs, such as decabromodiphenyl ethane and tribromophenol, that are increasingly used as replacements, but which may possess similar hazardous properties. This necessitates hazard and risk assessments of these compounds. For a set of 36 alternative BFRs, we searched 25 databases for chemical property data that are needed as input for a PBT assessment. These properties are degradation half-life, bioconcentration factor (BCF), octanol–water partition coefficient (Kow), and toxic effect concentrations in aquatic organisms. For 17 of the 36 substances, no data at all were found for these properties. Too few persistence data were available to even assess the quality of these data in a systematic way. The available data for Kow and toxicity show surprisingly high variability, which makes it difficult to identify the most reliable values. We propose methods for systematic evaluations of PBT-related chemical property data that should be performed before data are included in publicly available databases. Using these methods, we evaluated the data for Kow and toxicity in more detail and identified several inaccurate values. For most of the 36 alternative BFRs, the amount and the quality of the PBT-related property data need to be improved before reliable hazard and risk assessments of these substances can be performed. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Brominated flame retardants (BFRs) comprise a wide range of brominated aromatic and aliphatic compounds (Bergman et al. 2012). Two types of BFRs widely used in the last decades are polybrominated diphenylethers (PBDEs) and hexabromocyclododecane (HBCDD). Because of their hazardous properties regarding persistence, bioaccumulation potential and toxicity (PBT properties), the use of these substances is now restricted or prohibited under various national and international legislations (UNEP, 2013; ECHA, 2013a). However, PBDEs and HBCDD are often replaced by other BFRs such as decabromodiphenyl ethane, pentabromotoluene, pentabromoethylbenzene, and many more. Often, the chemical properties of these substitutes are only poorly known; thus concerns about their PBT characteristics remain (Covaci et al., 2011). PBT assessment is a component of chemicals assessment that focuses on three hazardous properties of organic chemicals. These ⇑ Corresponding author. Tel.: +41 44 632 3062; fax: +41 44 632 1189.
properties include persistence (P), potential for bioconcentration and bioaccumulation (B), and aquatic toxicity (T); Strempel et al. (2012) provide an overview of several PBT assessment schemes. Data required for a PBT assessment include degradation half-lives in water or soil for P, octanol–water partition coefficients, bioconcentration and, in some cases, also bioaccumulation factors for B, and toxic effect concentrations from chronic toxicity tests or, if no chronic toxicity data are available, acute tests for T (Strempel et al., 2012). Here we evaluate the availability and quality of chemical property data that are needed for a PBT assessment of alternative BFRs. We selected 36 BFRs that were listed by Bergman et al. (2012) and also found in the PBT database compiled by Strempel et al. (2012). To retrieve chemical property data for these chemicals, we extensively searched 25 publicly accessible databases. On this basis, we analyze the number of data points for each property and chemical, discuss the variability and scatter in the data, and evaluate the data for the octanol–water partition coefficient (Kow) and the aquatic toxicity in more detail.
E-mail address: [email protected] (M. Scheringer). http://dx.doi.org/10.1016/j.chemosphere.2014.01.083 0045-6535/Ó 2014 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
2
G. Stieger et al. / Chemosphere xxx (2014) xxx–xxx
2. Materials and methods The BFRs investigated are listed in Table 1, which is based on information from Bergman et al. (2012). For these 36 chemicals we retrieved chemical property data from 25 publicly accessible databases, including the database of the European Chemicals Agency (ECHA), http://www.echa.eu; and the OECD eChemPortal, which provides links to 24 participating data sources, http:// www.echemportal.org. The property data include biodegradation half-life in soil, t1/2,soil, for persistence, the octanol–water partition coefficient, Kow, and the bioconcentration factor, BCF, for bioaccumulation, and chronic and acute effect concentrations, Tc and Ta, for aquatic organisms (algae, daphnia, fish) for toxicity. The databases were searched by CAS number for each of the 36 BFRs and all available data entries for the PBT properties, including both experimental and estimated values, were collected and stored. The data were retrieved in November and December 2012. We determined the numbers of data points that were found for each chemical and property. For chemicals with several data points per property (Ta, BCF, Kow), we plotted the data in order to visualize the variability of the data. For the Kow and toxicity data in particular, we evaluated the plausibility of individual data points. The Kow of organic chemicals is largely determined by the chemicals’ solubility in water, Sw (Schwarzenbach et al., 2003). Sw is determined by the energy that is required to accommodate the solute in water, i.e. to break hydrogen bonds between water molecules and form a cavity that can take up the solute molecule. For neutral or weakly polar organic chemicals, this amount of energy is directly proportional to the molecular weight. In the case of polychlorinated biphenyls (PCBs), log Kow follows a highly significant positive linear relationship with increasing degree of chlorination and thus
increasing molecular weight (Schenker et al., 2005). Many brominated aromatic compounds in our set are also non-ionizing organic substances; for substances with phenolic groups, the Kow is measured at sufficiently low pH where ionization is suppressed and the substances are predominantly present in their neutral form (Sotomatsu et al., 1993; Kuramochi et al., 2004). Therefore, we expect that a similar relationship can be derived for the brominated aromatic substances in our set of 36 alternative BFRs. To establish this relationship, we looked for series of compounds with similar aromatic backbones, in analogy to the set of PCB congeners. From the set of 36 BFRs, two such subsets were available: (i) 14 monoaromatic BFRs and (ii) 9 di-aromatic structures, including TBBPA, 6 TBBPA derivatives, as well as DBDPE and BTBPE. Together, these two sets include 23 of the 36 BFRs. Because Kow values were available for only 11 of these 23 substances, we included the log Kow values collected by Strempel et al. (2012) and Bergman et al. (2012) in our analysis. Because the log Kow values for these 23 compounds showed a high degree of variability, we implemented a two-step process to find the best relationship between log Kow and molecular weight. First, we used all of the log Kow data to derive a linear relationship with molecular weight. We then used this linear relationship to determine, for each substance, which log Kow values seemed implausible because they were either too low or too high. In general, for chemicals where the variability of the Kow data was high, the lower log Kow values were (much) below the linear relationship whereas most of the higher ones were in line with it. In the second step, we excluded those log Kow values that were clearly below or above the linear relationship and calculated for each chemical the ‘‘best guess’’ log Kow as the average of the remaining log Kow values (these ‘‘best guess’’ estimates are the data points shown in Fig. 1, left).
Table 1 CAS numbers, common names, and practical abbreviations proposed by Bergman et al. (2012) of the 36 BFRs investigated in this work. Count
CAS number
Common name
Practical abbreviation
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
79-94-7 632-79-1 52434-90-9 66710-97-2 59447-55-1 34571-16-9 26040-51-7 39635-79-5 32588-76-4 38521-51-6 20566-35-2 87-82-1 3194-57-8 3278-89-5 25327-89-3 21850-44-2 87-83-2 3296-90-0 3322-93-8 85-22-3 25713-60-4 37853-59-1 84852-53-9 615-58-7 35109-60-5 37853-61-5 33798-02-6 3555-11-1 126-72-7 4162-45-2 42757-55-1 23488-38-2 608-71-9 51936-55-1 118-79-6 39569-21-6
Tetrabromobisphenol A 3,4,5,6-Tetrabromophthalic anhydride 1,3,5-Tris(2,3-dibromopropyl)-1,3,5-triazine-2,4,6-trione Tetrabromobisphenol A bis(2-hydroxyethyl)ether bisacrylate Pentabromobenzyl acrylate 1,2,3,4,7,7-Hexachloro-5-(2,3,4,5-tetrabromophenyl)-bicyclo[2.2.1]hept-2-ene Bis(2-ethylhexyl) tetrabromophthalate Tetrabromobisphenol S N,N0 -Ethylenebis (tetrabromophthalimide) Pentabromobenzyl bromide 2-(2-Hydroxyethoxy)ethyl 2-hydroxypropyl 3,4,5,6-tetrabromophthalate Hexabromobenzene 1,2,5,6-Tetrabromo-cyclooctane 2,4,6-Tribromophenyl allyl ether Tetrabromobisphenol A bis(allyl) ether Tetrabromobisphenol A bis(2,3-dibromopropyl) ether Pentabromotoluene Dibromoneopentyl glycol 4-(1,2-Dibromoethyl)-1,2-dibromocyclohexane Pentabromoethylbenzene Tris(2,4,6-tribromophenoxy)-s-triazine 1,2-Bis(2,4,6-tribromophenoxy)ethane Decabromodiphenyl ethane 2,4-Dibromophenol 2,4,6-Tribromophenyl 2,3-dibromopropyl ether Tetrabromobisphenol A bismethyl ether 3,30 ,5,50 -Tetrabromobisphenol A bisacetate Pentabromophenol allyl ether Tris(2,3-dibromopropyl) phosphate Tetrabromobisphenol A bis(2-hydroxyethyl) ether Tetrabromobisphenol S bis(2,3-dibromopropyl ether) 1,2,4,5-Tetrabromo-3,6-dimethylbenzene Pentabromophenol 5,6-Dibromo-1,10,11,12,13,13-hexachloro-11-tricyclo[8.2.1.02,9]tridecene 2,4,6-Tribromophenol 2,3,4,5-Tetrabromo-6-chlorotoluene
TBBPA TEBP-Anh TDBP-TAZTO TBBPA-BHEEBA PBB-Acr HCTBPH BEH-TEBP TBBPS EBTEBPI PBBB HEEHP-TEBP HBB TBCO TBP-AE TBBPA-BAE TBBPA-BDBPE PBT DBNPB DBE-DBCH PBEB TTBP-TAZ BTBPE DBDPE DBP TBP-DBPE TBBPA-BME TBBPA-BOAc PBP-AE TDBPP TBBPA-BHEE TBBPS-BDBPE TBX PBP DBHCTD TBP TBCT
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
3
G. Stieger et al. / Chemosphere xxx (2014) xxx–xxx
Excluded were 7 estimated values (4 too low, 3 too high), 7 measured values (all too low), and 4 values without any information about the method used (all 4 too low). Finally, we made a linear regression through the 23 best-guess log Kow values (see Fig. 1, left) and found the following equation for the correlation: log Kow = 0.0125MW + 0.3034, R2 = 0.88. For the 23 brominated aromatic substances in our set, we determined the deviation of individual data points from the value predicted by the relationship between log Kow and MW. Data points that deviate by several orders of magnitude from the value derived from the molecular weight were considered unreliable. To evaluate the data for aquatic toxicity, we used established relationships between LC50 for baseline toxicity and log Kow. Baseline toxicity or narcosis is caused by the partitioning of a chemical into the cell membrane lipids; it is an unspecific effect and represents the minimum toxicity that every chemical exerts (Escher and Hermens, 2002). Because it is determined by a chemical’s tendency to partition from water into lipids, the baseline toxicity of chemicals increases with increasing log Kow. The LC50 for baseline toxicity (denoted by LC50,bl in the following) can, therefore, be derived from the log Kow. If a chemical acts according to a specific mode of toxic action, this is an additional effect on top of the chemical’s baseline toxicity. In such a case, the actual LC50 of the chemical is lower than the LC50 for baseline toxicity derived from the log Kow. For a plot of LC50 data vs. log Kow, this implies that measured LC50 data can be expected to be close to the line defined by the relationship between log Kow and LC50,bl or, if the chemical acts according to a specific mode of toxic action, below this line. In contrast, LC50 data that are much higher than the LC50,bl derived from the log Kow are likely to be unreliable or incorrect. Fig. 1, right, shows six relationships between LC50 for baseline toxicity in fish and log Kow from the literature (Könemann, 1981; Veith et al., 1983; McCarty et al., 1992; Verhaar et al., 1995; Maeder et al., 2004). As pointed out by Mayer and Reichenberg (2006), the relationship between LC50,bl and log Kow also applies to high-Kow chemicals. Also shown in Fig. 1, right, are three domains of LC50 data: more toxic than baseline toxicity (more than 1 order of magnitude below the lowest baseline toxicity relationship); baseline toxicity (less than 1 order of magnitude below the lowest or above the highest baseline toxicity relationship); and inconclusive/incorrect (more than 1 order of magnitude above the highest baseline toxicity relationship).
values are persistence data (see Fig. 2, left). Values for Kow and BCF were found for all 19 of the 36 BFRs for which any data were available; values for t1/2,soil were found for 11 of the 36 BFRs, and values for Tc and Ta for 14 and 18 of the 36 BFRs, respectively. For all of these PBT properties, there are a few BFRs for which multiple values for the same property can be found, whereas for the other BFRs only one or no values are available for any of the properties (see Fig. 2, right). Among the few BFRs for which multiple entries are available for all PBT properties are TBBPA, CAS 79-94-7, and TBP, CAS 118-79-6. TBBPA is one of the most commonly used BFRs today (Bergman et al., 2012) and TBP is produced in high production volumes (>1000 t yr1) (WHO, 2005). In terms of data availability, we conclude that property data are scarce for persistence. Data on bioaccumulation potential and toxicity are more abundant; however, the data are unequally distributed: for some BFRs, in particular ones with high production volumes and extensive applications, many data are available whereas for most alternative BFRs only very little information is available. 3.2. Data variability Where multiple database entries were found for the same BFR and PBT property, the variability of the values is often substantial. As examples, the Kow values found for TBBPA range over five orders of magnitude (see Fig. 3, left, CAS 79-94-7) and the Ta values found for pentabromophenol (PBP) range over almost five orders of magnitude (see Fig. 3, right, CAS 608-71-9). This high variability and thus inconsistency of the available data raises questions about the quality of the data. 3.3. Data plausibility Plotting log Kow values against the BFRs’ molecular weight makes it possible to judge the plausibility of the available data, especially in cases where multiple and widely different data points are available for the same substance. The log Kow values retrieved from the various databases for the set of 23 aromatic BFRs are largely consistent with the linear relationship between log Kow and molecular weight. However, there are a few exceptions, see Fig. 4. One exception is the log Kow value of 3.55 provided by the ECHA database for DBDPE (CAS 84852-539, MW = 971 g mol1), which is too low by more than 8 log units (i.e. 8 orders of magnitude for the Kow) and was removed from the calculation of the best-guess log Kow for DBDPE, which is 12.3. For DBDPE, a high bioaccumulation factor, BAF, with log BAF values between 6.1 and 7.1 has been reported (He et al., 2012), which is consistent with a high log Kow. It is well known that high
3. Results 3.1. Data quantity Overall, data were available for 19 of the 36 BFRs. 47% of all collected data points are for Ta, whereas only 5% of the available 2
domain 3: effect concentrations above baseline toxicity: suspect
log K ow
log LC50,bl (mmol/L)
0
Verhaar et al. (1995) Könemann (1981) Verhaar et al. (1995) McCarthy et al. (1992) Veith et al. (1983) Maeder et al. (2004)
–2 –4 –6 domain 1: effect concentrations below baseline toxicity: specific toxicity
–8
domain 2: baseline toxicity
– 10
0 molecular weight, MW
2
4
6 8 log Kow
10
12
14
Fig. 1. Left: log Kow vs. molecular weight for 23 aromatic BFRs; symbols indicate best-guess values. Red line: relationship obtained by linear regression, log Kow = 0.0125MW + 0.3034. Right: six relationships between LC50 for baseline toxicity (in mmol L1) and log Kow as well as three domains of LC50 data in the plot of LC50 vs. log Kow.
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
G. Stieger et al. / Chemosphere xxx (2014) xxx–xxx
total number of datapoints found
4
46.7% 13.6%
19.9% 15.5%
4.3%
Kow
B
P
Tc
Ta CAS number
log Kow
effect concentration for acute aquatic toxicity (mg/L)
Fig. 2. Left: distribution of the total number of database entries found among the different PBT properties. Right: total number of database entries found for each of the 36 BFRs.
CAS number
TBBPA
CAS number
PBP
Fig. 3. Left: variability of database entries found for log Kow per chemical (chemicals indicated by CAS number). Right: variability of database entries found for acute aquatic toxicity (in mg L1) per chemical (chemicals indicated by CAS number). TBBPA (left) and PBP (right) are highlighted in gray as examples (see text).
14
Br
Br
DBDPE
log K ow
Br
Br
Br Br Br
Br Br
Br
ECHA: database of substances registered under REACH
ECHA H3C
Br
O
Br
CH3
Br
Br
HO
OH
HPVIS
Br
Br
TEBP-Anh
Br
Br
TBBPA
O
ECHA
Br
O Br
ECHA
Br
O Br
O
HPVIS: High Production Volume Information System (US EPA)
Br
BTBPE Br
molecular weight Fig. 4. log Kow of brominated aromatic compounds as a function of molecular weight. Red dots indicate values where the deviation from the linear relationship from Fig. 1, left, is well over 2 log units (sources: ECHA (2013b), EPA (2013)).
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
G. Stieger et al. / Chemosphere xxx (2014) xxx–xxx
log Kow values are difficult to measure because the water solubility of these compounds is very low (Pontolillo and Eganhouse, 2001). What is more surprising is that in the ECHA database the value of 3.55 for the log Kow of DBDPE is assigned the highest reliability score (‘‘reliable without restriction’’) (ECHA, 2013b). Three other cases are very low log Kow values that were found for tetrabromophthalic anhydride (TEBP-Anh), TBBPA, and BTBPE, see Fig. 4. Again, experimental difficulties in measuring the low concentrations in the water phase are the most likely explanation of these data. The data on aquatic toxicity can be tested for plausibility by relating them to effect concentrations for baseline toxicity as indicated by the relationships between LC50,bl and log Kow in Fig. 1, right. In order to test the Ta data that we were able to find for 18 of the 36 BFRs and to judge their plausibility, we plotted the log of the effect concentration for acute aquatic toxicity in mmol L1 against our best-guess log Kow values of these 18 BFRs (Fig. 5). Also indicated in this plot are the three domains defined in Fig. 1, right (below baseline relationship by more than one log unit; consistent with baseline relationship; above baseline relationship by more than one log unit). Baseline toxicants can be expected to be located in the middle domain, whereas specifically acting chemicals exhibit so-called excess toxicity in addition to their baseline toxicity and are therefore expected to lie below the baseline toxicity relationship. The majority of the BFR Ta data follows these trends, i.e. lie below the baseline toxicity relationship or in the domain defined by the baseline toxicity relationships in Fig. 1, right. However, there are several chemicals, in particular substances with a log Kow above 7, whose Ta data are located much above the baseline toxicity relationships. From a conceptual point of view, this is not possible (every chemical exerts at least its baseline toxicity). Closer inspection of several of these cases shows different possible sources of error. In the case of PBP, there are two effect concentrations reported in different databases that are different exactly by a factor of 1000 (blue dots connected by vertical line in Fig. 5) and the higher one clearly exceeds all other effect concentrations found for this substance. The higher effect concentration is 93 mg L1, the lower one is 0.093 mg L1. It is well possible that the higher value is a typing error where the decimal point has been lost.
2
BTBPE
EBTEBPI
PBP
DBDPE
log LC50 (fish) in mmol/L
0
–2
TBBPA-BAE
–4
domain 3
–6
rainbow trout fathead minnow Japanese killfish bluegill sunfish common carp zebra fish not specified
–8
– 10
0
2
domain 1
4
6
domain 2
8
10
12
14
log K ow Fig. 5. LC50 for acute toxicity to fish of 18 alternative BFRs vs. log Kow. The three domains separated by the dashed lines are from Fig. 1, right. Black line: relationship log LC50,bl = – log Kow + 2 (Maeder et al., 2004). Colored symbols indicate different fish species tested.
5
Another case is TBBPA allyl ether (TBBPA-BAE): here the effect concentration reported for TBBPA was used without any adjustment although the Kow of TBBPA-BAE exceeds that of TBBPA by three orders of magnitude (dashed horizontal line in Fig. 5). In principle, it is an established technique to extrapolate from the properties of known substances to the properties of structurally related substances. However, in this case the structural difference (two allyl ether groups instead of two hydroxyl groups) was completely disregarded although the difference between the log Kow values of these two substances clearly indicates that this structural difference is substantial. Finally, for two (EBTEBPI and DBDPE) of the three substances with very high effect concentrations in Fig. 5, BTBPE, EBTEBPI, and DBDPE, it is stated in the databases that the values represent nominal concentrations of solid substances instead of actual concentrations of dissolved substances. For the third substance, BTBPE, this is not stated in the database but is also a likely explanation of the very high values.
4. Discussion Conceptually, the Kow is a well-defined physicochemical quantity that has a single true value. However, very low solubilities in water are difficult to measure and therefore some scatter in experimentally derived log Kow values is to be expected. However, discrepancies of several log units are too big to be accepted as scatter caused by measurement uncertainties and it is important to identify and remove inaccurate values. A key point here is to evaluate Kow values reported for a chemical in the context of values for structurally similar chemicals. For neutral organic compounds, this can be done by establishing a relationship between log Kow and molecular weight. However, these relationships are not universal but need to be derived specifically for different groups of chemicals. Another important insight from our analysis is that log Kow values are not necessarily more reliable because they were determined experimentally and according to, for example, GLP standards. Especially for (very) high log Kow values, it is advisable to use estimates derived from quantitative structure–property relationships such as KOWWIN (Meylan and Howard, 1995). KOWWIN was trained on a set of 2351 chemicals and validated on a set of 6055 chemicals (Meylan and Howard, 1995). For non-ionizing organic compounds, log Kow values estimated with KOWWIN should always be included as a reference point. The toxicity data shown in Fig. 5 also show more scatter than could be attributed to normal measurement uncertainties. To some extent, this finding may be caused by interspecies variability (the Ta data shown in Fig. 3, right, represent a range of different fish species) and by different test conditions. However, the fact that the Ta data deviate most strongly from the baseline toxicity relationship for chemicals with high Kow, i.e. low water solubility, points to some systematic problems associated with the toxicity testing of these chemicals. Effect concentrations much above the baseline toxicity do not represent the actual toxicity of these chemicals. This has been pointed out by Mayer and Reichenberg (2006), who investigated the toxicity of super-hydrophobic chemicals. They concluded that the observation of a ‘‘solubility cut-off’’ for toxicity, i.e. a threshold below which chemicals may not be soluble enough to exert any appreciable toxicity, is an artifact that occurs when the solubility of the solid chemical is considered instead of the solubility of the sub-cooled liquid. It is the sub-cooled liquid solubility that needs to be considered in toxicity assessments, and this solubility is always, even for chemicals with very high Kow, high enough for the chemicals to reach concentrations where toxic effects are caused. In the light of this analysis by Mayer and
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
6
G. Stieger et al. / Chemosphere xxx (2014) xxx–xxx
Reichenberg (2006), our Ta data compilation indicates a need for improved toxicity testing schemes for highly hydrophobic chemicals. For these chemicals, it is not appropriate to test the solid substance; a promising technique is to use passive dosing systems (Smith et al., 2010) from which the chemicals can directly diffuse into the water or the tissue of the test organism without dissolution of the solid. In any case, nominal concentrations of solid high-Kow substances do not provide any information about the toxicity of these substances and should not be provided as test results. Another reason why Ta data of high-Kow chemicals may not reflect the actual toxicity of these substances is that high-Kow chemicals have long times-to-steady-state in aquatic organisms (Escher and Hermens, 2002). The time-to-steady-state increases with increasing Kow and increasing size of the organism and may well reach several weeks (Mayer and Reichenberg, 2006). In acute toxicity tests, which last for 48–96 h, these chemicals reach only a small fraction of their steady-state concentration in the organism. Accordingly, there is a strong need for more chronic toxicity tests for high-Kow chemicals. Currently, many chemical property data are collected in large databases and made publicly available, in particular in the context of REACH. It is highly desirable that these chemical property data be critically evaluated before they are incorporated into any database. For this purpose, relationships as we used them in the current work can be used. This may require more financial resources than initially anticipated, but without this kind of evaluation, there clearly is a danger that the value of new databases is seriously reduced because of an unknown and unidentified fraction of inaccurate chemical property data. Regarding the alternative BFRs, our analysis leads to several observations: First, the availability of chemical property data for these substances is still limited; for 17 of the 36 substances, no data at all were found in the 25 databases. This implies that for these chemicals there is no sufficient basis for their use as replacements of PBDEs or HBCDD. Also for 17 of the 19 chemicals for which there were some data, the number of data is too low for a reliable hazard and risk assessment. In particular, data on degradability are scarce; half-life estimates were found for only 11 chemicals, and the number of datapoints was very low. Second, in several cases where many data points were found, the data exhibt very wide scatter. This makes identification of the most reliable values difficult and, thereby, impedes hazard and risk assessment of the alternative BFRs. Third, for Kow and aquatic toxicity in particular, there are indications that Kow data that are too low by several orders of magnitude are reported and that effect concentrations exceeding the baseline toxicity of the chemicals by several orders of magnitude are listed. These data are inaccurate and will lead to erroneous conclusions if they are used in hazard and risk assessments. Such values should be removed from the databases with high priority. Finally, before regulatory decisions can be made that alternative BFRs are safe replacements of PBDEs and HBCDD, both the amount and the quality of chemical property data for these substances need to be substantially improved.
References Bergman, Å., Rydén, A., Law, R.J., de Boer, J., Covaci, A., Alaee, M., et al., 2012. A novel abbreviation standard for organobromine, organochlorine and organophosphorus flame retardants and some characteristics of the chemicals. Environ. Int. 49, 57–82. Covaci, A., Harrad, S., Abdallah, M.A.E., Ali, N., Law, R.J., Herzke, D., de Wit, C.A., 2011. Novel brominated flame retardants: a review of their analysis, environmental fate and behaviour. Environ. Int. 37, 532–556. ECHA, 2013a. Candidate list of Substances of Very High Concern for Authorisation. European Chemicals Agency, Helsinki, . ECHA, 2013b. Registered Substances. European Chemicals Agency, Helsinki, . EPA, 2013. High Production Volume Information System. US Environmenal Protection Agency, Washington DC, . Escher, B.I., Hermens, J.L.M., 2002. Modes of action in ecotoxicology: their role in body burdens, species sensitivity, QSARs, and mixture effects. Environ. Sci. Technol. 36, 4201–4217. He, M.-J., Luo, X.-J., Chen, M.-Y., Sun, Y.-X., Chen, S.-J., Mai, B.-X., 2012. Bioaccumulation of polybrominated diphenyl ethers and decabromodiphenyl ethane in fish from a river system in a highly industrialized area, South China. Sci. Total Environ. 419, 109–115. Könemann, H., 1981. Quantitative structure-activity relationships in fish toxicity studies. Toxicol 19, 209–221. Kuramochi, H., Maeda, K., Kawamoto, K., 2004. Water solubility and partitioning behavior of brominated phenols. Environ. Toxicol. Chem. 23, 1386–1393. Maeder, V., Escher, B.I., Scheringer, M., Hungerbühler, K., 2004. Toxic ratio as an indicator of the intrinsic toxicity in the assessment of persistent, bioaccumulative, and toxic chemicals. Environ. Sci. Technol. 38, 3659–3666. Mayer, P., Reichenberg, F., 2006. Can highly hydrophobic organic substances cause aquatic baseline toxicity and can they contribute to mixture toxicity? Environ. Toxicol. Chem. 25, 2639–2644. McCarty, L.S., Mackay, D., Smith, A.D., Ozburn, A.D., Dixon, D.G., 1992. Residuebased interpretation of toxicity and bioconcentration QSARs from aquatic bioassays: neutral narcotic organics. Environ. Toxicol. Chem. 11, 917–930. Meylan, W.M., Howard, P.H., 1995. Atom/fragment contribution method for estimating octanol–water partition coefficients. J. Pharm. Sci. 84, 83–92. Pontolillo, J., Eganhouse, R.P., 2001. The search for reliable aqueous solubility (Sw) and octanol–water partition coefficient (Kow) data for hydrophobic organic compounds: DDT and DDE as a case study. U.S. Geological Survey, Washington, DC. Schenker, U., MacLeod, M., Scheringer, M., Hungerbühler, K., 2005. Improving data quality for environmental fate models: a least-squares adjustment procedure for harmonizing physicochemical properties of organic compounds. Environ. Sci. Technol. 39, 8434–8441. Schwarzenbach, R.P., Gschwend, P.M., Imboden, D.M., 2003. Environmental Organic Chemistry. John Wiley, New York. Smith, K.E.C., Dom, N., Blust, R., Mayer, P., 2010. Controlling and maintaining exposure of hydrophobic organic compounds in aquatic toxicity tests by passive dosing. Aquat. Toxicol. 98, 15–24. Sotomatsu, T., Shigemura, M., Murata, Y., Fujita, T., 1993. Octanol–water partition coefficient of ortho-substituted aromatic solutes. J. Pharm. Sci. 82, 776–781. Strempel, S., Scheringer, M., Ng, C.A., Hungerbühler, K., 2012. Screening for PBT chemicals among the ‘‘existing’’ and ‘‘new’’ chemicals of the EU. Environ. Sci. Technol. 46, 5680–5687. UNEP (2013) Stockholm convention on persistent organic pollutants. . Veith, G.D., Call, D.J., Brooke, L.T., 1983. Structure-toxicity relationships for the fathead minnow, Pimephales promelas: narcotic industrial chemicals. Can. J. Fish. Aquat. Sci. 40, 743–748. Verhaar, H.J.M., Mulder, W., Hermens, J.L.M., Rorije, E., Langenberg, J.H., Peijnenburg, W.J.G.M., et al. 1995. Overview of structure-activity relationships for environmental endpoints, Part 1. Report of the EU-DG-XII Project QSAR for predicting fate and effects of chemicals in the environment. WHO, 2005, 2005. Concise International Chemical Assessment Document 66. 2,4,6Tribromophenol and other Simple Brominated Phenols. World Health Organization, Geneva.
Please cite this article in press as: Stieger, G., et al. Assessing the persistence, bioaccumulation potential and toxicity of brominated flame retardants: Data availability and quality for 36 alternative brominated flame retardants. Chemosphere (2014), http://dx.doi.org/10.1016/j.chemosphere.2014.01.083
