Environ Sci Pollut Res DOI 10.1007/s11356-014-3262-4
RESEARCH ARTICLE
Occurrence and fate of four novel brominated flame retardants in wastewater treatment plants M. Kim & P. Guerra & M. Alaee & S. A. Smyth
Received: 17 October 2013 / Accepted: 25 June 2014 # Her Majesty the Queen in Right of Canada 2014
Abstract Four novel brominated flame retardants (NBFRs), i.e., decabromodiphenylethane (DBDPE), 1,2-bis(2,4,6tribromophenoxy) ethane (BTBPE), pentabromoethylbenzene (PBEB), and hexabromobenzene (HBB) were studied in 377 liquid samples and 288 solid samples collected from 20 wastewater treatment plants. Lagoon, primary, secondary, and advanced treatment processes were included, in order to investigate NBFR occurrence and the effects of WWTP operational conditions on NBFR removal. Median influent and effluent levels were 14 to 3,700 and 1.0 to 180 pg/L respectively, with DBDPE being the highest in both. Overall median removal efficiencies for DBDPE, BTBPE, HBB, and PBEB across all process types were 81 to 93, 76 to 98, 61 to 97, and 54 to 97 %, respectively with advanced treatment processes obtaining the best removals. NBFRs removal was related to retention time, surface loading rate, and biomass concentration. Median NBFR levels in treated biosolids were 80 to 32,000 pg/g, influenced by solids treatment processes. Keywords Wastewater treatment plants . Biosolids . Brominated flame retardants . DBDPE . BTBPE . PBEB . HBB
Introduction Brominated flame retardants (BFRs) are extensively used in commercial products to delay combustion; their production Responsible editor: Philippe Garrigues Electronic supplementary material The online version of this article (doi:10.1007/s11356-014-3262-4) contains supplementary material, which is available to authorized users. M. Kim : P. Guerra : M. Alaee : S. A. Smyth (*) Science and Technology Branch, Environment Canada, 867 Lakeshore Road, P.O. Box 5050, Burlington, Canada L7R 4A6 e-mail: [email protected]
accounts for 20 to 25 % of the global production volume of flame retardants (Harju et al. 2009). Novel BFRs such as decabromodiphenylethane (DBDPE; C14H4Br10) and 1,2-bis (2,4,6-tribromophenoxy) ethane (BTBPE; C14H8Br6O2) are available for use in Canada. DBDPE is used in high impact polystyrene (HIPS) and cable coverings, while BTBPE has been applied to high impact polystyrenes and thermoplastics (de Wit et al. 2010). In addition to these two compounds, some formerly manufactured minor BFRs have reappeared on the global market, including pentabromoethylbenzene (PBEB; C8H5Br5) for textile, cable coating, and polyurethane foam application and hexabromobenzene (HBB; C6Br6) for uses in paper, woods, textiles, electronic, and plastic goods (Harju et al. 2009; Covaci et al. 2011; Salamova and Hites 2011). The production volume of these four novel BFRs (NBFRs) varied between compounds and region: for DBDPE, it was 4,500 to 23,000 tonnes (USA in 2005) and 12,000 tonnes (China in 2006); for BTBPE, it was 50 % accounted for 12 to 34 % of the dataset for liquid samples and 11 to 20 % for solid samples; therefore, variability in NBFR concentrations was generally less than 50 % for both wastewater and solids. NBFRs in the liquid stream NBFR levels The statistical distributions of the four NBFRs in raw influent, primary effluent, and final effluents are summarized in Table 1. There were 145, 86, and 146 samples analyzed for influents, primary effluents, and final effluents, respectively. NBFR levels were below DL in 35 to 80 % of total DBDPE analyses, 0 to 21 % of BTBPE analyses, 1 to 29 % of HBB analyses, and 8 to 61 % of PBEB analyses. Therefore, statistical methods handling for censored data, i.e., ROS and Kaplan–Meier were used to estimate statistical distributions of NBFR levels. The frequency of nondetects varied widely from 0 to 80 % over NBFR and sample type, which may be caused by varying concentrations, high DL, and matrix effects. The large dataset produced in this study adds to the reliability of the results available in the literature by reducing the dominance of nondetects in the dataset and producing more representative statistical summaries. To assess the effectiveness of the use of statistical methods instead of substitutions, statistical summaries of datasets with nondetection frequencies of 21 to 80 % were compared between different substitution methods using zero, 1/3 DL, 1/2 DL, 2/3 DL, and DL itself (Online resource Table S4). Results indicated that with increasing nondetection frequency, mean, median, and percentiles varied widely between the substitution methods. For instance, substituting DL for the data estimated 1.2 and 17 times higher mean values than substituting zero for 21 and 80 % nondetection frequencies, respectively. This indicated wider difference of means with higher
nondetection frequencies. This occurred because the use of fabricated numbers along with detected values did not consider the distribution of the data, resulting in significant differences in the estimates, which could also be far from their true values (Helsel 2012). The use of these different estimates would also encounter a problem in a decision-making situation for remediation and legislation (Helsel 2012). Therefore, in order to produce more accurate data interpretation, we adopted statistical approaches, used in medical and industrial studies as a standard procedure. Statistical summaries in this study (Table 1) showed that median influent concentrations were 3,700, 1,000, 74, and 14 pg/L for DBDPE, BTBPE, HBB, and PBEB, respectively, whereas median final effluent concentrations were 200, 120, 6.6, and 0.94 pg/L, respectively. Median primary effluent concentrations in secondary and advanced treatment processes were 3,100, 880, 61, and 13 pg/L for DBDPE, BTBPE, HBB, and PBEB, respectively. DBDPE and BTBPE were predominant and their levels were two orders of magnitude higher than those of HBB and PBEB. In the literature, the few cases that reported influent levels showed that DBDPE from a Swedish WWTP (Ricklund et al. 2008a) and HBB from three Norwegian WWTPs (Arp et al. 2011) were one or two orders of magnitude higher than the levels found in the current study. Comparing final effluent levels in this study, the Swedish study reported DBDPE in the same order of magnitude (Ricklund et al. 2008a) while HBB in the Norwegian investigation showed two orders of magnitude higher (Arp et al. 2011). The influent wastewater in this study was collected in two different seasons and from plants with different community sizes and industrial inputs. The potential effects of those conditions on influent levels were examined. The occurrence of some compounds varied seasonally: as seen in Table 1, BTBPE and HBB levels were 1.3- to 2-fold higher in summer than in winter (p
RESEARCH ARTICLE
Occurrence and fate of four novel brominated flame retardants in wastewater treatment plants M. Kim & P. Guerra & M. Alaee & S. A. Smyth
Received: 17 October 2013 / Accepted: 25 June 2014 # Her Majesty the Queen in Right of Canada 2014
Abstract Four novel brominated flame retardants (NBFRs), i.e., decabromodiphenylethane (DBDPE), 1,2-bis(2,4,6tribromophenoxy) ethane (BTBPE), pentabromoethylbenzene (PBEB), and hexabromobenzene (HBB) were studied in 377 liquid samples and 288 solid samples collected from 20 wastewater treatment plants. Lagoon, primary, secondary, and advanced treatment processes were included, in order to investigate NBFR occurrence and the effects of WWTP operational conditions on NBFR removal. Median influent and effluent levels were 14 to 3,700 and 1.0 to 180 pg/L respectively, with DBDPE being the highest in both. Overall median removal efficiencies for DBDPE, BTBPE, HBB, and PBEB across all process types were 81 to 93, 76 to 98, 61 to 97, and 54 to 97 %, respectively with advanced treatment processes obtaining the best removals. NBFRs removal was related to retention time, surface loading rate, and biomass concentration. Median NBFR levels in treated biosolids were 80 to 32,000 pg/g, influenced by solids treatment processes. Keywords Wastewater treatment plants . Biosolids . Brominated flame retardants . DBDPE . BTBPE . PBEB . HBB
Introduction Brominated flame retardants (BFRs) are extensively used in commercial products to delay combustion; their production Responsible editor: Philippe Garrigues Electronic supplementary material The online version of this article (doi:10.1007/s11356-014-3262-4) contains supplementary material, which is available to authorized users. M. Kim : P. Guerra : M. Alaee : S. A. Smyth (*) Science and Technology Branch, Environment Canada, 867 Lakeshore Road, P.O. Box 5050, Burlington, Canada L7R 4A6 e-mail: [email protected]
accounts for 20 to 25 % of the global production volume of flame retardants (Harju et al. 2009). Novel BFRs such as decabromodiphenylethane (DBDPE; C14H4Br10) and 1,2-bis (2,4,6-tribromophenoxy) ethane (BTBPE; C14H8Br6O2) are available for use in Canada. DBDPE is used in high impact polystyrene (HIPS) and cable coverings, while BTBPE has been applied to high impact polystyrenes and thermoplastics (de Wit et al. 2010). In addition to these two compounds, some formerly manufactured minor BFRs have reappeared on the global market, including pentabromoethylbenzene (PBEB; C8H5Br5) for textile, cable coating, and polyurethane foam application and hexabromobenzene (HBB; C6Br6) for uses in paper, woods, textiles, electronic, and plastic goods (Harju et al. 2009; Covaci et al. 2011; Salamova and Hites 2011). The production volume of these four novel BFRs (NBFRs) varied between compounds and region: for DBDPE, it was 4,500 to 23,000 tonnes (USA in 2005) and 12,000 tonnes (China in 2006); for BTBPE, it was 50 % accounted for 12 to 34 % of the dataset for liquid samples and 11 to 20 % for solid samples; therefore, variability in NBFR concentrations was generally less than 50 % for both wastewater and solids. NBFRs in the liquid stream NBFR levels The statistical distributions of the four NBFRs in raw influent, primary effluent, and final effluents are summarized in Table 1. There were 145, 86, and 146 samples analyzed for influents, primary effluents, and final effluents, respectively. NBFR levels were below DL in 35 to 80 % of total DBDPE analyses, 0 to 21 % of BTBPE analyses, 1 to 29 % of HBB analyses, and 8 to 61 % of PBEB analyses. Therefore, statistical methods handling for censored data, i.e., ROS and Kaplan–Meier were used to estimate statistical distributions of NBFR levels. The frequency of nondetects varied widely from 0 to 80 % over NBFR and sample type, which may be caused by varying concentrations, high DL, and matrix effects. The large dataset produced in this study adds to the reliability of the results available in the literature by reducing the dominance of nondetects in the dataset and producing more representative statistical summaries. To assess the effectiveness of the use of statistical methods instead of substitutions, statistical summaries of datasets with nondetection frequencies of 21 to 80 % were compared between different substitution methods using zero, 1/3 DL, 1/2 DL, 2/3 DL, and DL itself (Online resource Table S4). Results indicated that with increasing nondetection frequency, mean, median, and percentiles varied widely between the substitution methods. For instance, substituting DL for the data estimated 1.2 and 17 times higher mean values than substituting zero for 21 and 80 % nondetection frequencies, respectively. This indicated wider difference of means with higher
nondetection frequencies. This occurred because the use of fabricated numbers along with detected values did not consider the distribution of the data, resulting in significant differences in the estimates, which could also be far from their true values (Helsel 2012). The use of these different estimates would also encounter a problem in a decision-making situation for remediation and legislation (Helsel 2012). Therefore, in order to produce more accurate data interpretation, we adopted statistical approaches, used in medical and industrial studies as a standard procedure. Statistical summaries in this study (Table 1) showed that median influent concentrations were 3,700, 1,000, 74, and 14 pg/L for DBDPE, BTBPE, HBB, and PBEB, respectively, whereas median final effluent concentrations were 200, 120, 6.6, and 0.94 pg/L, respectively. Median primary effluent concentrations in secondary and advanced treatment processes were 3,100, 880, 61, and 13 pg/L for DBDPE, BTBPE, HBB, and PBEB, respectively. DBDPE and BTBPE were predominant and their levels were two orders of magnitude higher than those of HBB and PBEB. In the literature, the few cases that reported influent levels showed that DBDPE from a Swedish WWTP (Ricklund et al. 2008a) and HBB from three Norwegian WWTPs (Arp et al. 2011) were one or two orders of magnitude higher than the levels found in the current study. Comparing final effluent levels in this study, the Swedish study reported DBDPE in the same order of magnitude (Ricklund et al. 2008a) while HBB in the Norwegian investigation showed two orders of magnitude higher (Arp et al. 2011). The influent wastewater in this study was collected in two different seasons and from plants with different community sizes and industrial inputs. The potential effects of those conditions on influent levels were examined. The occurrence of some compounds varied seasonally: as seen in Table 1, BTBPE and HBB levels were 1.3- to 2-fold higher in summer than in winter (p
