PDF Archive

Easily share your PDF documents with your contacts, on the Web and Social Networks.

Share a file Manage my documents Convert Recover PDF Search Help Contact

J Environ Eng Sci 4, 2005, 353 367 .pdf

Original filename: J Environ Eng Sci 4, 2005, 353-367.pdf
Title: i:\
Author: Judy Riopelle (NRC Ottawa ON) 3031 2000 Jun 14 16:02:34

This PDF 1.5 document has been generated by DVIPSONE 2.2.3 http://www.YandY.com / Acrobat Distiller 6.0.1 (Windows), and has been sent on pdf-archive.com on 01/11/2015 at 18:03, from IP address 71.17.x.x. The current document download page has been viewed 757 times.
File size: 475 KB (15 pages).
Privacy: public file

Download original PDF file

Document preview


Polybrominated diphenyl ethers in an
advanced wastewater treatment plant.
Part 1: Concentrations, patterns, and influence
of treatment processes
Sierra Rayne and Michael G. Ikonomou

Abstract: Concentrations and patterns of the mono- through deca-substituted polybrominated diphenyl ether (PBDE) flame
retardants were determined in all major unit operations and processes within a tertiary-level wastewater treatment plant
(WWTP) having post-filtration ultraviolet light disinfection. The results show that PBDEs do not appear to be degraded
substantially or otherwise removed by wastewater treatment processes such as anaerobic, anoxic, and aerobic biological
treatment, anaerobic digestion, dissolved air flotation, or sand–anthracite filtration. An overall removal efficiency of 93%
was observed for PBDEs in the WWTP due to sorption onto wastewater sludges, well below that predicted by equilibrium
partitioning models. High levels observed in the resulting WWTP biosolids (∼2.4 mg·kg−1 dry weight) may contaminate a
wider environment through their use as a soil amendment. Lower concentrations of PBDEs contained within high volumes
of aqueous WWTP effluent (∼26 ng·L−1 ) may result in a large PBDE flux into receiving waters, posing a potential threat
to drinking water supplies and fisheries resources.
Key words: polybrominated diphenyl ethers (PBDEs), flame retardants, municipal wastewater treatment, domestic sewage,
mass balance, congener patterns.
Résumé : Les concentrations et les patrons de produits ignifugeants à base d’éthers diphényliques, polybromés, monosubstitués à décasubstitués, ont été déterminés dans toutes les opérations–procédés unitaires majeurs d’une station d’épuration
des eaux usées de niveau tertiaire munie d’un système de désinfection UV post-filtration. Les résultats montrent que les
éthers diphényliques polybromés ne semblent pas être substantiellement dégradés ou autrement éliminés par les procédés
de traitement des eaux usées tels que le traitement biologique anaérobie, anoxique et aérobie, la digestion anaérobie, la
flottation à l’air dissous ou la filtration sur sable – anthracite. Une efficacité globale d’élimination de 93 % a été observée
pour les éthers diphényliques polybromés dans la station d’épuration en raison de la sorption dans les boues d’épuration,
bien en deçà de ce qui est prédit par les modèles de partage à l’équilibre. Les niveaux élevés observés dans les biosolides
de la station d’épuration qui en découlent (∼2,4 mg·kg−1 poids sec) pourraient contaminer un environnement plus vaste
par leur utilisation en tant qu’amendement de sol. Des concentrations plus faibles des éthers diphényliques polybromés
contenus dans des grands volumes d’effluent de la station d’épuration (∼26 ng·L−1 ) peuvent engendrer un grand débit
d’éthers diphényliques polybromés dans les eaux réceptrices, posant ainsi un risque potentiel aux réserves d’eau potable et
aux ressources halieutiques.
Mots clés : éthers diphényliques polybromés, produits ignifugeants, système d’épuration des eaux usées municipales, eaux
usées domestiques, bilan massique, patrons des congénères.
[Traduit par la Rédaction]

Water is a critical resource in the Okanagan Basin located in
the southern interior of British Columbia and northern Washington State. The Okanagan Basin extends 204 km in Canada
and 118 km in the USA in a north–south direction with a catchment area of ∼8220 km2 (Fig. 1) (Truscott and Kelso 1979) and

contains a main valley lake system comprising six lakes connected by stream and groundwater flow in a structural trench
partially filled with >350 m of unconsolidated material overlying a system of subparallel, linked faults (Anon. 1974). The
Okanagan Basin is broadly split into two topographic features:
(i) a highland plateau (>1000 m above sea level (ASL)) with

Received 22 September 2003. Revision accepted 15 October 2004. Published on the NRC Research Press Web site at http://jees.nrc.ca/ on
2 September 2006.
S. Rayne. Department of Chemistry, University of Victoria, Victoria, BC V8W 3V6, Canada.
M.G. Ikonomou.1 Marine Environment and Habitat Science Division, Pacific Region, Institute of Ocean Sciences, Fisheries and Oceans
Canada, 9860 West Saanich Road, P.O. Box 6000, Sidney, BC V8L 4B2, Canada.
Written discussion of this article is welcomed and will be received by the Editor until 31 January 2006.

Corresponding author (e-mail: IkonomouM@pac.dfo-mpo.gc.ca).

J. Environ. Eng. Sci. 4: 353–367 (2005)

doi: 10.1139/S04-071

© 2005 NRC Canada


J. Environ. Eng. Sci. Vol. 4, 2005

Fig. 1. Map of the study area showing locations of major population centers and hydrologic features.


Mabel Lake
Salmon Arm

Deep Creek


Swan Lake
Vernon Creek


5 10
Scale in km


Coldstream Creek
Kalamalka Lake

Okanagan Basin
Watershed Boundary

Wood Lake
Kelowna Creek


Squally Point

Mission Creek

Okanagan Lake
Trout Creek
Skaha Lake
Vaseaux Lake

Similkameen River

Osoyoos Lake
United States

Okanagan River



relatively shallow, humic lakes that drain into (ii) the main lake
system lying in the Okanagan Valley. This valley results from
Pleistocene glaciation and is ∼160 km in length and is U-shaped
with valley sides from 1220 to 2134 m ASL and bench lands

∼30–60 m above the lakes (Truscott and Kelso 1979). Several
hundred metres of unconsolidated materials were deposited during the Pleistocene epoch from glacial outwash, direct glaciation, and lacustrine fluvial sedimentation; these line the valley
© 2005 NRC Canada

Rayne and Ikonomou

bottom and give rise to well-drained, fertile soils that support
the region’s extensive horticultural and agricultural industries
and that are also major determinants in subsurface contaminant transport within the region. From its position in the rain
shadow of the Coast and Cascade mountain ranges, which line
the western front of North America, the Okanagan Valley has
a dry continental climate where average precipitation is typically <350 mm/year, of which ∼85% is lost through evaporation from local lakes and evapotranspiration, with summertime temperatures often exceeding 35–38 ◦ C and a mean annual
temperature of 7.8 ◦ C (Anon. 1974). Early evidence also indicates that climate and streamflow in the Pacific Northwest are
influenced by global-scale climate fluctuations such as the El
Niño southern oscillation phenomenon (Andrusak et al. 2000),
suggesting that climate change may contribute further to water
scarcity in this semi-arid region (Cohen and Kulkami 2001).
Among the potential threats to the water quality of Okanagan Lake, which is the main water body in the Okanagan Basin,
the largest point source of organic contaminants is likely from
the City of Kelowna wastewater treatment plant (WWTP) (see
Figs. 2 and 3 for plant schematics). This WWTP was the first (the
current configuration opened in 1983) in North America specifically designed to include a sequence of anaerobic, anoxic, and
aerobic environments for the removal of carbon, nitrogen, and
phosphorus (Oldham and Stevens 1984); the authors of a previous report claimed the honor for Palmetto, Florida, in 1989
(Zitomer and Speece 1993).At present, the Kelowna wastewater
system serves 1050 institutional users and an estimated population of ∼70 000 (City of Kelowna 2003) with 28 pumping
stations, >400 km of sewage pipe, and a 10 ha WWTP capable
of treating 40 ML/d. Upon arriving at the WWTP, wastewater is
first bar-screened to remove large materials, after which a grit
chamber removes sand and other small, primarily inorganic,
solids. Both screenings and grit are sent to the municipal landfill without further processing. Following the grit chamber, the
wastewater is mixed with recycled subnatant from the dissolved
air flotation (DAF) unit, backwash from the sand–anthracite
filter, and centrate from the centrifuge. The combined process
flow then goes to the primary clarifiers, from which the primary
sludge is sent to an anaerobic fermenter and the settled primary effluent travels to the biological nutrient removal (BNR)
reactor. The BNR reactor (bioreactor) was formerly a world
leader in the Bardenpho system, utilizing a sequential anaerobic → anoxic → aerobic → anoxic → aerobic arrangement
with internal and external recycle for removal of nitrogen and
phosphorus from the aqueous stream. Over the last few years,
the reactor arrangement was changed to that of a parallel train
modified BNR process (A2 O; anaerobic → anoxic → aerobic)
because it was observed that the last two stages of the Bardenpho system were not removing additional nutrients and were
at times even increasing nutrient concentrations in the effluent.
Effluent from the BNR reactor is sent to a suite of secondary
clarifiers where the settled sludge is returned to the anaerobic
stage of the bioreactor as recycled activated sludge (RAS) and
the over-weir effluent travels through a sand–anthracite filter


for removal of residual suspended solids (SS) and finally to an
ultraviolet (UV) light treatment system (λem = 254 nm) to disinfect the wastewater prior to deep water (65 m depth) release
into Okanagan Lake.
Among the potential emerging contaminants in domestic
wastewater streams, the polybrominated diphenyl ether (PBDE)
flame retardants are now well known as ubiquitous environment pollutants with levels reaching up into the milligram per
kilogram range for many compartments such as marine mammals, sediments, and sewage residuals (Alaee 2002; de Boer
2000; Letcher and Behnisch 2003; World Health Organization
1994). Recent reports of PBDEs in domestic wastewaters (de
Boer et al. 2003; Oberg et al. 2002) demonstrate that the presence of these compounds must arise, at least in part, from the
widespread household use of PBDE-containing foams, plastics,
and textile-based consumer products. However, to our knowledge, no studies have comprehensively investigated both levels
and detailed congener patterns of PBDEs within a WWTP with
an emphasis on a full accounting (i.e., mass balance) of PBDEs
in all process streams and with regard to the WWTP as a whole.
To help fill this current knowledge gap in an important contaminant class, the present study was undertaken to determine the
fate of PBDEs in an advanced municipal WWTP and to examine
whether any abiotic or biotic treatment processes can degrade
PBDEs on the timescales typically observed in WWTPs (i.e.,
hours to weeks, depending on the process train).

Materials and methods
Sample collection
Samples were collected on 25 June 2002 between 0800 and
1600 from 23 sites within the City of Kelowna WWTP using
pre-cleaned 500-mL and 4-L amber glass jars. Whereas polychlorinated biphenyl (PCB) concentrations have been previously shown to display diurnal patterns with the highest levels
observed between 1100 and 1800 (Lawrence and Tosine 1977),
we chose not to sample outside the typical work hours of WWTP
staff for logistical and safety reasons. However, samples were
collected in a methodical manner starting at the plant influent,
working through the process train according to how a control
mass of wastewater would travel through the WWTP, and ending with the plant solid and liquid effluent streams. Single grab
samples were obtained at each of the sampling locations shown
in Fig. 3 with pre-cleaned 500-mL Teflon bottles and a 3-m
sampling arm. Labels on sampling locations shown in Fig. 3
correspond with the detailed sample information given in Supplementary material Tables D1 and D2.2 Samples were acidified
on site to pH <2 using 1 mol/L HCl and stored at 4 ◦ C for the
following 36–48 h during transport back to our laboratory in

Supplementary data for this article are available on the Web
site or may be purchased from the Depository of Unpublished
Data, Document Delivery, CISTI, National Research Council Canada, Ottawa, ON K1A 0R6, Canada. DUD 4017. For more
information on obtaining material refer to http://cisti-icist.nrccnrc.gc.ca/irm/unpub_e.shtml.
© 2005 NRC Canada


J. Environ. Eng. Sci. Vol. 4, 2005

Fig. 2. Aerial photograph and schematic of the advanced wastewater treatment plant under study.

Effluent Pumphouse

UV Building
Filtration Building
RAS Pumphouse

Cl2 Contact Chambers (unused)

2° Clarifiers (5 total)

2° Clarifiers (5 total)
Odor Control

(2 Trains)

Sludge Dewatering

1° Clarifiers
DAF & Alum Room

Equalization Basin/
Thickener Pumphouse

Raw Pumping
Administration &
Control Building

Sidney, British Columbia. Following arrival at the laboratory,
the 500-mL samples were stored at –20 ◦ C, whereas the four
4-L samples (DAF subnatant, secondary clarifier effluent, sand–
anthracite filter effluent, and UV treatment effluent) were stored
at 4 ◦ C during the 63-d period between sampling and analysis.
Sample analysis
For percent solids analysis, performed within 48 h upon return to the laboratory, ∼5 g aliquots of sample were placed
in a pre-cleaned aluminum foil dish and weighed accurately.
Samples were then placed in a vented oven at 105 ◦ C for 24 h,
removed to room temperature, and allowed to cool to a constant
weight, and weighed accurately. Percent solids was calculated
as ((mass remaining after drying at 105 ◦ C for 24 h)/(mass
before drying)) × 100%. The percent solids measured for the
23 samples agreed well with similar data collected by WWTP
staff. Values for percent solids were then converted to a sample specific gravity using the following relationship, 1/S =
Ps /Ss + Pw /Sw , where S is the specific gravity of the sample,
Ps is the percentage of solids in the sample, Ss is the specific
gravity of solids (taken as 1.32 for primary solids and 1.60 for
digested solids), Pw is the percentage of water in the sample
(Pw = 1 − Ps ), and Sw is the specific gravity of water (taken as
1.0) (Tchobanoglous and Burton 1991). The calculated values
are within the range normally reported for primarily domestic
wastewaters (Hammer 2001; Tchobanoglous and Burton 1991).

Because of high variation in the percent solid content of the
samples, novel sample extraction experiments were devised to
maximize PBDE extraction efficiency while minimizing potential sources of sample contamination. All samples were first
spiked with a suite of 13 C-labeled PBDE (except 13 C-labeled
BDE-209, which was spiked into the samples prior to extraction) procedural internal standards (Cambridge Isotope Laboratories, Andover, Massachusetts), subsequently transferred into
a porcelain Buchner funnel (126-mm diameter), and filtered under vacuum over Whatman No. 7 GF/C filters (11-cm diameter,
nominal pore size 1.2 µm) into a 1-L vacuum filter flask. Sampling containers were rinsed several times with deionized MilliQ grade water to remove deposits, and all the rinsate was passed
through the filter. Two samples that were difficult to filter (DAF
sludge and fermented primary sludge) were centrifuged prior
to filtration. The sludge sample was transferred into a solventrinsed Teflon tube and centrifuged at 1000 rpm for 10 min. The
centrifuged solids were then filtered as above, and the filtrate
was combined with the centrate for subsequent liquid–liquid
extraction. The flask of the Buchner filtering system was rinsed
with toluene at least three times, and the rinsate was combined
with the filtrate of the same sample. All glassware and apparatus used to handle the samples were baked and rinsed with
solvents before and after baking. The Whatman filter paper used
was Soxhlet extracted overnight using toluene–acetone (4:1 v/v)
and air dried before use.
© 2005 NRC Canada



Anaerobic Fermenter
Qin=Qout=1.56 ML/d
Min=2.0 g/d; Mout=1.9 g/d
Mass Balance=94%

Q=431 L/d
M=0.022 g/d

DAF Subnatant
Q=2.2 ML/d
M=0.18 g/d

Raw Influent
Q=28.8 ML/d
M=11.3 g/d

1° Clarifier
Qin=Qout=38.5 ML/d
Min=11.9 g/d; M out=11.4 g/d
Mass Balance=96%

1° Effluent
Q=36.9 ML/d
M=9.4 g/d







DAF Sludge
Q=108 900 L/d
M=12.9 g/d

MLSS Wasting
Q=2.3 ML/d; M=13.2 g/d

Q=24.0 ML/d; M=383 g/d


Dissolved Air Flotation
Qin=Qout=2.3 ML/d
Min=13.2 g/d; Mout=13.1 g/d
Mass Balance=99%

Fermenter Supernatant
Q=1.53 ML/d
M=0.90 g/d

1° Sludge
Q=1.56 ML/d
M=2.0 g/d

Internal Recycle
Q=250 ML/d; M=1 451 g/d





Anoxic Stage
Aerobic Stage
2° Biological Treatment
2° Biological Treatment
Qin=Qout=312 ML/d
Qin=Qout=312 ML/d
Min=1 958 g/d; Mout=1 773 g/d Min=1 773 g/d; Mout=1 795 g/d
Mass Balance=91%
Mass Balance=101%

Fermenter Sludge
Q=34 300 L/d
M=0.98 g/d



Anaerobic Stage
2° Biological Treatment
Qin=Qout=62.5 ML/d
Min=393 g/d; Mout=507 g/d
Mass Balance=129%


Sand-Anthracite Filter
Qin=Qout=36.2 ML/d
Min=0.95 g/d; Mout=0.97 g/d
Mass Balance=102%

Filter Backwash
Q=7.4 ML/d
M=0.23 g/d

Q=34 800 L/d
M=14.4 g/d

Overall Solids Stream
Min=11.3 g/d
Mout(avg)=12.0 g/d
% Raw Influent Mass=106%

Overall Aqueous Stream
Min=11.3 g/d
Mout=0.76 g/d
% Raw Influent Mass=6.7%

UV 3° Treatment
Qin=Qout=28.8 ML/d
Min=0.74 g/d; Mout=0.76 g/d
Mass Balance=103%


Qin=Qout=0.16 ML/d
Min=13.9 g/d; Mout=14.6 g/d
Mass Balance=105%


Q=0.12 ML/d
M=0.22 g/d


2° Clarifier
Qin=Qout=60.2 ML/d
Min=344 g/d; Mout=355 g/d
Mass Balance=103%

Fig. 3. Process flows and PBDE mass balances within the wastewater treatment plant. Labels on sampling locations correspond with the detailed sample information given in
Supplementary material Tables D1 and D2.2

Rayne and Ikonomou

© 2005 NRC Canada


This procedure yielded two sub-samples from each sludge
sample: (i) a filtrate (or combined filtrate and centrate) that was
processed using liquid–liquid extraction and (ii) the retained
solids that were processed together with the filter using Soxhlet
extraction. A separatory funnel was used to extract all filtrates
three times with approximately equal volumes of toluene. The
retained solids and filter paper were dried with Na2 SO4 in a
mortar, transferred into the glass thimble of the Soxhlet apparatus, and extracted for 16 h using toluene–acetone (4:1 v/v).
Extracts from the filtrate and filter-retained solids for each sample were combined; successively washed with 40 mL of KOH,
80 mL of high performance liquid chromatography grade water,
and 10 mL of H2 SO4 ; reduced in volume by rotary evaporation;
and finally reconstituted in 10 mL of CH2 Cl2 –hexane (1:1 v/v).
Next, the extracts were subjected to previously published sample cleanup and analysis methods, which are described in brief
below. Sample cleanup took place in three steps. In the first step,
aliquots were passed through a multilayer silica column packed
with successive layers of silica gel (basic, neutral, acidic, neutral) and eluted with CH2 Cl2 –hexane (1:1 v/v). The second step
used a glass column filled with copper filings and Na2 SO4 to remove sulfur and residual water and eluted with CH2 Cl2 –hexane
(1:1 v/v). The third step involved a neutral activated alumina column capped with anhydrous sodium sulfate and washed with
hexane followed by CH2 Cl2 –hexane (1:1 v/v) to elute the analytes of interest. Eluants from the alumina column were concentrated to <10 µL and spiked with 1 ng of a 13 C-3,3 ,4,4 tetrabromodiphenyl ether method performance standard prior to
congener-specific PBDE analyses by high-resolution gas chromatography and high-resolution mass spectrometry. Details on
the composition of the internal standards, the solvents and conditions used in all the clean-up steps, the instrumental analysis
conditions, the quality assurance quality control protocols, and
the criteria for congener identification and quantification are
reported elsewhere (Ikonomou et al. 2002a, 2002b; Rayne et
al. 2003a, 2003b).
Data analysis
Polybrominated diphenyl ether removal and degradation efficiencies and loss processes within a “generic” secondary level
WWTP were estimated for the major PBDE congeners using
the environmental modeling software EPI SuiteTM v.3.11 (EPA
2001). Data compilation and graphing were performed using
MicrosoftTM Excel 2002 (Redmond, Oregon). Differences between sampling groups were investigated using single factor
ANOVA in SPSS v.10.0 (Chicago, Illinois). Cluster analysis
(with the standardized Euclidean measure and Ward clustering
method) was performed using KyPlot v.2.0 b.9 (Tokyo, Japan).

Results and discussion
Polybrominated diphenyl ether concentrations in the
wastewater treatment plant
A comprehensive PBDE sampling and analysis program for
a tertiary-level WWTP with UV disinfection demonstrated that

J. Environ. Eng. Sci. Vol. 4, 2005

PBDEs do not appear to be substantially degraded by advanced
wastewater treatment processes but that the high levels observed
in the resulting biosolids may pose an environmental threat
through use as a soil amendment. In addition, the lower concentrations of PBDEs contained within high volumes of aqueous effluent may result in a large PBDE flux into receiving waters over
the coming decades, posing a potential threat to drinking water
supplies and the local fisheries resource. Of the 63 congeners
analyzed for, 46 had ≥30% of all values above the method detection limit (MDL). These were included in calculations of
total PBDEs ( PBDE, sum of the 46 congeners) and used in
the cluster analysis and further discussions below. Concentrations of PBDE ranged over 4.2 orders of magnitude within
the WWTP, from ∼413 000 ng·L−1 in the centrifuged biosolids
effluent leaving the WWTP for disposal to ∼26 ng·L−1 in the
sand–anthracite filter and UV treatment effluents. See Supplementary material Tables D1 and D2 for congener-specific PBDE
concentrations at each of the 23 locations sampled within the
Potential impacts of polybrominated diphenyl ethercontaminated biosolids disposal
The PBDE concentrations in the resulting biosolids, which
are subsequently mixed and composted with other organic material for domestic and commercial use as a soil amendment, at
∼413 µg·L−1 (or ∼2429 µg·kg−1 dry weight (dw)) are in the
range previously reported for some European sewage sludges
(0.4–572 µg·kg−1 wet weight) (Hagenmaier et al. 1992; Hale
et al. 2003; Palm et al. 2002) and from stabilized biosolids samples from across the USA (466–7000 µg·kg−1 dw) (Hale et al.
2003, 2001). Previous work has shown that municipal sewage
wastes are increasingly used as fertilizers and soil conditioners,
either as “raw” or composted products, and that similarly hydrophobic contaminants, such as PCBs, are typically retained
in the upper soil horizons following application. This soil retention can then pose a risk by such contaminants entering the
food chain either through direct ingestion of soil and (or) sludge
or trophic transfer from soil → plant → herbivore → carnivore
→ humans (via consumption of contaminated dairy or meat
products) (Bergh and Peoples 1977; Naylor and Loehr 1982).
With such highly hydrophobic contaminants as PBDEs, humic
materials in soils may retain these contaminants to such a degree as to largely inhibit uptake by plants (Mcintyre and Lester
Assuming an application rate of 10 t dw of biosolids per
hectare of arable soil (bulk density is 1.0 g·cm−3 ) ploughed to
a depth of 15 cm (Alcock and Jones 1993), PBDE concentrations would be expected to increase by ∼16 mg·kg−1 . Alternatively, if the same mass of biosolids were applied to the surface
of a 1-ha unploughed pasture grassland, PBDE concentrations would be expected to increase by ∼250 mg·kg−1 . There
are very few reports of PBDE concentrations in ambient soils,
and only one report of ∼0.5 mg·kg−1 dw appears to be available
in the literature (Palm et al. 2002; World Health Organization
1994). Thus, if this value is assumed to be fairly representative
© 2005 NRC Canada

Rayne and Ikonomou

of ambient PBDE levels in industrialized nations, biosolids additions to tilled and untilled agricultural soils would be expected
to increase PBDE concentrations by ∼32- and 500-fold, respectively, over background levels. Such increases may be of
concern for livestock grazing on forage crops where bioaccumulation and biomagnification may take place, with subsequent
human consumption, and also for domestic users who may be
exposed to elevated PBDE concentrations via dermal contact
with, and ingestion of, contaminated soils.
By comparison, PCB concentrations in sewage sludges vary
widely (from <0.1 to 185 mg·kg−1 dw) (West and Hatcher
1980), with an average value of 5 mg·kg−1 dw in American
WWTPs during the late 1970s (Furr et al. 1976). In the WWTP
under study, the sludges (including biosolids) are routinely analyzed for PCBs, and no values above the ∼1 mg·kg−1 MDL
have been reported. Thus, PBDE concentrations in biosolids
at the Kelowna WWTP appear to be at least 2–3-fold higher
than those of PCBs and within the range of PCB levels reported in a variety of American WWTPs. Furthermore, much
of the published PCB data from WWTPs is >20 years old, and
given the general decline in PCB levels over this period, current
PBDE:PCB ratios in WWTPs are likely near the upper end of the
ranges reported above. Although the plant staff do not analyze
for polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs), a recent study at the nearby
metropolis of Vancouver, British Columbia, found P4−8 CDD
and P4−8 CDF (sum of tetra- through octa-chlorinated congeners) levels of ∼10 ng·g−1 dw (Bright and Healey 2003),
or ∼2.4 orders of magnitude lower than those of PBDE in
the Kelowna WWTP biosolids. Assuming similar per capita
contaminant loadings between these two cities, it appears as
though PBDE concentrations (2.4 µg·g−1 dw) in Kelowna
biosolids exceed those of all other individual organic contaminants (including individual polycyclic aromatic hydrocarbon
components) in the Vancouver biosolids, except that of the plasticizing agent bis(2-ethylhexyl) phthalate at 2.7 µg·g−1 dw.
Thus, it is likely that PBDEs can now be considered one of
the dominant organic contaminants in North American sewage
sludges, consistent with our recent report that PBDE concentrations are overtaking those of PCBs in fish from another region
of the Columbia River watershed (Rayne et al. 2003a).
Polybrominated diphenyl ether removal in the
wastewater treatment plant
Raw influent PBDE concentrations were estimated at
392 ng·L−1 using a calculated grit production rate of 431 L·d−1
based on an average normalized grit production rate of
14.96 L·ML−1 (Tchobanoglous and Burton 1991). Using this
calculated raw influent PBDE concentration of 392 ng·L−1
and the observed UV treatment effluent (i.e., plant discharge
to Okanagan Lake) concentration of 26.4 ng·L−1 , the WWTP
removes ∼93% of influent PBDE from the aqueous stream,
yet the high wastewater flow rates result in a PBDE loading of 0.76 g·d−1 into receiving waters. Because of the small


quantity of grit produced per volume of raw influent, the grit
chamber effluent PBDE concentration was lowered by only
1 ng·L−1 to 391 ng·L−1 (a 0.3% reduction, Fig. 3). It is expected that the large materials (e.g., rocks, branches, lumber,
paper, plastics, rags) retained on the bar-screen unit upstream of
the grit chamber would remove a similarly negligible amount
of influent PBDEs and thus not introduce significant error into
our reconstructed raw influent concentrations and profiles. For
reasons that are not clear, this reduction in the grit chamber is
far less than has been reported previously for PCBs (up to 94%)
(Bergh and Peoples 1977).
The higher Kow values for PBDEs versus PCBs suggest that
PBDEs should be removed more efficiently on WWTP solids,
yet we observed a negligible PBDE removal on grit. PBDE removal efficiencies for the primary (17.4%) and secondary clarifiers (99.5%) and tertiary sand–anthracite filter (2.3%) varied
widely. Previous studies have shown typically higher removal
efficiencies for PCBs and organochlorine pesticides in primary
clarifiers, generally in the range of 33% to 99% (Bergh and
Peoples 1977; Gutierrez et al. 1984; Hannah et al. 1986; Mcintyre et al. 1981; Petrasek et al. 1983; Pham and Proulx 1997),
although removal efficiencies as low as 9% have been observed
(Rogers 1996), with no dependence on hydraulic loadings or
SS removals. Indeed, ratios of PCB concentrations between the
primary sludge and primary effluent have been reported in the
range of 100 to 1000 (Mcintyre et al. 1981; Petrasek et al. 1983),
far above our observed ratio of ∼5 for PBDEs. In contrast, the
analogous sludge concentration ratios of ∼50 for PCBs in the
RAS and secondary clarifier effluent have been reported (Petrasek et al. 1983) and are significantly lower than our similar
concentration ratio of 570 for PBDEs.
This suggests that PBDEs may have different partitioning
patterns within WWTPs than would be expected by comparison with PCBs and other halogenated aromatics. By the time
the wastewater had undergone secondary clarification, ∼93% of
PBDE had been removed from the aqueous stream by sorption
to sludge, with negligible reductions during sand–anthracite filtration and UV disinfection. The removal efficiency up to the
secondary clarification is comparable with previous reported
efficiencies for PCB and organochlorine pesticides (75% to
>99%) (Bergh and Peoples 1977; Gutierrez et al. 1984; Horning
et al. 1984; Martin and Gosselin 1978; Mcintyre et al. 1981; Petrasek et al. 1983; van Luin and van Starkenburg 1984). Whereas
a greater removal efficiency for the secondary versus the primary clarifiers is expected because of the lower organic content
in the resulting secondary effluent (i.e., the majority of PBDEs
in primary influent are associated with nonsettleable solids or
are in solution), as has been reported for PCBs and organochlorine pesticides (Gutierrez et al. 1984)), the almost negligible removal in the sand–anthracite filter is puzzling. One would intuitively expect high PBDE-removal efficiencies for carbon-based
adsorption processes (e.g., granular activated carbon (GAC))
because GAC is well recognized for removing hydrophobic
organic compounds from aqueous streams, even at low concentrations (Chaudhary et al. 2003; Janssens et al. 1997). Fur© 2005 NRC Canada


thermore, compounds with log Kow values >4.0 are believed
to have high sorption potential (Rogers 1996), which is why
we would also expect higher PBDE removals in the adsorption
and partitioning dominated biomass sedimentation processes.
With log Kow values >3.0, competition for adsorption sites by
hydrophobic contaminants in a multicomponent waste stream
such as domestic wastewater is believed to be negligible, and
in the case of biomass sorption, adsorption can generally be
ignored in favor of partitioning (Wang et al. 1993).
Thus, the apparent lack of sorption equilibrium for PBDEs
in the biomass sedimentation processes of a WWTP appears
to result from non-equilibrium conditions related to partitioning because the adsorption step can be ignored as a potential
mass transport limitation. However, low removal efficiencies
of PCBs (∼35%) by GAC have been observed for drinking
water treatment plants (with subsequent ozonation having negligible effect) (Chevreuil et al. 1990); hence, it appears that
once organic contaminants such as PBDEs accumulate in the
water column of drinking water sources, they may be difficult
to remove by conventional treatment processes. As well, the
quantity of suspended sediments passing through the filter is
much reduced compared with the filter influent, so that PBDE
retention and passage through the filter on small particulates
would explain only a small quantity of the refractory PBDEs.
Two other possible explanations for the low GAC removal rate
are that (i) sufficient dissolved organic carbon (DOC) may remain in the filter effluent to allow substantial dissolution of
PBDEs in the aqueous portion and (ii) the larger molecular size
of many PBDEs compared with other hydrophobic contaminants (e.g., PCBs, PCDDs, and PCDFs) may prevent efficient
GAC adsorption under the high flow rates (i.e., kinetic limitations) observed in the WWTP. Hence, should future investigators choose to examine PBDE adsorption by activated carbon
as a potential treatment process, equilibrium-based partitioning
(i.e., conventional Freundlich and Langmuir models (Wang et
al. 1993)) should not be automatically assumed in laboratory
and pilot-scale trials and a kinetic parameter should be included
in any scale-up attempts.
Logarithmic partitioning factors between the solid and liquid
streams (log KP,Obs = log[CS /CL ] where CS is the PBDE
concentration on a dw basis in the solids stream in milligrams
per kilogram, and CL is the PBDE concentration in the liquid
stream in milligrams per litre) were also calculated to determine
the extent to which equilibrium-based partitioning controlled
PBDE distributions within the WWTP. PBDE removal in the
grit chamber gave a log KP,Obs = 2.8. Because even the monosubstituted PBDEs have log Kow values (∼5.08 for BDEs 1, 2,
and 3 (Wania and Dugani 2003)) greater than this value (up to
a log Kow > 11 for the fully brominated BDE209 (Palm et al.
2002)), it appears as though partitioning between grit and the
other components of the raw influent (i.e., the aqueous portion,
dissolved and colloidal organic particulates, smaller detritus,
etc.) is not at equilibrium because of a less than “optimum”
residence time in the distribution system. What is more likely,
however, is that sufficient dissolved and particulate organic ma-

J. Environ. Eng. Sci. Vol. 4, 2005

terial remains in the grit chamber effluent with a sufficiently
high affinity for PBDEs to prevent a higher removal rate on the
larger particles.
To further examine this apparent lack of equilibrium-based
PBDE partitioning in WWTPs, “expected” partitioning factors (log KP,Pred ) were calculated using the following equation,
KP = 0.63foc Kow , where foc is the fraction of organic carbon
in the solids and is assumed to be 0.20 for wastewater solids
(Sawyer et al. 1994) and Kow is the “overall” mass-normalized
octanol–water partition coefficient of the wastewater stream.
A mass normalized overall log Kow of 9.14 for the raw influent was obtained using the regression-based log Kow prediction
equation presented elsewhere (Wania and Dugani 2003), substituting a value of 11.15 for BDE209 (Palm et al. 2002) because
the regression equation appears to significantly underestimate
the log Kow of BDE209 at 8.70, and using a mass normalized
raw influent congener composition equal to that of the primary
clarifier influent. This overall Kow value is strongly influenced
by the assumed log Kow of 11.15 for BDE209 and is reduced to
7.79 when a log Kow of 8.70 is used for BDE209. In either case,
and using this range of overall Kow values, the partitioning between grit and grit chamber effluent (log KP,Obs ) appears to be
∼4.1–5.4 orders of magnitude less than that predicted by equilibrium partitioning (log KP,Pred = 6.9–8.2). Similar log KP,Pred
partitioning factors between the solid and liquid streams were
calculated for the primary clarifier, the anaerobic fermenter,
the secondary clarifier, the sand–anthracite filter, the DAF unit,
and the centrifuge. See Supplementary material Table D3 for
calculated log KP,Pred values for various unit processes.2 For
the centrifuge, where sludge from both the anaerobic fermenter
and the DAF unit is combined as the influent and where sampling was not taken in the mixed influent, an influent congener
pattern was constructed as the volume-weighted addition of
both influent streams. For all unit processes, log KP,Obs values are from 1.3 to 5.4 orders of magnitude lower than that
predicted by equilibrium-based partitioning theory. By comparison, PCBs have lower log Kow values than PBDEs, but the
log KP,Obs values of PCBs (3.0–5.4, depending on the environmental compartment (Chevreuil et al. 1990; Horzempa and
Ditoro 1983a, 1983b)) typically exceed those of PBDEs. This
suggests that PBDEs in a WWTP behave in a manner not readily predicted by a comparison of their physico-chemical properties with similar halogenated contaminants. The lower values
observed for the primary clarifier (KP,Obs = 3.3) and the anaerobic fermenter (KP,Obs = 2.7) may be readily rationalized on
the observably high DOC and particulate organic carbon (POC)
concentrations in the respective effluents such that, even near
equilibrium, a partitioning factor much lower than the solution
average Kow may be expected. However, for the secondary clarifier (KP,Obs = 4.9), sand–anthracite filter (KP,Obs = 4.7), the
DAF unit (KP,Obs = 4.5), and the centrifuge (KP,Obs = 4.9),
effluent liquid streams have relatively little DOC and POC compared with the effluent solids streams, and perhaps the discrepancy between log KP,Obs and log KP,Pred more clearly shows a
lack of equilibrium-based partitioning.
© 2005 NRC Canada

Rayne and Ikonomou

Hence, whereas equilibrium-based partitioning approaches
may be satisfactory in predicting the fate of highly hydrophobic compounds in natural systems or treatment processes where
the difference in organic content is quite substantial between the
two “phases”, it does not appear to be as useful in assessing their
fate within WWTPs. This finding is in contrast to fugacity-based
approaches for modeling chemical fate in WWTPs, where equilibrium partitioning is assumed and believed to be reasonable
because chemical equilibrium is thought to occur within the
treatment residence time of WWTPs (Clark et al. 1995). The
fugacity-based model was calibrated using compounds with
log Kow values from 1.5 to 5.2; perhaps compounds such as
PBDEs, with log Kow values ranging from 5.5 to >8.7, have
a greater driving force for partitioning into organic phases and
(or) adsorbing onto solids but a slow rate for achieving equilibrium. Should equilibrium-based partitioning apply to WWTPs,
we would expect a log 6.9–8.2 reduction in PBDE concentrations between raw influent and the plant effluent. The log 2.0
reduction we observe is consistent with values among individual unit processes where the PBDE concentrations in the solid
stream are not corrected for moisture content and are ∼5–6 orders of magnitude lower than what would have been predicted
had an equilibrium-based partitioning approach been used to
assess the fate of PBDEs in WWTPs.
Such findings call attention to the need for more robust partitioning models for WWTPs, such as the environmental modeling software EPI SuiteTM v.3.11 that predicted overall removal
efficiencies of ∼94% between the raw influent and final water
effluent (see Supplementary material Table D4 for results of the
EPI SuiteTM modeling on PBDEs in the WWTP2 ), which is in
excellent agreement with the observed value of 93%. However,
EPI SuiteTM was not able to predict removal by primary sedimentation accurately and consistently overestimated removal
efficiencies for seven individual congeners (BDEs 15, 28, 47,
99, 100, 153, 183, and 209) by 2.5–4.8 fold. EPI SuiteTM also
predicted negligible losses within the plant due to volatilization
and biodegradation, consistent with the continuity of a mass
balance throughout the WWTP presented below.
Further evidence for the lack of equilibrium-based partitioning is provided by the removal efficiencies for individual congeners within these unit processes. Removal efficiencies were
calculated as percent removal equal to (CE − CI )/CI , where CE
and CI are the concentrations in the liquid effluent and influent
streams, respectively. Only weak negative linear regressions
(R 2 = 0.00–0.29) were observed between these removal efficiencies and the estimated log Kow values for the suite of congeners (see Supplementary material Table D5 and Fig. D1 for
the relationships between estimated log Kow values and removal
efficiencies for the various solid–liquid separation processes),2
in contrast to previous work with PCBs and other hydrophobic
aromatic contaminants (Petrasek et al. 1983; Pham and Proulx
1997). Nonlinear regression analysis did not provide better correlations between the two parameters. If equilibrium-based partitioning were taking place, we would expect a strong negative
correlation between removal efficiency (where increasing re-


moval efficiency is represented by increasingly negative values)
and log Kow values (i.e., more efficient removal of the more hydrophobic congeners into the solids stream).
Polybrominated diphenyl ether mass balances in the
wastewater treatment plant
The measured process flows on the day of sampling were used
to set up control volumes around each unit process and calculate
mass balances based on the observed PBDE concentrations
within the plant (Fig. 3). Control volumes are indicated by dotted lines surrounding each unit process in Fig. 3. Masses (M)
are presented as fluxes of total PBDEs in grams per day and
flows (Q) in millions of litres per day. Mass balances on all unit
processes were ∼100% (91%–129%), suggesting that PBDEs
act conservatively within a WWTP (i.e., neither created nor
destroyed). Except for the anaerobic stage of the BNR reactor,
mass balances ranged from 91% to 105%, well within the inherent error for sampling and analysis. The higher mass balance for
the anaerobic stage (129%) likely arises from the variability in
solids content of the RAS between each of the four operational
clarifiers during the day of sampling. Suspended solids concentrations were 3000, 5130, 6190, and 5890 mg·L−1 for RAS
lines 1 through 4, with flows of 8.1, 5.2, 5.1, and 5.5 ML·d−1 ,
respectively. The RAS line 3 was analyzed for PBDEs, and this
line also had the highest SS concentrations and the lowest flow
contribution to the anaerobic stage. A relative flow × concentration weighting of 0.92 was applied to the estimated PBDE
heading back to anaerobic stage using the total RAS → bioreactor flow of 23.9 ML·d−1 and the PBDE concentration from
RAS line 3. This weighting factor assumes a linear relationship
between PBDE and SS concentrations; such a relationship is
an “educated guess” that has not been investigated elsewhere.
Hence, the mass balance on the anaerobic stage is highly dependent on the estimated overall PBDE flux from the RAS.
Given the inherent variability in both quality and quantity of
mixed liquor suspended solids (MLSS), any estimates of the
relationship between PBDE concentrations and SS concentrations will have a relatively large amount of uncertainty. To
ensure that mass balances were robust and incorporated the inherent variability in wastewater flows, solids production, and
residence times within the system, the mass balances on each
unit process were calculated using three time periods: the date
of sampling (25 June 2002), the 3-d period centered on the sampling date (24–26 June 2002), and the 7-d period centered on
the sampling date (22–28 June 2002). Mass balances did not
vary significantly (p = 0.96) among these three periods, and
as noted above, the 25 June 2002 flows were used in calculating all mass balances and PBDE fluxes discussed in the paper.
See Supplementary material Table D6 for mass balances over
the three time periods on each major unit process.2 Molar balances were also calculated for all unit processes and the three
sampling periods. In the case of treatment-induced debromination processes, the mass balance may be reduced, but the molar
balance should remain near 100% as long as other degradative
pathways and loss processes are not active. All molar balances
© 2005 NRC Canada

Related documents

j environ eng sci 4 2005 353 367
j environ sci health 40 2005 535 551
beginners guide to wastewater treatment for paper industry
j environ eng sci 4 2005 369 383
treatment of wastewater from pharmaceutical industry

Related keywords