Upload
ram-prasad
View
213
Download
0
Embed Size (px)
Citation preview
Available at www.sciencedirect.com
.els
ind
Ram Prasad, Gary C. Schafran
a r t i c l e i n f o
Article history:
Received 10 August 2004
Received in revised form
26 October 2005
Accepted 11 November 2005
Available online 6 January 2006
the full-scale plant and may be associated with polymer carry through.
oceangoing vessels estimated to use TBT-containing paints
us in
and
and
t al.,
Concern about the presence of TBT in the environment has
ARTICLE IN PRESS
WAT E R R E S E A R C H 40 ( 2006 ) 453 462Corresponding author. Tel.: +1 757 683 3753; fax: +1 757683 5354.(de Mora, 1996; Champ and Seligman, 1996). As an anti-
fouling agent, it inhibits the settlement and attachment of
marine organisms (e.g. barnacles) to ship hulls by continu-
ously releasing the paint into the surrounding water. It is
grown over the past three decades due to its known
toxicological characteristics. To date, TBT has been the
subject of over 400 studies of its occurrence and the negative
health impacts on various marine organisms associated with
0043-1354/$ - see front matter & 2005 Elsevier Ltd. All rights reserved.doi:10.1016/j.watres.2005.11.024
E-mail addresses: [email protected] (R. Prasad), [email protected] (G.C. Schafran).(biocide) in marine paints applied to ship hulls for more than
30 years. Its use has been widespread with up to 70% of 2004).1. Introduction
Tributyltin (TBT) has been used as an anti-fouling agent
widely present in marine coastal systems and ubiquito
the water column and sediments in regions near ports
shipping activities (Kram et al., 1989; Quevauviller
Donard, 1990; Dowson et al., 1992; Hoch, 2001; Wade e& 2005 Elsevier Ltd. All rights reserved.Keywords:
Tributyltin (TBT)
Coagulation
Treatment
Granular activated carbonent, Old Dominion University, Norfolk, VA 23529, USA
A B S T R A C T
Characterization and treatment studies were conducted in an effort to evaluate treatment
options capable of removing tributyltin (TBT) in shipyard waters from above 1,000,000ng/L
to effluent concentrations below 50ng/L. Laboratory studies and operation of a full-scale
treatment plant were used to examine treatment options for TBT removal and included
physicochemical treatment processes of coagulation-clarification, filtration, and granular
activated carbon (GAC) adsorption.
Significant variability was observed in TBT-containing shipyard waters (generated from
different ships) in terms of their particulate solids, conductivity, and TBT and dissolved
organic carbon concentrations. Laboratory tests with aluminum sulfate and ferric sulfate
showed that on average 90% of TBT in shipyard waters could be removed by coagulation-
flocculation-clarification under optimum conditions. No statistically significant difference
was found in TBT removal capabilities between the two metal salts when compared at
equivalent metal doses and coagulation pH. Much lower removals were observed for the
coagulation-flocculation-clarification portion of the full-scale plant while the complete
full-scale treatment plant averaged 99.8% TBT removal over a period of 3 years. While
relatively high % removals were achieved, the total treatment process did not consistently
remove TBT to levels that wouldmeet the regulatory requirements proposed (50ng/L) at the
time the study was conducted. Based on the results from limited efforts to characterize the
dissolved and particulate TBT fractions in the full-scale treatment plant effluent,
particulate TBT was observed to be the dominant component of the effluent TBT fromCivil and Environmental Engineering Departmjournal homepage: www
Characterization of tributyltinremoval through laboratory anevier.com/locate/watres
shipyard waters andfull-scale treatment
and high concentrations of organic particulate material. The
ability to degrade TBT by ultraviolet (UV) photolysis using a
ARTICLE IN PRESS
0 (its presence in the aquatic environment (Alden et al., 1996). In
situ and ex situ studies have been conducted in many waters
and under various conditions and consistent observations of
growth inhibition and interference with organism reproduc-
tion have been reported (Hoch, 2001). Recent studies have
shown that TBT is a potent endocrine disruptor for lower level
organisms while it has been observed to be a dopamine
inhibitor in mammals (Kim et al., 2002). Lower level organ-
isms typically have shown impairment at concentrations as
low as 1ngTBT/L while higher-level organisms show health
impairment at concentrations as low as 1mgTBT/L (Hoch,
2001). The ability to cause imposex in molluscs at concentra-
tions in water of 1ngTBT/L has been reported repeatedly
(IPCS, 1999 (TBTO)) and particularly illustrates the potential
consequence of TBT being present in the aquatic environ-
ment.
TBT is an organometallic molecule that exhibits character-
istics of both its organic and metallic moiety and these
characteristics influence its fate in the environment. The
molecule carries a monovalent charge in the absence of
complexation with an anionic ligand and exhibits pH-
dependent solubility with minimum solubility at approxi-
mately pH 7 (Inaba et al., 1995). It has a high affinity for
organic matter and strongly partitions to the surface of
particulate matter present in water (Unger et al., 1988; Harris
et al., 1996). Because of these characteristics, its presence in
dissolved form in water is generally thought to be brief and it
has been observed to accumulate in sediments (Watanabe et
al., 1995; Langston and Pope, 1995).
National governments and regulatory agencies have moved
to limit or ban the release of TBTwhere the need for aquatic
life protection has been indicated. A number of governments
have instituted outright bans on the use of TBT-containing
paints while others such as the US have instituted a ban on its
use on vessels under 25m in length. Within the US, the
Commonwealth of Virginia was the first state-level govern-
ment to impose controls on its release by instituting a limit on
water discharges of 50ngTBT/L under the state-administered
National Pollutant Discharge Elimination System (NPDES)
permit. The imposition of this limit prompted Virginia
shipyards, beginning in the mid-1990s, to examine options
to reduce TBT in shipyard discharges to less than 50ngTBT/L.
Activities in shipyards, primarily ship repair/maintenance
activities, can produce TBT-containing water in several ways.
When a ship is placed in dry dock, washing the hull to remove
any attached organisms/slime layer is conducted so that the
condition of the hull paint can be inspected. Water generated
by this lower pressure cleaning is referred to as washwater in
this paper. After cleaning and inspection, if the hull is to be
painted, hydroblasting (high-pressure water spray) is often
used to remove either the surface paint coating or remove all
the paint layers to bare metal. Hydroblasting pressures can
range from 1000 lb in2 (7MPa) to approximately 30,000 lb in2
(210MPa) but were not recorded in this study. Volumes of
water generated depend on the size of the ship and the
amount of hull work to be conducted but well over a million
liters of water can be generated by these operations on large
(300-m long) vessels. In addition, any precipitation that falls
WAT E R R E S E A R CH 4454at any point during the time when a ship with TBT-paint is in
dry dock produces TBT-containing water. TBT-containingmercury lamp was examined and confirmed by Fletcher and
Lewis (1999) in a laboratory study where synthetic and dry
dock samples were treated with a low output UV light. They
observed solution pH, turbidity, organic matter content, and
composition of the anti-fouling paint as important variables
affecting removal.
In the study reported here, characterization of shipyard
TBT-containing waters, bench-scale coagulation studies, and
evaluation of TBT removal with a full-scale process train
consisting of coagulation-flocculation-clarification, filtration
and granular activated carbon (GAC) are presented. The
studies reported here were conducted to characterize the
composition of TBT-containing shipyard waters collected
from different shipyards and different regions, to evaluate
in bench-scale studies the effectiveness of a ferric salt and
alum, and to evaluate the long-term performance (ability to
meet a 50ng/L discharge standard) of a full-scale treatment
plant used to treat all TBT-containing waters discharged from
Virginia shipyards during the course of this study.
2. Materials and methods
In Phase 1 of the study, waters from a number of ships at
various shipyards were collected and analyzed. The TBT-
containing waters were collected in one or two 20-L poly-
ethylene containers from shipyards in southeastern Virginia,
the Gulf Coast, and the Pacific Northwest during this effort
from November 1999 through June 2000.
Samples generated in Virginia were packed in ice and
returned to the laboratory within a few hours of sample
collection. For samples collected outside of the region, they
were returned on ice by overnight mail for analysis at Old
Dominion University. In addition to the water samples
generated from ship hull washing or paint removal, one
sample was generated from washing down a steel plate that
had been freshly painted and then dried for 3 days before
washing. Total and dissolved (determined following filtra-water can also be generated in smaller volumes (o100,000L)at shipyards through replacement of water inside certain
hull-mounted radar domes. Additional details on the genera-
tion of TBT-containing water at shipyards can be found
elsewhere (Schafran et al., 2003).
The promulgation of an effluent TBT discharge standard in
Virginia prompted the studies reported in this paper. While
considerable research has been conducted on occurrence and
health effects of TBT, little has been published on treatment
processes to remove TBT from solution (Messing et al., 1997).
Argamen et al. (1984) and Fent and Muller (1991) examined
TBT in domestic wastewater and observed that 8090% in
untreated wastewater was associated with particulate matter
and was removed when the particulate material was retained
during treatment. Donard et al. (1993) and Fent (1996)
confirmed the findings of Fent and Muller but all of these
studies examined low TBT concentrations (o 200ngTBT/L)
2006 ) 453 462tion with a 0.45mm polycarbonate filter) fractions of TBT
were determined, as well as total suspended solids (TSS),
ARTICLE IN PRESS
0 (conductivity, pH, and dissolved organic matter measured as
dissolved organic carbon (DOC).
The Phase I waters used for characterization were also
utilized for laboratory-scale treatment evaluation that in-
cluded coagulation. Since most TBT-containing shipyard
waters contain high concentrations of suspended solids and
TBT has a propensity for partitioning to particles, efforts to
remove particulate matter were determined to be essential
for high levels of TBT removal. Removal of dissolved TBTwas
targeted to occur primarily on GAC, since it was observed that
TBT had a high affinity for this material based on preliminary
testing. In this paper, TBT removal on GAC is presented only
for the full-scale treatment process. Laboratory observations
of TBT removal by GAC and destruction by advanced
oxidation process studies will be presented elsewhere.
Two coagulants were utilized during the laboratory coagu-
lation studies to assess the removal of TBT; commodity-grade
alum (Al2(SO4)3 14H2O, 58gAl/L, General Chemical) and ferricsulfate (Fe2(SO4)3, 190 gFe/L, Midland Resources). Both coagu-
lants were used as received and without dilution before
addition to wash water during the coagulation studies. Prior
to conducting a jar test for each water, portions of the sample
were poured out and dosed at the desired coagulant doses
and then subsequently titrated with base to the desired pH
values (pH 5.510.0) for subsequent coagulation evaluations
(i.e. jar tests). This procedure was followed so that the
amount of base needed to obtain a target coagulation pH
could be determined. Following the base dose requirement
determination, a 2-L aliquot of sample was poured into a
square beaker (gator jar) and placed on a gang stirrer (Phipps
and Bird) where mixing was begun. The previously deter-
mined base dose was then added and allowed to mix for
several seconds and then followed by coagulant addition.
Coagulant doses for alum ranged from 22 to 242mg/L as alum
and for ferric sulfate 15164mg/L as ferric sulfate and were
based on preliminary treatment of shipyard waters prior to
this study. Rapid mixing was conducted at 200 rpm for 2min
followed by slow mixing (simulating flocculation) for 20min
at 20 rpm. While it is common to conduct multiple coagulant
tests simultaneously with a gang stirrer, in this study only
one sample at a time was treated. This procedure allowed for
continuous monitoring of the solution pH during each test
and allowed for small additions of acid (0.01N HCl) or base
(0.01N NaOH) as needed to maintain the target pH. Following
the 20-min slow mixing period, mixing was stopped and the
flocculated suspension was allowed to settle for 3h. A 500mL
sample was then removed from each jar and analyzed for
TBT, turbidity, and DOC.
A laboratory study involving coagulant aids (i.e. organic
polymers) to evaluate the potential affinity of TBT for these
molecules was also conducted. Five polymers (one cationic,
three anionic, and one non-ionic) acquired from a single
manufacturer were used in a modified jar test to evaluate
whether TBTpolymer interactions occur and whether a
polymer migrating through the treatment process train could
cause TBT to similarly carry through the process train. A
single TBT solution made to a dose of 2400ng/L was divided
into 2-L solutions and dosed with polymer at 1mg/L as
WATER RESEARCH 4product with one jar serving as a control (no polymer
addition). Sample aliquots were removed from each jar withone being filtered through a 0.45mm polycarbonate filter and
the other aliquot unfiltered. The polymers used in coagulation
were all large (long-chain, high molecular weight) molecules
capable of being substantially removed through filtration with
a 0.45mm membrane filter. Consequently, if polymer-treated
waters exhibited low concentrations of TBT after filtration, it
was deemed to be the result of removal of the polymer and
associated TBT.
A major component of this research effort was the
demonstration of TBT removal using a full-scale treatment
plant that was mounted on a barge and transported among
the shipyards as needed to provide treatment of TBT-contain-
ing waters. The treatment plant was designed and con-
structed for this study with a capability of treating water at
380L/min but was operated during the study at 190225L/
min. The treatment process train consisted of rapid mix and
flocculation units followed by a dissolved air flotation (DAF)
tank for clarification and a sand filter and two activated
carbon columns (GAC1 and GAC2) operated in series. Coagu-
lation was studied using alum for a limited period of time and
ferric sulfate the majority of the time. Hydrated lime was
added in the rapid mix tank with the coagulant while an
organic polymer was added at the entrance to the flocculation
tank to promote floc development. Retention times in the
treatment plant were approximately 2min in the rapid mix
tank, 20min in flocculation, and 75min in the DAF tank based
on the theoretical hydraulic residence time calculations and
16min in the sand filter, GAC1 and GAC2 based on tracer
study results at 190 L/min. Water passing through the full-
scale treatment plant was typically coagulated at higher pH
values (between 9 and 10) based on optimizing particle
removal in the DAF. During the early part of the study, the
water was applied to the filter and activated carbon columns
at this higher pH while after October 2000 the water was
acidified to pH 7 before passing through the sand filter and
the activated carbon columns. During all phases of the study,
a state-licensed wastewater treatment plant operator oper-
ated the full-scale plant in compliance with state discharge
permit requirements.
TBT sample concentrations varied by over six-orders of
magnitude in this study necessitating extreme care to prevent
cross-contamination of samples. A quality assurance protec-
tion plan (QAPP) and quality management plan (QMP) for the
study following EPA guidelines was developed at the begin-
ning of the study (Champ et al., 2000) and included standard
operating procedures for all aspects of sample collection,
handling, processing, analytical analyses, and data manage-
ment. All samples were collected in pre-cleaned polycarbo-
nate bottles since TBT has been shown not to adsorb to this
material (Unger, 1999). All sample containers were washed
with soap, methanol, and HCl and copiously rinsed with
dionized distilled water between each wash. Only new
sample containers were used for collection of samples from
the effluents of the first (GAC1) and second stage (GAC2) GAC
contactors as a precaution to prevent contamination. During
field sampling, samples collected from the influent, post-DAF,
and post-sand filter were placed in separate coolers on ice
from the samples collected after GAC1 and GAC2. This
2006) 453 462 455separation continued at the laboratory where separate cold
storage facilities (in separate rooms) were used prior to
sample analysis. Sample preparation prior to analysis also
occurred in these separate rooms prior to transporting the
samples to the analytical laboratory. After returning field
samples to the laboratory, the outside of the sample contain-
ers was washed with tap water and dried before opening.
Aliquots of sample were then removed for TSS, DOC, and
conductivity analysis, and then the samples were acidified to
pHo2 with trace metal grade HCl. Influent and post-DAFsamples were placed on a shaking table for 24h after
acidification to promote dissolution of particulate TBT.
After 24h of shaking, these samples were removed and an
aliquot of sample was filtered through a 0.45mm polycarbo-
nate filter and then saved for TBT analysis.
Use of field and trip blanks were incorporated into the study
and no incidents of contamination were detected during the
study. Analytical blanks and spike recoveries were examined
on a daily basis and any spike recovery outside of 720% was
cause for rejection of all analyses conducted on that date.
Similar care and procedures were exercised when handling
depending on the pH range of samples. Calibration was
conducted and Nernstian response (D5973mV/DpH) checked
daily.
Conductivity was measured with a YSI Model 35 conduc-
tance meter. The instrument was calibrated using Standard
KCl solution of 0.00100M (Clesceri et al., 1998; Method 2510B).
TSS measurements were conducted following SM 2540D
(Clesceri et al., 1998) with copious amounts of DI water used
to rinse the filter following sample filtration.
3. Results and discussion
3.1. Characterization of TBT-containing shipyard waters
Shipyard TBT-containing waters in the characterization and
laboratory treatment study (Phase I; 19992000) were ob-
served to be highly variable in TBT concentration spanning a
ARTICLE IN PRESS
stic
tivi
,300 7.6 696 21.5
610 6.5 2130 45.5
350
,00
,50
287
56
108
89
000
251
WAT E R R E S E A R CH 40 ( 2006 ) 453 462456untreated water and treated samples generated in the
laboratory (e.g. following coagulation studies). All samples
were acidified and held in cold storage (4 1C) prior to analysis.
TBTwas analyzed by the hydride generation-AAS detection
method of Hodge et al. (1979) using an atomic absorption
spectrophotometer (Buck Scientific Model 210 VGP) fitted with
a custom quartz burner. Standards were made by serial
dilution of stock solutions (all from Sigma-Aldrich) of TBT
chloride (TBT, 96%), dibutyltin dichloride (DBT, 96%), and MBT
(butyltin trichloride, 95%) to quantify organotin concentra-
tions. The analytical detection limit was determined to be
1243ng/L (mean 23ng/L). Analytical determination of DOCwas conducted using a high-temperature catalyzed oxidation
procedure (Clesceri et al., 1998; SM 5310B) with a Shimadzu
TOC 5000 with four standards bracketing the range of
samples. Turbidity was measured with a Monitek TA1
nephelometer calibrated daily. pH values were determined
with an Orion Model 520 pH meter with Ross electrode and
calibrated with pH 4 and 7 or 7 and 10 buffer solutions
Table 1 Phase I shipyard TBT-containing water characteri
Ship ID TBT (ng/L) Conduc
Total Dissolved
CV1Fa 1,150,000 n.a. 11
CV1Ha 1,440,000 n.a.
CV3a 1,830,000 12,900 1
CV4 53,400 5050 11
CV5 8290 740 18
CV6a 6,260,000 926,000
CV7 5500 330
CV8 41,200 390
CV9 39,400 160
CV10 20,900 3138 9
CV11b 28,700 9220
n.a.not analyzed.
a Hydroblast water; all others are washdown (lower pressure) waters.b Freshly painted steel plate.6.8 574 13.9
0 7.4 119 7.1
0 7.2 14 3.4
7.1 4340 24.8
7.2 17 3.0
7.3 540 10.2
7.3 1030 13.4
7.3 17 7.4
7.4 15 4.3range of less than 55006,260,000ngTBT/L (Table 1). Higher
TBT concentrations were observed in water generated
from hydroblast operations (median concentration 1.6106ngTBT/L) while TBT concentrations in wash water were
lower in comparison (median concentration 2.9105).Based on the limited observations presented here it appears
that elevated concentrations (41,000,000ngTBT/L) can occur
through the removal of TBT hull coatings particularly when a
minimum of water is used and dilution with other waters (e.g.
rain, non-contact cooling waters) is minimized or prevented.
The TBT in these waters was primarily associated with
particles as dissolved TBT averaged only 9.8% of total TBT.
This large proportion of TBT associated with particulate
material clearly illustrates the need for effective particle
removal to lower TBT concentrations in these waters. It also
illustrates the likelihood that the fate of previous TBT
discharges from these sources to waterways surrounding
shipyards was removal from the water column through
settling.
s
ty (m
A
) pH TSS (mg/L) DOC (mg/L)
Concentrations of organic matter (DOC) in these waters
were also variable ranging from 3.0mgC/L (only slightly
higher than the source potable water) to 45.5mgC/L). Elevated
concentrations of DOC can potentially impact treatment
since aqueous organic matter has been observed to complex
with TBT potentially making it more difficult to remove from
solution (OLoughlin et al., 2000). A distinct difference in the
concentration of DOC was observed between hydroblast-
generated waters (mean7s.d. 26.4713.5mgC/LDOC) andwashwaters (mean 7.474.0mgC/LDOC). A similar differ-ence between source waters was noted for TSS with higher
concentrations of TSS associated with hydroblast waters
4. Coagulation studies
The large proportion of TBT associated with particulate
material in TBT-containing shipyard waters necessitates a
high level of particle removal to achieve low TBT concentra-
tions before these waters can be discharged to a receiving
water. To determine the influence of coagulation conditions
on TBT removal, 10 of the 11 waters (Table 1) collected for the
Phase I characterization and laboratory treatability study
were coagulated under laboratory conditions and examined
for TBTremoval. In this effort, coagulant type, coagulant dose,
and coagulation pH value were the variables examined for
their influence on the removal of TBT. As noted previously,
coagulant doses for alum and ferric sulfate ranged from 22 to
121mg/L and 15164mg/L (0.372.0104M as metal for each
ARTICLE IN PRESS
Fig. 2 Comparison of TBT removal by ferric sulfate and
WATER RESEARCH 40 (2006) 453 462 457(1940mg/L) compared to wash waters (250mg/L).
The higher DOC concentrations in the hydroblast waters
are likely the result of solubilization of organic material from
the paint matrix (present in the water as fine particulate
material) after removal of the paint from the ship hulls. The
significance of higher DOC concentrations in waters with
higher TBT concentrations is that any treatment process that
employs a sorption process where both TBT and organic
matter both compete for the same surface sites will have
greater difficulty in achieving lower effluent TBT concentra-
tions. The interference of DOC in the sorption of TBT on GAC
has been clearly shown in batch isotherm studies (Unger and
Schafran, 2003) and similar interference would be expected in
a continuous flow process.
A greater number of TBT-containing water samples was
collected and analyzed during full-scale treatment efforts
over a 3-year period (December 1999 through November 2002)
and the samples collected during this effort can be used to
further characterize TBT-containing waters generated at
shipyards. During this effort, there were 136 days of treatment
in which a discharge occurred to a receiving water and 119
influent samples were analyzed. Influent concentrations
ranged from a low of 800 up to 1.3106ngTBT/L with meanand median values of 253,000 and 137,000ngTBT/L, respec-
tively. A number of the wash and hydroblast waters were co-
mingled before treatment so no attempt was made to
characterize concentrations by the water generation activity.
The distribution of concentrations for these waters is
illustrated in Fig. 1. Based on these influent concentrations
the treatment requirement to reach 50ng/L was observed to
vary between minimum and maximum values of 93.7% and
99.996%, respectively, with a median value of 99.96% TBT
removal required to reach 50ng/L.
0
0.2
0.4
0.6
0.8
1
1.2
100 1000 10000 100000 1000000 10000000
Frac
tion
of V
alue
s Le
ss th
an In
dica
ted
Conc
entra
tionInfluent Total TBT (ng TBT/L)
Fig. 1 Influent TBT distribution for the full-scale treatment.coagulant), while coagulation pH values ranged from 5.4 to
10.0. The influence of a coagulant aid (organic polymer) was
not examined in the laboratory studies, in part due to
limitation in the amount of water available for coagulation
studies.
A significant finding of the coagulation study was that there
was no statistically significant difference (paired Students t-
test, 95% confidence interval, p 0:4428, n 47) in TBTremoval when comparing alum and ferric sulfate treatment
under similar coagulation conditions (Fig. 2). Mean values for
TBTremoval were 87.1% and 86.8% for alum and ferric sulfate,
respectively, and most likely reflected removal of TBT
associated with particulate matter in these solutions. Mean
turbidity removal values of 96.0% and 95.0% for alum and
ferric sulfate indicate a higher level of particle removal than
TBT in these waters. These results likely reflect dissolved TBT
remaining in solution after coagulation and/or a greater
proportion of TBT associated with small colloidal material
remained than the particulate material removed via coagula-
tion. Overall, no statistically significant correlation between
TBT and turbidity removal was observed.
The influent characteristics of the waters treated in the
coagulation study were widely variant in terms of DOC
concentrations and conductivity (i.e. salinity) as well as
particulate matter and total and dissolved TBT (Table 1).
It was expected that DOC removal would increase with
40
50
60
70
80
90
100
40 50 60 70 80 90 100% TBT Removal with Alum
% T
BT R
emov
al w
ith F
erric
aluminum sulfate under comparable coagulation
conditions.
near the end of each treatment event is presented in terms of
ARTICLE IN PRESS
0 (increased coagulant dose and conditions that favored TBT
removal, since DOC is frequently removed to a significant
extent when natural waters are coagulated (Dempsey et al.,
1984). However, DOC removal from these waters was often
poor and under many conditions an increase in DOC was
detected following coagulation (not shown). These observa-
tions may reflect the formulation of the paints that are
designed to ablate (dissolve) and possibly that a significant
hydrophilic fraction of organic matter that is not well
removed by coagulation was present in these waters.
Solution conductivity in these waters exceeded two-orders
of magnitude and potentially might play a role in TBTremoval
since it can influence electrostatic conditions near the surface
of particles, the moiety of ionic organic compounds, and the
binding that can occur between TBT and natural organic
matter (OLoughlin et al., 2000). Examination of these para-
meters revealed no correlation between conductivity (sali-
nity) and TBT, DOC, or turbidity removal in the laboratory
coagulation studies.
Of particular interest in understanding the fate of TBT
during coagulation is whether dissolved TBT is removed
through interaction with the precipitate of the metal coagu-
lant formed during coagulation. Many dissolved organic
molecules can adsorb to or form surface complexes with
ferric and aluminum hydrous oxides and be removed from
solution by this mechanism (Pommerenk and Schafran, 2005).
To evaluate whether dissolved TBT could be removed via a
similar mechanism, a jar test study evaluating TBT removal
from a synthetic solution (100,000ng/L TBT, 40g/L NaCL) was
conducted using ferric sulfate at coagulant doses of
14326mg/L and at coagulation pH values of 6.5, 7.5, and
8.5. Removal of TBT was similar for pH 6.5 and 7.5
(mean 63% and 64% TBT removal, respectively) while TBTremoval did not occur at pH 8.5 (mean 5% TBT removal).These results indicate the importance of coagulation pH in
terms of removing dissolved TBT.
The lower removal of TBTat pH 8.5 may be related to effects
on the coagulant metal or the solubility of the TBT molecule
or a combination of the two. The net electrical surface charge
on the precipitated ferric hydrous oxide would be expected to
be near zero or slightly negative at pH 8.5 while at pH 6.5 and
7.5 the charge on the precipitate would be positive (Dzombak
and Morel, 1990). This condition suggests that the interaction
(sorption) between dissolved TBT and hydrous oxide is
favored when a hydrous metal oxide is positively charged.
The influence of the coagulation pH value on TBT removal
may also be related to the form of TBT present in solution.
TBT solubility has been reported to be minimum in the
neutral pH range (pH 68) at approximately 1mgTBT/L and
more soluble at lower and higher pH values (Inaba et al.,
1995). TBT solubility in solution has also been reported to be
negatively correlated to salinity (Inaba et al., 1995) but little
change occurred in total ionic composition under the
coagulation conditions examined here so it is expected that
variations in ionic strength had little influence on solubility.
Hydroxide complexes of TBT (TBTOH) can occur and
would be favored at higher pH values while at decreasing
pH values it would be expected that cationic TBT species
WAT E R R E S E A R CH 4458would become dominant. A similar influence of pH on the
sorbent surface charge would be observed with a positivelya probability distribution to illustrate the effectiveness of the
full-scale treatment plant in meeting the 50ng/L treatment
goal (Fig. 3). During the 136 days of treatment the o50ng/Lgoal was achieved 41% of the time and exceeded 59%. When
examined in terms of the level of treatment achieved the
treatment plant was able to treat consistently to greater than
90% TBT removal and achieved 99.9% TBT removal in more
than 75% of the time. There was no statistically significant
relationship observed between paired influent and effluent
TBT concentrations (Fig. 4) and there was generally a two-
orders of magnitude difference in TBT removal observed
across the range of influent concentrations treated. It is not
surprising that there is scatter in this relationship since thecharged surface at lower pH and negatively charged surface at
higher pH (Stumm, 1992). If ionic interaction is important in
TBTremoval, the change in speciation of both the sorbent and
the sorbing species may favor removal in a narrow pH range
where charges are opposite or where they are substantially
lowered allowing non-ionic forces to affect attraction be-
tween the TBT molecule and the sorbent surface. Hoch et al.
(2002) observed this narrow pH range effect for TBT adsorp-
tion with four mineralogically distinct clay-enriched sedi-
ments and postulated a zone of maximum adsorption in the
pH 67 due to favorable ionic charges between the soluble
species and the sorbent in this pH range.
5. TBT removal in the full-scale treatmentplant
The primary focus of the full-scale treatment study was to
determine whether a conventional treatment process train
that is present in many shipyards to treat waters containing
trace metals would be capable while operating in shipyards
under shipyard conditions of meeting the treatment
objective of o50ng/L TBT. The full-scale treatment plantwas operated in this study from December 1999 until
November 2002 treating TBT-containing waters typically
shortly after they were generated. During this effort a number
of operational and treatment conditions were evaluated
including dose and type of coagulant, coagulation pH, water
flow rate (and contact time in GAC contactor columns), use of
hydrogen peroxide, and use of powdered activated carbon
(PAC). While there were multiple treatment variables exam-
ined in this study, these conditions were varied over short
durations and the majority of the time the treatment plant
was operated using ferric sulfate (2872mg/L; mean 52mg/L) at a coagulation pH of 910, using a coagulant aid
(2.810.2mg/L; mean 7.5mg/L polymer), no PAC, no hydro-gen peroxide, and operated for a duration of less than 12h.
One significant change that was made in October 2000 was
the depression of pH from 910 to approximately 7 as water
exited the DAF unit entering the filter. This change was made
in an effort to improve TBT removal on the GAC columns
based on observations in batch laboratory studies (not
reported here).
The TBT concentrations measured in samples collected
2006 ) 453 462waters being treated varied greatly in composition (e.g.
2.136.9mgC/L DOC and 42020,400 m
A
/cm conductivity),
and the distribution between dissolved and particulate TBT
likely also varied.
If improvements are to be made in the removal of TBT from
these waters it is imperative to understand the fate of TBT in
the treatment process train. Toward this goal, samples were
collected on a number of occasions immediately following
each unit process and examined to determine the effective-
ness of each unit process in removing TBT. The results for one
period of treatment when the influent concentrations were in
the upper 20% of observed values during the study are
provided in Table 2. The first removal process in the treatment
process train is the DAF clarifier and during this period an
average of 46.1% TBT removal occurred within the DAF unit.
Since this process removes the majority of particulate
material contained in the influent water it is apparent that
a substantial amount of dissolved TBTor TBT associated with
fine particulate material is carried through this process. The
performance of the DAF clarifier was well below the removals
observed in the Phase I coagulation studies but this perfor-
mance is likely due to the optimal conditions and extended
settling time provided in laboratory jar tests (corresponding to
a loading rate of 0.033Mh1) versus the considerably higher
loading rate occurring in the DAF (1.7Mh1). Wave action on
the barge was also observed to periodically cause water in the
DAF to slosh back and forth and cause floating solids on the
surface to carry over. This may also have decreased the TBT
removal efficiency of the DAF.
concentrations in the GAC2 influent, a lower concentration
gradient, and a lower resultant mass flux to the GAC surface.
ARTICLE IN PRESS
0
0.2
0.4
0.6
0.8
1
1.2
1 10 100 1000 10000
Effluent TBT (ng/L)
Frac
tion
of V
alue
s Le
ss th
an
Indi
cate
d Co
ncen
tratio
n
50
Fig. 3 Effluent TBT concentrations from the full-scale
treatment plant measured for regulatory
compliance.
100
1000
10000
BT (n
g/L)
99% Removal 99.9% Removal99.99% Removal
erc
WATER RESEARCH 40 (2006) 453 462 4591
10
100 1000 10000 100000 1000000 10000000
Influent TBT (ng/L)
Efflu
ent T
Fig. 4 Relationship between influent and effluent TBT
concentrations for the full-scale treatment plant.
Table 2 Summary statistics for TBT concentrations and p30December 7, 2000 Period
Influent DAF effluent
Mean conc. (ng/L) 514,500 277,167
Median conc. (ng/L) 405,000 235,000
% Removal based on mean N/A 46.1
% Removal based on median N/A 42.0N/Anot applicable.It was suspected that the TBT in the influent to GAC2 may
also include a fraction of TBT not as amenable to removal as
the TBT entering the GAC1 column and that this TBT fraction
carried through to the effluent. The passage of particulate
TBT through the TBT treatment plant and the GAC2 contactor
column was suspected and efforts to investigate this source
issue are described below.
6. Fractionation of TBT in the full-scaletreatment plant effluent
TSS concentrations were monitored in the effluent of the full-
scale treatment plant under requirements of the discharge
permit and were measured for the 136 days of full-scale
treatment (effluent requirement was o60mg/L TSS). The TSSconcentration in the effluent ranged from less than 1mg/L up
to a maximum of 30mg/L with a median value of 2.3mg/L.
While the potential of particulate TBT to contribute to the
final effluent exists, a regression analysis of effluent TBT
and TSS concentrations showed there was no statistically
ent removal by individual unit process for the November
Sand filter effluent GAC1 effluent GAC2 effluent
257,000 1420 368
190,000 428 262
3.9 99.40 74.1
11.1 99.80 38.80The sand filter was capable of removing additional TBT of
3.9% (mean for the period) with this removal likely represent-
ing retention of TBT associated with fine particulate matter.
The GAC1 contactor provided the highest level of TBTremoval
in the treatment process train removing over 99% of the TBT
contained in the influent to the GAC1 column. This removal is
most likely related to the removal of dissolved TBT though
additional fine particulate TBT may also be removed at this
stage. The GAC2 contactor column removed 74.1% (mean) of
the TBT entering the column but on average only 0.4% (1050
vs. 256,000ng/L) of the amount of TBT removed by the GAC1
column. This low removal of TBT in the GAC2 column was
observed throughout the study and is consistent with low
through the DAF, sand filter, and two carbon contactors in
tion with TBT, it is clear that TBT can associate (e.g. sorb,
ing TBT generated at shipyards illustrates that TBT concen-
trations vary widely (total TBT concentrations between 5500
and 6,260,000ngTBT/L and that TBT concentrations are
dominated by the particulate fraction. Dissolved organic
matter and solution conductivity also were observed to vary
dramatically, which can potentially impact TBT removal via
various treatment processes.
Both ferric sulfate and aluminum sulfate (alum) were used
in coagulation studies of TBTremoval and both were observed
capable of substantial particulate solids and TBT removal.
When compared at comparable coagulation conditions (coa-
gulation pH andmolar metal dose) no significant difference in
treatment was observed between the two coagulants.
Treatment of shipyard TBT-containing waters with the full-
scale treatment plant demonstrated that TBT concentrations
could be lowered by typically over three-orders of magnitude
(99.9%). The DAF and GAC1 were the primary sites of removal
with the DAF unit removing primarily particulate TBT and
ARTICLE IN PRESS
0 (series. Consistent with these observations were those re-
ported by Rowland et al. (2000) and de Rosemond and Liber
(2004) where toxicity in the effluent of municipal and
industrial wastewater treatment processes was observed
and attributed to the passage of polymers, added during the
treatment process, through to the final effluent.
A brief study to discern whether it was possible for TBT to
associate with (e.g. sorb, complex) organic polymers was
conducted by evaluating five different polymers made up in
separate solutions and exposed to a concentration of 2400ng/
L TBT. Both filtered and unfiltered aliquots were analyzed
from each solution to assess whether TBT interacted with the
polymer used in each solution. The TBT concentrations in
unfiltered aliquots taken from the control (no polymer) and
all waters dosed with polymer were similar in TBT concen-
tration indicating that the presence of the polymer in solution
did not inhibit the analytical determination of TBT (Fig. 5).
The filtered control sample was slightly lower (8.5%) than the
unfiltered control and within the analytical uncertainty of the
method. The filtered aliquots of the polymer-treated waters
were all substantially lower than the unfiltered aliquots
(2791% lower) indicating that the polymer removed during
filtration was clearly also causing the removal of TBT. Thesignificant relationship, indicating that TSS cannot be used as
a surrogate indicator of TBT concentrations.
To determine to what extent TBT concentrations in the
effluent could be associated with particulate matter, samples
collected from the effluent of the full-scale plant during the
first 2 weeks of November 2000 were analyzed for both total
and dissolved TBT (n 10). Effluent total TBT concentrationsranged from 67 to 1470ng/L, dissolved TBT concentrations
ranged from 52 to 276ng/L and total TBT removals across the
treatment plant ranged from a minimum of 99.7% to a
maximum of 99.97% (median removal 99.88%). During theperiod of study particulate TBTwas consistently observed to
be the dominant fraction in the effluent. Particulate TBT
represented 090% of the total TBT concentration during this
period with mean and median values of 63% and 73%. TSS
concentrations during this effort ranged from 2 to 24mg/L
(median 4mg/L) and exhibited no trend with regard toparticulate TBT concentrations. The results indicate that a
significant proportion of the TBT in the final effluent of the
full-scale treatment plant was associated with particulate
matter, but also that TSS does not correlate well with the
particulate TBT fraction.
7. Polymer carry-through
The possibility that polymer added during the coagulation
process could potentially influence the movement of TBT
through the treatment process train to the final effluent was
investigated following a short study where a microfiltration
membrane (0.1mm) was used to treat effluent from the full-
scale treatment plant. The membrane repeatedly fouled over
a short period of time (days) and upon visual examination
appeared to be coated with polymer and iron that had passed
WAT E R R E S E A R CH 4460non-ionic polymer showed the lowest polymer-effect on TBT
removal but, because of the few polymer types examined,complex) with the polymer and that passage of polymer
through the treatment process train may cause TBT to also
pass through the process train. Because the polymerTBT
molecule can be filtered out with a 0.45mm filter, it is possible
that some of the particulate TBT determined in the
November 2000 fractionation effort may have been polymer-
associated TBT.
Analysis of the test polymers found one polymer to contain
TBT at a concentration of 120ngTBT/mg polymer; this
polymer was excluded from these experiments. This un-
expected finding suggests a possible external source of TBT to
the treatment process train but analysis of the polymer used
in the full-scale treatment plant confirmed it to be TBT-free.
8. Conclusions
The analysis of wash waters and hydroblast waters contain-there is insufficient information to conclude the influence on
TBT removal by a particular polymer type. Regardless of
whether a particular polymer type shows a greater associa-
0
500
1000
1500
2000
2500
Control Anionic E-42
Nonionic E-30
Cationic ZE 7873
Magnafloc-1011
(Anionic)
Selfloc-2250
(Anionic)
TBT
(ng/L)
UnfilteredFiltered
Fig. 5 Fate of TBT in solutions dosed with organic polymer.
2006 ) 453 462GAC1 removing what would be expected to be primarily
dissolved TBT. The final effluent from the full-scale treatment
mine effluent. Environmental Toxicology and Chemistry 23,22342242.
in Natural Systems. Wiley, New York, p. 428.
ARTICLE IN PRESS
0 (Donard, O.F.X., Quevauviller, P., Bruchet, A., 1993. Tin andorganotin speciation during wastewater and sludge treat-ment processes. Water Res. 27 (6), 10851089.
Dowson, P.H., Bubb, J.M., Lester, J.N., 1992. Organotin distributionin sediments and waters of selected east coast estuaries in theUK. Mar. Pollut. Bull. 24 (10), 492498.
Dzombak, D.A., Morel, F.M.M., 1990. Surface ComplexationModeling: Hydrous Ferric Oxide. Wiley, New York.plant exceeded the o50ng/L treatment goal 59% of the timeand may in part be due to fine, particulate TBT passing
through the treatment system. Evidence of organic polymers
being able to interact with TBT in a manner that could cause
carry through in the treatment system was observed in
laboratory studies.
Acknowledgements
Funding for portions of the work presented in this paper has
been provided through the USEPA, Virginia Department of
Environmental Quality, the Maritime Administration, Na-
tional Shipbuilding Research Program (NSRP), and the Virginia
Center for Innovative Technology. We wish to thank the NSRP
SP-1 Environmental Panel for assisting with acquiring TBT-
containing shipyard waters for the characterization and
laboratory treatment efforts. We specifically acknowledge
the many and varied contributions of T. Fox, M.A. Champ, T.
Tekleab, M. Unger, S. Tanaka, A. Diaz, J. Hirschman, A. Quick,
R. Sherman, P. Ford, K. Ogburn, F. Thorn, J. Soles, M. Ewing, F.
Wheatley, J. Bingham, and A. Rainsberger with this project.
R E F E R E N C E S
Alden, R.W., McDaniel, R.J., Ribeiro, T.A., Ramirez, L.M., Rose, J.H.,1996. A literature search for the effects of tributyltin onaquatic organisms and current tributyltin regulatory limitsamong the coastal states of the Unites States. TechnicalReport No. 3008. Applied Marine Research Laboratory, OldDominion University.
Argamen, Y., Hucks, C.E., Shelby, S.E., 1984. The effects oforganotin on the activated sludge process. Water Res. 18 (5),535542.
Champ, M.A., Seligman, P.F. (Eds.), 1996. Organotin, Environmen-tal Fate and Effects. Chapman & Hall, London 674pp.
Champ, M.A., Schafran, G.C., Fox, T.J., Cutter, G.A., Unger, M.A.,Salata, G., 2000. Quality assurance project plan for the analysisand treatment of TBT in waste waters from Virginia shipyardsand drydocks, Norfolk, VA, 170pp. (Revision 4.1.2000).
Clesceri, L.S., Greenberg, A.E., Eaton, A.D. (Eds.), 1998. StandardMethods for the Examination of Water and Wastewater, 20thed. APHA/WEF/AWWA.
de Mora, S.J., 1996. Tributyltin: Case Study of an EnvironmentalContaminant. Cambridge Environmental Chemistry Series.Cambridge University Press, New York, NY.
Dempsey, B.A., Ganho, R.M., OMelia, C.R., 1984. The coagulationof humic substances by means of aluminum salts. Journal ofthe American Water Works Association 76 (4), 141150.
de Rosemond, S.J.C., Liber, K., 2004. Wastewater treatmentpolymers identified as the toxic component of diamond
WATER RESEARCH 4Fent, K., 1996. Organotin compounds in municipal wastewaterand sewerage sludge: contamination, fate in treatmentprocess and ecotoxicological consequences. Sci. Total Environ.185, 151159.
Fent, K., Muller, M.D., 1991. Occurrence of organotins in municipalwastewater and sewage sludge and behavior in a treatmentplant. Environ. Sci. Technol. 25, 489493.
Fletcher, L.E., Lewis, J.A., 1999. Regulation of shipyard dischargesin Australia and the potential of UV oxidation for TBTdegradation in washdown waste water. In: Proceedings of theSpecialty Session on Treatment of Regulated Discharges fromShipyards and Drydocks. Oceans 99 Conference, Seattle, WA,September 1316, 1999.
Harris, J.R.W., Cleary, J.J., Valkirs, A.O., 1996. Particlewaterpartitioning and the role of sediments as a sink andsecondary source of TBT. In: Champ, M.A., Seligman, P.F. (Eds.),Organotin. London, ISBN:0412-582406.
Hoch, M., 2001. Organotin compounds in the environmentanoverview. Appl. Geochem. 16 (8), 719743.
Hoch, M., Alonso-Azcarante, J., Lischick, M., 2002. Adsorptionbehavior of toxic tributyltin to clay-rich sediments undervarious environmental conditions. Environ. Toxicol. Chem. 21,13901397.
Hodge, V.F., Seidel, S.L., Goldberf, E.D., 1979. Determination ofTin(IV) and organotin compounds in natural waters, coastalsediments and macro algae by atomic absorption spectro-metry. Anal. Chem. 51, 12561259.
Inaba, K., Shiraishi, H., Soma, U., 1995. Effects of salinity, pH, andtemperature on aqueous solubility of four organotin com-pounds. Water Res. 29 (1), 14151417.
International Programme on Chemical safety (IPCS), 1999. Con-cise International Chemical Assesment Documents, No. 14,Tributyltin oxide. World Health Organization, Geneva, Swit-zerland.
Kim, Y.M., Joon Lee, J., Yu Yin, S., Kim, Y., Lee, J.K., Yoon, Y.P., Kang,M.H., Lee, M.K., 2002. Inhibitory effects of tributyltin ondopamine biosynthesis in rat PC12 cells. Neurosci. Lett. 332 (1),1316.
Kram, M.L., Stang, P.M., Seligman, P.F., 1989. Adsorption anddesorption of tributyltin in sediments of San Diego bay andPearl Harbor. Appl. Organometall. Chem. 3, 523536.
Langston, W.J., Pope, N.D., 1995. Determinants of TBT adsorptionand desorption in estuarine sediments. Mar. Pollut. Bull. 31(13), 3243.
Messing, A.W., Ramirez, L.M., Fox, T., 1997. Document technolo-gies available to clean brackish waters to 50ppt TBT levels.NSRP Report 0508, US Department of the Navy, CarderockDivision, Naval Surface Warfare Center.
OLoughlin, Edward, J., Samuel, J.T., Yu-Ping, C., 2000. Associationof organotin compounds with aquatic and terrestrial humicsubstances. Environ. Toxicol. Chem. 19 (8), 20152021.
Pommerenk, P., Schafran, G.C., 2005. Adsorption of inorganic andorganic ligands onto hydrous aluminum oxide: evaluation ofsurface charge and the impacts on particle and NOM removalduring water treatment. Environ. Sci. Technol. 39, 64296434.
Quevauviller, P., Donard, O.F.X., 1990. Variability of butyltindetermination in water and sediment samples from Europeancoastal environments. Appl. Organometall. Chem. 4, 353367.
Rowland, C.D., Burton Jr., G.A., Morrison, S.M., 2000. Implication ofpolymer toxicity in a municipal wastewater effluent. Environ.Toxicol. Chem. 19, 21362139.
Schafran, G.C., Prasad, R., Thorn, F.H., Ewing, R.M., Soles, J., 2003.Removal of tributyltin in shipyard waters: characterizationand treatment to meet low parts per trillion levels. J. Ship Prod.19 (3), 179186.
Stumm, W., 1992. Chemistry at the Solid Water Interface:Processes at the MineralWater and ParticleWater Interface
2006) 453 462 461Unger, M.A., 1999. Department of Environmental Science. VirginiaInstitute of Marine Science, Gloucester Point, VA.
Unger, M.A., Schafran, G.C., 2003. Enhancing the remediation ofTBT-contaminated wastewaters through mechanistic investi-gations and treatment development. Final Report to the SeaGrant Technology Program.
Unger, M.A., MacIntyre, W.G., Huggett, R.J., 1988. Sorptionbehavior of tributyltin on estuarine and freshwater sedi-ments. Environ. Toxicol. Chem. 7, 907915.
Wade, T.L., Sweet, S.T., Quinn, J.G., Cairns, R.W., King, J.W., 2004.Tributyltin in environmental samples from the Former De-recktor Shipyard, Coddington Cove, Newport, RI. Environ.Pollut. 129 (2), 315320.
Watanabe, N., Sakai, S., Takatsuki, H., 1995. Release and degrada-tion half lives of tributyltin in sediment. Chemosphere 31 (3),28092816.
ARTICLE IN PRESS
WAT E R R E S E A R CH 40 ( 2006 ) 453 462462
Characterization of tributyltin in shipyard waters and removal through laboratory and full-scale treatmentIntroductionMaterials and methodsResults and discussionCharacterization of TBT-containing shipyard waters
Coagulation studiesTBT removal in the full-scale treatment plantFractionation of TBT in the full-scale treatment plant effluentPolymer carry-throughConclusionsAcknowledgementsReferences