Content uploaded by Arne R D Verliefde
Author content
All content in this area was uploaded by Arne R D Verliefde on Jan 10, 2015
Content may be subject to copyright.
Pharmaceutical removal during managed aquifer recharge
with pretreatment by advanced oxidation
K. Lekkerkerker-Teunissen, E. T. Chekol, S. K. Maeng, K. Ghebremichael,
C. J. Houtman, A. R. D. Verliefde, J. Q. J. C. Verberk, G. L. Amy
and J. C. van Dijk
ABSTRACT
Organic micropollutants (OMPs) are detected in sources for drinking water and treatment
possibilities are investigated. Innovative removal technologies are available such as membrane
filtration and advanced oxidation, but also biological treatment should be considered. By combining
an advanced oxidation process with managed aquifer recharge (MAR), two complementary
processes are expected to provide a hybrid system for OMP removal, according to the multiple
barrier approach. Laboratory scale batch reactor experiments were conducted to investigate the
removal of dissolved organic carbon (DOC) and 14 different pharmaceutically active compounds
(PhACs) from MAR influent water and water subjected to oxidation, under different process
conditions. A DOC removal of 10% was found in water under oxic (aerobic) conditions for batch
reactor experiments, a similar value for DOC removal was observed in the field. Batch reactor
experiments for the removal of PhACs showed that the removal of pharmaceuticals ranged from
negligible to more than 90%. Under oxic conditions, seven out of 14 pharmaceuticals were removed
over 90% and 12 out of 14 pharmaceuticals were removed at more than 50% during 30 days of
experiments. Under anoxic conditions, four out of 14 pharmaceuticals were removed over 90% and
eight out of 14 pharmaceuticals were removed at more than 50% over 30 days’experiments.
Carbamazepine and phenazone were persistent both under oxic and anoxic conditions. The PhACs
removal efficiency with oxidized water was, for most compounds, comparable to the removal with
MAR influent water.
K. Lekkerkerker-Teunissen (corresponding
author)
Dunea, PO 34, 2270 AA, Voorburg,
The Netherlands
E-mail: k.lekkerkerker@dunea.nl
K. Lekkerkerker-Teunissen
A. R. D. Verliefde
J. Q. J. C. Verberk
G. L. Amy
J. C. van Dijk
Delft University of Technology,
PO 5048, 2600 GA, Delft, The Netherlands
E. T. Chekol
G. L. Amy
UNESCO-IHE Institute for Water Education,
Westvest 7, 2601 DA Delft, The Netherlands
K. Ghebremichael
Patel School of Global Sustainability,
University of South Florida,
Tampa, FL, USA
S. K. Maeng
Sejong University,
98 Gunja-Dong, Gwangjin-Gu,
Seoul 143-747, South Korea
C. J. Houtman
The Water Laboratory, PO Box 734,
2300 RS Haarlem, The Netherlands
G. L. Amy
King Abdullah University of Science and
Technology, 4700 KAUST,
Thuwal 23955-6900,
Kingdom of Saudi Arabia
A. R. D. Verliefde
Particle and Interfacial Technology Group,
Faculty of Bioscience Engineering,
Ghent University, Coupure Links 653,
B-9000 Gent, Belgium
Key words |drinking water, MAR, organic micropollutants, oxic conditions, ozonation, soil passage
INTRODUCTION
Organic micropollutants (OMPs), such as pesticides,
pharmaceutically active compounds (PhACs), endocrine dis-
rupting compounds, X-ray contrast media and personal care
products, have been found at ng/L to low μg/L concentrations
in surface waters throughout the world ( Jurgens et al.;
Kolpin et al.;Stolker et al.;Kasprzyk Hordern
et al.). PhACs are also detected in most Western Euro-
pean rivers (Houtman ). Amongst other compounds,
PhACs are micropollutants of concern to drinking water uti-
lities (Ray et al.) because of their biological activity,
possible long term effects caused by a mixture of PhACs,
and sensitivity in the public media. The treatment provided
755 © IWA Publishing 2012 Water Science & Technology: Water Supply |12.6 |2012
doi: 10.2166/ws.2012.050
by Dunea (The Hague, The Netherlands) is a typical multiple
barrier treatment, characterized by an extensive infrastruc-
ture to apply managed aquifer recharge (MAR) in the
coastal dune area and successive post-treatment processes.
MAR processes are robust and cost-effective systems for
obtaining a safe water supply, and they include a wide var-
iety of systems for different applications (Dillon ).
MAR is an engineered process in which surface water or
stormwater is infiltrated into the aquifer through dug wells,
ponds (basins), injection wells, etc., to augment groundwater
and is subsequently abstracted by recovery wells (Bouwer
;Ray et al.). Due to the physical, chemical and bio-
logical processes involved, MAR acts as a purification step
in water treatment processes (Massmann et al.).
The removal of PhACs by MAR
Because of the biological and chemical processes occurring,
MAR has the potential to (sustainably) remove PhACs.
Howard ()found seven different structural properties
that determine the biodegradability of a solute. By the combi-
nation of microbial activity and PhAC properties, MAR is, in
addition to the removals of bacteria, viruses and suspended
solids, also able to remove (many) OMPs. Few studies exist
on this subject, and most of them deal with bank filtration
(BF) (Grünheid et al.) and not dune infiltration. Grün-
heid et al.()compared a lake bank filtration (LBF) site
and an artificial recharge and recovery (ARR) site, at Lake
Tegel near Berlin. Both sites differed in retention times and
in the predominant redox status (ARR: aerobic, 50 days;
LBF: after short aerobic zone mostly anoxic, 4–5 months).
In both cases the concentrations of OMPs was measured
via a series of monitoring wells. The monitored compounds
included: iopromide, sulfamethoxazole and three isomers
of the naphthalenedisulfonates. This monitoring program
showed that these trace organic compounds, representative
of different groups of persistent pollutants, behaved differ-
ently during infiltration as compared to direct abstraction.
For some solutes an influence of redox conditions on the
degradability was observed. Again, the removal processes
depended on the contaminant itself, but also on the hydraulic
and chemical characteristics of the bottom sediment and the
aquifer, the local recharge/discharge conditions, and bio-
chemical processes (Grünheid et al.).
Segers & Stuyfzand ()investigated the fate of OMPs
during the MAR system of Dunea. In this system pretreated
river water recharges the groundwater in the dune area by
open infiltration. The average residence time is 60 days,
but some water travels more than 10 years before recovery.
The shortest travel distance is 50 m. Removal efficiency
during MAR was observed to depend on the influent con-
centrations, residence time, media sorption characteristics,
water temperature and redox conditions. They found 61%
of the OMPs measured after MAR were below the detection
limit. X-ray contrast media, except iopamidol, were well
removed under oxic conditions. Amidotrizoic acid and car-
bendazim were well removed under anoxic conditions.
However, other substances (MTBE, diglyme, bentazon and
1,4-dioxane) were barely removed during MAR; these sub-
stances were poorly biodegradable and had poor sorption
characteristics.
An extensive review focused on the fate of PhACs in
soil/aquifer-based natural treatment processes (Maeng
et al. ). It was concluded that different classes of
PhACs behave differently during BF and ARR. Antibiotics,
Non-Steroidal Anti-Inflammatory Drugs (NSAIDs), beta
blockers and steroid hormones generally exhibited good
removal efficiencies, especially for compounds having
hydrophobic-neutral characteristics. However, anticonvul-
sants showed a persistent behavior during soil passage.
There were also some PhACs for which removal was
strongly redox dependent. For example, X-ray contrast
agents and sulfamethoxazole (an antibiotic) were degraded
more favorably under anoxic conditions than under oxic
conditions. On the other hand, phenazone-type pharma-
ceuticals (NSAIDs) exhibited better removal efficiencies
under oxic conditions (except for 1-Acetyl-1-methyl-2-
dimethyl-oxymoyl-2-phenylhydrazide (AMDOPH) which
was persistent under all conditions during BF and ARR).
Carbamazepine, which has been extensively studied because
it is the most frequently detected anticonvulsant in the
environment, showed persistence in almost all studies
(Drewes et al.;Heberer ;Cordy et al.;Heberer
& Adam ;Massmann et al.;Schmidt et al.).
Carbamazepine has been shown to be poorly biodegradable
(<10% removal) in most wastewater treatment plants
(Ternes ;Stamatelatou et al.;Zhang et al.;
Kasprzyk-Hordern et al.).
756 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
AOP and MAR: a synergistic hybrid system
Advanced oxidation processes (AOP) and MAR have comp-
lementary modes of action and are expected to remove a
different yet complementary set of solutes. Additionally, it
is expected that these two processes, when installed in
series with the AOP in front of the MAR, will enhance
each other. AOP generates assimilable organic carbon
(AOC), a part of biodegradable dissolved organic carbon
(BDOC). This additional BDOC is usually a limiting factor
for biological growth, and thus for biodegradation of
OMPs (Maeng et al.). The additional BDOC produced
by the AOP could enhance the biodegradation of OMPs
during MAR by acting as an external carbon source to
enhance the removal of OMPs through co-metabolism.
Therefore there is special interest in the interaction between
AOP and MAR. Can these two processes, oxidation and soil
passage, provide a synergistic hybrid for the removal of
OMPs? Will the changes in water matrix caused by oxi-
dation result in an increased biodegradation of solutes
during soil passage?
One purpose of this study was to investigate the removal
efficiency of PhACs during MAR. The MAR site of Dunea
was simulated by batch reactors filled with (bio)acclimated
soil material from the dunes. A second purpose was to inves-
tigate one aspect of the synergistic effect between AOP and
MAR. Only the effect of oxidation of the water matrix on the
removal of PhACs during MAR was investigated, and not
the oxidation of the PhACs itself which is being studied in
ongoing work. The influence of oxidizing the water matrix
was assessed by oxidation of the MAR influent water by
applying an ozone dosage to the water before adding the
PhACs.
MATERIALS AND METHODS
Batch experiments set-up
The water used in this study originates from a dead end
side stream of the Meuse River, and is treated by coagu-
lation, micro-straining and dual layer rapid sand
filtration before being transported towards the MAR site
of Dunea. The sand used in the batch reactors originated
from the dune infiltration area of Dunea and was col-
lected at 1 m depth and 1 m next to an infiltration pond.
Before the experiment was started, the sand was accli-
mated to the laboratory settings. Batches were placed in
a dark and temperature (10 WC) controlled room. One hun-
dredgramsofdunesandwereplacedintoglassbatch
reactors with a capacity of 0.5 L. The reactors were then
filled with 450 mL oxic or anoxic MAR influent water or
oxidized water, placed on a shaker table at 100 rpm and
the water was refreshed every 5 days, until steady state
conditions were reached with respect to BDOC removal.
Every 5 days, the DOC concentration of influent (c
in
)
and effluent (c
ef
) water was measured, and the DOC
removal was calculated as c
ef
/c
in
.
After ripening of the reactors, MAR influent water and
oxidized water, both spiked with a set of PhACs (see
below under Pharmaceutically active compounds), were
added to the batch reactors. Samples were taken at regular
intervals, this time without changing the influent water.
Twenty milliliters of samples were taken from each batch
reactor, at 0, 10 and 30 days, and analyzed for DOC and
PhACs. The 30 days’retention time follows from the field
data. Although 50% of the water is recovered after 120
days, 5% of the water is recovered after 35 days. A 30 day
retention time is considered as a worst case, but determining
for treatment objectives.
The experiments have been performed in winter (ripen-
ing period October till February, spiking in March), when
the general water composition was characterized by the par-
ameters as given in Table 1.
Different experimental conditions were chosen, and all
were tested in duplicate. Additional batches were used for
DOC removal measurements and controls. Figure 1 pro-
vides an overview of the experimental conditions.
Afirst set of batch reactors was used to simulate the
present field conditions during the MAR at Dunea and
real MAR influent water was used. Oxic conditions
were maintained by leaving the reactors uncovered so
air could enter the batch reactors. Anoxic conditions
were imposed by stripping off dissolved oxygen from the
influent water by purging with nitrogen gas until the
oxygen concentration was <0.5 mg/L. Fine diffusers
were used to distribute the nitrogen gas evenly over the
reactors.
757 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
In order to investigate the influence of oxidation by AOP
on the removal of PhACs during subsequent MAR, the water
was oxidized by ozonation with an ozone:DOC ratio of
1:1 mg/mg. Ozone gas was diffused through the water
until the desired applied ozone dose (between 3 and 4 mg/L,
depending on the DOC concentration) was applied. A
common ozonization time was about 12 minutes.
Pharmaceutically active compounds
To investigate the fate and removal of PhACs during MAR,
a set of 14 pharmaceuticals listed in Table 2 were spiked
into the eight batch reactors. The pharmaceuticals were
chosen for their wide range of physico-chemical properties,
and their occurrence in Dutch surface water. In case of oxi-
dation, the pharmaceuticals were spiked after oxidation so
only the possible synergistic effect of an oxidized water
matrix was studied. Spiking concentrations (between 1 and
10 μg/L) were chosen, low enough to represent environ-
mental levels and high enough to enable analysis without
having to extract or preconcentrate the samples. All com-
pounds were obtained from Sigma-Aldrich, Zwijndrecht,
The Netherlands. A 1 L spike solution was prepared in
Milli-Q water, containing 5 mg/L of each solute. An aliquot
of 450 μL of the spike solution was added to each batch
reactor.
Analyses and measurements
Micro parameters
Analysis of the selected pharmaceuticals was performed by
injection of 20 μL glass-filtered sample on an Ultra Perform-
ance Liquid Chromatograph (UPLC, Waters Acquity),
equipped with a quaternary pump, a UPLC BEH C18
column (5 cm, particle size 1.7 μm, internal diameter
2.1 mm, Waters Acquity) and combined with a Quattro
Xevo triple quadrupole Mass Selective Detector (Waters
Micromass) (Houtman et al., in preparation).
Macro parameters
The water in the different reactors was characterized for
DOC, UV-absorbance, selected trace organics (pharmaceuti-
cals) and other water parameters such as pH, temperature,
electric conductivity (EC), turbidity, nitrate and dissolved
oxygen.
DOC was measured using a Schimadzu TOC-VCPN
total organic carbon analyzer. The samples were filtered
through a 0.45 μm cellulose acetate membrane. For each
sample, the average DOC concentration is reported. DOC
values were measured as low as 0.01 mg/L and as high as
20 mg/L. UV absorbance at a wavelength of 254 nm was
Figure 1 |Experimental conditions in batch reactors.
Table 1 |Water quality MAR influent water, used in batch reactor experiments
Parameter O
2
pH Temp. EC NH
4
-N NO
3
-N PO
4
-P SO
4
DOC
Unit mg/L –
WCμS/cm mg/L mg/L mg/L mg/L mg/L
MAR influent water 10.4 ±1.2 7.9 ±0.2 7.1 ±1.2 511 ±8<0.1 3.7 ±0.1 0.17 48.0 3.9 ±0.7
758 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
measured by a Perkin Elmer UV/VIS Spectrophotometer
with 1 cm quartz cuvettes. The samples were measured
after filtering the undiluted samples through a 0.45 μm cellu-
lose acetate membrane.
RESULTS
Bulk parameters
DOC removal
Figure 2 shows results for normalized DOC removal
during the ripening process in the batch reactors. The
DOC effluent concentration is plotted as fraction of the
DOC influent concentration. After 60 days, the DOC
removal is stabilized and a steady-state 10% DOC removal
is achieved.
The influence of spiking PhACs in low concentrations
on the removal of DOC was investigated by DOC analyses
from water with and without spiked PhACs. For the
pretreated water without spiked PhACs, the concentration
of DOC was reduced from 3.8 to 3.4 mg/L, resulting
in a 9.8% removal. For the pretreated water with PhACs
spiked, the concentration of DOC was reduced from 4.4
to 3.9 mg/L, resulting in a 9.9% removal. It was con-
cluded that spiking low concentrations of PhACs did not
influence the DOC removal in the batch reactors.
DOC changes during PhACs study
During the 30 days of PhACs removal experiments (with-
out refreshment of feed water), DOC removal was
Figure 2 |DOC removal by batches stabilizes after 2 months.
Table 2 |Pharmaceuticals and their properties used in experiments
Pharmaceutical MW (g/mol) Log K
ow
(–) Log D
a
at pH 8(–) Hydrophobicity
b
Therapeutic use
Bezafibrate 361.8 4.25 0.31 Hydrophilic Lipid regulator
Carbamazepine 236.3 2.45 na Hydrophobic Anticonsulvant
Clofibric acid 214.7 2.57 2.08 Hydrophilic Lipid regulator (metab.)
Cyclophosphamide 261.1 0.63 na Hydrophilic Cytostaticum
Diclofenac 296.2 4.51 0.59 Hydrophilic NSAID
Gemfibrozil 250.3 4.77 1.22 Hydrophobic Lipid regulator
Ibuprofen 206.3 3.97 0.44 Hydrophilic NSAID
Ketoprofen 254.3 3.12 0.59 Hydrophilic NSAID
Metoprolol 267.4 1.88 0.38 Hydrophilic Beta blockers
Naproxen 242.2 3.90 0.05 Hydrophilic NSAID
Pentoxifylline 278.3 0.29 na Hydrophilic Blood thinner
Phenazon 188.2 0.38 na Hydrophilic NSAID
Propranolol 259.4 3.48 1.89 Hydrophilic Beta blockers
Sotalol 272.4 0.24 1.2 Hydrophilic Beta blockers
a
Log D is the distribution coefficient at pH 8, for ionized and unionized compounds.
b
Hydrophobicity is defined as: log K
ow
>3 is hydrophobic, log K
ow
<3 means hydrophilic.
759 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
monitored at the same time when samples were taken for
PhAC analyses. From the results plotted in Figure 3,itis
observed that the DOC consumption under anoxic con-
ditions was lower compared to the DOC consumption
during oxic conditions. This can or be explained by a
smaller microbial community of denitrifiers compared to
the aerobic bacteria community. Or it can be explained
by the different bacterial populations present under differ-
ent conditions. The denitrifying bacteria present under
anoxic conditions usually show slower organic matter
degradation as also seen by McNally et al.().Todis-
tinguish between these two explanations, ATP
measurements or nitrate concentrations had to be
measured during the experiment, which was not foreseen
and is recommended in future studies.
The DOC removal with MAR influent water was com-
pared to the DOC removal when the MAR influent water
was oxidized (called oxidized water). The lower DOC
concentration at day 0 for oxidized water is due to some
mineralization of DOC by ozone. However, in absolute
terms, the DOC removal is comparable to the removal
in MAR influent water, both under oxic as well as
anoxic conditions. After 30 days, the DOC removals for
MAR influent water were 0.76 and 0.47 mg/L for oxic
and anoxic conditions, respectively. For oxidized water,
the 30 days’DOC removals were 0.97 and 0.51 mg/L
for oxic and anoxic conditions, respectively. DOC
removal can be regarded as a measure for biological
activity. Because of the comparable DOC removals, it is
thus expected that the biological activity during MAR
will be comparable for MAR influent water and oxidized
water, and that the removal of PhACs by biodegradation
will be comparable.
PhACs removal
PhACs removal during batch experiments with MAR
influent water
The batch reactors with MAR influent water simulate the cur-
rent field conditions at Dunea, The Netherlands. The
measured PhACs concentrations at day 0 and after 30 days,
for oxic and anoxic conditions, are presented in Table 3.
During oxic conditions the compounds bezafibrate, ibupro-
fen, ketoptofen, naproxen, pentoxifylline and propanolol
are virtually completely removed (i.e. below detection limit)
after 30 days’contact time. The compounds clofibric acid,
metoprolol, and sotalol showed more than 90% removal.
The compounds cyclophosphamide, diclofenac and gemfi-
brozil showed a removal between 50 and 90%. Only the
compounds carbamazepine and phenazon showed a persist-
ent behavior under these conditions and were not removed.
During anoxic conditions only bezafibrate and
ibuprofen showed complete removal. The compounds car-
bamazepine, gemfibrozil, naproxen and phenazon showed
persistent behavior under these conditions and were not
removed.
Batch reactor experiments thus showed that, under
oxic conditions, seven out of 14 PhACs were removed
from MAR influent water over 90% after 30 days; 12 out
of 14 PhACs were removed over 50%. Under anoxic con-
ditions, four out of 14 PhACs were removed over 90%
after 30 days; eight out of 14 PhACs were removed over
50%. It can be concluded that the removal of PhACs
during MAR occurs more preferentially under oxic con-
ditions, or at least that the rate of biological degradation
is higher under oxic conditions.
Figure 3 |DOC concentration during batch experiments for four experimental conditions.
760 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
PhACs removal with oxidized water
For this experiment, the PhACs were spiked into the water
1 h after oxidation with ozone. The measured concen-
trations at day 0 and after 30 days for oxic and anoxic
conditions are presented in Table 4. In the same table,
the two most right columns repeat the percentage removal
obtained with MAR influent water for oxic and anoxic con-
ditions (from Table 3). From these results it can be
concluded that after oxidation, removal percentages of
the PhACs are comparable to normal MAR conditions.
Only three compounds showed significant differences in
removal between MAR influent and oxidized water. Ibu-
profen and naproxen were no longer completely removed
after oxidation and gemfibrozil showed persistent and
was not removed (along with carbamazepine and
phenazon).
Removal over time
PhAC samples in the MAR influent were not only taken at
day 0 and after 30 days, but also after 10 days’retention
time (Table 5). For oxic conditions with MAR influent
water, most absolute removal occurred between days 0
and 10; for six out of 12 PhACs. For anoxic conditions, how-
ever, a shift in absolute removal over time was observed.
During anoxic conditions, only three solutes showed most
absolute removal between day 0 and 10. Four solutes
showed minimal removal between day 0 and 10, but signifi-
cant removal between day 10 and 30. This indicates that the
lower removal of solutes under anoxic conditions does not
necessarily imply that biological degradability of the solutes
under anoxic conditions is limited, but more that the reac-
tion kinetics are slower under anoxic conditions.
DISCUSSION
PhACs removal during batch experiments
The two most important mechanisms for PhACs removal
during MAR at Dunea’s site are found to be sorption and
biodegradation (Segers & Stuyfzand ). A third mechan-
ism, hydrolysis of the compound in (natural) water, is less
feasible in this study with contact times of 30 days, since
the half-life of PhACs in water, although they vary a great
Table 3 |The measured PhACs concentrations and calculated removal percentages for oxic and anoxic conditions with MAR influent water
MAR influent water
Concentration at day 30 (μg/L) Removal over 30 days (%)
Pharmaceutical Concentration at day 0 (μg/L) Oxic Anoxic Oxic Anoxic
Bezafibrate 1.56 0.03 0.03 98 98
Carbamazepine 4.36 4.48 3.77 313
Clofibric acid 4.33 0.71 0.97 84 78
Cyclophosphamide 6.06 1.93 3.11 68 49
Diclofenac 8.96 1.97 6.05 78 32
Gemfibrozil 9.59 2.66 10.03 72 4
Ibuprofen 7.96 <0.01 <0.01 99 99
Ketoprofen 5.24 0.08 0.20 98 96
Metoprolol 3.70 0.29 1.38 92 63
Naproxen 9.94 <0.01 11.35 99 14
Pentoxifylline 4.45 0.06 0.19 99 96
Phenazon 5.32 5.64 6.07 614
Propranolol 2.84 0.05 0.88 98 69
Sotalol 4.98 0.59 2.12 88 57
761 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
Table 4 |The measured PhACs concentration and calculated removal percentage for oxic and anoxic conditions with oxidized water –compared to MAR influent water
Oxidized water MAR influent water
Concentration at day 30 (μg/L) Removal percentage (%) Removal percentage (%)
Pharmaceutical Concentration at day 0 (μg/L) Oxic Anoxic Oxic Anoxic Oxic Anoxic
Bezafibrate 1.36 0.03 0.03 98 98 98 98
Carbamazepine 4.21 4.58 3.72 912313
Clofibric acid 4.38 1.03 0.74 77 83 84 78
Cyclophosphamide 5.86 3.69 3.16 37 46 68 49
Diclofenac 7.42 4.13 5.37 44 28 78 32
Gemfibrozil 7.54 6.48 7.06 14 6 72 4
Ibuprofen 9.32 2.76 6.30 70 32 100 100
Ketoprofen 5.58 0.07 0.18 99 97 98 96
Metoprolol 3.67 0.40 2.53 89 31 92 63
Naproxen 10.76 7.34 4.42 32 59 100 14
Pentoxifylline 4.76 0.08 0.11 98 98 99 96
Phenazon 5.14 6.23 5.72 21 11 614
Propranolol 2.93 0.14 2.15 95 27 98 69
Sotalol 4.97 1.24 2.70 75 46 88 57
Table 5 |Measured PhACs concentrations over time for oxic and anoxic conditions with MAR influent water
MAR influent water, removal over time
Concentration at
Oxic conditions, concentration at Anoxic conditions, concentration at
Pharmaceutical day 0 (μg/L) day 10 (μg/L) day 30 (μg/L) day 10 (μg/L) day 30 (μg/L)
Bezafibrate 1.36 0.69 0.03 0.73 0.03
Carbamazepine 4.21 4.13 4.48 4.02 3.77
Clofibric acid 4.38 3.76 0.71 3.96 0.97
Cyclophosphamide 5.86 4.88 1.93 4.61 3.11
Diclofenac 7.42 2.96 1.97 6.49 6.05
Gemfibrozil 7.54 5.28 2.66 10.56 10.03
Ibuprofen 9.32 3.79 0.00 7.46 0.00
Ketoprofen 5.58 0.75 0.08 4.72 0.20
Metoprolol 3.67 1.39 0.29 2.28 1.38
Naproxen 10.76 4.94 0.00 8.95 11.35
Pentoxifylline 4.76 1.65 0.06 3.64 0.19
Phenazon 5.14 5.51 5.64 4.72 6.07
Propranolol 2.93 0.78 0.05 1.36 0.88
Sotalol 4.97 3.50 0.59 3.87 2.12
762 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
deal, are in the order of magnitude of years. EPI Suite v.3.20
HYDROWIN model gives half life times by hydrolysis, for
example the half life time of bezafibrate in water, based on
the chloroacetamide group, is 1.46 years (US EPA ).
Although the effect of hydrolysis in this study of 30 days is
not considered, in the field with longer contact times it
can have an influence, depending on the half-life time of
the specific compounds looked at.
Sorption
A parameter often used to predict sorption capacity of
different organic solutes is the octanol-water partition coef-
ficient, log K
ow
(Verliefde et al.;De Ridder et al. ).
Compounds with a high log K
ow
are most often less polar
and therefore more easily removed from the water by sorp-
tion. Because some PhACs are charged compounds, and
charge influences the hydrophobicity, log Dis used as a
descriptor of solute hydrophobicity in this study. Log D
values are corrected for charge and also depicted in
Table 2.InTable 6, compounds with a log D higher than
2 are considered to be able to sorb onto the soil or natural
organic matter (NOM) present in the water. These com-
pounds are highlighted dark grey. Compounds with a log
D between 1 and 2 are highlighted in light grey, indicating
minor adsorption is expected.
Biodegradation
Biodegradation strongly depends on the molecular structure
of a solute, but it is difficult to identify one or two most rel-
evant compound properties. EPI Suite v.3.20 BIOWIN
model (US EPA, http://www.epa.gov/oppt/exposure/pubs/
episuitedl.htm) predicts biodegradation potential using the
group contribution approach (US EPA ). BIOWIN con-
sist of seven models: linear and non-linear probability
models, ultimate and primary biodegradation models, field
data models and an anaerobic model. BIOWIN model 4 pre-
dicts the primary biodegradation in a timeframe. The primary
biodegradation is defined as the transformation of a parent
solute to an initial metabolite. The models 3 and 4 are
based on estimates of primary and ultimate biodegradation
rates for 200 chemicals, gathered in a survey of 17 biodegra-
dation experts, conducted by the US EPA. The methodology
of primary and ultimate biodegradation is published by
Boethling et al. (). The values generated by BIOWIN
model 4 are given in Table 6. Compounds with a BIOWIN
value above 3.7 are highlighted in dark grey and are
Table 6 |Sorption and biodegradation values for compounds compared with removal in this study with MAR influent water under oxic conditions
Pharmaceutical log K
ow
pK
a
Charge log D
a
at pH 8 BIOWIN model 4 Removal in this study [%]
Bezafibrate 4.25 3.44 0.31 3.6065 98
Carbamazepine 2.45 2.3 þ/2.45 3.5068 3
Clofibric acid 2.57 3.35 2.08 3.6820 84
Cyclophosphamide 0.63 2.84 þ/0.63 3.2698 68
Diclofenac 4.51 4.08 0.59 3.2984 78
Gemfibrozil 4.77 4.45 1.22 3.6587 72
Ibuprofen 3.97 4.47 0.44 3.7986 100
Ketoprofen 3.12 4.29 0.59 3.7806 98
Metoprolol 1.88 9.49 þ0.38 3.6336 92
Naproxen 3.90 4.15 0.05 3.9097 100
Pentoxifylline 0.29 6 þ/0.29 3.4240 99
Phenazon 0.38 1.4 þ/0.38 3.5811 6
Propranolol 3.48 9.58 þ1.89 3.7234 98
Sotalol 0.24 9.44 þ1.22 3.6492 88
a
Log D values calculated at http://www.raell.demon.co.uk/chem/calcs/LogP/logD.htm.
763 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
considered to be easily biodegradable. Compounds with a
value above 3.6 are highlighted in light grey and considered
to exhibit average biodegradability. Solutes with lower
BIOWIN values are considered only slightly biodegradable.
Experimental results compared to theoretical sorption and
biodegradation values
The two parameters described above, namely log D and
BIOWIN model 4, are considered here as indicators for
removal by sorption and biodegradation, respectively.
Together, they can give an indication of whether the results
observed in this study are consistent with qualitative predic-
tions. Compounds with average removal above 90% are
highlighted in dark grey in Table 6 and compounds with a
removal between 50 and 90% are highlighted in light grey.
From Table 6, it is apparent that most removal is
expected by biodegradation, since more compounds show
high biodegradation values than high log Dvalues. Com-
pounds with average or high biodegradation value mostly
showed good or average removal in this study. Compounds
with a high BIOWIN 4 value (namely ibuprofen, ketoprofen,
naproxen and propranolol) are all well removed (>90%)
during the batch reactor experiments. These compounds
are probably mainly removed by biodegradation. Proprano-
lol, which has a log Dvalue of 1.89, might be partly
removed by sorption as well. Gemfibrozil has average
values for both log Dand BIOWIN 4 and also shows aver-
age removal (72%), probably explained by partial removal
by sorption and partial removal by biodegradation. Sotalol,
with a low log Dand an average biodegradation value,
showed 88% removal during batch reactor experiments,
and removal is expected to be mainly due to biodegradation.
Carbamazepine and phenazon, the two persistent com-
pounds during the batch experiments, have low
biodegradation values and especially phenazon also has a
low log D. The persistent behavior of carbamazepine can
be explained by its low biodegradability, caused by three
fused rings in its molecular structure. This results in limited
possibilities for electron transfer and thus biological degra-
dation reactions, especially under oxic conditions.
Carbamazepine, although having a relatively high log D,is
known to be poorly adsorbing. Persistent behavior therefore
was expected.
It can thus be concluded that log D and BIOWIN model
4 values are simple, quick parameters to give an indication
on removal of organic solutes during MAR.
Only the (relatively) good removal (68 and 99%) of
cyclophosphamide and pentoxifylline could not be explained
by these two parameters. These compounds both have a low
log Dand a low biodegradation value, but showed good
removal during the 30-day batch reactor experiments.
PhACs removal with oxidized water
The removal of compounds in batches of oxidized water
appears to be largely comparable to the removals observed
with MAR influent water. This seems logical since both con-
ditions give comparable absolute DOC removal. However,
three compounds showed significantly lower removal after
the water was oxidized, which was not expected. These
three solutes are ibuprofen, naproxen and gemfibrozil.
The differences can not be explained by oxidation of the
PhACs, since the water was pre-oxidized before adding
the PhACs, and sufficient time passed before addition of
the PhACs. It is also not expected that these differences
result from analytical errors, since the lower removal in oxi-
dized water is observed both under oxic, as well as anoxic
conditions.
An explanation for the observed lower removal of
naproxen, ibuprofen and gemfibrozil in pre-oxidized water
might be found in the biodegradation mechanism: all three
compounds have average to high biodegradation values
according to BIOWIN model 4. Figure 3, in the DOC
removal section above, showed a decrease in DOC concen-
tration after pre-oxidation. Less NOM in the feed water
could result in less absolute NOM removal and therefore
less removal of PhACs that are removed by cometabolism
during the degradation of NOM. This explanation is not
very likely since the absolute DOC consumption is compar-
able for non-oxidized MAR influent water and pre-oxidized
water.
Another explanation can be found in indirect biodegra-
dation via adsorption onto NOM. Oxidation of the NOM
could thus also influence the indirect biodegradation. It
might be possible that compounds first adsorb onto
NOM, which is then adsorbed onto the soil and biode-
graded, resulting in biodegradation of the adsorbed
764 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
solutes via co-metabolism. After oxidation, the NOM
becomes more polar and the compounds will tend to
adsorb less onto the NOM and will thus not be degraded
together with the NOM.
These experiments do not show a synergistic effect of
oxidizing the water matrix on PhACs removal during
MAR. If there is a synergistic effect between AOP
and MAR, this will be in the oxidation of the OMPs before
MAR. OMPs will be oxidized into transformation products
(smaller structures) with an expected higher biodegradabil-
ity. This hypothesis will be tested in future research.
PhACs removal during MAR in the field
Pharmaceuticals are often OMPs of concern for water utili-
ties, because perceptions by customers and the uncertainty
concerning their implications for human health. This study
showed that most PhACs are well removed during natural
treatment, in this case MAR. Most compounds showed
good removal after 30 days, which is sufficient given the
long retention times during most MAR sites. At the real
system of Dunea, 5% of the water is recovered after 35
days, but 50% of the water is recovered after 120 days. Aver-
age removal efficiencies in the field are expected to be
higher than found in this study, which illustrates a worst
case for Dunea’s system. MAR is therefore an adsorptive/
biological treatment step that is applicable for PhACs
removal. Only a few compounds were persistent during
MAR. In this research, two compounds showed persistent
behavior under all conditions, namely carbamazepine and
phenazon. These compounds are advised to be included in
regular monitoring programs.
Carbamazepine has already been shown to be a persist-
ent solute during biological treatment; Kasprzyk-Hordern
et al.()found that bank filtration can be regarded as a
useful tool but does not guarantee the complete removal
of carbamazepine. Special care should thus be taken when
residues of carbamazepine or other antiepileptic drugs are
present in the raw water. The poor biodegradability of carba-
mazepine is caused by the three fused rings in its molecular
structure, resulting in a very stable chemical structure with
very few possibilities for electron transfer. Compounds simi-
lar to carbamazepine, such as other antiepileptic drugs
(topimarate, lamotrigine) or other compounds with more
than two fused ring structures need special attention when
assessing a biological treatment step for their removal.
Phenazon was also found to be persistent in this study,
under all conditions investigated. Phenazon has a low mol-
ecular weight, low pK
a
value and a low log Dvalue. Since
it is quite hydrophilic, it is not likely to be removed by sorp-
tion. Although BIOWIN model 4 predicts a relative low
primary biodegradation for phenazon (comparable with
results observed in this study), some literature has indicated
that phenazon is possibly biodegradable under oxic con-
ditions (Massmann et al. ;Schmidt et al. ;Snyder
et al. ;Massmann et al. ). This emphasizes that
removal is not only depending on the solute properties,
but also on the characteristics of the MAR site.
In conclusion, MAR is considered to be a useful, econ-
omic and sustainable tool for the removal of PhACs and a
biological treatment step should be considered when look-
ing at the available toolbox for drinking water
purification. The combination of sorption and biodegrada-
tion results in a treatment step with two removal
mechanisms. In addition, the sorption and perhaps later
de-sorption results in retardation, which results in a longer
residence time of the PhACs for biodegradation. Most
PhACs are removed during MAR, but special attention is
needed for compounds with multiple fused ring structures
or poor sorption characteristics. Since MAR is able to
remove a large part of PhACs and is sustainable, but does
not guarantee complete removal of all compounds, its use
is advised, but it should preferably be used in a multiple
barrier treatment concept. In this case, MAR can be used
as pretreatment (e.g. river bank filtration followed by mem-
brane filtration) or as post-treatment (e.g. advanced
oxidation followed by dune filtration). In the latter case,
MAR can remove byproducts (e.g. AOC, BDOC) and trans-
formation products (e.g. partly oxidized PhACs) resulting
from an oxidative treatment step, and so preventing them
from entering the drinking water. Results from the AOP
study at Dunea’s site are reported in Lekkerkerker et al.
()and Scheideler et al. ().
In addition to being a multiple barrier concept,
advanced oxidation as pretreatment to MAR might even
result in an enhancement of PhACs removal in the MAR
process. In that case, a synergistic hybrid system occurs.
Advanced oxidation followed by MAR is believed to be a
765 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
synergistic hybrid system. However, this study investigated
the influence of oxidizing the water matrix on the removal
of PhACs during MAR, and no significant evidence of this
synergy between the two systems was found. It must
be stated that in this study, the PhACs were spiked after
oxidation of the water matrix. Therefore, only the influence
of a different water matrix on the removal of PhACs was
investigated. When AOP are installed as pretreatment,
there will not only be an effect on the water matrix, but
the PhACs themselves will also be (partly) degraded by
the AOP. It is therefore expected that the total removal
efficiency of PhACs will increase if AOP is installed as
pretreatment before MAR, since most compounds will
already be (partly) oxidized. In general, smaller, partly
oxidized, transformation products are expected to be
more easily biodegradable. This will be studied in future
research.
CONCLUSIONS
In this study, the removal of PhACs by MAR was simulated
by batch reactor experiments.The main conclusion are:
•The removal of pharmaceuticals ranged from negligible
to more than 90%, but under oxic conditions; seven out
of 14 pharmaceuticals were removed over 90% and 12
out of 14 pharmaceuticals were removed at more than
50% during 30 day-experiments. Under anoxic con-
ditions, only four out of 14 pharmaceuticals were
removed over 90% and eight out of 14 pharmaceuticals
were removed at more than 50%.
•Carbamazepine and phenazone were persistent both
under oxic and anoxic conditions.
•Under oxic conditions, the compounds show highest
removal efficiency between day 0 and 10. Under anoxic
conditions, the removal efficiency is highest between
day 10 and 30, indicating slower degradation, probably
caused by slower co-metabolism.
•The PhACs removal efficiency with oxidized water was,
for most compounds, comparable to the removal with
MAR influent water.
•Spiking PhACs in low concentrations did not affect the
DOC removal efficiency of the sand in the batch reactors.
•The absolute DOC removal during batch experiments
was comparable for MAR influent water and oxidized
water.
•From the results it could be observed that the DOC
removal under anoxic conditions is lower compared to
the DOC removal during oxic conditions.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the NWO Casimir
program for financial support for this research.
REFERENCES
Boethling, R. S., Howard, P. H., Meylan, W., Stiteler, W.,
Beauman, J. & Tirado, N. Group contribution method
for predicting probability and rate of aerobic biodegradation.
Environmental Science and Technology 28 (3), 459–465.
Bouwer, H. Artificial recharge of groundwater: hydrogeology
and engineering.Hydrogeology Journal 10 (1), 121–142.
Cordy, G. E., Duran, N. L., Bouwer, H., Rice, R. C., Furlong, E. T.,
Zaugg, S. D., Meyer, M. T., Barber, L. B. & Kolpin, D. W.
Do pharmaceuticals, pathogens, and other organic
waste water compounds persist when waste water is used for
recharge? Ground Water Monitoring and Remediation 24
(2), 58–69.
De Ridder, D. J., Verliefde, A. R. D., Heijman, S. G. J., Verberk,
J. Q. J. C., Rietveld, L. C., Van Der Aa, L. T. J., Amy, G. L. &
Van Dijk, J. C. Influence of natural organic matter on
equilibrium adsorption of neutral and charged
pharmaceuticals onto activated carbon.Water Science and
Technology 63 (3), 416–423.
Dillon, P. Future management of aquifer recharge.
Hydrogeology Journal 13 (1), 313–316.
Drewes, J. E., Heberer, T. & Reddersen, K. Fate of
pharmaceuticals during indirect potable reuse. Water Science
and Technology 46 (3), 73–80.
Grünheid, S., Amy, G. & Jekel, M. Removal of bulk dissolved
organic carbon (DOC) and trace organic compounds by
bank filtration and artificial recharge.Water Research 39,
3219–3228.
Heberer, T. Occurrence, fate, and removal of pharmaceutical
residues in the aquatic environment: a review of recent
research data.Toxicology Letters 131,5–17.
Heberer, T. & Adam, M. Transport and attenuation of
pharmaceutical residues during artificial groundwater
replenishment.Environmental Chemistry 1,22–25.
766 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
Houtman, C. J. Emerging contaminants in surface waters and
their relevance for the production of drinking water in
Europe.Journal of Integrative Environmental Sciences 7(4),
271–295.
Howard, P. H. Biodegradation. Handbook of Property
Estimation Methods for Chemicals and Health Sciences. CRC
Press LLC, Boca Raton.
Jurgens, M. D., Holthaus, K. I. E., Johnson, A. C., Smith, J. J. L.,
Hetheridge, M. & Williams, R. J. The potential
for estradiol and ethinylestradiol degradation in English
rivers.Environmental Toxicology and Chemistry 21 (3),
480–488.
Kasprzyk-Hordern, B., Dinsdale, R. M. & Guwy, A. J. The
occurrence of pharmaceuticals, personal care products,
endocrine disruptors and illicit drugs in surface water in
South Wales, UK.Water Research 42, 3498–3518.
Kasprzyk-Hordern, B., Dinsdale, R. M. & Guwy, A. J. The
removal of pharmaceuticals, personal care products,
endocrine disruptors and illicit drugs during wastewater
treatment and its impact on the quality of receiving waters.
Water Research 43, 363–380.
Kolpin, D. W., Furlong, E. T., Meyer, M. T., Thurman, E. M.,
Zaugg, S. D., Barber, L. B. & Buxton, H. T.
Pharmaceuticals, hormones, and other organic wastewater
contaminants in U.S. streams, 1999–2000: A national
reconnaissance.Environmental Science and Technology 36
(6), 1202–1211.
Lekkerkerker, K., Scheideler, J., Meang, S. K., Ried, A., Verberk,
J. Q. J. C., Knol, A. H., Amy, G. & van Dijk, J. C.
Advanced oxidation and artificial recharge: a synergistic
hybrid system for removal of organic micropollutants.
Water Science and Technology –Water Supply 9(6),
643–651.
Maeng, S. K., Sharma, S. K., Lekkerkerker-Teunissen, K. & Amy,
G. L. Occurrence and fate of bulk organic matter and
pharmaceutically active compounds in soil/aquifer-based
natural treatment processes: a review.Water Research 45
(10), 3015–3033.
Massmann, G., Greskowiak, J., Dünnbier, U. & Zuehlke, S.
The impact of variable temperatures on the redox conditions
and the behavior of pharmaceutical residues during artificial
recharge.Journal of Hydrology 328, 141–156.
Massmann, G., Dunnbier, U., Heberer, T. & Taute, T.
Behaviour and redox sensitivity of pharmaceutical residues
during bank filtration –investigation of residues of
phenazone-type analgesics.Chemosphere 71, 1476–1485.
McNally, D. L., Mihelcic, J. L. & Lucking, D. R.
Biodegradation of mixtures of polycyclic aromatic
hydrocarbons under aerobic and nitrate-reducing conditions.
Chemosphere 38 (6), 1313–1321.
Ray, C., Soong, T. W., Lian, Y. Q. & Roadcap, G. S. Effect of
flood-induced chemical load on filtrate quality at bank
filtration sites.Journal of Hydrology 266, 235–258.
Segers, W. C. J. & Stuyfzand, P. J. Appearance and behaviour
of emerging substances during dune filtration (in Dutch).
M.Sc. Thesis, VU University Amsterdam, The Netherlands.
Scheideler, J., Lekkerkerker-Teunissen, K., Knol, T., Ried, A.,
Verberk, J. & van Dijk, H. Combination of O3/H
2
O
2
and
UV for multiple barrier micropollutant treatment and
bromate formation control –an economic attractive option.
Water Practice and Technology 6(4), 1–8.
Schmidt, C. K., Lange, F. T. & Brauch, H. J. Characteristics
and evaluation of natural attenuation processes for organic
micropollutant removal during riverbank filtration. In
Regional IWA Conference on Groundwater Management in
the Danube River Basin and Other Large River Basins,7–9
June 2007, Belgrade, Serbia, pp. 231–236.
Snyder, S. A., Wert, E. C., Lei, H. X., Westerhoff, P. & Yoon, Y.
Removal of EDCs 599 and Pharmaceuticals in Drinking
and Reuse Treatment Processes. Awwa 600 Research
Foundation, Denver, CO, 331.
Stolker, A. A., Niesing, W., Hogendoorn, E. A., Versteegh, J. F.,
Fuchs, R. & Brinkman, U. A. Liquid chromatography
with triple-quadrupole or quadrupole-time of flight mass
spectrometry for screening and confirmation of residues of
pharmaceuticals in water.Analytical and Bioanalytical
Chemistry 378 (4), 995–963.
Stamatelatou, K., Frouda, C., Fountoulakis, M. S., Drillia, P.,
Kornaros, M. & Lyberatos, G. Pharmaceuticals and
health care products in wastewater effluents: the example of
carbamazepine. Water Science and Technology: Water
Supply 3(4), 131–137.
Ternes, T. A. Occurrence of drugs in German sewage
treatment plants and rivers.Water Research 32, 3245–3260.
US EPA Estimation Programs Interface Suite
TM
for Microsoft
®
Windows, v 4.10. United States Environmental Protection
Agency, Washington, DC, USA.
Verliefde, A. R. D., Cornelissen, E. R., Heijman, S. G. J., Amy, G.
L., Van der Bruggen, B. & van Dijk, J. C. Influence of
electrostatic interactions on the rejection with NF and
assessment of the removal efficiency during NF/GAC
treatment of pharmaceutically active compounds in surface
water.Water Research 41 (15), 3227–32240.
Zhang, Y., Geissen, S. U. & Gal, C. Carbamazepine
and diclofenac: removal in wastewater treatment plants
and occurrence in water bodies.Chemosphere 73, 1151–1161.
First received 21 February 2012; accepted in revised form 2 May 2012
767 K. Lekkerkerker-Teunissen et al.|Pharmaceutical removal during MAR, combined with AOP pretreatment Water Science & Technology: Water Supply |12.6 |2012
