treatment

technologies

Granular activated carbon (GAC) filtration experiments demonstrated that perchlorate was

removed by ion exchange rather than by chemical reduction. A column ion-exchange capacity
of 0.172 mg perchlorate per g GAC was calculated. Batch tests confirmed that ion exchange
was the dominant mechanism of perchlorate removal by GAC. When influent dissolved oxygen

(DO) was 2.5 mg/L and an electron donor solution was supplied, efficient reduction of 50 µg/L
perchlorate was achieved using biologically active carbon (BAC) filtration. Perchlorate
reduction decreased as the concentration of nitrate in the filter increased. With low influent

DO and nitrate concentrations and excess electron donor added, the BAC filter removed 50
µg/L perchlorate to below detection for 103 days of flow at a 25-min empty bed contact time.
Batch biological experiments verified the occurrence of biological reduction and demonstrated

that microorganisms in the BAC filter converted perchlorate completely to chloride.

ABIOTIC AND BIOTIC

PERCHLORATE removal
in an activated carbon filter
n April 1997, the California Department of Health Services (CDHS) com-

I
pleted the development of an ion chromatographic method that decreased
the limit of detection (LOD) for aqueous perchlorate (ClO4–) from 400
µg/L to 4 µg/L (CDHS, 1997a). Since then, perchlorate concentrations rang-
ing from a few micrograms per litre to hundreds of micrograms per litre
have been found in drinking water supplies across the United States. Ammo-
nium perchlorate (NH4ClO4) is an oxidizer that helps solid rocket fuel ignite
and burn and is often a component of munitions (Urbansky, 1998). Because
ammonium perchlorate has a limited shelf life, it is frequently washed out of
BY JESS C. BROWN,
the nation’s solid rocket supply. Contamination is likely the result of the large vol-
VERNON L. SNOEYINK, umes of perchlorate-containing water generated during this process. Perchlorate
contamination also could be attributable to wastes produced by the munitions
AND MARY JO KIRISITS
industry or from mining sites and other areas subjected to substantial use of
explosives.
Ammonium perchlorate is highly soluble in water and dissociates completely
to the ammonium (NH4+) and perchlorate ions. Though perchlorate is a very strong
oxidant from a thermodynamic perspective, it is kinetically inert in aqueous
solutions (Urbansky, 1998). These characteristics make perchlorate both highly
mobile and persistent in the environment.

A full report of this project, Removal of Bromate and Perchlorate in Conventional Ozone/GAC Systems
(90836), is available from AWWA Customer Service (1-800-926-7337). Reports are free to AWWA
Research Foundation subscribers by calling 303-347-6121.

70 FEBRUARY 2002 | JOURNAL AWWA • 94:2 | PEER-REVIEWED | BROWN ET AL

 treatment
technologies

BACKGROUND
Perchlorate contamination. Exten- FIGURE 1 Schematic of GAC filter
sive perchlorate contamination has
been found in drinking water supplies
in the western United States. For
example, perchlorate concentrations
of > 100 µg/L were detected in the
Glass column
San Gabriel Basin in Southern Cali-
fornia, one of the largest groundwa- Support
ter basins in the United States (Najm material
et al, 1999). The state of California Influent
has closed more than 18 wells because reservoir GAC

of perchlorate contamination up to
280 µg/L (WATERWEEK, 1997). Per-
Stir plate
chlorate has also been detected in
Lake Mead, one portion of the lower (Not to scale)

Colorado River system, which sup-

plies drinking water to more than 12 Pump
million people in Southern Califor- GAC—granular activated carbon
nia, Nevada, and Arizona (USEPA,
2001). Ten states have been found to
have perchlorate-contaminated drink- FIGURE 2 Ion exchange of perchlorate to GAC
ing water supplies; this number is
likely to increase because only six
Fraction Perchlorate Remaining—C/C0

2.5
states do not have users or manufac-
turers of perchlorate within their bor-
2.0
ders (Renner, 1999; Betts, 1998).
Health risks of perchlorate. Per- 1.5
chlorate is of concern because of its
ability to disrupt the thyroid gland’s 1.0
use of iodine in the generation of
metabolic hormones. When perchlo- 0.5
rate is taken up by the thyroid gland
instead of iodide, the thyroid gland 0.0
loses some hormone production 0 1,000 2,000 3,000 4,000 5,000 6,000
capacity (Book, 1999). This develop- Bed Volumes

ment is of particular concern for Initial conditions—Champaign–Urbana tap water, 50.4 µg/L perchlorate, 2.5 mg/L
infants and children, because it could dissolved oxygen, 4.5 mg/L nitrate, pH 7.5, 25-min empty bed contact time,
GAC—granular activated carbon
significantly hinder their normal phys-
ical growth and development.
Perchlorate regulation. Because con-
cern over perchlorate contamination
in the United States has only recently come to the fore- ther sampling and testing for perchlorate continue across
front, no state or federal standard for perchlorate exists. the nation, the number of known communities affected by
However, on the basis of a 1992 health study conducted perchlorate contamination will likely increase.
by the US Environmental Protection Agency, the CDHS
established a provisional perchlorate action level of 18 PERCHLORATE REMOVAL MECHANISMS
µg/L (CDHS, 1997b). Several source waters for drinking Past studies. The removal of perchlorate using gran-
water utilities currently contain perchlorate concentra- ular activated carbon (GAC) filtration has not been
tions that are greater than the 18-µg/L action level. As fur- well-documented. However, the literature supports the

BROWN ET AL | PEER-REVIEWED | 94:2 • JOURNAL AWWA | FEBRUARY 2002 71


FIGURE 3
Batch perchlorate removal in Champaign–Urbana tap water
GAC 1 GAC 2 GAC 3 GAC 4
1.0
Fraction Perchlorate Remaining—C/C0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 50 100 150 200
Time—min
Initial conditions—50 µg/L perchlorate, pH 7.5, 500 mg/L GAC. GAC 1—virgin RO 0.8,
GAC 2—virgin F-400, GAC 3—AWOG (acid-washed, outgassed) RO 0.8, GAC 4—AWOG
F-400, GAC—granular activated carbon
use of GAC filtration for the removal of several other
inorganic compounds. The chemical reduction of bro-
mate to bromide by activated carbon has been demon-
strated by several researchers (Asami et al, 1999; Bao
et al, 1999; Kirisits, 1997; Miller et al, 1996; Siddiqui
et al, 1996). It was generally observed that the effi-
ciency of bromate reduction was dependent on water
quality characteristics and the surface chemistry of the
carbon. Suidan and colleagues (1977) noted that
hypochlorous acid (HOCl) and the hypochlorite ion
(OCl–) are also chemically reduced in the presence of
activated carbon. It was found that HOCl reacts faster
with GAC than OCl– and that increased aqueous tem-
perature improved reactivity (Suidan et al, 1977).
Voudrias et al (1983) found that chlorite (ClO2–) is
reduced to chloride in the presence of GAC, whereas
Gonce and Voudrias (1994) observed that chlorate
(ClO3–) was reversibly sorbed to and not chemically
reduced by GAC. This finding suggests that the chem-
ical reduction of perchlorate by GAC may not be pos-
sible because perchlorate is closely related to chlorate.
Abiotic methods have been investigated for the
removal of perchlorate from drinking water systems. Of
these methods, ion exchange has received the most atten-
tion. It has been shown that various ion-exchange resins
and system configurations are effective at removing per-
chlorate (Tripp & Clifford, 2000; Batista et al, 1999;
Brown et al, 1999; Najm et al, 1999; Venkatesh et al,
1999). Najm and co-workers (1999) found that the effi-
ciency of perchlorate removal by ion exchange decreases
in the presence of sulfate or nitrate and that resin regen-
eration salt loading rates need to be at least 480 kg/m3
72 FEBRUARY 2002 | JOURNAL AWWA • 94:2 | PEER-REVIEWED | BROWN ET AL
(30 lb/cu ft) for efficient resin regen-
eration. However, Gu et al (2001)
have demonstrated that regenerating
perchlorate-laden ion-exchange resins
using tetrachloroferrate (FeCl 4 – )
anions can be highly efficient and may
be much more cost-effective than
using traditional regeneration tech-
niques. Ultimately, all ion-exchange
processes produce a concentrated per-
chlorate solution, which must be dis-
posed of or further treated.
Perchlorate reduction via biologi-
cal catalysis. Biologically catalyzed
perchlorate reduction is a promis-
ing treatment alternative for the
250 300 350 removal of perchlorate. Because the
chlorine atom within the perchlo-
rate molecule is in its highest oxi-
dation state (+7), the reduction of
perchlorate is highly thermodynam-
ically favorable. Rikken et al (1996)
provided the free-energy equations
for the stoichiometric reaction be-
tween acetate and oxygen, acetate and nitrate, and
acetate and perchlorate:
CH3COO– + 2O2 → 2 HCO3– + H+; Go’
(1)
= –844 KJ/mol acetate
CH3COO– + 3/5 NO3– + 13/5 H+ → 2 HCO3–
+ 4/5 H2O + 4/5 N2; Go’ (2)
= –792 KJ/mol acetate
fiCH3COO– + ClO4– → HCO3– + fiH+ + ClO2–; Go’
(3)
= –801 KJ/mol acetate
In Eq 3, ClO2– is used as the perchlorate reduction
end product because it has been demonstrated that the
microbially mediated transformation of ClO2– to Cl– is a
dismutation reaction that yields no energy for perchlorate-
reducing microorganisms (Coates et al, 1999; Rikken et
al, 1996).
The thermodynamic data indicate that perchlorate is
a strong oxidant (i.e., accepts electrons readily) and could
provide a large amount of energy to microorganisms as
an electron acceptor. Natural organic matter or an exter-
nally added compound could serve as an electron donor.
Although biologically based treatment of perchlorate-
contaminated wastewater has been practiced since the
1970s (Korenkov et al, 1976; Romanenko et al, 1976;
Yakovlev et al, 1973), biological perchlorate removal
technologies have only recently been applied to drinking
water. Using a 3-h residence time and 300 mg/L sodium
acetate, Herman and Frankenberger
(1999) demonstrated the reduction of FIGURE 4 Batch perchlorate removal in deionized, distilled water
perchlorate from 130 µg/L to < 5 µg/L
using sand-packed columns inoculated
GAC 1 GAC 2 GAC 3 GAC 4
with bacterial isolate strain perc1ace. 1.0

Fraction Perchlorate Remaining—C/C0

Miller and Logan (2000) used perchlo- 0.9
rate-reducing microorganisms isolated 0.8
from a wastewater treatment plant to
0.7
reduce perchlorate in an autotrophic,
0.6
gas-phase, packed-bed bioreactor. With
an average influent perchlorate concen- 0.5

tration of 740 µg/L and detention times 0.4

of 1.1–1.3 min, they achieved 38% ± 0.3
9% perchlorate removal over a 140-day 0.2
period. In a hollow-fiber membrane 0.1
biofilm reactor with a hydrogen pres- 0.0
sure of 2.5 psi and a 44-min detention 0 50
time, Nerenberg and Rittmann (2001)
observed the reduction of 6, 50, and 100 Initial conditions—50 µg/L perchlorate, pH 7.5, 500 mg/L GAC. GAC 1—virgin RO 0.8,
GAC 2—virgin F-400, GAC 3—AWOG (acid-washed, outgassed) RO 0.8, GAC 4—AWOG
µg/L perchlorate to between nondetection F-400, GAC—granular activated carbon
(<2 µg/L) and 4.5 µg/L.
Feasibility of biological systems for
drinking water treatment. Like perchlo-
rate, nitrate is an inorganic, anionic con-
taminant that can be found in drinking water sources
across the United States. A number of different processes
are used to treat nitrate-contaminated drinking water,
including ion exchange, chemical reduction, reverse osmo-
sis, electrodialysis, and biological denitrification (Kapoor
& Viraraghavan, 1997). Several full-scale, biological den-
itrification processes are operating in France and Ger-
many (Dahab & Woodbury, 1998; Richard, 1989). In
Guernes-Dennemont, France, a 2.5 mgd (9.5 ML/d) plant
removes nitrate using a fixed-film, down-flow reactor.
Expanded schist serves as the biofilm support medium, and
ethanol is the electron donor. Treated water is polished
using a sequence of biologically active carbon (BAC) fil-
tration, slow sand filtration, ozonation, and chlorination
(Dahab & Woodbury, 1998). Other full-scale biological
denitrification plants are located in Chateau Landon and
Ergany, France, and Monchengladbach, Germany.
Although biological filtration is a fairly new technol-
ogy in the United States, several US treatment plants use
biological filtration after ozonation (Najm & Trussell,
1998). These filtration systems reduce the concentration
of assimilable organic carbon in the effluent to provide bio-
logically stable product water. Additionally, the first full-
scale biological denitrification plant in the United States
went on line in 1998 in Coyle, Okla., and has been oper-
ating effectively ever since (Silverstein, 2000).
OBJECTIVES
A few promising technologies are being investigated
for the remediation of perchlorate-contaminated drink-
ing water. However, the literature has yet to report on a
treatment process that (1) reduces perchlorate from con-
100 150 200 250 300 350
Time— min
centrations typically found in drinking water sources
(<100 µg/L) to below the provisional action level (18
µg/L), (2) requires contact times characteristic of a drink-
ing water treatment plant, and (3) avoids the generation
of a concentrated perchlorate waste stream. The overall
objective of this research, therefore, was to evaluate the
use of activated carbon filtration to meet these three
treatment goals.
The specific objectives of the project were
• to determine whether the chemical reduction of
perchlorate could be facilitated on the surface of acti-
vated carbon,
• to determine the ion-exchange capacity of GAC for
perchlorate if chemical reduction does not occur, and
• to demonstrate that microbially mediated perchlo-
rate reduction can be achieved and sustained using BAC
filtration.
MATERIALS AND METHODS
Water. The natural water used for this research was
Champaign–Urbana tap water (CUW). It is a ground-
water that has been softened and typically contains 8
mg/L chloride, < 1 mg/L nitrate, < 1 mg/L sulfate, 0.8–1.2
mg/L ammonia, and 1.5 mg/L dissolved organic carbon
(DOC).
Organic-free water. Deionizing and distilling CUW pro-
duced a water with a DOC  0.3 mg/L. Deionized, dis-
tilled water (DDW) used for any perchlorate experimen-
tation was buffered with 1 mM phosphate buffer.
Dry chemicals. The chemicals used for this project were
all reagent-grade. When appropriate, chemicals were oven-
dried (105oC) overnight and then stored in a desiccator.

BROWN ET AL | PEER-REVIEWED | 94:2 • JOURNAL AWWA | FEBRUARY 2002 73

 surface oxygen groups and has a higher
FIGURE 5 Comparison of perchlorate removal in CUW and DDW surface pH (Miller, 1996).
GAC filters. Carbon filters (Figure
1), made using virgin RO 0.8 carbon,
GAC 2, DDW GAC 2, CUW were constructed in 25 mm (1 in.)
1.0
Fraction Perchlorate Remaining—C/C0

0.9
inner diameter glass pipes.‡ The pipes
were capped with PTFE and steel end-
0.8
caps. A 25 mm (1 in.) layer of 3 mm
0.7
(0.1 in) glass beads was placed above
0.6 and below the carbon bed to promote
0.5 uniform flow and prevent channeling.
0.4 Water was pumped from the influent
0.3 reservoir to the filters through PTFE
0.2 tubing using peristaltic pumps.§ The
0.1
influent consisted of CUW spiked with
50 µg/L ClO4– and 4.5 mg/L NO3–
0.0
0 50 100 150 200 250 300 350 and was buffered at pH 7.5 using 1
Time—min mM phosphate buffer. Samples were
Initial conditions—50 µg/L perchlorate, pH 7.5, 500 mg/L GAC. GAC 2—virgin F-400; taken from the influent and effluent
GAC—granular activated carbon; DDW—deionized, distilled water;
CUW—Champaign–Urbana tap water
of each column.
Abiotic batch experiments. A jar-test
apparatus** was used to conduct a
series of batch tests to determine
GAC. Two carbons were used for this research: RO 0.8 whether chemical reduction of perchlorate could be
GAC* and F-400 GAC.† The RO 0.8 carbon is an achieved using GAC under various conditions. Four GACs
extruded peat with mean pellet diameter of 0.8 mm (0.03 were used: virgin RO 0.8 (GAC 1), virgin F-400 (GAC 2),
in.) and length of 3.4 mm (0.13 in.). F-400 is generated AWOG RO 0.8 (GAC 3), and AWOG F-400 (GAC 4).
from bituminous coal and has a mean particle diameter After the carbon was weighed, it was stored in DDW
of ~ 0.5 mm (0.02 in.). Prior to use, the carbons were overnight to remove carbon fines. For the batch-test stud-
rinsed with DDW to remove carbon fines, placed in a ies, 2 L of DDW or CUW was placed in each jar. The
105 o C oven overnight, and then stored in a desiccator. pH of each solution was adjusted, 1 mM phosphate buffer
“Virgin” carbon was used to describe any carbon that was added, and 50 µg/L of perchlorate was spiked into the
(1) had not been used previously and (2) had not been batch-test jars. Nine different experiments were run. The
treated by an acid-wash outgassing (AWOG) procedure behavior of each of the four carbons was examined in
(Miller, 1996). both CUW and DDW at pH 7.5. The ninth experiment
Both the RO 0.8 and the F-400 car-
bons were used in their virgin and
AWOG forms. To acid-wash a carbon,
TABLE 1 Abiotic perchlorate batch removal and wash tests
approximately 10 g of virgin carbon
was placed in a column. The carbon
was rinsed with 10 bed volumes (BV) Mass of Perchlorate Mass of Perchlorate
of DDW, 10 BV of 1 normal Removed Washed Off Carbon
GAC* Water pH µg µg
hydrochloric acid (HCl), and 20 BV
of DDW, with an empty bed contact GAC 1 CUW† 7.5 42.0 Not measured

time (EBCT) of ~ 2.5–5.5 min. The GAC 2 CUW 7.5 71.8 70.2

carbon was dried overnight in a 105oC GAC 3 CUW 7.5 80.5 79.0

oven and then placed in a fluidized GAC 4 CUW 7.5 72.3 72.1

bed furnace to be outgassed. The car- GAC 1 DDW‡ 7.5 60.4 Not measured

bon was heated to a temperature of GAC 2 DDW 7.5 90.7 Not measured

about 900oC for 3 h. A nitrogen flow GAC 3 DDW 7.5 95.5 Not measured

rate of approximately 50 mL/min was GAC 4 DDW 7.5 95.7 Not measured

used in the furnace. The carbon was GAC 1 DDW 2.0 85.3 83.1

allowed to cool for 30 min in the inert *GAC—granular activated carbon; GAC 1—virgin RO 0.8; GAC 2—virgin F-400; GAC 3—AWOG (acid-
washed, outgassed) RO 0.8; GAC 4—AWOG F-400
nitrogen atmosphere and then stored †CUW—Champaign–Urbana tap water
in a desiccator. AWOG carbon typi- ‡DDW—deionized, distilled water

cally contains lower concentrations of

74 FEBRUARY 2002 | JOURNAL AWWA • 94:2 | PEER-REVIEWED | BROWN ET AL

used DDW, GAC 1, and pH 2. After
FIGURE 6 Comparison of batch perchlorate removal at pH 7.5 and 2.0 in deionized,
the addition of 500 mg/L of GAC, the
distilled water
jars were stirred to suspend the GAC,
and samples were taken regularly over
a 5-h period. GAC 1, pH 7.5 GAC 1, pH 2.0
Batch-test wash experiments. Upon 1.0

Fraction Perchlorate Remaining—C/C0

completion of the batch-test experi- 0.9
ments, the GAC from the four CUW 0.8
experiments was removed, placed in 0.7
separate glass columns equipped with 0.6
stopcocks and glass wool to prevent
0.5
the carbon from escaping into the efflu-
0.4
ent, and washed with a solution con-
0.3
taining 200 mg/L of nitrate. The efflu-
ent volume was measured and analyzed 0.2

for perchlorate. The GAC was washed 0.1

until no perchlorate was detected in 0.0
0 50 100 150 200 250 300 350
the effluent.
Time—min
BAC filters. BAC filters were struc-
turally identical to the GAC filters Initial conditions: 50 µg/L perchlorate, pH 7.5, 500 mg/L GAC. GAC 1—virgin RO 0.8,
described previously. Biological activ- GAC—granular activated carbon
ity was developed in the filters by con-
tacting the carbon with dechlorinated
CUW for 6,000 BV using a 25-min
EBCT. Gaseous nitrogen was used to lower influent dis- thiamine HCl), 448 mg/L perchlorate, 295 mg/L acetate,
solved oxygen (DO) concentrations to 2.5 mg/L, and 445 mg/L lactate, and 435 mg/L pyruvate.
reservoirs were fitted with clear acrylic, headspace-free Instead of the DDW solution used for BAC filtration
covers to prevent oxygen diffusion into the influent. An experiments, a basal medium was used to ensure that nutri-
electron donor solution composed of acetate, lactate, ent limitations were not factors for these “high” perchlo-
and pyruvate was added to the influent. rate concentration experiments. Organisms were exposed
Once the filters were biologically active, the influent to perchlorate concentrations that were approximately
base water was switched from CUW to DDW to allow for 9,000 times greater than concentrations used during BAC
better control over influent conditions. The influent pH filtration. There were also low initial cell numbers in the
was adjusted to 7.5, 1 mM phosphate buffer was added, batch tests, decreasing the availability of endogenous res-
and the appropriate concentrations of perchlorate and piration as a nutrient source. The medium was boiled to
nitrate were spiked to the influent. Samples were taken at remove DO, adjusted to pH 7.0, autoclaved, and cooled.
the influent and effluent of the filter. The BAC filter was stirred using a stainless-steel rod, and
Batch biological experiments. Batch biological experi- 1 mL of the resulting suspended biomass was inoculated into
ments were performed to confirm that biological reduction a 160-mL serum bottle that contained the sterilized basal
was the mechanism of perchlorate removal in the BAC medium. The bottle was placed on a shaker table and
filter. A basal medium was prepared for the batch bio- shaken at 200 rpm. A serum bottle containing no inoculum
logical experiments. The medium consisted of DDW spiked was prepared and placed on the shaker table as a control.
with buffer (0.25 g/L potassium diphosphate [KH2 4 PO ], Analytical methods. Nitrate, nitrite, perchlorate, chlo-
0.4 g/L potassium hydrogen phosphate [K2HPO4]), salts rate, chlorite, and chloride were analyzed by an ion chro-
(0.015 g/L calcium chloride [CaCl2]·2H2O, 0.02 g/L mag- matograph†† with an anion self-regenerating suppresser‡‡
nesium chloride [MgCl2]·6H2O, 0.007 g/L iron sulfate and conductivity meter. Three columns§§—including one
[FeSO4]·7H2O, 0.005 g/L sodium bisulfate [NaSO4]), that adsorbed organic material to prevent fouling of the
nitrogen (0.8 g/L ammonium chloride [NH4Cl]), trace analytical column—were used in series for nitrate, nitrite,
metals (5 mg/L manganese chloride [MnCl2]· 4H20, 0.5 chlorate, chlorite, and chloride analyses.
mg/L boric acid [H3BO3], 0.5 mg/L zinc chloride [ZnCl2],
0.5 mg/L cobalt chloride [CoCl2]· 6H2O, 0.5 mg/L nickel *Norit RO 0.8, lot no. 72303-5, Norit Americas, Atlanta, Ga.
sulfate [NiSO 4 ]· 6H 2 O, 0.3 mg/L copper chloride †Calgon F-400, lot no. 5925-LG, Calgon, Pittsburgh, Pa.
‡Ace Glass, Vineland, N.J.
[CuCl2]· 2H2O, 0.01 mg/L NaMoO2 · 2H2O), vitamins §Masterflex, Cole-Parmer, Vernon Hills, Ill.
(0.001 mg/L B12, 0.020 mg/L biotin, 0.050 mg/L calcium **B-KER2, Model 7790-400, Phipps and Bird, Richmond, Va.
††Series 300, Dionex, Sunnyvale, Calif.
pantothenate, 0.020 mg/L folic acid, 0.050 nicotinamide, ‡‡ASRS-1, Dionex, Sunnyvale, Calif.
0.100 mg/L pyridoxine HCl, 0.050 mg/L riboflavin, 0.050 §§Ionpac NG1, AG9-HC, and AS9-HC, Dionex, Sunnyvale, Calif.

BROWN ET AL | PEER-REVIEWED | 94:2 • JOURNAL AWWA | FEBRUARY 2002 75

 influent concentration at 1,600 BV.
FIGURE 7 Perchlorate removal versus nitrate remaining during syringe pump After 1,600 BV, perchlorate often
experiments
leached from the GAC, as indicated
by C/C0 values > 1. A mass balance
–
C/C 0 Effluent NO3 calculation demonstrated that > 99%
2.0 3.0 of the perchlorate mass applied to the
1.8
filter for 5,800 BV had been detected
Effluent Perchlorate—C/C0

2.5
1.6

Effluent NO3 —mg/L

in the filter effluent.
1.4 2.0 Perchlorate was displaced by other
1.2 4.5 1.5 mg/L 2.5 mg/L 2.0 mg/L

–
ions present in the influent that were
1.0 mg/L – 1.5
Influent NO3 more strongly attached to the GAC
0.8
1.0 than perchlorate. Therefore, any re-
0.6
0.4
moval of perchlorate up to 5,800 BV
0.5
0.2
was the result of ion exchange. To cal-
0.0 0.0 culate the column ion-exchange capac-
0 500 1,000 1,500 2,000 2,500 3,000 ity of the GAC for perchlorate under
Bed Volumes the given water quality conditions, the
Initial conditions—deionized, distilled water, 52.5 µg/L perchlorate, 2.5 mg/L authors integrated the area above the
–
dissolved oxygen, pH 7.5, 25-min empty bed contact time. NO3 —nitrate curve and below C/C 0 = 1 from
0BV1,600 (shaded area of Figure
2). This integration showed a column
ion-exchange capacity of 0.172 mg
The analytical method included a 9.0-mM sodium ClO – (1.7  10 –6 mol) per gram GAC. For an EBCT of
4
carbonate eluent, a 1.0-mL/min flow rate, and a 250- 15 min, this capacity would allow fewer than seven days of
µL injection loop. The analytical method for perchlo- filter operation before perchlorate breakthrough would
rate analyses included a 100-mM sodium hydroxide elu- occur and slightly more than two weeks of operation before
ent, a 1.0-mL/min flow rate, and a 250-µL injection loop; the removal capacity of the GAC was completely exhausted.
three columns* in series were used. All anions were Abiotic batch experiments. Batch experiments were per-
detected by suppressed conductivity. A calibration curve formed to investigate perchlorate removal using various car-
using five standards was run for each ion chromato- bons, water qualities, and pH conditions. Figures 3 and 4
graphic analysis, and new standards were prepared for show that each of the four carbons removed perchlorate
each curve. If the R2 correlation for a particular cali- in CUW and DDW, respectively. In both DDW and CUW,
bration curve was < 0.99, the calibration curve was GAC 2 performed better than GAC 1. This is likely attrib-
rerun. The standard curve was used both for calibration utable to GAC 1’s larger particle size. As particle size
and as an indicator of expected perchlorate retention increases, the distance over which a perchlorate molecule
times. Samples were stored in a 4oC cold room for five must diffuse before reaching an active sorption site on the
or fewer days before they were analyzed. Blank samples carbon also increases. Figures 3 and 4 also show that the
(DDW), which allowed for the detection of false-positives, AWOG-treated carbons performed better than the virgin
were also analyzed with each set of unknown samples. carbons. This is in agreement with earlier findings (Kirisits,
The LOD for perchlorate was calculated by the method 1997; Miller et al, 1996), which attributed the differences
outlined by Skoog and Leary (1992) and was found to be in virgin and AWOG carbon reactivity with bromate to
approximately 2 µg/L. Any perchlorate concentration changes in surface chemistry generated during the processes
that appeared to be < 2 µg/L was recorded as being at one of acid-washing and outgassing.
half of the LOD. A comparison of perchlorate removal by each of the
DO was measured using a DO meter and probe.† A four carbons in CUW and DDW showed that performance
pH/ion selective electrode meter with pH probe‡ was used was better in the lower organic content water. Figure 5 plots
to measure pH. DOC was analyzed with a total organic this comparison for GAC 2. Perchlorate is likely compet-
carbon analyzer by ultraviolet–persulfate oxidation.§ ing with other anions in CUW for active sites on the car-
bon, thus diminishing both the rate of perchlorate removal
RESULTS AND DISCUSSION as well as the overall removal capacity.
GAC filtration and ion-exchange capacity. The fraction of Analyses of perchlorate removal performance at a neu-
perchlorate remaining (C/C0) in the effluent of the 25- tral and low pH were also performed. GAC 1 exhibited
min CUW filter is plotted in Figure 2 for the first 5,800 faster removal kinetics and a greater removal capacity
BV of flow. Perchlorate breakthrough—defined as any when the influent pH was 2, compared with the pH 7.5
ion chromatographic detection of perchlorate—was condition (Figure 6). At a lower solution pH, more of
observed at 600 BV; effluent concentration was equal to the carbon’s surface functional groups will be protonated,

76 FEBRUARY 2002 | JOURNAL AWWA • 94:2 | PEER-REVIEWED | BROWN ET AL

thus increasing the amount of posi-
FIGURE 8 Sustained perchlorate removal in a biologically active carbon filter
tive charge on the carbon (Miller,
1996). Therefore, the lower solution
1.0
pH should favor sorption of negatively
0.9

Effluent Perchlorate—C/C0
charged perchlorate molecules. 0.8
Wash tests. Using 200-mg/L nitrate 0.7
solutions, wash tests were performed 0.6
to ascertain the mechanism of per- 0.5
chlorate removal during the batch 0.4
experiments (Table 1). It was observed 0.3
that within the limits of detection for 0.2

the ion chromatographic analysis, the 0.1

0.0
entire mass of perchlorate removed by 0 1,000 2,000 3,000 4,000 5,000 6,000
the GAC was washed off the carbons. Bed Volumes
Thus, ion exchange again proved to Initial conditions—52.5 µg/L perchlorate, 2.5 mg/L dissolved oxygen, 0.0–0.5
mg/L nitrate, pH 7.5, 25-min empty bed contact time
be the dominant mechanism of per-
chlorate removal by the GAC.
Biological filtration. Given that (1)
perchlorate was not substantially reduced by GAC and gesting that nitrate was outcompeting perchlorate for use
(2) the filter demonstrated a relatively low column ion- as an electron acceptor.
exchange capacity for perchlorate, it is evident that abiotic Additionally, Figure 7 shows that even when influent
GAC filtration would not be an effective means of reme- nitrate concentrations were constant, effluent nitrate con-
diating perchlorate-contaminated water. Therefore, the centrations fluctuated widely. This was a result of unin-
authors evaluated microbially catalyzed perchlorate reduc- tentional fluctuations in influent DO concentrations. The
tion and the use of biologically active carbon filtration. influent DO profile, which varied between 2.5 and 4.5
The GAC filter was contacted extensively with dechlo- mg/L, closely paralleled the effluent perchlorate and nitrate
rinated CUW to develop biological activity in the filter. An profiles shown in Figure 7. The electron donor solution
electron donor mixture of acetate, lactate, and pyruvate was was added at a concentration sufficient only to reduce
added to the influent matrix to expedite the growth of the influent nitrate and 2.5 mg/L DO (1.5–1.7 mg/L as C).
microorganisms. During the acclimation period, DOC Given that effluent DO concentrations were always at
removal was initially significant, but subsequently decreased the LOD (~0.1 mg/L), any increase in the influent DO
as adsorption capacity was approached. Toward the end concentration diminished the availability of electron donor
of the acclimation period, however, DOC removal increased for nitrate and perchlorate reduction. DOC measure-
and remained essentially constant at approximately 60%, ments confirmed that electron donor availability was a lim-
indicating that the filters had been rendered biologically iting variable during this experiment.
active. After the acclimation period, the influent base Sustainability of perchlorate reduction. To determine
water was switched from CUW to DDW to allow for bet- whether efficient perchlorate reduction was sustainable for
ter control over influent conditions. A syringe pump** long periods of flow, adjustments were made so that elec-
was installed to add a sterile electron donor solution at a tron donor availability was not limiting. A new headspace-
point just before the influent reached the filter. The acetate– free cover was constructed that fit the reservoir tighter and
lactate–pyruvate mixture was added at a concentration was more effective at preventing oxygen diffusion. In addi-
required to stoichiometrically reduce all influent DO and tion, the electron donor solution was supplied at a con-
nitrate. (For calculation purposes, it was assumed that (1) centration sufficient to completely reduce 1.5 mg/L nitrate
nitrate was reduced completely to N2 gas and (2) the frac- in excess of that required to stoichiometrically remove all
tion of electrons used for energy [fe] = 1.) influent DO and nitrate (~1.7 mg/L as C). Influent nitrate
The results of the first BAC–syringe pump experiment was maintained between 0.03 mg/L (ambient) and 0.5
are shown in Figure 7. Initially, 4.5 mg/L of nitrate was mg/L, and a 25-min EBCT was used. Under these condi-
added to the influent. At 300 BV, the influent nitrate con- tions, 50 µg/L of influent perchlorate was reduced to below
centration was decreased to 1.5 mg/L, resulting in an im- the LOD for 103 days of operation (Figure 8). Perchlo-
mediate decrease in effluent perchlorate concentrations. rate was detected in the effluent (at 5 µg/L) only once dur-
The BAC filter achieved at least 80% perchlorate removal ing this period, and there were no indications that per-
for more than 1,000 BV of flow. It was also observed
that when DO concentrations were low, the biological *Ionpac NG1, AG11, and AS11, Dionex, Sunnyvale, Calif.
reduction of perchlorate was highly sensitive to the con- †Models 58 and 5905, YSI, Yellow Springs, Ohio
‡Model 720A, Orion, Beverly, Mass.
centration of effluent nitrate. As effluent nitrate concen- §Phoenix 8000, Tekmar-Dohrmann, Cincinnati, Ohio
tration increased, perchlorate removal decreased, sug- **kd Scientific, Boston, Mass.

BROWN ET AL | PEER-REVIEWED | 94:2 • JOURNAL AWWA | FEBRUARY 2002 77

chlorate removal would not have continued past 103 days
had the experiment continued. Effluent DO, nitrate, and
DOC concentrations were approximately 0.25 mg/L, 0.03
mg/L, and 0.35 mg/L as C, respectively.
On day 57 (3,300 BV), head loss across the BAC pro-
duced a leak in the tubing leading into the filter. To remove
some of the excess biofilm that was causing the head loss,
the BAC was extracted from the glass pipe, placed in a large
beaker containing DDW, and agitated for 30 s using a stain-
less-steel rod. The resulting suspended biomass was decanted,
and the BAC was returned to its original position in the
glass pipe. No change in perchlorate removal performance
was observed prior to or after this cleaning procedure.
Batch biological experiments. The suspended biomass
that was decanted during the filter-cleaning procedure
was utilized in batch experiments to verify the assump-
tion that biological reduction was responsible for the
removal of perchlorate during BAC filtration. After a 51-
day incubation, the initial 447.5 mg/L of perchlorate in
the serum bottle had been reduced to 23.9 mg/L. The
serum bottle that was prepared without the microbial
inoculum (control) showed no perchlorate reduction
over the same incubation period. Therefore, it can be
concluded that 423.6 mg/L (4.26 mM) perchlorate had
been biologically reduced. Furthermore, an end-prod-
uct analysis showed that perchlorate was completely
reduced to chloride. Although no chlorate or chlorite
accumulated in the bottle, the chloride concentration
increased by 143.2 mg/L (4.04 mM), accounting for
95% of the perchlorate decrease in the bottle. This is
consistent with other biological perchlorate reduction
research found in the literature, which has demon-
strated complete microbial conversion of perchlorate
to chloride (Coates et al, 1999; Herman & Franken-
berger, 1999; Rikken et al, 1996; Attaway & Smith,
1993; Malmqvist et al, 1991).
CONCLUSIONS
• Abiotic GAC filter experiments demonstrated that
perchlorate was removed by ion exchange rather than
by chemical reduction. A column ion-exchange capacity
of virgin GAC for perchlorate was calculated to be 0.172
mg perchlorate per g GAC. Batch experiments using CUW
and DDW, four carbons, and two pH values confirmed
that ion exchange was the mechanism of perchlorate
removal by GAC.
• Biological reduction of perchlorate in the low-micro-
gram-per-litre concentration range was achieved using
BAC filtration. Perchlorate reduction was highly sensitive
to the concentration of nitrate in the filter. As nitrate con-
centrations increased, perchlorate removal decreased.
When influent DO and nitrate concentrations were kept
low and excess electron donor was added (~1.7 mg/L as
C), the BAC filter removed 50 µg/L perchlorate to below
detection for 103 days of flow at a 25-min EBCT (~6,000
BV). There were no indications that perchlorate removal
would not have continued past 103 days had the exper-
iment continued.

REFERENCES
Asami, M. et al, 1999. Bromate Removal During
Transition From New Granular Activated
Carbon (GAC) to Biological Activated Car-
bon (BAC). Water Res., 33:12:2797.
Attaway, H. & Smith, M., 1993. Reduction of
Perchlorate by an Anaerobic Enrichment
Culture. Jour. Indus. Microbiol., 12:408.
Bao, M.L. et al, 1999. Removal of Bromate Ion
From Water Using Granular Activated
Carbon. Water Res., 33:13:2959.
Batista, J R.; McGarvey, F.X.; & Vieira, A.R.,
1999. Removal of Perchlorate From
Waters Using Ion Exchange Resins. Proc.
ACS Natl. Mtg., New Orleans.
Betts, K.S., 1998. Rotating Ion Exchange Sys-
tem Removes Perchlorate. Envir. Sci. &
Technol., 32:19:454A.
Book, S., 1999. Perchlorate MCL. Presentation
at multisponsored national Perchlorate
Conference, Ontario, Calif. (Mar. 18–19).
Brown, G.M. et al, 1999. Design of Selective
Resins for the Removal of Pertechnetate
and Perchlorate From Groundwater. Proc.
ACS Natl. Mtg., New Orleans.
CDHS (California Department of Health Ser-
vices), 1997a. Determination of Perchlo-
rate by Ion Chromatography, Rev. 0. Sani-
tation and Radiation Laboratories Branch,
Berkeley, Calif.
CDHS, 1997b. Perchlorate in Drinking Water—
Basis for Provisional Action Level. Proc.
AWWA Res. Fdn. Perchlorate Research
Issue Group Conf., Ontario, Calif.
Coates, J.D. et al, 1999. Ubiquity and Diversity
of Dissimilatory (Per)chlorate-Reducing
Bacteria. Applied & Envir. Microbiol.,
12:5234.
Dahab, M.F. & Woodbury, B.L., 1998. Biological
Options for Nitrate Removal From Drink-
ing Water. Proc. AWWA Inorganic Conta-
minants Workshop, San Antonio, Texas.
Gonce, N. & Voudrias, E.A., 1994. Removal of
Chlorite and Chlorate Ions From Water
Using Granular Activated Carbon. Water
Res., 28:5:1059.
Gu, B. et al, 2001. Regeneration of Perchlorate
(ClO4–)-loaded Anion Exchange Resins by
a Novel Tetrachloroferrate (FeCl4–) Dis-
placement Technique. Envir. Sci. & Tech-
nol., 35:3363.
Herman, D.C. & Frankenberger, W.T. Jr., 1999.
Bacterial Reduction of Perchlorate and
Nitrate in Water. Jour. Envir. Qual.,
28:1018.
Kapoor, A. & Viraraghavan, T., 1997. Nitrate
Removal From Drinking Water. Jour. Envir.
Engrg., 4:371.
Kirisits, M.J., 1997. Removal of Bromate by
Granular Activated Carbon Filtration.
Master’s thesis, Univ. of Illinois at
Urbana–Champaign.
Korenkov, V.N. et al, 1976. Process for Purifica-
tion of Industrial Waste Waters From Per-
chlorates and Chlorates. US Patent
3,943,055.
Malmqvist, A.; Welander, T.; & Gunnarsson, L.,
1991. Anaerobic Growth of Microorgan-
isms With Chlorate as an Electron Accep-
tor. Applied & Envir. Microbiol., 57:2229.
Miller, J.A., 1996. Reduction of Bromate in
Drinking Water by Activated Carbon. Doc-
toral dissertation, Univ. of Illinois at
Urbana–Champaign.
Miller, J.; Snoeyink, V.L.; & Harrell, S., 1996.
Effect of Granular Activated Carbon Sur-
face Chemistry on Bromate Reduction.

78 FEBRUARY 2002 | JOURNAL AWWA • 94:2 | PEER-REVIEWED | BROWN ET AL

 • Batch biological experiments verified the occurrence
of biological reduction and demonstrated that microor-
ganisms in the BAC filter convert perchlorate to chloride.
• BAC filtration is a promising treatment process for
the remediation of perchlorate-contaminated drinking
water. Using customary contact times, BAC filtration can
reduce concentrations of perchlorate typically found in
drinking water to below the detection limit (~2 µg/L). In
addition, biological filtration completely destroys the per-
chlorate molecule, converting it to innocuous chloride.
The conditions required to achieve biological perchlo-
rate reduction are similar to those required to achieve bio-
logical denitrification. Given that full-scale biological den-
itrification is being applied throughout Europe and is now
being considered in the United States, it is clear that BAC
filtration may be a feasible option for perchlorate treatment
in drinking water. Before BAC filtration can be applied
to full-scale perchlorate remediation, however, the process
must be optimized with respect to operating conditions and
water quality characteristics. In addition, the production
of any undesirable taste and odor compounds, the bio-
logical stability and disinfectant demand of product water,
and any reaeration requirements must be examined to
ensure potability of the product water from the plant,
through the distribution system, to the tap.
support for this work was provided by the AWWA
Research Foundation (AWWARF grant 2535).
ABOUT THE AUTHORS:
Jess C. Brown* is a doctoral candidate in the Depart-
ment of Civil and Environmental Engineering, Uni-
versity of Illinois at Urbana–Champaign, 3230 New
Mark Civil Engineering Laboratory,
205 N. Mathews Ave., Urbana IL
61801, <jcbrown1@uiuc.edu>. He
has MS and BS degrees from the
University of Illinois at
Urbana–Champaign and a BA
degree from Harvard University,
Cambridge, Mass. He is the 2001 recipient of
AWWA’s Larson Aquatic Research Support Scholar-
ship. Vernon L. Snoeyink is the Racheff Professor of
Environmental Engineering at the University of Illi-
nois at Urbana–Champaign. Mary Jo Kirisits is a
postdoctoral research associate at Northwestern
University.
*To whom correspondence should be addressed

ACKNOWLEDGMENT If you have a comment about this article, please contact

The authors thank Lutgarde Raskin, Joanne Chee- us at <journal@awwa.org>.
Sanford, and Robert Sanford for their input. Financial

Disinfection By-products in Water Treat-
ment—The Chemistry of Their Formation
and Control (R.A. Minear & G.L. Amy, edi-
tors). Lewis Publ., Boca Raton, Fla.
Miller, J.P. & Logan, B.E., 2000. Sustained Per-
chlorate Degradation in an Autotrophic,
Gas-phase, Packed-bed Bioreactor. Envir.
Sci. & Technol., 34:14:3018.
Najm. I. et al, 1999. Application of Ion
Exchange Technology for Perchlorate
Removal From Drinking Water. Proc. 1999
AWWA Ann. Conf., Chicago.
Najm. I. & Trussell, R.R., 1998. New and Emerg-
ing Drinking Water Treatment Technolo-
gies. Proc. NSF Workshop on Emerging
Drinking Water Contaminants, Washington.
Nerenberg, R. & Rittman, B.E., 2001. Concur-
rent Perchlorate and Nitrate Reduction in
a Novel, Hollow-fiber-membrane Biofilm
Reactor. Proc. 2001 AWWA Ann. Conf.,
Washington.
Renner, R., 1999. EPA Draft Almost Doubles
Safe Dose of Perchlorate in Water. Envir.
Sci. & Technol., 33:5:110A.
Richard, Y.R., 1989. Operating Experiences of
Full-Scale Biological and Ion- Exchange
Denitrification Plants in France. Jour.
Instit. Water & Envir. Mgmt. (IWEM),
4:3:154.
Rikken, G.B.M.; Kroon, A.G.; & van Ginkel, C.G.,
1996. Transformation of (Per)chlorate Into
Chloride by a Newly Isolated Bacterium:
Reduction and Dismutation. Appl. Micro-
biol. Biotechnol., 45:420.
Romanenko, V.I.; Korenkov, V.N.; & Kuznetsov,
S.I., 1976. Bacterial Decomposition of
Ammonium Perchlorate. Mikrobiologiya,
61: 347.
Siddiqui, M. et al, 1996. Bromate Ion Removal
by Activated Carbon. Water Res.,
30:7:1651.
Silverstein, J., 2000. Personal communication.
Skoog, D.A. & Leary, J.J., 1992 (4th ed.). Princi-
ples of Instrumental Analysis. Harcourt
Brace College Publ., Fort Worth, Texas.
Suidan, M.T.; Snoeyink, V.L.; & Schmitz, R.A.,
1977. Reduction of Aqueous HOCl With
Activated Carbon. ASCE, Jour. Envir.
Engrg. Div., 103:EE4:677.
Tripp, A.R. & Clifford, D.A., 2000. Treatability of
Perchlorate in Groundwater Using Ion
Exchange Technology. Proc. 2000
AWWA–AWWA Res. Fdn. Inorganic Cont-
aminants Workshop, Albuquerque, N.M.
Urbansky, E., 1998. Review and Discussion of
Perchlorate Chemistry as Related to
Analysis and Remediation. Bioremed.
Jour., 2:81.
USEPA (US Environmental Protection Agency),
2001. Perchlorate, <www.epa.
gov/OGWDW/ccl/perchlor/perchlo.html>.
Ofce. of Ground Water and Drinking
Water, Washington (updated July 26).
Venkatesh, K.R. et al, 1999. Removal and
Destruction of Perchlorate and Other
Anions From Ground Water Using
ISEP+TM System. Proc. ACS Natl. Mtg.,
New Orleans.
Voudrias, E.A. et al, 1983. Reactions of Chlorite
With Activated Carbon and With Vanillic
Acid and Indan Adsorbed on Activated
Carbon. Water Res., 17:9:1107.
WATERWEEK, 1997. Perchlorate Detected in Lake
Mead Water. AWWA, Denver (Sept. 1).
Yakovlev, S.V. et al, 1973. Method for Biochemi-
cal Treatment of Industrial Waste Water.
US Patent 3,755,156.

BROWN ET AL | PEER-REVIEWED | 94:2 • JOURNAL AWWA | FEBRUARY 2002 79