&EPA
United States
Environmental Protection
Agency
Water Engineering
Research Laboratory
Cincinnati, OH 45268
Research and Development EPA/600/M-87/003 June 1987
ENVIRONMENTAL
RESEARCH BRIEF
Bacteria Attached To Granular
Activated Carbon In Drinking Water
Gordon A. McFeters, Anne K. Camper, Mark W. LeChevallier,
Susan C. Broadaway, and David G. Davies
Laboratory and field studies were undertaken to answer
basic questions about the influence of granular activated
carbon (GAC) on the bacteriological quality of drinking
water. A sampling apparatus consisting of a 47-mm
Swmnex* and a 16-tayer gauze filter was developed to
trap filter fines from large volumes of water. A desorption
technique (Zwittergent 3-12, 10 6M; EGTA, 1Q-3M,
peptone, 0.01%; Tns buffer, pH 70, DIM; homogenized
at 4°C for 3 mm at 16,000 rpm) combined with optimal
cultunng procedures (heterotrophs, R2A medium at 28°C
for 7 days; coliforms, mT7 medium MF procedure and an
MPN with lauryl sulfate added after 4 hr of incubation)
allowed for the enumeration of particle-associated
bacteria
GAC-attached bacteria were resistant to 2 0 mg/L
chlorine after 1 hr of exposure. Enteric pathogens were
capable of colonizing laboratory-scale GAC filters Their
colonization potential and longevity depended on the
presence of autochthonous river water organisms GAC
filter particles were found in effluents from properly
operated treatment facilities. More than 40% of the
samples obtained contained particles significantly
colonized with heterotrophic plate count bacteria, 17%
were populated with coliforms. The appearance of
colonized fines was not related to a specific time in filter
operation. Increases in the breakthrough of bacteria-
laden particles were seen in the spring and fall Several
operational variables (increased bed depth, turbidity of
applied water, and filtration rate) did correlate positively
with the presence of fines in filter effluents Bed age was
not associated with breakthrough
* Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
This Research Brief was developed by the principal
investigators and EPA's Water Engineering Research
Laboratory, Cincinnati, OH, to announce key findings of
the research project that is fully documented in the
reports and publications listed at the end.
Introduction
The deterioration of surface water quality and increasing
taste and odor problems in drinking water have resulted
in the widespread use of activated carbon filtration
medium The active surface area of this substance makes
it ideal for the removal of organic molecules, including
trihalomethanes (THM's). However, certain characteristics
of this compound also make it an ideal substance for the
concentration of bacterial nutrients Bacteria have been
shown to adsorb to and extensively colonize the surface
of activated carbon particles in granular activated carbon
(GAC) filter beds The seeding of distribution systems
with bacteria can occur if colonized particles pass
treatment barriers or if the organisms are sloughed or
sheared from filters. These cells are not deleterious to
water quality if disinfection is adequate, as previous
studies have suggested However, the chlorine resistance
of GAC-adsorbed cells has not been adequately
addressed
Prior investigations into the bacteriological impact of GAC
filtration have relied on standard grab sample and
enumeration procedures Thus it is possible that a heavily
colonized particle would yield only one colony. Some
researchers have reported large nmbers of bacteria in
GAC-filtered drinking water and others have not. The
lack of appropriate methods to enumerate adsorbed cells
has contributed to this problem The development of
procedures to release cells from particles, deaggregate
-------
them, and effectively enumerate them is necessary to
adequately study the influence of GAG on the
bacteriological quality of treated water
If bacteria from GAG filter beds can reach drinking water,
the public health significance of these organisms must be
considered. Earlier research has identified potential or
opportunistic pathogens in GAC-treated water. Research
into the ability of pathogens to colonize GAG filters and
survive disinfection was undertaken to determine whether
pathogenic bacteria could enter drinking water from this
source.
If GAG filter material can penetrate treatment barriers, it is
important to define how this phenomenon is influenced by
variables in the operation of a drinking water filter. Though
other investigators have implicated certain operational
procedures in the appearance of organisms in finished
water, no attempt has been made to identify variables
involved in the occurrence of populated filter fines
Information from these studies could help plant operators
select procedures to minimize breakthrough of colonized
filter material.
This summary addresses the following questions related
to research conducted in these areas: (1) How can GAC-
borne bacteria be accurately enumerated, (2) how
susceptible are adsorbed bacteria to disinfection by
chlorine, (3) if bacteria of public health significance can
colonize GAG filter beds, do colonized particles of filtration
media actually appear in finished water, (4) what
operational variables contribute to the occurrence of
colonized particles in filtered water, and (5) what
physiological advantage, if any, do GAC-attached cells
have over planktonic cells.
Results and Discussion
Desorption and Enumeration Techniques
Fundamental to this research was the establishment of a
procedure to remove bacteria adsorbed to the surface of
GAG and to prevent reattachment without compromising
bacterial viability. The dispersal of individual cells would
then allow for a more accurate determination of the actual
bacterial load on waterborne GAG particles
Initial experiments were performed to determine which
physical means of interrupting cell-surface interactions
were most efficient. Sonication and blending decreased
cell viability as a result of heat generation. Optimal counts
were attained when the sample was homogenized for 3
min at 16,000 rpm in a container immersed in an ice bath
(ca. 4°C). In addition, specific chemicals and enzymes
selected for their ability to disrupt extracellular polymeric
materials or to act as surface-charge interactors were
tested in conjunction with homogenization.
Of the 35 chemicals, combinations, or concentrations
evaluated, the greatest numbers of bacteria were detected
with a solution of Zwittergent 3-12 (10-6 M),
ethyleneglycol-bis-(beta amino-ethyl ether)-N,N1-
tetra acetic acid (EGTA) (10-3 M), peptone (0.1 %), and
Tris buffer (0.1 mM, pH 7.0). The efficacy of the technique
was tested using a known number of cells adsorbed to
activated carbon. Approximately 90% of the bacteria were
recovered, as determined by plate counts. This result was
supported by observation of acridine-orange-treated
GAG particles with epifluorescence microscopy.
The medium and growth conditions under which
maximum cell counts could be obtained were then
determined. Four media (plate count agar, 0.1 plate count
agar, mSPC, and R2A) were incubated at 28°C for varying
lengths of time (2 to 7 days) for the enumeration of
heterotrophic plate count (HPC) bacteria. R2A medium
incubated for 7 days at 28°C consistently provided the
highest counts. mT7 agar was used as the medium of
choice for enumerating desorbed coliforms by the
membrane filter (MF) procedure. A modified MPN (3
tubes, 3 dilutions, and lauryl sulfate addition after 4 hr of
incubation) was also used as a coliform enumeration
technique, as it has been shown that sample turbidity can
interfere with MF detection of these organisms.
Susceptibility of GAC-Attached Bacteria to
Chlorine
Particle-associated bacteria are reported to be more
resistant to disinfection. Thus because colonized GAG
filter bed particles could be released from the filter, the
chlorine resistance and public health significance of these
attached bacteria merit investigation.
Experiments were conducted with GAG removed from an
operating drinking water filter and maintained in a column
in the laboratory. Planktonic ceils cultured from this
column were also used. GAC-attached and planktonic
cells of an Escherichia coli river isolate and the pathogens
Salmonella typhimurium, Yersinia enterocolitica, and
Shigella sonnei were also tested. For these experiments,
the attachment was accomplished by exposing virgin GAG
to suspensions of each bacterial species for 20 min
followed by gentle rinsing. In this time, the bacteria had
little opportunity to produce extracellular material. Cells
were also grown in the presence of GAC to evaluate the
chlorine resistance of the GAC-attached biofilms.
Scanning electron micrographs of these particles and of
those from the drinking water filter revealed colonies of
bacteria and the presence of extracellular polymer. The
exposure of planktonic HPC cells to chlorine (2.0 mg/L)
resulted in a rapid decrease in viability (within 5 mm),
whereas GAC-associated cells experienced little decline
in numbers after exposure for 1 hr (Figure 1). No decrease
in viability was observed within 1 hr for the GAC-grown
coliforms. Some injury did occur with GAC-attached
cells, suggesting that the extracellular polymer produced
by the grown'cells or the integrity of the colony afforded
some amount of protection from chlorine.
These data suggest a means by which bacteria, including
pathogens, can breach disinfection barriers and enter
distribution systems.
Growth and Persistence of Enteric Pathogens on
GAC Filters
The potential for colonization of GAC by enteric pathogens
(Yersinia enterocolitica. Salmonella typhimurium, and a
human enterotoxigenic Escherichia coli) was investigated.
Laboratory GAC columns were inoculated with each
organism in the presence or absence of autochthonous
river water organisms. Core samples were removed from
the columns at regular intervals and homogenized.
-------
Figure 1. Survival of naturally occurring helerotrophic
plate count bacteria exposed to chlorine at 2.0
mg/L for 1 hour (free chlorine residual after 1
hour was 1.7 mg/L).
Viable Count /gram
105
Time (mm)
Bacteria were enumerated by the spread plate technique
on plate count agar (HPC) and a selective medium that
had been shown to provide maximum counts of each
pathogen. When each pathogen was suspended in sterile
river water and introduced to a sterile GAC column, the
GAC was rapidly colonized Maximum colonization (ca.
105 to 107 cfu g~1 GAC) had occurred by the first
sampling time (2 days) and remained for the duration of
the experiment (14 to 20 days). When these columns were
then exposed to nonstenle river water, the population of
the pathogen declined gradually (0.08 to 0.14 log day"1),
and pathogens existed (104 to 106 cfu g~1 GAC) at the
termination of the experiment. The addition of pathogens
to nonstenle river water circulated through an initially
sterile column resulted in colonization by pathogens at
rates similar to those obtained with sterile water. However,
the pathogen numbers declined at a more rapid rate (0.10
to 0.22 log day1), than when colonization was
established before the addition of HPC bacteria. If the
pathogens were introduced into a column supporting a
mature biofilm of HPC bacteria, there was least
attachment (104 cfu g~1 GAC) In this case, pathogen
cells attached to the GAC declined at a more rapid rate
(011 to 0.70 log day~1).
Data from the experiments with Salmonella typhimurium
are shown in Figure 2. The results demonstrate the
Figure 2.
CFU Per Gram Carbon
Attachment and persistence of S. typhimurium
on GAC Columns.
703'
102
12
14
• Sterile River Water, Sterile Carbon
O Continuation of Above, Nonstenle River
Water Added
L Nonstenle River Water, Sterile Carbon
• Nonstenle River Water, Precolon/zed Carbon
importance of indigenous surface water organisms in the
control of human enteropathogenic organisms on GAC. An
established biofilm of heterotrophic bacteria appears to be
beneficial in controlling the attachment and longevity of
pathogens on GAC filters.
Colonized Filter Fines in Drinking Water
The occurrence of populated GAC filter fines in drinking
water was substantiated at nine operating drinking water
treatment facilities A sampling device was developed to
allow the testing of large volumes of drinking water The
apparatus consisted of a 47-mm Swinnex filter holder
with the ends bored to a 6-mm inside diameter. A sterile,
16-layer gauze filter was enclosed in the unit The
samplers were installed in each treatment facility at a
point after GAC filtration and before final chlormation. GAC
filter effluent was passed through the gauze for the entire
filter cycle 1 hr before backwash or 4 hr after backwash.
The gauze was then removed and shipped to the lab
Particles were chlorinated (2.0 ppm, 30 mm) to inactivate
planktonic cells, since attached cells were more resistant.
Following dechlormation, the sample was split. Half was
-------
homogenized, and the other half was handshaken. A
greater-than-twofold-mcrease in colony-forming units
(cfu)/ml_ in homogenized over handshaken samples was
used as an index of significant particle colonization More
than 200 gauze filters were received. The trapped
particles were examined for attached HPC and cohform
bacteria
Forty-one percent of the samples contained GAG
particles colonized by HPC bacteria Cohforms were found
in association with fines from GAC-filtered water in 17%
of the samples (Table 1) Of these, nearly 28% exhibited
the fecal biotype Scanning electron micrographs of
particles clearly demonstrated bacterial cells and
associated extracellular material in surface pits and cracks
of the GAC particles Analyses also showed that colonized
filter material was released throughout the filter cycle and
was not related to turbidity spikes just before and after
backwashmg. Evaluation of the data on the basis of time
of year revealed a distinct seasonal trend in the
occurrence of attached cohforms (spring and autumn)
This trend was not seen with HPC bacteria Image
analysis of the particles released from GAC drinking water
filters provided information as to their size and shape.
Most particles were nearly spherical. The sizes varied
from 1.0 urn to 3 5 x 103 \im
Note that all these results were based on samples
received from drinking water treatement facilities that were
run well and in compliance with established operation
regulations. The data show that bacteria attached to
carbon fines may be an important mechanism by which
microorganisms can pass treatment barriers and enter
finished water. Indicator bacteria and potential or
opportunistic pathogens were observed on these particles
that would not have been enumerated by conventional
analysis of drinking water samples
Influence of Operating Variables of GAC-Filters
on the Occurrence of Populated Fines in
Drinking Water
The effects of treatment differences on the release of filter
material were studied at two drinking water treatment
facilities. Sampling procedures described previously were
used Statistical analyses of data obtained during 1 year of
bimonthly sampling revealed that GAC filter bed age
(virgin and 1, 2, and 3 years old) does not affect the
release of colonized particles. At Plant 1 (Table 2),
significantly more HPC-populated particles were
observed as GAC bed depth increased from 60 cm to 1.5
m. Coliform bacteria also increased with depth from 6% of
the 60-cm samples to 25% of the 1.5-m samples.
Colonized particle breakthrough correlated positively with
increased mean turbidity of applied water (mean 4.3 ntu
versus 2 9 ntu). Filtration rate also proved to be important.
As the flow rate doubled from 4.9 to 9.8 m/hr (2 to 4
gpm/ft2), more filter fines were released, and these
particles were populated to a greater extent.
Also investigated was the relative contribution of various
filtration media to the appearance of populated particles in
finished water Studies conducted in the laboratory
showed that columns of sand, anthracite, and three brands
of GAC were all colonized to the same level by HPC
bacteria. However, GAC columns (regardless of
manufacturer) supported nearly a 1-log higher cohform
Table 1.
Analysis of Particles Collected from GAC-Treated Effluents.
Coliform
Item
Total Number of Samples
Number Showing > 2x Increase
Meanfold lncreaseb
Maximum Increase
neieruirupinu
Plate Count
198
82 (41.4)a
8.6
50.0
MF
201
14 (7.0)
124.3
1194.0
MPN
191
33 (17.2)
24.5
122.2
a Numbers in parenthesis indicate percentage of total samples.
b Homogenized versus handshaken analyses
Table 2. Effect of GAC Filter Bed Depth, Applied Water Quality, and Filtration Rate on the Release of Populated GAC
Particles into Drinking Water at Plant 1.
Depth
Turbidity (ntu)
Flow Rate (Lpm/m2)
Item
60 cm
1.5 m
2.92
4.32
0.72
1.44
Number of Samples
Mean Filter Rating
Median HPC Ratio3
p Value
16
3.6
1.39
0.002
16
2.3
2.21
0.002
16
2.1
1.65
0.004
16
2.1
3.04
0.004
17
2.0
1.53
0.001
16
3.6
3.86
0.001
a Ratio of colony-forming units from homogenized values divided by handshaken values from split samples.
-------
(Klebsiella oxytoca) load than sand or anthracite
Sampling devices were installed at treatment facilities with
anthracite and sand filters. When compared with results
from GAC-filtered water, the GAC-treated effluents
contained more particles colonized with chlorine-
resistant organisms.
Physiology of GAC-Grown Klebsiella oxytoca
Compared with Planktonic Growth
Laboratory studies showed that the growth rate of
Klebsiella oxytoca adsorbed on GAG was enhanced up to
10 times that of planktonic cells when the organisms were
provided with a negatively charged substrate (glutamate)
that could adsorb to the particle surface. No differences
were observed when the uncharged substrate glucose was
used
[3H]-thymidme was used to assess DMA biosynthesis.
GAC-attached cells grown on glutamate (20 0 mg/L) took
up to five times more [3H]-thymidine than did unattached
cells grown in liquid medium. When [3H]-undme was
used to measure RNA turnover, the GAC-attached cells
took up 11 times more [3H]-uridine per cell than their
planktonic counterparts
Cell size measurements were performed by differential
filtration Planktonic cells grown on glutamate (20.0 mg/L)
decreased in size and 62% could pass through a 1.0-pm
filter after 9 days. Only 39% of the GAC-attached cells
passed through a 1.0-nm filter. The studies indicated that
GAC provided an enhanced environment for the growth of
Klebsiella oxytoca when a charged substrate (glutamate)
was present that was adsorbed by the GAC.
Conclusions
The following conclusions can be drawn from this
research:
1. GAC-attached bacteria were effectively removed by
homogenization at 16,000 rpm at 4°C in a solution of
Zwittergent 3-12 (1Q-6M), EGTA (1Q-3M), peptone
(0.01%) and Tris buffer (10-1M) at pH 7.0 for 3 min.
HPC bacteria were best enumerated on R2A medium
incubated for 7 days at 28°C. Conforms were
effectively quantified with mT7 agar and a modified
MPN procedure.
2. HPC, coliform, and enteropathogenic bacteria grown on
GAC or attached for less than a generation time were
not killed by exposure to chlorine (2 mg/L) for 1 hr.
3. Enteropathogenic bacteria were capable of colonizing
laboratory-scale GAC filters. Persistence of the
pathogens depended on the presence of
autochthonous surface water organisms.
4. Populated GAC filter fines were found in drinking water
from properly operated treatment facilities. HPC and
coliform bacteria were detected on particles that were
released throughout the filter cycle.
5. Increasing the applied water turbidity, flow rate, and
filter depth all caused an appearance of (1) a higher
number of released particles, (2) increased bacterial
colonization of the particles, or (3) elevated adsorbed
coliforms. GAC supported more conforms than sand or
anthracite in laboratory experiments.
6. GAC-attached Klebsiella oxytoca had a greater growth
rate than planktonic cells in the presence of a charged
substrate (glutamate). Other physiological indices
showed greater activity in adsorbed cells
References
The following publications collectively contain the
complete findings of this research project.
1. Camper, A K., S.C. Broadaway, M.W LeChevallier, and
G A. McFeters. 1986. Operational variables and the
release of colonized granular activated carbon particles
in drinking water. (Submitted, JAWWA).
2 Camper, A.K., S.C Broadaway, M.W. LeChevallier, and
G.A. McFeters. 1985. Evaluation of procedures to
desorb bacteria from granular activated carbon. J.
Microbiol. Methods. 3187-198.
3. Camper, A K., M W. LeChevallier, S.C Broadaway, and
GA McFeters. 1985 Growth and persistence of
pathogens on granular activated carbon filters. Appl.
Environ. Microbiol 501378-1382.
4. Camper, A.K., M.W. LeChevallier, S.C. Broadaway, and
G.A. McFeters. 1986. Bacteria associated with granular
activated carbon particles in drinking water. Appl.
Environ. Microbiol. 52434-438
5. Davies, D.G. and G.A. McFeters. 1986. Growth and
comparative physiology of Klebsiella oxytoca attached
to granular activated carbon particles and in liquid
media. (Submitted, Microbial Ecology).
6. McFeters, G.A., A.K Camper, D.G. Davies, S.C.
Broadaway, and M.W. LeChevallier. 1986. Enumeration,
transport and survival of bacteria attached to granular
activated carbon in drinking water. In: Proceedings of
American Water Works Association Water Quality
Technology Conference, Houston, Texas, Dec 8-12,
1985.
7. McFeters, G.A., J.S. Kippin, and M.W. LeChevallier.
1986. Injured coliforms in drinking water. Appl. Environ.
Microbiol. 51:1-5.
8. LeChevallier, M.W., T.S. Hassenauer, A.K. Camper, and
G.A. McFeters. 1984. Disinfection of bacteria attached
to granular activated carbon. Appl. Environ. Microbiol.
48:918-923.
Gordon A. McFeters, Anne K. Camper, Susan C.
Broadaway, and David G. Davies are with Montana State
University, Bozeman, Montana. Mark W. LeChevallier is
currently with the American Water Works Service Co.,
Belleville, Illinois.
------- |