APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Aug. 1984, p. 395-403 Vol. 48, No.

3
0099-2240/84/080395-09$02.00/0
Copyright ©D 1984, American Society for Microbiology

Evaluation of Cleaning Strategies for Removal of Biofilms from

Reverse-Osmosis Membranes
C. WHITTAKER,1 H. RIDGWAY,2 AND B. H. OLSON'*
Environmental Analysis, Program in Social Ecology, Univ'ersity of California, Irvine, California 927171 and Water Factory
21, Orange County Water District, Fountain Valley, California 927082
Received 29 February 1984/Accepted 30 May 1984

An evaluation was made of the efficiency of five classes of chemical cleaning agents for removing biofilm from
spirally wound cellulose acetate reverse-osmosis membranes receiving influent with high or low levels of
combined chlorine. Each cleaning regimen utilized one or more of the following types of chemical: (i)
surfactants and detergents, (ii) chaotropic agents, (iii) bactericidal agents, (iv) enzymes, and (v) antiprecipi-
tants. Cleaning efficiency was tested in the laboratory on membrane material removed from operations at
various intervals (2 to 74 days). Cleaning effectiveness was evaluated against nontreated control membranes
and was scored by scanning electron microscopy and enumeration of surviving bacteria after treatment of the
membranes. The combinations of classes which were most effective in biofilm removal were the anionic and
chaotropic agent combination and combinations involving enzyme-containing preparations. Membranes
receiving influent with high levels of combined chlorine were easier to clean but more susceptible to structural
damage from prolonged exposure to combined chlorine. No treatment or combination of treatments was
completely effective or effective at all stages of biofilm development.

The development and use of the cellulose acetate reverse-
osmosis (RO) membrane have expanded greatly over the
past 30 years. The first practical and efficient membrane was
developed by S. Loeb and S. Sourirajan in the early 1960s (8;
U.S. patents 3, 133, 132 and 3, 133, 137, 1964). It was
improved dramatically by Loeb and Manjikian in 1967 (7)
and has since been incorporated into several industrial
processes, the most common being the demineralization and
purification of water from brackish and wastewater sources
(7). Several physical arrangements of the membrane unit
structure have been designed to maximize the surface area
per unit of volume (packing density) and to maintain high
pressure across the membrane (1). The spirally wound
membrane configuration (Fig. 1) has the advantage of a high
surface area per unit of volume, it also has the disadvantage
of relatively high pressure losses across the membrane and
difficulty in handling large concentrations of solids. An
additional disadvantage is the difficulty in cleaning modules
of this type of design owing to inefficiency in recirculating
the concentrate and subsequent resuspension of the fouling
layer (4).
Biofouling of surfaces over or through which water must
pass has been of major economic importance because the
formation of microbial biofilms can decrease water velocity,
clog pipes, increase energy utilization, and decrease the
efficiency of operations. Biofouling has been a major prob-
lem in cooling systems for many years. Research on the
fouling of cooling units has led to extensive characterization
of fouling layers on pipe and tank surfaces and other
segments of each system (9, 14,15). The growth of microor-
ganisms and the precipitation of inorganic chemical constitu-
ents have been recognized as the major constituents of these
fouling layers. To combat biofouling on surfaces of this type,
bactericidal agents as well as various dispersants and surfac-
tants have been successfully applied in numerous laboratory
and field trials (9, 14).
Cleaning cooling systems requires physical or chemical
*
Corresponding author.
395
removal of accumulated materials from pipes and reservoirs.
Cleaning cellulose acetate RO membranes is similar in
approach to removing biofilms from outer surfaces, but it is
more complex. The integrity of the cellulose acetate mem-
brane is easily destroyed by strong chemical oxidants,
enzymes, pH shifts, and temperature fluctuations. The de-
sign configuration of spirally wound membranes makes
physical methods of cleaning ineffectual or impossible. Pre-
viously, the fouling layer of RO membranes has been consid-
ered to be primarily inorganic in composition (16-18). Re-
cent work (3, 12), however, has shown that the fouling layer
may be either microbial or inorganic and that the extent to
which either type of material predominates is the result of
such interacting factors as the chemical quality of the water,
the microbial quality of the water, and the presence and
amounts of calcium and magnesium ions. Microorganisms
composing the biofilm may accumulate owing to the adhe-
sion of organisms present in the feedwater or to the prolifera-
tion of a few opportunistic organisms at the membrane
surface. Some of the fouling bacteria involved in the earliest
stages of biofilm formation have been shown to exhibit a
marked affinity for the cellulose acetate membrane surfaces
(13). The differences in the accumulation mechanisms in the
formation of a biofilm may be important in determining the
best way to clean membranes or maintain their efficiency
with bacteriostatic or other agents.
This study examined a variety of compounds by utilizing a
screening procedure to determine appropriate and potential-
ly effective classes of chemicals which could be employed in
cleaning spirally wound cellulose acetate RO membranes
(Table 1). Chemical cleaning agents and combinations of
these agents were classified as (i) surfactants and detergents,
which neutralize charged colloidal particles and resolubilize
or resuspend them; (ii) chaotropic agents, which denature
proteins and therefore readily solubilize organic constitu-
ents; (iii) bactericides, which kill microorganisms through
cell lysis or dissolution; (iv) enzymes, which hydrolyze the
proteinaceous and glycoprotein exopolymers surrounding
microorganisms; and (v) antiprecipitants, which remove
396 WHITTAKER, RIDGWAY, AND OLSON APPL. ENVIRON. MICROBIOL.

MODULE SEAL (SEALS AGAINST THE INSIDE WALL OF A PRESSURE

/VESSEL TO FORCE THE FEED SOLUTION THROUGH THE MODULE)
/ v PERMEATE COLLECTION HOLES

FEED SOLUTION

FEED CHANNEL
FEED SOLUTION it- SPACER
;D - MEMBRANE
PERMEATE COLLECTION
9- MATEfRlAL

>E- MEMBRANE
> FEED CHANNEL
- SPACER
PERMEATE FLOW (AFTER PASSAGE
THROUGH MEMBRANE INTO PERMEATE
COLLECTION MATERIAL)

FIG. 1. Interior structure of a spirally wound RO membrane (13).

metals and other precipitated ions from the layer of fouling. moved, the layers of the membrane were carefully unwound
This experimental procedure allowed the screening of a large and spread upon a sterile surface. A measured section of the
number of possible cleaning agents and considered the membrane (600 cm2) was scraped with a sterile razor blade
chemical principle or mechanism of the selected cleaning to remove the adhering biofilm. This fouling layer was
compound. In this way, classes of chemicals or combina-
tions of classes were studied, and their effectiveness was
generalized to other compounds of the same class. Visual
inspection by scanning electron microscopy (SEM) and TABLE 1. Classes of compounds used in cleaning treatments for
counting of the bacterial population remaining on the mem- RO membranes
brane were used to assess the efficiency of the treatment in Class Compounds"
the removal of biofilm. Antiprecipitants ..................... EDTA (disodium salt)
STP
MATERIALS AND METHODS TSP
Sampling. Miniature spirally wound RO membrane units Bactericides ..................... MBTC
(diameter, 6.35 cm) were provided by Water Factory 21, GuHCI
Fountain Valley, Calif. This facility has been described in a Urea
previous paper (12). These units were put into service in an ZDDC
on-site testing module which received the same water as the CTAB
full-scale (diameter, 20.3 cm) counterparts to simulate actual Detergents-sequestrants ............... Triton X-100
operating conditions. Six membranes units in the module Biz"
were available: four received water containing 3 to 5 mg of CTAB
combined chlorine per liter (designated hereafter as the low- TSP
chlorine [LC] membranes), and two received water with a Sodium tripolyphosphate
combined chlorine concentration of 15 to 20 mg/liter (high- SDS
chlorine [HC] membranes). The feedwater entering the test
modules was secondary (activated sludge) treated municipal Enzymes ............. ........ Trypsin
wastewater from the Orange County Sanitation District Protease
Thermolysin
facility. This water was treated by lime clarification, ammo- Papain
nia air stripping, recarbonation, and multimedia filtration Esterase
before RO. The membrane units were removed from the test Pancreatin
module and replaced by new units at each sampling time Biz
(Table 2). Membrane units were removed at intervals span-
ning 3 to 74 days. During each sampling, the membrane unit Chaotropic-denaturing agents .......... Urea
was removed from the module, placed directly on ice, and GuHC
transported to the laboratory (elapsed time between removal SDS
of the unit and processing was less than 2 h). At the " Abbreviations: CTAB, cetyltrimethylammonium bromide, GuHCI. guani-
laboratory, the spun-glass protective coating was removed. dine hydrochloride; MBTC. methylene bisthiocyanate: SDS. sodium dodecyl
At this point, aseptic conditions were introduced and main- sulfate; STP, sodium triphosphate: TSP. trisodium phosphate; ZDDC. zinc
dimethyidithiocarbamate.
tained throughout the rest of the processing. ' Biz is the trade name for a laundry presoaking detergent containing broad-
Processing. After the spun-glass covering had been re- spectrum enzymes and bleaching compounds.
VOL. 48. 1984 REMOVAL OF BIOFILM FROM RO MEMBRANES 397

TABLE 2. Sampling schedule of RO membrane units TABLE 3. Chemical compounds and dosages used to remove
Days elapsed at time of sampling" biofilm from RO membranes
Membrane unit Exposure
Initial unit Replacement unit Treat-
Chemical(s) Dose time
ment no.
1 3 43 (min)
2 5 57 1 Esterase 10 ,ug/ml 30
3 11 EDTA 311 FLl/ml (10 mM)
4 20
Sb 3 22 Trypsin 100 ,ug/ml 30
6" 11 63 Triton X-100 1 Ill/ml
Values show days elapsed after the beginning of the experiment. After
each membrane unit was removed, it was replaced by a new unit.
"
Received 15 to 20 mg of combined chlorine per liter in feedwater. 2 EDTA 311 ,ul/ml (10 mM) 60
Urea 120 mg/ml (2 M)
TSP 10 mg/ml
suspended in 10 ml of 100 mM phosphate buffer. pH 7.1.
Appropriate serial dilutions were spread plated onto m- 4 GuHCI 120 mg/ml (2 M) 60
standard plate count (m-SPC) agar (16) plates to determine CTAB 10 mg/ml
bacterial counts. Of the remaining membranes, two strips (1
by 12 cm) were placed in sterile screw-cap tubes, one for 5 Urea 120 mg/ml (2 M) 60
experimental cleaning treatment and one for a control. The SDS 10 mg/ml
control section received 100 mM phosphate buffer. pH 7.0.
Experimental treatments and dosages are summarized in
Table 3. The tubes were sealed, placed horizontally in a 6 Urea 120 mg/ml (2 M) 60
rack, and shaken at 110 rpm in a Lab-line Orbit Environ- CTAB 10 mg/ml
Shaker at 35°C for 1 h. The treated and control strips were
then removed from the tubes and were sectioned for ATP 7 Papain 100 ,ug/ml 30
analysis, protein analysis, bacterial counting, and SEM. Triton X-100 1 ,ul/ml
Bacterial counts and isolates. Serial dilutions of the mem- EDTA 311 p.1/ml (10 mM)
brane scrapings in phosphate buffer were spread plated onto
m-SPC agar. After 5 to 7 days of incubation at 28°C, colonies Protease 100 ,ug/ml 30
were counted. All plates containing 40 colonies or fewer Triton X-100 1 IlU/mI
were selected for identification. On plates with more than 40 EDTA 311 [LI/ml (10 mM)
bacterial colonies, 40 colonies were randomly selected from
the plate. Colonies were purified and stored on m-SPC agar 8 Biz 10 mg/ml 6()
slants at 4°C until identification. Identification of the isolates
was accomplished by the three-tube Lassen method for
gram-negative rods (6) and the acid-alcohol fast determina- 9 Pancreatin 100 p.g/ml 60
tion for mycobacteria with additional tests as recommended Triton X-100 1 p.1/ml
by Berge 's Maniual of Deterininativse Bacteriologv (2) for EDTA 311 ,ul/ml (10 mM)
gram-positive bacteria.
The 1-cm2 section removed from each strip was vortexed 10 Thermolysin 100 ,ug/ml 60
vigorously for 1 min in 5 ml of m-SPC broth to remove the Triton X-100 1 ,ul/ml
bacterial film. Serial dilutions were then plated onto m-SPC EDTA 311 ,ul/ml (10 mM)
agar plates. CFU were counted after 5 to 7 days at 28°C, and
15 colonies were picked, isolated, and identified as described
above. 11 GuHCI 2M 30
CTAB 10 mg/ml
SEM. The pieces of membrane to be studied by SEM were Protease 100 p.g/ml
placed in 5 ml of 2.5% glutaraldehyde in phosphate buffer,
pH 7.0, and were refrigerated overnight. The membranes Papain 100 ,ug/ml 30
were then dehydrated in a series of ethanol solutions: 30, 50, Esterase 10 p.g/ml
75, 85, 95, and 100% ethanol (the last twice), ethanol-Freon EDTA 311 p.l/ml (10 mM)
113 (1:1), and 100% Freon 113 (twice). The samples were
then air dried and mounted on aluminum stubs with a 12 CTAB 10 mg/ml 60
collodial silver adhesive. The stubs were then coated with
gold-palladium (60:40) and viewed on a Hitachi model SU
500 scanning electron microscope at 15 kev with a working 13 SDS 10 mg/ml 60
distance of 10 mm and a stage tilt angle of 00. Black and
white photographs were made with Polaroid P/N55 film. The
effectiveness of each cleaning treatment was scored semi- 14 Triton X-100 10 mg/ml 60
quantitatively according to the presence and appearance of
the biofilm. A score of 0 was given if no removal was evident 15 MBTC 10 mg/ml 60
(compared with an untreated control), and a score of 4 was CTAB 10 mg/ml
assigned if removal was complete. For the criteria utilized
for scoring by SEM, see Table 4, footnote b.
Chemical cleaning treatments. The chemicals used in each Continiued on following page
398 WHITTAKER, RIDGWAY, AND OLSON APPL. ENVIRON. MICROBIOL.

TABLE 3-Continued of the 24 treatments and doses applied are presented in Table
Treat- Exposure 3. Doses of previously investigated chemicals were chosen
Chemical(s) Dose time to approximate or exceed values reported in the literature (7,
ment no. (min) 11, 19). All of the chemicals were dissolved in 100 mM
phosphate buffer, pH 7.0. The control consisted of a strip of
16 ZDDC 10 mg/ml 60 the membrane treated in the same manner as test membranes
Triton X-100 0-1 ,V/mi
but exposed to 100 mM phosphate buffer, pH 7.0, alone.
This was used to control for the physical action of the liquid
17 STP 10 mg/ml 60 on the membrane.
EDTA 10 mg/ml
TSP 10 mg/ml RESULTS
Observation of the membranes by SEM provided the most
18 EDTA 311 RI/ml (10 mM) 60 direct evidence for biofilm removal. In Table 4, results of
SEM scores of the treated membranes are presented in
19 Trypsin 100 Rg/ml 60 ranked order, as described above, indicating the amount of
EDTA 311 ,u/ml (10 mM) biofilm removed compared with that removed from a treated
Triton X-100 1 ±1/ml control. The control corrected for the removal of any biofim
by mechanical effects during the treatment period. Table 4
shows the scores of the treated LC and HC membranes and
20 Esterase 10 ,g/ml 60 the mean score for all days sampled for each treatment.
EDTA 311 ±1/ml (10 mM) Finally, the overall score for each treatment was calculated
Triton X-100 1 Rl/ml as follows: n n

z XLC + z XHC
21 Urea 120mg/ml (2 M) 60 Overall score = Xtotal = i=1 i=1
nLC + nHC
22 TSP 10 mg/ml 60 Overall rankings of the treated membranes indicated that
treatments with a mean score of -2.5 were very good
23 Pancreatin 100 ,ug/ml 30 cleaners. These treatments included esterase-EDTA/trypsin-
EDTA 311 RI/ml (10 mM) Triton (i.e., treatment with esterase-EDTA followed by
Triton X-100 1 RI/ml trypsin-Triton), urea-sodium dodecyl sulfate (SDS), papaih-
Triton-EDTA/protease-Triton-EDTA, Biz, and pancreatin-
CTAB 10 mg/ml 30 Triton-EDTA. Cleaners with an overall score of 2.0 to 2.5
were considered to be good and included SDS, sodium
triphosphate (STP)-EDTA-trisodium phosphate (TSP), tryp-
Control Potassium phosphate 17.4 mg/ml (100 mM) 60 sin-EDTA-Triton, and Triton. The remainder of the treat-
(pH 7.0) ments (poor cleaners) were consistently unable to reduce the
biofilm by 50% and are not included in Table 4.

TABLE 4. SEM evaluation of biofilm reduction by various cleaning agents on RO membranes receiving HC or LC doses of feedwater
Cleaning effectivenessb
Treatment' LC membranes on day: HC membranes on day:
Overall'
3 5 11 20 43 57 Mean 3 11 22 63 Mean
Urea-SDS 2+ 3+ 3+ 3+ 2+ 3+ 2.7 3+ 3+ 4+ 4+ 3.5 3.0
Biz 3+ - 4+ 3+ 2+ 1+ 2.6 3+ 4+ 3+ 4+ 3.5 3.0
Esterase-EDTA/ 2+ 2+ 3+ 2+ 2+ 2+ 2.2 3+ 3+ 3+ 4+ 3.2 2.6
trypsin-Triton
Pancreatin-Triton-EDTA 2+ - 2+ 3+ 2+ 1+ 2.0 3+ 3+ 3+ 4+ 3.2 2.6
Papain-Triton-EDTA/ 2+ 3+ 3+ 2+ 2+ 1+ 2.2 2+ 3+ 3+ 4+ 3.0 2.5
protease-Triton-EDTA
SDS - 3+ 1+ 1+ 1+ 2.0 3+ 3+ 4+ 3.3 2.3
Triton - 2+ 0 - 1+ 1.0 - 2+ 3+ 4+ 3.0 2.0
STP-EDTA-TSP - - - 1+ 1+ 1.0 - - 4+ 4.0 2.0
Tiypsin-EDTA-Triton - - 1+ 1+ 1.0 4+ 4.0 2.0
Pancreatin-EDTA/Triton-CTAB - 2+ 1+ 1.5 3+ 3.0 2.0
aTreatments with an overall rank of <2.0 are not included.
b Scores, based on comparison with treated control, were assigned as follows. 0, No removal of biofilm apparent; multilayer biofilm (MLB, two or more cell
layers thick) and single-layer biofilm (SLB) both present. 1+, Poor biofilm renmoval, usually less than 20 to 30%; MLB and SLB both present. 2+, Fair biofilm re-
moval, usually 50 to 70%; MLB absent, SLB present. 3+, Good biofilm removal, usually 70 to 80%; MLB absent, SLB present. 4+, Excellent biofilm removal,
usually 90 to 100%o; a few scattered cells might remain, but there were no patches; no MLB or SLB present.-, No sample was treated with the indicated cleaning
agent.
c Overall score was determined by the formula given in Results.
VOL. 48, 1984 REMOVAL OF BIOFILM FROM RO MEMBRANES 399

Representative SEM micrographs in Fig. 2 and 3 show the
relative rankings of selected treatments compared with the
control. The biofilm formed a confluent layer over the
membrane on both the HC and LC controls. The micro-
graphs depict samples from the LC membrane (Fig. 2) and
the HC membrane (Fig. 3) after 11 days of operation. In
these figures, the range of cleaning effectiveness is demon-
strated from no differences from the control (Fig. 2h and 3h)
to >90% removal of biofilm (Fig. 2f and 3f0.
The various chemical strategies and combinations of strat-
egies illustrated in Fig. 2 and 3 included enzyme-surfactant-
chelator combinations (b, e, and g), denaturant-detergent
combinations (c and d), an enzyme-detergent combination (f)
and a biocide-detergent combination (h). The methylene
bisthiocyanate (MBTC)-cetyltrimethylammonium bromide
(CTAB) treatment (Fig. 2h and 3h), representing a biocide-
detergent combination, did not reduce the biofilm when
compared with the treated control. On the other hand, the
Biz treatment (Fig. 2f and 30) showed excellent biofilm
reduction on both LC and HC membranes. Some variability
in cleaning efficiencies between the LC and HC membranes
is seen in the urea-CTAB treatment (Fig. 2d and 3d) and the
pancreatin-Triton-EDTA treatment (Fig. 2g and 3g). The HC
membranes appeared to be cleaned with higher efficiency in
nearly all cases. This variability was demonstrated at other
sampling times with several treatments (Table 4).
A number of the experimental cleaning formulations ap-
peared to exhibit bactericidal activity. For several of the
enzymatic cleaning treatments, a problem with the sterility
of the solutions made counts of CFU inaccurate, and some
treatment schemas, such as SDS and Triton, were initiated
after the study had begun.
Table 5 shows the percentage of reduction of numbers of
bacteria compared with the control for the compounds found
to have a rating of .2.0 in the SEM evaluation. The urea-
SDS combination was the most effective as a bactericide and
was also a successful cleaning treatment. Biz consistently
showed good reduction in numbers of bacteria, but many of
the other cleaning strategies produced more uneven results.
Some of this variability may have been due to the lack of a
highly reproducible method for removing all bacteria from
the treated and control membranes.
The cleaning effectiveness of the highly bactericidal com-
pounds (>90% reduction in numbers of bacteria) is shown in

FIG. 2. Representative SEM micrographs showing the relative effectiveness of various cleaning treatments in removal of biofilm from LC
RO membranes after 11 days in operation. (a) Treated control membrane, given a score of 0 (no biofilm removal). (b) Membrane treated with
esterase-EDTA/trypsin-Triton and given a score of 3+ (good biofilm removal, 70 to 80% reduction). (c) Membrane treated with urea-SDS and
given a score of 3+ (good biofilm removal). (d) Membrane treated with urea-CTAB and given a score of 1+ (poor biofilm removal, <20 to 30%
reduction). (e) Membrane treated with papain-Triton-EDTA/protease-Triton-EDTA and given a score of 3+ (good biofilm removal). (f)
Membrane treated with Biz and given a score of 4+ (excellent biofilm removal, 90 to 100% reduction). (g) Membrane treated with pancreatin-
Triton-EDTA and given a score of 2+ (fair biofilm removal, -50 to 70% reduction). (h) Membrane treated with MBTC-CTAB and given a
score of 0 (no biofilm removal discernible). Bars, 5 p.m.
400 WHITTAKER, RIDGWAY, AND OLSON

_I
_I
_I
_I
_I
_I
_I
_I

_I
_I
_II
_III
I s,
APPL. ENVIRON. MICROBIOL.

_x _S 11

FIG. 3. Representative SEM micrographs showing the relative effectiveness of several cleaning agents in removal of biofilm from HC RO
membranes after 11 days in operation. (a) Treated control membrane, given a score of 0 (no biofilm removal). (b) Membrane treated with
esterase-EDTA/trypsin-Triton and given a score of 3+ (good biofilm removal, -70 to 80%4 reduction). (c) Membrane treated with urea-SDS
and given a score of 3+ (good biofilm removal). (d) Membrane treated with urea-CTAB and given a score of 2+ (fair biofilm removal, -50 to
70% reduction. (e) Membrane treated with papain-Triton-EDTA/protease-Triton-EDTA and given a score of 3+ (good biofilm removal). (f)
Membrane treated with Biz and given a score of 4+ (excellent biofilm removal, 90 to 100% reduction). (g) Membrane treated with pancreatin-
Triton-EDTA and given a score of 3+ (good biofilm removal). (h) Membrane treated with MBTC-CTAB and given a score of 0 (no biofilm re-
moval). Bars, 5 p.m.

TABLE 5. Effects of cleaning agents on reduction of counts of viable bacteria from LC RO membranes"
SEM % Reduction from control on day:
Treatment score
(range) 3 5 11 20 43 57
Biz >2.5 83.5 70.5 69.5 58.4 71.2 32.7
Urea-SDS 94.1 100.0 100.0 100.0 99.9 100.0
Pancreatin-Triton-EDTA NE" NE NE NE 88.9 18.3
Esterase-Triton/ NE NE NE NE 32.6 19.2
trypsin-EDTA
Papain-Triton-EDTA/ 97.8 NE NE NE 69.0 92.1
protease-Triton-EDTA
SDS 2.0-2.5 ND" ND 97.5 48.9 68.7 0.0
Triton ND ND 87.6 0.0 0.0 49.0
TSP-EDTA-STP ND ND ND ND 0.0 0.0
Trypsin-EDTA-Triton ND ND ND ND 31.0 0.0
' Treatments shown are those which received a score of at least 2.0 in the SEM evaluation (Table 4). There were no colonies isolated from the HC membranes
until 63 days had elapsed, when 2.0 x 102 colonies per cm2 were isolated from the control membranes. Colonies were not enumerated from the treated HC
membranes.
*NE, Not enumerated; contamination from enzyme preparations.
ND, Not determined.
VOL. 48, 1984 REMOVAL OF BIOFILM FROM RO MEMBRANES 401

Table 6. Pancreatin-EDTA-Triton/CTAB showed some
cleaning potential, and urea-SDS ranked as a very good
cleaner. For the most part, however, the highly rated
bactericidal combinations (those containing CTAB or urea)
were not very effective at removing biofilm.
Identification of the bacteria recovered from the fouled
RO membranes indicated that >95% of isolates were mem-
bers of the genus Mycobacterium, an acid-fast, gram-posi-
tive group of rod-shaped microorganisms (2). Mycobacteria
form tenacious mucoid colonies on m-SPC agar, and at least
one strain has been recently demonstrated to exhibit a
marked affinity for the cellulose acetate RO membrane
surface (13).
DISCUSSION
Innovations in cleaning techniques for RO membranes
have been sporadic and somewhat haphazard in their devel-
opment. Most techniques have been developed under actual
operating conditions by simply substituting new procedures
for old ones and passing along the insight gained. Many new
cleaning agents have been utilized on the assumption that the
predominant fouling material is inorganic scale or precipitate
and that any organic material present is primarily humic
acids (3).
The most common treatments for removing the fouling
layer of RO membranes are periodic flushing with mixtures
of detergents and chelating agents, changes in pH, and the
addition of antiprecipitants such as sodium hexametaphos-
phate or citric acid (3, 10, 19).
In this study, we have characterized the biofilm micro-
scopically and bacteriologically to examine the effectiveness
of different cleaning strategies in the removal of a developing
biofilm. Our cleaning strategies, contrary to those in previ-
ous work, have assumed that the major fouling entity is
bacterial in nature with its associated organic and inorganic
by-products.
The present studies indicate that the most reliable evi-
dence for the degree of biofilm removable was direct SEM
examination. Bacterial counts did not correlate well with the
visual observations via SEM. Strongly bactericidal com-
pounds were not necessarily effective in removing biofouling
layers, and cleaning solutions that were effective in biofilm
removal were not necessarily bactericidal. This dichotomy
of effects eliminates one of the proposed strategies as an
effective mechanism of attack on the biofilm; many bacteri-
cidal compounds probably do not lyse and dissolve the
fouling layers.
Other evidence that lends support to this hypothesis is that
although no bacterial colonies were isolated from the HC
membrane until 63 days had elapsed, the HC and LC
membranes appeared under SEM to be equally fouled by
bacteria at each sampling time in the experiment. This
evidence suggests that chlorine-inactivated bacteria may
also produce a biofouling layer on the RO membrane sur-
face.
Some chemicals consistently reduced the numbers of
bacteria by 30 to 80%. These compounds may not act
through bactericidal activity; the reduction in numbers may
be due rather to the actual removal of the bacteria from the
surface of the membrane. This effect would be preferred
over bactericidal activity. Chemical treatments which exhib-
ited this type of action included Biz, papain-Triton-EDTA/
protease-Triton-EDTA, pancreatin-Triton-EDTA, and SDS
(Table 5).
The SEM photographs show some dramatic decreases in
biofilm that were not well reflected by corresponding counts
of bacteria. Some of the membranes appeared virtually bare,
yet CFU measurement revealed that a substantial number of
bacteria remained associated with the membranes. This
effect may be due to one of several causes. The confluency
of biofilm may have been so disrupted after cleaning that the
samples that were measured were too small (1 cm2) to be
representative of the whole sample. It is also possible that
the processes of fixation and dehydration for SEM may
actually alter the biofilm, in this case enhancing the removal
of bacteria from the membrane and exaggerating the report-
ed effectiveness of the cleaning treatments. Since SEM is the
most important parameter assessed in this study, it is critical
that the limitations of this method be understood. However,
assuming that the effects of fixation and dehydration are
approximately equal across the treatments, SEM evaluation
should still be a good measure of the relative effectiveness of
each treatment, although it may not be quantitative.
Analysis of effective cleaning treatments. Each treatment
incorporated one or several cleaning strategies. Using the
criteria from Table 4 to determine the effectiveness of
biofilm removal, it is apparent that cleaning treatments that
were most successful (mean score, >2.5) had constituents in
each of the following categories: enzymes, antiprecipitants,
bactericides, and denaturing agents. We have already deter-
mined that bactericidal compounds are not necessarily effec-
tive in biofilm removal, so the remaining constituents require
closer examination.
Preparations containing enzymes. The enzyme-antiprecipi-
tant-dispersant combination formed most of the very good
cleaners. EDTA, a chelator, was added to the enzyme

TABLE 6. Effectiveness of cleaning combinations that reduced counts of viable bacteria by at least 90% on HC and LC RO membranes
Cleaning effectiveness"
Treatment LC membranes on day: HC membranes on day:
Overall''
3 5 11 20 43 57 Mean 3 11 22 63 Mean
EDTA-urea-TSP 0 1+ 2+ 2+ 1+ 3+ 1.5 2+ 2+ -h 1+ 1.7 1.6
Urea-SDS 2+ 3+ 3+ 3+ 2+ 3+ 2.7 3+ 3+ 4+ 4+ 3.5 3.0
Urea-CTAB 2+ 0 1+ 1+ 0 2+ 1.0 2+ 2+ 3+ 4+ 2.7 1.7
CTAB 0 0 1+ 0 0 0.2 1+ 1+ 3+ 1+ 1.5 0.8
MBTC-CTAB - 0 1+ 0.5 - 0 1+ - 0.5 0.5
Urea - 0 2+ 1.0 2+ 2.0 1.3
Pancreatin-EDTA/ 2+ 1+ 1.5 3+ 3.0 2.0
Triton-CTAB
a Measured by SEM examination with scopes assigned as explained in Table 4, footnote b.
b_, No sample was treated with the indicated cleaning agent.
C Overall score was determined by the formula given in Results.
402 WHITTAKER, RIDGWAY, AND OLSON
solutions to chelate any toxic metals that could retard
enzymatic action. Triton, a nonionic detergent, was used
because of its surfactant activity to facilitate penetration by
the enzyme of the biofouling layer. In these combinations,
the enzyme itself is believed to be the active agent, but
enzymes used alone did not fare as well. Perhaps this was
because substances of such large molecular weight are
unable to penetrate the biofilm effectively without the aid of
surfactants of chelating agents. The enzymatic mechanism of
action is the proteolytic and glycolytic (esterase) action,
which disrupts the glycocalyx that probably surrounds the
bacteria, encouraging sloughing of the material. By attacking
the exopolymers that are implicated in mediating bacterial
attachment, it is possible to solubilize the biofilm and clean
the membrane.
Use of enzymes under operating conditions has two major
drawbacks, expense and lack of stability. The cost of
enzymes varies widely, depending on the source of the
enzyme. Some can be quite expensive, and even crude
preparations of enzymes are more expensive than cleaning
products currently used on RO membranes. Before enzymes
can be considered as a viable alternative to present cleaning
methods, more evidence should be collected to support their
effectiveness against biofilms. Furthermore, each biofouling
problem should be analyzed to determine the nature of the
fouling layer (bacterial, organic, or inorganic), and a cost-
benefit assessment should be sought.
Preparations containing detergents. Another successful
treatment in our schema was the urea-SDS treatment. This
treatment combined a chaotropic and bactericidal agent,
urea, with an anionic detergent, SDS. Some evidence has
been collected that cationic detergents are quite effective in
solubilizing gram-negative bacteria, whereas anionic deter-
gents have more success in solubilizing gram-positive bacte-
ria (9). Furthermore, the bacteria isolated from the mem-
branes belonged almost exclusively to the genus
Mycobacterium, which has a gram-positive wall structure.
SDS by itself ranked as a good cleaner, although its cleaning
abilities were greatly enhanced when coupled with urea.
Urea by itself did not rank high as a cleaning agent, possibly
because it was applied at too low a concentration. Typically,
urea concentrations of at least 6 M are required to solubilize
most polypeptides (5). In other experiments, urea at concen-
trations of 6 to 8 M did in fact exhibit excellent biofilm
removal (unpublished data).
The three types of detergents which were utilized-cation-
ic (CTAB), anionic (SDS), and nonionic (Triton X-100)
were each tried alone and in combination with other
compounds. SDS in combination with urea was the most
successful.
CTAB, the cationic detergent, although an excellent bac-
tericide, had a very low cleaning score when used alone.
Triton, the nonionic detergent, merited a score of 2.0 and
was labeled a good cleaner. This was probably due to its
surfactant activities, which were also utilized in the enzy-
matic treatments.
Both urea and SDS are fairly inexpensive ingredients. A
major drawback in using SDS (and other anionic detergents),
however, is its sudsing properties. SDS has tremendous
sudsing action, which may be unsuitable in RO membrane
systems. One possible way of resolving this problem is to
identify a detergent with the anionic properties of SDS and
with low sudsing potential. This type of compound in
combination with urea may prove very effective in combat-
ing bacterial biofouling problems.
Commercial mixed-component preparation. Biz tied for the
APPL. ENVIRON. MICROBIOL.
highest score in the SEM evaluation. Biz is a commercial
laundry presoaking detergent which contains broad-spec-
trum enzymes, bleaching compounds, and detergents. The
exact mechanism of action of this compound, of course, is
not known because Biz combines so many different com-
pounds. However, this combination of enzymatic, oxidizing,
and detergent compounds appears very efficient in removing
biofilm, especially when the film is less extensive, as evi-
denced by greater effectiveness at the earlier sampling times.
The major drawback in using Biz as a cleaning agent is its
oxidizing properties, which endanger the integrity of the
membrane. For several days before the sampling time on day
12, the ammonia concentration was accidentally lowered,
exposing one membrane unit to 15 to 20 mg of free chlorine
per liter. The membrane unit analyzed shortly after this
event showed signs of membrane failure and loss in salt
rejection ability, and under SEM, elliptical depressions and
holes appeared in the membrane. The biofilm also seemed to
be removed much more easily than at previous times. It is
also evident that HC membranes were cleaned more effi-
ciently at all times than LC membranes were (Table 4).
Because chlorination destroys the cellulose acetate mem-
brane, such oxidizing compounds must be used with care.
Repeated use of such products as Biz may cause deteriora-
tion of membrane integrity.
In summary, the strategies which were most efficient in
removing biofilm composed primarily of mycobacteria and
their organic matrix were an enzyme-chelator-dispersant
combination, an anionic detergent-denaturant combination,
and Biz. None of these was effective at all sampling times,
and all have some drawbacks in use, but the elimination of
alternatives (such as bactericides) as effective means of
biofilm removal may ultimately save costs to industrial
ventures. This study indicates that research into the funda-
mental mechanisms of biofouling of RO membranes is war-
ranted to determine which classes of compounds and en-
zymes will be most efficient in cleaning. In this manner,
classes of compounds can be eliminated as inappropriate,
and efforts can be concentrated on members of the classes of
compounds that are effective for the particular fouling prob-
lem at hand.
ACKNOWLEDGMENTS
This project was funded by the Office of Water Research and
Technology contract 4107-160-81 through the Orange County Water
District.
The authors thank Carol Justice for her technical assistance
during the experimental phase of this project.
LITERATURE CITED
1. Bailey, D. A., K. Jones, and C. Mitchell. 1971. The reclamation
of water from sewage effluents by reverse osmosis. J. Water
Pollut. Control Fed. 73:353-366.
2. Buchanan, R. E., and N. E. Gibbons (ed.). 1974. Bergey's
manual of determinative bacteriology. 8th ed. The Williams &
Wilkins Co., Baltimore.
3. Burns and Roe Industrial Services Corp. 1979. Reverse osmosis
technical manual. OWRT TT/80 2 for the Office of Water
Research and Technology, U.S. Department of the Interior.
Burns and Roe Industrial Services Corp.. Paramus, N.J.
4. Golomb, A., and F. Besik. 1970. Reverse osmosis-a review of
its applications to waste treatment. Water Sewage Works 117:
R-81-R-89.
5. Gordon, J. A., and W. P. Jencks. 1963. The relationship of
structure to the effectiveness of denaturing agents for proteins.
Biochemistry 2:47-57.
6. Lassen, J. 1975. Rapid identification of gram negative rods using
a three tube method combined with a dichotomic key. J.
Microbiol. Scand. Sect. B 83:525-533.
VOL. 48, 1984
7. Loeb, S., and S. Manjikian. 1963. Brackish water desalinization
by an osmotic membrane. UCLA Department of Engineering
report no. 63-22. University of California, Los Angeles.
8. Loeb, S., and S. Sourirajan. 1963. Sea water demineralization
by means of an osmotic membrane. Saline conversion II. Amer.
Chem. Soc. Adv. in Chem. Series 38 p. 177-132.
9. McCoy, J. W. 1974. The chemical treatment of cooling water.
Chemical Publishing Co., New York.
10. McCutchan, J., and J. Johnson. 1970. Reverse osmosis at
Coalinga, California. J. Am. Water Works Assoc. 62:346-353.
11. Richter, V. J. 1971. Performance value of enzymes in presoak
formulations. Dev. Ind. Microbiol. 12:36-41.
12. Ridgway, H. F., A. Kelly, C. Justice, and B. H. Olson. 1983.
Microbial fouling of reverse-osmosis membranes used in ad-
vanced wastewater treatment technology: chemical, bacterio-
logical, and ultrastructural analyses. Appi. Environ. Microbiol.
45:1066-1084.
13. Ridgway, H. F., M. G. Rigby, and D. G. Argo. 1984. Adhesion of
REMOVAL OF BIOFILM FROM RO MEMBRANES 403
a Mycobacterium sp. to cellulose diacetate membranes used in
reverse osmosis. Appl. Environ. Microbiol. 47:61-67.
14. Shair, S. 1971. Microbiocide treatment of recirculating cooling
water. Ind. Water Eng. 8:26-30.
15. Shair, S. 1973. Cooling system defense against microbiological
attack. Power Eng. 77:68-71.
16. Taylor, R. E., and E. E. Geldreich. 1979. Standard plate count
methodology: a new membrane filter procedure for potable
water and swimming pools. J. Am. Water Works Assoc. 71:402-
405.
17. Wechsler, R. 1976. Reverse osmosis on secondary sewage
effluent: the effect of recovery. Water Res. 11:379-385.
18. Winfield, B. A. 1979. A study of the factors affecting the rate of
fouling of reverse osmosis membranes treating secondary sew-
age effluents. Water Res. 13:565-569.
19. Winfield, B. A. 1979. The treatment of sewage effluents by
reverse osmosis-pH based studies of the fouling layer and its
removal. Water Res. 13:561-564.