XEA1
IV/MI
WATER POLLUTION CONTROL RESEARCH SERIES • 17O2ODUDO9/7O
NEW TECHNOLOGY
FOR TREATMENT OF
WASTEWATER BY REVERSE OSMOSIS
ENVIRONMENTAL PROTECTION AGENCY • WATER QUALITY OFFICE
-------
WATER POLLUTION CONTROL RESEARCH SERIES
The Water Pollution Control Research Reports describe the results
and progress in the control and abatement of pollution in our
Nation's waters. They provide a central source of information on
the research, development, and demonstration activities in the
Water Quality Office, in the Environmental Protection Agency,
through inhouse research and grants and contracts with Federal,
State, and local agencies, research institutions, and industrial
organizations.
Inquiries pertaining to Water Pollution Control Research Reports
should be directed to the Head, Project Reports System, Planning
and Resources Office, Office of Research and Development, Environ-
mental Protection Agency, Water Quality Office, Room 1108, Wash-
ington, D. C. 202*12.
-------
NEW TECHNOLOGY FOR TREATMENT OF WASTEWATER BY REVERSE OSMOSIS
ENVIROGENICS COMPANY
A Division of
Aerojet-General Corporat ion
El Monte, California 9173^
for the
WATER QUALITY OFFICE
ENVIRONMENTAL PROTECTION AGENCY
Program # 17020 DUD
Contract #14-12-553
WQO Project Officer, J. M. Smith
Advanced Waste Treatment Research Laboratory
Cincinnati, Ohio
September, 1970
For sale by the Superintendent of Documents, U.S. Government Printing Office
Washington, B.C., 20402 - Price 70 cents
-------
WQO Review Notice
This report has been reviewed by the Water
Quality Office and approved for publication.
Approval does not signify that the contents
necessarily reflect the views and policies
of the Water Quality Office, nor does mention
of trade names or commercial products con-
stitute endorsement or recommendation for use.
ii
-------
ABSTRACT
Stable, high-flux membranes were sought for use in the renovation of
wastewater by reverse osmosis. Cellulose ester membranes were formu-
lated to produce fluxes greater than 60 gal/ft^-day which would not
decrease by more than 20$ after the first year of operation, and re-
ject at least 60$ of sodium chloride and 95$ of sodium sulfate when
tested at 600 psi with 1000 ppm feed solutions.
The target osmotic performance was achieved with each of three mem-
brane types: A cellulose diacetate of moderately-low acetyl content,
a cellulose triacetate-diacetate blend, and crosslinked cellulose
acetate methacrylate. The intrinsic flux stabilities of these mem-
branes extrapolated to flux losses of only 12 to 18% after the first
year of operation.
The fluxes of these high-performance membranes declined rapidly in
bench-scale tests with secondary sewage effluent but were restored
to within 80 to 90$ of the initial values after cleaning with an en-
zymatic laundry presoak (Biz). Daily cleaning by this technique main-
tained the fluxes at a nearly constant level over a 5-day test period.
The rejection of sewage components by the high-flux membranes was ex-
cellent, the best of them rejecting 90 to 97$ of TDS, 70 to 100$ of
COD, 86 to 96$ of ammonium ion, 72 to 99$ of nitrate ion and 97 to 99$
of total phosphate.
Techniques were explored for attachment of proteolytic enzymes to cel-
lulose acetate membranes tojrender them resistanti&toacolloid fouling.
The proteolytic enzyme trypsin was chemically attached to the active
layer surface of a membrane prepared from the N-hydroxysuccinimide
ester of cellulose acetate hydrogen succinate. The resulting enzymatic
membrane displayed hydrolytic activity toward an ester substrate of
low molecular weight. The enzyme-polymer compound (in bulk rather than
membrane form) also hydrolyzed casein, a model for proteinaceous col-
loids present in sewage.
This report was submitted in fulfillment of Contract No. lU-12-553,
Program No. 17020 DUD, between the Federal Water Quality Administration
and Aerojet-General Corporation, and covers the period June 1969 to
July 1970.
iii
-------
CONTENTS
Page
I. CONCLUSIONS
II. RECOMMENDATIONS_
III. INTRODUCTION
IV. STABLED HIGH-FLUX MEMBRANES 7
Membranes from Cellulose Diacetate 7
Cellulose Triacetate-Diacetate Blend Membranes 11
Cellulose Acetate Methacrylate Membranes l6
V. TESTING OF HIGH-FLUX MEMBRANES WITH SEWAGE 27
VI. MEMBRANES RESISTANT TO COLLOID FOULING 53
Methods of Attachment of Enzymes to Cellulosic
Membranes . 53
Attachment of Enzymes to Bulk Polymeric Supports 55
Preparation of Enzyme-Coupled Membranes 57
Membranes from Cellulose Acetate Hydrogen Succinate 58
REFERENCES 60
APPENDIX 62
iv
-------
FICUBES
Figure Page
1 Void Structure of CAM-360 Membranes _ 23
2 Flux Decline of E-383-^0 Membranes During
Test No. 2 _ 32
3 Flux Decline of B-Series E-383-40 Membranes
with Different Effluents _ 38
k Effect of Periodic Biz Cleaning on Performance
of B-Series E-383-40 Membranes with
Secondary Effluent 1^.3
Effect of Membrane Initial Flux on Flux
Decline During Test No. k with Pomona
Secondary Effluent
6 Flux Decline of DS 2.55 Blend Membranes
During Test No. 5
7 Flux Decline of the Three Membrane Types
During Test No. 6 _ 5!
-------
TABLES
Table
1 Osmotic Properties of E-383-40 Cellulose
Acetate Membranes
2 Long-Term Osmotic Properties of E-383-^0
Cellulose Acetate Membranes _ 10
3 Osmotic Properties of Unannealed DS 2.63
Blend Membrane _ _ 13
k Osmotic Properties of DS 2.55 Blend Membranes _ 15
5 Osmotic Properties of Crosslinked CAM-360
Membrane s T Cast at -10°C_ _ _ _ 18
6 Reformulation of CAM-360 Membranes for
Room-Temperature Casting _ 19
1 Effect of Annealing Temperature on Osmotic
Properties of CAM-360 Membranes Cast at
Room Temperature _ - _ 21
8 Long-Term Osmotic Properties of Crosslinked
CAM-360 Membranes Cast at Room Temperature _ 25
9 Testing of E-383-^0 Cellulose Acetate Membranes
with Pomona Secondary Effluent r Test Number 1 _ 28
10 Testing of E-383-1K) Cellulose Acetate Membranes
with Pomona Secondary Effluent - Test Number 2 _ 29
11 Composition of Sewage Feeds in Test Numbers 1
and 2 _ 30
12 Cleaning of Fouled Membranes from Test Number 2
with Biz 3^
13 Testing of E-383-^0 Cellulose Acetate Membranes
with Pomona Carbon-Treated Secondary Effluent -
Test Number 3 _ _ 35
1^- Feed Composition - Test Number 3 with Pomona Carbon-
Treated Secondary Effluent _ 36
15A Testing of E-SSS-^O Membranes with Sewage Utilizing
Periodic Biz Cleaning ^ Test Number k -
Membrane A _ 39
vi
-------
TABLES (Continued)
Table
15B Testing of E-383-40 Membranes with Sewage
Utilizing Periodic Biz Cleaning - Test
Number h - Membrane B
16 Composition of Primary and Secondary
Effluent Feeds in Test Number 4
17 Testing of DS 2.55 Blend Membranes with
Pomona Secondary Effluent - Test Number 5
18 Comparative Test of Three Membrane Types
with Pomona Secondary Effluent - Test
Number 6 _ _ __ . 50
vii
-------
I. CONCLUSIONS
The work carried out during this program has shown that flat-sheet mem-
branes with greatly improved fluxes and excellent flux stability can be
fabricated from cellulose esters. At the same time, rejection of or-
ganic and ionic constituents adequate for renovation of municipal waste-
water by reverse-osmosis was achieved. The best of the high-flux mem-
branes rejected 90 to 91% of TDS, TO to 100$ of COD, 86 to 96% of
ammonium ion (at pH 5.00 +_ 0.25), 72 to 99$ of nitrate ion, and 97 to
99% of total phosphate from secondary effluent at initial fluxes of 50
to 100 gal/ft2-day (gfd).
The new membranes exhibited flux decline slopes of -0.011 to -0.021 with
initial fluxes of 50 to 70 gfd in 200-hr reverse-osmosis tests conducted
at 600 psi with 1000 mg/£ sodium chloride solution. Flux losses of only
9 to 18$ are projected from these flux decline rates after the first
year of operation at 600 psi. On the basis of their demonstrated intrin-
sic flux and flux stability, and excluding the effects of fouling, these
membranes are calculated to produce 16,500 to 21,000 gal/ft2 of product
water during the first year of reverse-osmosis operation at 600 psi.
These values compare with first-year total water productivity (TWP) of
7000 to 9000 gal/ft2 calculated for state-of-the-art cellulose diacetate
brackish-water membranes produced in these and other laboratories.
Certain of the new membranes exhibited osmotic performance suitable for
low-pressure reverse-osmosis operation, which may offer economic advan-
tages over operation at higher pressure. A crosslinked cellulose ace-
tate methacrylate membrane produced a flux of 36 gfd at 200 psi, with no
appreciable flux decline, along with J0% salt rejection in a 200-hr test
with 1000 mg/4 sodium chloride solution. The total water productivity
calculated for this membrane over the first year of operation at this
pressure is 13,000 gal/ft2.
Although the new high-flux membranes exhibited excellent flux stability
with salt solutions, tests on various sewage effluents demonstrated that
membrane fouling is the single most important factor responsible for re-
duced membrane productivity in the treatment of wastewater. The rate
and extent of flux decline appears to depend on the quality of the efflu-
ent used, the initial membrane flux, and to a lesser extent, the type of
polymer used to fabricate the membrane.
Periodic cleaning with the enzymatic presoak, Biz, during bench-scale
reverse-osmosis operation on secondary sewage effluent was shown to be
an effective means of maintaining the high flux of the new membranes.
For example, in a test conducted at 600 psi with secondary effluent, the
fluxes of specially formulated cellulose diacetate membranes were main-
tained at k-0 to 70 gfd over a 6-day period during which the membranes
were cleaned every 2k to k8 hrs with Biz solution.
-------
Three membrane types exhibited the excellent performance desired for
wastewater treatment. These were Eastman Type E-383-M) cellulose
diacetate; a blend of commercial cellulose triacetate (Eastman Type
A-^32-130B) and diacetate (Eastman Type E-398-3) having an average
degree of substitution of 2.55) and a crosslinked cellulose acetate
methacrylate.
Chemical attachment of proteolytic enzymes to cellulose ester membranes
was investigated as'..a novel technique for making them resistant to foul-
ing by sewage components. Methods were developed which allow the prepa-
ration of membranes from a cellulose acetate 'containing a reactive func-
tional group capable of surviving the casting and annealing process;-and
subsequently binding an enzyme covalently. Membranes prepared from the
N-hydroxysuccinimide ester of cellulose acetate hydrogen succinate (CAHS-
NHS), then treated with trypsin, exhibited 2 to 3% of the esterolytic
activity of the native (unattached) enzyme. This degree of activity may
be greater than is apparent since the exposed area of the membrane is
very small compared with that of the finely divided bulk material. A
bulk fibrous preparation of CAHS^-trypsin displayed 13 to 17$ of the" es-
terolytic activity of the native enzyme and 3$ of the activity of the
native enzyme toward hydrolysis of casein, a model proteinaceous colloid
This indicated that the enzymatic membranes should display similar ac-
tivity in catalyzing the breakdown of proteinaceous colloids present in
sewage.
The results obtained in a reverse-osmosis test of CAHS membranes with
secondary effluent were indicative of an antifouling effect resulting
from the negative charge on the CAHS polymer. These results, while pre-
liminary in nature, indicate that membranes carrying a suitable negative
surface charge may-be resistant to colloid fouling.
-------
II. RECOMMENDATIONS
Excellent results were obtained during this program in the develop-
ment of new membranes in flat-sheet configuration. The fabrication
of high-flux membrane types in tubular form was beyond the scope of
the program. A tubular membrane configuration offers advantages for
scaled-up operation as well as for flow characteristics. It is
therefore recommended that the highly promising membranes based on
E-383-40 cellulose acetate and the ds 2". 55 blend be developed into
high-flux tubes for evaluation against various sewage effluents.
The development of crosslinkable cellulose acetate methacrylate (CAM)
membranes in flat-sheet form should be continued, with emphasis on
casting these on porous substrates of the type used to line the inside
of tubular supports. CAM prepared from the commercially available
E-383-^0 cellulose acetate should be investigated as a membrane polymer
to determine whether membranes fabricated from it will duplicate the
outstanding performance of CAM made from a special lot of cellulose di-
acetate (Type E-360-6o).
Further work should be conducted with enzymatic membranes to demonstrate
their antifouling properties with sewage effluents. Simple and inex-
pensive techniques for attachment of enzymes to membranes should also
be sought.
Weakly-charged membranes prepared from acidic or basic polymers, or
blends of these with cellulose acetate, should be examined for their
antifouling properties in tests with sewage effluents.
Some measures to control the effects of fouling, if not the process it-
self, must be taken in order to realize even a healthy fraction of the
high intrinsic water productivity obtainable with suitably formulated
cellulose ester membranes. It is therefore recommended that novel methods
for the characterization and removal of colloids from wastewater be in-
vestigated and that such studies, as well as the use of more conventional
techniques such as membrane cleaning and feed pretreatment, be coupled
with the testing of high-flux tubular membranes with sewage.
-------
Ill. INTRODUCTION
The overall objective of this program was to develop^improved mem-
branes for renovation of municipal wastewater;by.reverse osmosis; A
major effort was placed: on the. fabrication,of membranes having very
high flux with high flux:stability and sufficient rejection of\sewage
components to produce acceptable product water. The;goals of osmotic
performance were a.-flux of,-at least 60 gal/ft2-day (gfd): with a-, flux
decline of less than 20% (as projected from absolute values of meas-
ured flux decline slopes iOf.-less than 0.02^) in the first year of op-
eration at 600 psi along with moderate rejection of sodium chloride.
A secondary objective wasjto develop, membranes inherently, resistant to
colloid fouling,; the-factor which appears to be most responsible for
flux decline duringtoperation of reverse .osmosis units.with sewage ef-
fluent. Such a high flux and flux stability are needed for economical
treatment of•wastewater.
Until recently, attainment of fluxes in the desired range was relatively
easy, but maintaining them at acceptably high levels over long test
periods had not been achieved. Earlier efforts to produce high-flux
membranes for .desalination of brackish water consisted of reducing the
annealing temperatures.used in the preparation ofcellulose diacetate
membranes formulated ofpr the very high salt retention (99$+.) necessary
for single-pass desalination of seawater. The flux stability of these
seawater membranes, however, was dependent on high annealing tempera-
tures (e.g., 90 to 95°C). While the fluxes of such membranes could in-
deed be increased by lowering the annealing temperature, their flux
stability suffered greatly. The high fluxes produced initially by this
approach were only transitory and after very short periods of testing
declined to values comparable to those of the membranes annealed at
higher -temperatures.
More recently, a program was undertaken in these laboratories under
sponsorship of the Office of Saline _Water to investigate more effective
ways of attaining stable, high-flux brackish-water membranes. Two very
promising approaches to stabilization,of the high flux potentially
attainable with asymmetric cellulose ester membranes were found as .a
result of these efforts, and these approaches were applied very success-
fully during this program to the fabrication of high-flux membranes
suitable for wastewater treatment.
The first of these approaches was to attain stable high fluxes through
modification of the casting formulation. This method entailed the use
of larger amounts of, or more powerful, swelling agents in the formula-
tion than were employed in the case of the older cellulose diacetate sea-
water membranes. This is believed to effect a higher degree of swelling
during the gelation step, and the resulting membranes, due to their high
intrinsic water permeability, still produce high fluxes after annealing at
high temperatures. An improvement in the flux stability of cellulose di-
acetate brackish-water membranes was obtained with this fabrication pro-
cedure with an attendant increase in long-term total water productivity.
-------
In the second approach, flux stabilization of membranes was achieved
by chemical crosslinking.' It was felt that crosslinking would serve
to lock the polymer ^chains into position and thus reduce the tendency
of the membrane to compact under pressure during reverse-osmosis op-
eration. Such membrane compaction, due to both reversible elastic
compression and irreversible plastic creep, was believed to be: respon-
sible for-most of the flux decline observed with reverse-osmosis mem-
branes during operation under pressure.2>3 Crosslinking of membranes
prepared from cellulose acetate methacrylate (CAM), a mixed cellulose
ester containing pendant unsaturation in its chemical structure, was
shown to result in a more stable flux.1 The initial fluxes of the
crosslinked CAM membranes, moreover, were higher than those of the
corresponding uncrdsslinked CAM membranes indicating that resistance
to the rapid compaction which occurs immediately on pressurization was
increased by crosslinking.
The aim in the development of improved high-flux membranes for waste-
water treatment was to achieve maximum flux consistent with both high
flux stability and, consequently, high total water productivity over
long periods of operation, along with acceptable sewage solute rejec-
tion. Membrane productivity was evaluated on the basis of intrinsic
membrane flux stability independent of other factors, such as membrane
fouling and degradation, which affect flux. The requirements for mem-
brane rejection were felt to be less stringent in general than those
for flux and flux stability, because of the low total dissolved solids
(TDS) content of domestic sewage, such as sulfate, phosphate, and higher
molecular weight organic compounds; which are efficiently rejected" by
cellulose ester membranes. It was recognized, however, that although
moderate TDS rejection is acceptable, the requirements for removal of
certain constituents, such as nitrate, ammonia, and soluble COD'(which
includes lower molecular weight organ!cs) are more demanding. The
target rejection properties chosen as criteria'for the rapid screening
of candidate membranes were 60$ rejection of sodium chloride and 95$
rejection of sodium sulfate from 1000 mg/£ feed solutions. These goals
appeared reasonable in view of the high sulfate-to-chloride rejection
ratio usually obtained with cellulose acetate membranes, and permitted
considerable flexibility in both formulation and choice of polymer dur-
ing membrane development.
High-flux membranes meeting all of the osmotic performance goals were
prepared from each of three polymer'types during the program. The
choice of these polymer types was-based on certain desirable character-
istics shown by each, as well as promising results obtained with them
in other programs to develop brackish-water desalination membranes.
Eastman Type E-383-4-0 cellulose acetate was an attractive candidate be-
cause of the high flux potentially available by virtue of its moderately-
low degree of substitution (ds of 2.28 compared with 2.4l for,the more
commonly used Type E-398-3 cellulose diacetate), and its commercial
availability. Blends of cellulose triacetate and diacetate are-also
based on commercially available polymers (Eastman Type A-^32-130B tri-
acetate and E-398-3 diacetate) arid membranes produced from 'tlierd have
-------
exhibited generally higher flux stability and salt rejection than those
prepared from cellulose diacetate alone.*->^ Membranes prepared from
cellulose acetate methacrylate (CAM) were of particular interest, not
only because they could be crosslinked, but because of their very desir-
able flux and rejection properties. The CAM polymers, prepared by methac-
rylation of various cellulose acetates, are not now available in commer-
cial quantities but are readily synthesized by established techniques.!^
Covalent attachment of proteolytic enzymes to the surfaces of cellulose
acetate membranes was investigated as a novel approach to making them re-
sistant to fouling by colloidal deposits during operation with sewage ef-
fluents. It was expected that the attached enzymes could minimize the
buildup of flux-depressing deposits by catalyzing,the hydrolytic break- ,
down of colloidal particles present in wastewater when these contacted the
membrane surface, or by otherwise interfering with the process of adhesion
of the particles to the surface. Since the concentration of colloids in
wastewater is on the order of 5 mg/i, the amount and activity of enzymes
needed to minimize deposition may be very .small.5 The aim was to develop
techniques for the chemical attachment of the enzymes to cellulose ace-
tate without undue loss of enzymatic activity. For this purpose, modi-
fied cellulose acetate polymers containing reactive groups capable of
binding the enzymes were synthesized and these coupled with the enzymes.
The sensitive chemical nature of most enzymes imposes certain restrictions
on the methods used for preparing enzyme-coupled membranes. Most enzymes
are readily denatured by heat, although enzymes covalently attached to an
insoluble support often show increased heat stability." It therefore
seemed unadvisable to attach enzymes to membranes by a scheme which re-
quires the enzyme to undergo the membrane annealing process. Enzymes are
also denatured readily on contact with solvents such as alcohols and
acetone.' The membrane casting procedure would thus be expected to affect
adversely the activity of an enzyme coupled to the membrane material.
*"
j • i. >
To avoid the deleterious effects of heat and solvents, two alternatives
were considered. In the first of these, the polymer (a modified cellulose
acetate) would be prepared with reactive functional groups capable of sur-
viving the casting and annealing temperatures. The reactive membrane would
then be treated with an enzyme under mild, aqueous conditions to produce
the enzyme-coupled membrane. In the second alternative, a cellulose
acetate membrane (or substituted cellulose acetate) would be prepared in
the usual way, treated with an "activating" reagent or bifunctional cou-
pling agent, and then coupled with the enzyme. Bifunctional coupling
agents pose the problem of crosslinking the cellulosic polymer or the en-
zyme, while the "activating" reagents, such as cyanogen bromide, may pro-
duce a coupling reaction which has not been completely characterized."
It appeared particularly important to avoid the uncertainties present in
the latter methods since the literature at the outset of this program re-
vealed only one report of an enzyme-coupled membrane, which was prepared
by crosslinking papain adsorbed in a collodion membrane.9 Accordingly,
the more complex but better characterized scheme involving a reactive
polymer was adopted in the initial exploration.
-------
IV. STABLE, HIGH-FLUX MEMBRANES
Although membranes were fabricated and tested in flat-sheet form
throughout the program, fabrication conditions applicable to tube
preparation were used vhere possible with the flat-sheet membranes
in anticipation of later development of the most promising formula-
tions into tubes. In the process for producing tubular membranes,^
used in these laboratories, the membrane is formed by extruding the
casting solution onto the inside of a fabric or paper tube at room
temperature and gelling the wet film in cold water before any appreci-
able drying of it occurs. The flat-sheet membranes, accordingly, were
cast at room temperature and gelled in water at 1 C within several
seconds of casting. The casting and gelation were carried out on 6-in.-
wide aluminized Mylar film in a laboratory-scale casting machine. The
membranes were cast at a thickness of 10 mils and gelled for a period
of 10 min. Annealing (heat-treating) was conducted by immersing the
membranes for a period of 3 min in a water bath thermostatted at the
specified annealing temperature. After annealing, the membranes were
stored in a refrigerator under water in sealed containers until ready
for testing.
Reverse-osmosis testing of the flat-sheet membranes was conducted in
3-in. diameter flat-plate test cells* over the 200 to TOO psi pressure
range. The test cells and ancillary equipment used has been described
in previous reports.^Al Three specimens of each membrane were evalu-
ated in all of the tests carried out with salt solutions and sewage,
and test data reported are the averages of the individual test specimens.
MEMBRANES FROM CELLULOSE DIACETATE
Eastman Type £-383-^0 cellulose acetate was an attractive candidate
polymer for development of high-flux membranes because of its moder-
ately low degree of acetyl substitution (ds); its ds value is 2.28 in
comparison with a ds of about 2.k5 for the cellulose diacetate used in
state-of-the-art membranes. Since the intrinsic water permeability of
homogeneous films of cellulose acetate of comparable density has been
shown to increase with decreasing acetyl ds of the polymer, it was of
interest to exploit the high flux potential of this material in membrane
development. Cellulose acetate Type E-383-^0 has the lowest ds of any
of the commercially available cellulose acetate polymers and no previous
reports of its use in reverse-osmosis membranes have been found.
*
The cells provide a channel height of 0.007 in. over a 2.7-in. diameter
section of exposed membrane. The minimum flow velocity which occurs at
the center of the circular flat plate is calculated to be I.k ft/sec at
5 GPH circulation rate and 2.8 ft/sec at 10 GPH circulation rate.
-------
A formulation study was carried out with E-383-^0 cellulose acetate
using acetamide as the swelling agent; acetamide had been found to
be very effective in promoting high flux with cellulose triacetate-
diacetate blend membranes.1 In this study, the amounts of acetone
and acetamide in the casting solution were varied, and the membranes
annealed at temperatures ranging from 80 to 90°C. The reverse-osmosis
properties of the membranes prepared from the most promising of these
formulations are given in Table 1. The tests were conducted with 1000
mg/l solutions of sodium chloride and sodium sulfate, respectively, at
300 and 600 psi.
The short-term (2-hr) osmotic properties of the membranes cast from
formulation 33A and annealed at 80 to 90°C, and those cast from for-
mulation 33B and annealed at 80 to 85°C, were well within the target
performance range of the program, with fluxes of 6k to 105 gfd, sodium
chloride rejections of 6l to 67$ and sodium sulfate rejections of 95
to 98$ determined at 600 psi. The membranes annealed at 80°C showed
potential for low-pressure (200 to 300 psi) operation producing at 300
psi, fluxes of k2 to kS gfd with 60 to 66$ rejection of sodium chloride.
The relatively low sodium sulfate rejections obtained with the high-flux
membranes cast from formulation 33C indicated that they contained de-
fects possibly due to overswelling at the higher level of acetamide and
lower level of solvent. No further work was done with this formulation.
The intrinsic flux stability of the experimental membranes was assessed
by comparison of the flux decline slope, m, defined by the equation
log J = log J + m log t, where J is the flux after 1 hr of testing and
J is the flux at time t. A least means squares analysis of the flux
data measured during a nominal 200-hr reverse-osmosis test is used to
calculate the value of the slope. In the absence of fouling or membrane
degradation the flux decline behavior of membranes appears to obey the
linear relationship given by the above equation, on the basis of many
long-term tests conducted in these and other laboratories. Integration
of the logarithmic equation leads to the following expression for the
total water production (TWP) in gallons/ft^ expected during time t, where
t is expressed in hours.
J,t*1
TWP =
It should be noted that because of the exponential form of both
equations, the predicted decline in flux or productivity during the
second and subsequent years is very small. For example, a flux decline
slope of -0.021)- corresponds to a flux loss of 20$ during the first year
of operation but less than 2% additional loss during the second year.
Long-term (170-hr) tests were conducted at 600 psi with the most promis-
ing E-383-^0 cellulose acetate membranes to determine their rates of
flux decline. The results of these tests, summarized in Table 2, indi-
cate excellent flux stability for these membranes. On the basis of
8
-------
TABLE 1
OSMOTIC PROPERTIES 'OF E -383 -^0
CELLULOSE ACETATE MEMBRANES
Casting Solution Formulation (parts by weight)
Annealing
Temp., °C
80
85
90
E-383-40
Acetone
Acetamide
Water
Pyridine
Test
Pressure,
psi
300
600
300
600
600
33A
10
Uo
11
3
U
Formulation 33 A
Flux, „
gfd Wad 2 k
U2.2 66.3
105 63.0 96. ^
38.3 72.8
85.3 66.7 98.2
6^.3 - 97.2
33B
10
35
11
3
4
Q
Osmotic Properties
Formulation 33B
Flux,
gfd NaCl &2 k
U7.7 60.5
98.0 60.8 96.3
32.3 70.5
6^.7 65.9 9*f,5
32.0 - 99.0
33C
10
32.5
12
3
h
Formulation
33C
Flux,
gfd NaCl Na2S°U
^5-0 59-9
81.3 60.5
37.3 59-0
66.3 65.1
29.3
-
85.9
-
85.3
99.0
Tested for 2 hrs against 1000 mg/jC feed solutions.
-------
H
O
TABLE 2
LONG-TERM OSMOTIC PROPERTIES OF E-383-to CELLULOSE ACETATE MEMBRANES
First Year
Flux Total Water
One Hour Flux Decline d b
Casting Annealing , rfd*>>c be Production Salt Rejection, %
Formulation8 Temp., °C 1* g Slope, m' gal/sq. ft. 2 hr 170 hr
33A 85 63.07 + 0.78 -0.017 + .003 20,000 7^.3 ftA
33A 90 la.50 + 0.36 -0.016 + .002 13,too 66.8 72.2e
33B 85 66.92 +_0.88 -0.021 + .003 20,700 78.U 79.3e
o
Formulations given in Table 1.
Determined at 600 psi with 1000 mg/jfc sodium chloride solution. Duration of test was 170 hrs.
Given by the equation: log J = log J + m log t, where J = flux at time t (hrs) and J is the flux
after 1 hr. X
Calculated from the equation: TWP = J,tim' /2k (mfl).
pH of feed was maintained at 5«2 + 0.2.
-------
measured flux decline slopes of -0.016 to -0.021, flux losses due to
membrane "compaction" expected during the first year of operation are
13 to 18#. Values of 20,000 to 21,000 gal/ft2/yr were calculated for
the expected total water production (TWP) during the first year of
reverse-osmosis operation with the 85°C membranes. These are greatly
superior to the productivity values of 7000 to 9000 gal/ft2/yr calcu-
lated for state-of-the-art cellulose diacetate brackish-water membranes
from data obtained at 800 psi with 5000 mg/4 sodium chloride solution.
A number of test specimens of the 33A and 33B membranes were fabricated
and heat-treated at 85°C for testing against sewage effluents. The re-
sults of these tests are presented and discussed in a later section en-
titled "Testing of High-Flux Membranes with Sewage."
CELLULOSE TRIACETATE-DIACETATE BLEND MEMBRANES
The development of high-flux membranes from blends of cellulose tri-
acetate (Eastman Type A-1*32-130B) and diacetate (Eastman Type £-398-3)
was of interest because of the superior flux stability and salt rejec-
tion which has been shown by blend membranes formulated for desalination
of brackish water in comparison., to that of brackish-water membranes pre-
pared from the diacetate alone. Work in this program was directed
toward realization of the maximum flux of blend membranes through both
modification of the casting formulation and variation of the triacetate-
diacetate ratio (and accordingly the average ds), while retaining the
intrinsically high flux stability and good rejection properties.
Membranes were cast from the formulation of a 1:1 (by weight) blend of
triacetate and diacetate (average ds = 2.63) shown below.
Component Amount (parts by weight)
A-U32-130B 10
E-398-3 10
1,4-Dioxane 55
Acetone 35
Methanol 9
Maleic Acid 3
The casting and gelling procedure employed was the same as that de-
scribed at the beginning of this section except that the membranes
were gelled after a 30-sec drying period. Testing was conducted with
a 1000 mg/e TDS mixed feed solution of ionic composition characteristic
of that of a typical wastewater. This was done to establish a baseline
for further development of blend membranes for wastewater treatment.
The membranes were evaluated in the unannealed state to obtain maximum
flow.
11
-------
Both the feed composition and test results are shown in Table 3« The
test was conducted for 2-hr periods successively at 200 psi, TOO psi
and again at 200 psi. Considerable compaction of this membrane oc-
curred at 700 psi as evidenced by the very small increase in flux pro-
duced at this pressure over that produced initially at 200 psi, and by
comparison of the initial and final fluxes at 200 psi (before and after
testing at TOO psi, respectively). The unannealed blend membrane would
best be suited for low pressure operation, where it produces high flux
with low chloride rejection but moderate overall rejection. This for-
mulation was not an optimum one for producing stable, high-flux mem-
branes because a low annealing temperature was necessary to obtain the
high flux.
Efforts to reformulate the 1:1 blend (ds 2.63) to achieve high flux
with high annealing temperature through use of a more potent swelling
agent than maleic acid were not successful. Membranes were cast at
room temperature from the formulations shown below, gelled in ice water
within 1 sec of casting, annealed at 85°C and tested for 2-hr periods
at 200 and 600 psi against a mixed feed containing 300 mg/i of sodium
chloride and TOO mg/£ of sodium sulfate.
Component Amount (parts by weight)
A-U32-130B 10
E-398-3 10
1,1^-Dioxane ^5
Acetone 35
Msthanol 0, 5, 10
Propionamide 8, 10, 12
Most of the membranes appeared overswollen, producing excessive flux
with little or no rejection of either solute. The membranes prepared
from the formulations containing no methanol and 8 or 10 parts of
propionamide produced mediocre flux (T to 28 gfd at 200 psi and 13 to
53 gfd at 600 psi) and low rejection (9 to U2# of sodium chloride and
36 to T8£ of sodium sulfate). The low sodium sulfate rejection was in-
dicative of flaws in these latter membranes. In view of these results,
no further work with the 1:1 (ds 2.63) blend was carried out.
Considerably better results were obtained with a blend of 3 parts of
cellulose triacetate and T parts of cellulose diacetate having an aver-
age ds of 2.55. It was theorized that the ds 2.55 blend, by virtue of
its lower ds should have higher flux potential and still exhibit the
excellent flux stability and rejection behavior characteristic of 1 to
1 blends. This contention was borne out by results obtained with the ds
2.55 blend during the course of the program.
Two formulations of the ds 2.55 blend were examined using a combination
of maleic acid and acetamide or propionamide as swelling agents. A
12
-------
TABLE 3
OSMOTIC PROPERTIES OF UNANNEALED DS 2.63 BLEND MEMBRANES
Feed
Osmotic Properties
Pressure,
psi
200
(initial)
700
200
(final)
PH
8.0
7.6
8.9
Cl"
Cone . ,
PPm
178
180
180
Conductance ,
H mho cm
1180
1250
1150
Flux,
gfd
38.6
39-2
27.0
Cl"
50.0
60.0
56.2
Rejection, %
Electrical
Conductivity
73.1
77. ^
77.'2
Nominal feed composition (mg/4) : Na+, 180; Ca++, 80; Mg++, 33; Cl",
"b
Determined after 2-hr test periods at designated pressures.
; HCO,", ^00; SO, , 133.
j **•
-------
series of membranes was cast from these formulations and annealed
over the temperature range of 7k to 85°C utilizing the casting and
gelling procedure described at the beginning of this section. The
reverse-osmosis properties of these membranes tested with 1000 mg/>6
solutions of sodium chloride and sodium;sulfate at 300 and 600 psi
were sufficiently close to the target values,for the program to war-
rant further work with this blend. The formulations used!and test
results obtained are summarized in Table k. Fluxes of 52 to 72 gfd
with 62 to 66% rejection of sodium chloride produced at 600 psi were
within the target range, but the somewhat low sodium sulfate rejec-
tions of 89 to 91$ were indicative of membrane imperfections. !The
relatively low rejections, particularly of sodium sulfate, exhibited
by this first series of ds 2.55 blend membranes were apparently due
to the presence of rather large (approximately 100 y,m diameter) voids
or bubbles which could rupture under pressure producing pinholes.
Such holes provide the leakage paths for salt solution which lowers
the salt rejection of the membrane. Sodium sulfate rejection of less
than 99% is usually an indication of membrane imperfections, since
the intrinsic permeability of cellulose acetate membranes toward this
solute is very low.
The number and size of the voids in the ds 2.55 blend membranes were
reduced considerably by more careful drying of casting solution in-
gredients, particularly the cellulose triacetate, before preparing
the solution. A membrane prepared from a solution of carefully dried
ingredients contained very few voids, which were 20 to 30 ym in
diameter. This membrane, cast from the 35A formulation, annealed at
85°C and tested with 1000 mg/A salt solutions at 600 psi, produced
lower flux (36 gfd) but much higher salt rejection (95$ of sodium
chloride and 98% of sodium sulfate) than did previous membranes fabri-
cated and tested Under the same conditions, but cast from solutions
of less carefully dried ingredients.
The flux stability of the ds 2.55 blend membranes fabricated from both
of the casting formulations described in Table k and annealed at 85°C
determined in a long-term (172-hr) reverse-osmosis test conducted at
600 psi with 1000 mg/4 sodium chloride solution was outstandingly high.
Casting One-Hour Flux Decline Salt Rejection, %
Formulation Flux, gfd Slope 2 hr 172 hr
35A 59-11* + 0.60 -0.013 + 0.003 69.5 73-1
3.5C 1*9.39 + 0.58 -0.011 + 0.003 7^.5 79-7
The total water production for the first year of operation calculated
from this data was 18,500 gal/ft2 for the 35A membrane and 16,500 gal/
ft^ for the 35C membrane. The results of tests conducted with the ds
2.55 blend membranes on secondary sewage effluent are described later
in the section entitled "Testing of High-Flux Membranes with Sewage."
-------
TABLE 1*
OSMOTIC PROPERTIES OF DS 2.55 BLEND MEMBRANES
Casting Formulation (parts by vt)
35A 35C
A-1*32-130B Cellulose Triacetate 6 6
E-398-3 Cellulose Diacetate ll* ll*
Dioxane 1*5 1*5
Acetone 1*0 1*0
Maleic Acid 6 6
Acetamide 10
Propionamide - lU
•a
Osmotic Properties
Formulation 35A
Annealing
Temp.j °C
Determined
7l* 300 psi
600 psi
80 300 psi
600 psi
85 300 psi
600 psi
Determined
80 300 psi
600 psi
Flux,
gfd
with 1000 mg//
57.7
101
1*0.5
77.0
25.2
52.3
with 1000 mg//
38.0
73.0
Rej.,
JL_
Sodium Chloride
1*6.8
57.7
50.1
55A
65.6
65.7
Sodium Sulf ate
87.2
89.0
Formulation
Flux,
gfd
Solution
56.2
92.7
1*1*. 2
72.0
31.8
•5M
Solution
30.0
55.0
35C
Rej.,
-!_
56.0
1*7.6
61.9
56.1*
65.8
90.5
90.7
Determined after 1- to 2-hr test periods.
15
-------
CELLULOSE ACETATE METHACRYLATE MEMBRANES
One of the approaches to flux stabilization of membranes investigated
in this program was crosslinking. A considerable effort was devoted
to fabrication of high-flux membranes from cellulose acetate methacry-
late (CAM), a mixed cellulose ester containing unsaturated groups in
its chemical structure which serve as crosslink sites. The CAM polymer
used for membrane fabrication during the program was synthesized by
methacrylation of Eastman Type E-360-60 cellulose acetate (ds 2.09 to
2.12) by an established procedure. This polymer, designated herein as
CAM-360, was synthesized in 1000-g batches which had total ds ranging
between 2.3^ to 2.39 and methacrylyl ds ranging between 0.2 to 0.3. A
methacrylate ds in this range is considered sufficiently high to permit
adequate crosslinking. This particular CAM polymer was chosen because
of its relatively low total ds, compared with the CAM polymers derived
from higher ds cellulose acetates such as E-383-40 (ds 2.28) and E-398-3
(ds 2.Ul), and because of the excellent osmotic properties exhibited by
brackish water membranes prepared from it on another program.
As in the case of *vthe blend membranes, CAM-360 membranes formulated for
brackish water- desalination were tested using the 1000 mg/A six-ion
mixed feed described in Table 3, as a starting point in the wastewater
membrane development effort. Unlike the cellulose diacetate and blend
membranes described in this report, these membranes were cast at -10°C
on glass plates and allowed to dry in air for 3 min at this temperature
before gelling in water at 1°C. The following casting solution formula-
tion was employed:
Component Amount (parts by weight)
CAM-360 10
Acetone 40
Water 10
Maleic Acid Ik
The membranes were annealed at 80C and 83 C for 3 min then crosslinked
to 93$ acetone insolubility by immersion for 3 min at 90°C in a solution
containing 0.035 M of potassium persulfate and 0.0^0 M of sodium bisulfite
using a previously described procedure.^ The degree of acetone insolu-
bility of a crosslinked CAM membrane is an indication of how completely
the crosslinking reaction has taken place since crosslinking of a polymer
insolubilizes it. The acetone insolubility of a membrane is determined
by measuring the loss in weight occurring after it is stored for several
hours at room temperature in the solvent.
Short-term reverse osmosis tests were conducted on the CAM-360 membranes
at 200 and 700 psi with a 1000 mg/JL mixed feed solution similar to that
used with the unannealed blend membrane described previously (see Table
3). The osmotic properties of these crosslinked CAM membranes as shown -
16
-------
by the data in Table 5 were characterized by very high flux (70 to 90
gfd at 700 psi) with relatively high chloride rejection (80 to 90$)
and somewhat higher total dissolved solids (TDS) rejection as measured
by electrical conductivity. The nearly proportionate increase in flux
produced by the membranes with increased test pressure ,was indicative
of high resistance to compaction, particularly so with the membrane an-
nealed at 83°C.
It was of interest to determine the maximum flow obtainable with the
crosslinked CAM-360 membranes while maintaining salt rejection and flux
stability within an acceptable range. Such information would also estab-
lish the potential for development of very high flux CAM membranes suit-
able for low pressure operation. For this purpose, additional specimens
of membranes cast at -10°C from the formulation described above were
annealed at lower temperature, 7^ and 77°C, respectively, crosslinked
to 85 to 90ft acetone insolubility by the same procedure used with the
membranes annealed at 80°C and 83°C, and tested for 200 hrs at 200 psi
against a feed containing 292 mg/4 of sodium chloride and 708 mg/4 of
sodium sulfate. The test results, summarized below, indicate that very
high fluxes can be obtained at low pressures, together with suitable salt
retention.
Flux j
Annealing One-Hour Decline Rejection; ft
Temp., °C Flux, gfd Slope NaCl Ha2^1»
Ik 61.1 -0.015 67.2 96.0
77 ^9-0 -0.015 76.5 95.5
Although exhibiting excellent osmotic properties for wastewater treat-
ment, these CAM membranes were fabricated under conditions (cast at -10°C,
with long drying time before gelation) unsuitable for development of
tubular membranes. Consequently, CAM-360 was reformulated for room tem-
perature casting and gelling within several seconds of casting, the con-
ditions which prevail in the extrusion of tubular membranes. Two basic
formulations were examined, one based on maleic acid/water as the swell-
ing agent/non-solvent combination (similar to the formulation used in
preparation of the CAM membranes described above) and the other utilizing
acetamide/water for this purpose); A series ;of membranes was prepared
from each basic formulation in which ine amount of acetone was varied.
The solvent level (i.e., polymer concentration) in the casting solution
sometimes has a marked effect on the properties of the resulting mem-
branes. I The membranes were all annealed at jk°C and tested without
crosslinking at 300 psi and 600 psi against 1000 mg// sodium chloride
solution. The formulations and test results are summarized in Table 6.
The most notable observation with these results was the difference in
flux and rejection behavior with increased pressure between the maleic
acid and acetamide formulations. The membranes cast from the maleic acid
formulations (Nos. 32A through 32D, Table 6) exhibited a marked drop in
salt rejection when the test pressure was increased with a flux increase
proportionately higher than the pressure increase, particularly at the
17
-------
TABLE 5
00
Annealing
Temp., °C
80
80
83
83
OSMOTIC PROPERTIES OF CROSSLINKED CAM-360 MEMBRANES CAST AT -10°C
_ ,a h
F66* osmotic Properties
Cl" Conductance Rejection, %
Pressure, Cone., ^ Flux, Electrical
psi j>H ppm jimho cm gfd Cl" Conductivity
200 8.3 180 1190 32.7 85.2 88.5
700 7.^ 200 1180 91.3 81.1 8^.1
200 8.3 180 1190 22.2 89.5 93.0
700 7A 200 1180 70.0 88.0 91.8
aNominal feed composition(mg/jC): Na+j 180; Ca++, 80; Mg++, 33;' Cl", 171*-; HC03", JfOO; S0^~, 133-
Determined after 2-hr test periods at designated pressures.
-------
TABLE 6
REFORMULATION OF CAM-360 MEMBRANES FOR ROOM-TEMPERATURE CASTING
CAM-360
Acetone
Maleic Acid
Acetamide
Water
300 psi Flux, gfd
ReJ., %
600 psi Flux, gfd
ReJ., %
3§A
10
ho
Ik
-
10
23.5
85-9
57.8
69.1*
Casting
32B
10
35
Ik
-
10
25-5
87.8
55.3
78.7
Solution Formulation
32C
10
32.5
Ik
-
10
Osmotic
21.0
89.0
131
32.1
32D
10
30
1U
-
10
(parts
32E
10
kO
-
8
5
by wt.)a
32F
10
35
-
8
5
3§G
10
32.5
-
8
5
Properties
22.7
87.6
79.0
56.0
23-5
78.5
53-3
83.3
33-7
27.8
55-7
35-9
53.5
8.9
7^.7
11 .k
Membranes cast at room temperature, gelled at 1 C within 1 sec of casting and annealed at 7^ C.
Membranes were not crosslinked.
Determined after 2 hrs test time at the designated pressures with 1000 mg/4 sodium chloride solution.
-------
lower solvent levels. This is indicative of the rupture of voids at
the higher pressure. Formulation 32B, however, yielded a membrane
which decreased only moderately in salt rejection on high pressure op-
eration. In contrast, the salt rejection of all the acetamide mem-
branes (formulations 32E through 32G, Table 6) increased with pressure
and their fluxes increased with pressure in a more normal behavior for
reverse-osmosis membranes. Promising results were obtained with the
acetamide formulation at an acetone level of ^0 parts'per 10 parts of
polymer. The low rejections obtained with the membranes cast from solu-
tions containing less solvent were apparently due to pinhole defects
which formed during their fabrication rather than during operation since
the rejections were low even at the low pressure.
Additional CAM-360 membranes were cast from the maleic acid and acetamide-
based formulations 32B and 32E, respectively, to determine the range of
flux and salt rejection' obtainable with uncrosslinked membranes fabri-
cated under these conditions. In contrast to the first series, these
membranes were cast at room temperature, dried for a few seconds, and
annealed at low temperatures (50 to TO°C). This second series of mem-
branes exhibited very desirable osmotic properties as shown by the re-
sults of short-term tests conducted at 300 psi and 600 psi with 1000
mg/£ sodium chloride and sodium sulfate solutions, summarized in Table J.
Fluxes of 71 and 88 gfd with sodium chloride rejections of 75 and 65$
produced at 600 psi by the membranes annealed at 60°C and 50 C were well
within the program performance goals. The 60°C membranes when tested
against sodium,sulfate solutions gave high (90 to 97%) rejection of this
solute. These_membranes were cast from the same batch of casting solu-
tion used to prepare the 60°C membranes tested with sodium chloride solu-
tion but were annealed and tested separately from the latter. The higher
fluxes produced with the sodium sulfate feed may have been due to random
variations in annealing temperature and test pressure. The 50°C membrane
showed potential for low pressure operation, producing 56 gfd flux with
66% sodium chloride rejection at 300 psi.
Contrasting behavior of flux and salt rejection with increasing test
pressure was"again exhibited by the two formulations - the maleic acid
formulation showing a decrease in salt rejection while the acetamide
formulation showed a slight increase in rejection. This difference in
behavior is attributed to differences in the void structures of the two
types of membranes. Voids, as described in the previous section on blend
membranes, are bubbles or cavities present in the microporous membrane
substructure which are large in size compared to the thickness of the
dense active layer of the membrane, and which form during the gelation or
desolvation step, in membrane fabrication. They occur very commonly in
asymmetric reverse-osmosis membranes and can impair membrane performance
by rupturing when the membrane is pressurized. The microporous substruc-
ture provides ample mechanical support for the active layer at the pres-
sures commonly employed in reverse-osmosis operation. The presence of
large voids very close to the active layer surface, however, causes weak
spots in that surface. Failure at these weak points usually occurs under
pressure to produce holes or breaks in the active layer which leak salt
solution, and subsequently, reduces salt rejection.
20
-------
TABLE T
EFFECT OF ANNEALING TEMPERATURE ON OSMOTIC PROPERTIES OF CAM-360
MEMBRANES CAST AT ROOM TEMPERATURE
Osmotic Properties
Annealing Test
Temp., °C Pressure, psi
Determined with 1000
74 300
600
70 300
600
60 300
600
50 300
600
32Bb
Flux, Re j . ,
j£
-------
The CAM-360 membranes prepared from either formulation contained numer-
ous small (l to 10 micrometer diameter) and large (20 to 30 micrometer
diameter) voids, as shown in Figure 1 by the scanning electron micro-
graphs of the maleic acid (formulation 32B) and acetamide (formulation
32E) membranes. In the former membranes, large voids are seen to be
situated very close to the active layer (top) surface. These membranes
exhibited a drop in salt rejection and disproportionately large increase
in flux with increased pressure, which behavior is indicative of void
rupturing. In contrast, the voids, particularly the larger ones, in the
acetamide membrane were located further beneath the active layer surface,
and these membranes showed no evidence of void rupturing at the higher
test pressure (see data of Tables 6 and 7). These differences are
brought out particularly well by the data obtained with the 1000 mg/4
sodium sulfate feed and shown in Table 7- The rejection of this solute
decreased from 97 to 90$ with the 32B membrane while increasing from 9^
to 96$ for the 32E membrane when the test pressure was increased from
300 to 600 psi.
Although the mechanism of void formation is not yet well understood^ it
has been established from work accomplished under other programs, l^A^AS
that void frequency, size, shape and location in a membrane are greatly
dependent upon a number of factors, such as formulation, fabrication con-
ditions and polymer type. Much progress has been made in the control of
void formation through modifications in membrane formulation and fabrica-
tion conditions and the problem does not appear to be a limiting one in
the development of membranes.for wastewater treatment. A detailed study
of this phenomenon was beyond the scope of this program, but its impor-
tance to membrane performance is borne out by the observations described
above for the CAM-360 membranes, and in the previous section for the ds
2.55 blend membranes.
Crosslinking of the CAM-360 membranes cast at room temperature from the
326 (maleic acid-based) and 32E (acetamide-based) formulation described
above (see Tables 6 and 7) and annealed at 60°C was attempted using the
procedure employed previously with the membranes cast at -10°C and an-
nealed at higher temperature. This procedure was unsatisfactory for
treatment of the reformulated membranes because of excessive shrinkage
and deformation which'occurred. Treatment of the membranes for h to-12-
hr periods at room temperature in the crosslinking solution rather than
90°C was also unsatisfactory because of apparent membrane degradation,
evidenced by excessive flux and low salt rejection (e.g., 50$ rejection
of sodium sulfate).
Acceptable crosslinking was finally achieved by immersion of the membranes
in the persulfate-bisulfite bath at the temperature at which the membranes
had been annealed for periods of 5 to 20 min, depending upon the tempera-
ture. The results of 1-hr reverse-osmosis tests conducted at 300 and 600
psi with 1000 mg/J& sodium sulfate solution with uncrosslinked and cross-
linked CAM-360 membranes cast from the 32E formulation (at room tempera-
ture with short drying time before gelation) and annealed at 60°C are
summarized below. The membranes were crosslinked to 60$ acetone insolu-
bility by immersion for 15 min at 60 C in the persulfate-bisulfite bath.
22
-------
32B (Maleic Acid) Formulation
32E (Acetamide) Formulation
50 micrometer
10 micrometer
Figure 1. Void Structure of CAM-360 Membranes
-------
_Test ._ „, Na0SO, Rejection. %
Pressure, Flux, gfd 2 k °
psi Uncrosslinked Crosslinked Uncrosslinked Crosslinked
300 52.0 T8.0 93.8 98.5
600 81^.0 3M.O 96.3 98.3
Excellent osmotic properties were also obtained with similar membranes
annealed at TO°C and crosslinked to 60% acetone insolubility by treat-
ment in the bath for 10 min at TO°C. The results of 1-hr tests con-
ducted with these latter CAM-360 membranes both uncrosslinked and
crosslinked at 600 psi with 1000 mg/jfc sodium chloride are given below.
Flux, gfd Nad Rejection, #
Uncrosslinked Crosslinked Uncro sslinked Crosslinked
^3-3 73.3 89.6 91.7
The short-term osmotic properties of the crosslinked CAM-360 membranes
described above were extremely promising, particularly in view of the
high salt rejections obtained at high flux levels. The crosslinked 60°C
membranes appeared ideal for low-pressure (e.g., 200 psi) operation.
Crosslinking of these membranes is seen to have greatly increased their
fluxes (12% increase for the 60°C membrane and 69% increase for the 70°C
membrane at 600 psi) with a slight increase in salt rejection. This be-
havior is attributed to an increase in the compaction resistance of the
membranes as a result of crosslinking. The relatively low degree of
acetone insolubility (approximately 60$ compared with 85 to 93$ obtained
earlier with CAM-360 membranes crosslinked at 90°C) found for these mem-
branes after crosslinking indicated that the crosslinking procedure war-
rants further improvement. The degree of crosslinking achieved, however,
evidently provided adequate flux stabilization, as shown by the results
of long-term tests described below.
The osmotic properties of the crosslinked CAM-360 membranes fabricated
by casting at room temperature and gelled within seconds of casting
compared very favorably with those of the earlier membranes cast at -10°C
and gelled after relatively long drying times. Lower annealing tempera-
tures were required with the former than with the latter membranes to
achieve the same flux. The crosslinked CAM-360 membranes, in general,
exhibited higher salt rejection at a given flux level than did the
E-383-^0 cellulose acetate and ds 2.55 blend membranes.
The flux stability of" the crosslinked CAM-360 membranes fabricated at
room-temperature, under short drying time conditions was determined in
long-term (nominal 200-hr) tests conducted at 200 and 600 psi with 1000
mg/4 sodium chloride solution. The 32B and 32E membranes annealed and
crosslinked at 70°C were tested at 600 psi and a 32E membrane processed
at 60 C was tested at 200 psi. The results are summarized in Table 8. The
-------
ro
V/l
TABLE 8
LONG-TERM OSMDTIC PROPERTIES OF CROSSLINKED CAM-360 MEMBRANES CAST AT ROOM TEMPERATURE
Yer
_ .. One Hour Flux T
Percent Test Flux Decline Total. Water
Casting Annealing Acetone Pressure, ^' ^ Production, Rejection,
5 °
F'ormulationa Temp., °C Insoluble psi gfd Slope gal/sq. f t. 2 hr 175 hr
32E 60C 58 200 35.52 + 0.51 -0.001 + 0.004 12,800 70.1 68.8
32E 70d 65 600 63.19+0.72 -0.020+0.003 19,700 87.2 87.0
32B 70d 29 600 73.36+1.19 -0.054+0.004 17,400 74.7 75-5
Formulations given in Table 6
Determined with 1000 mg/4 sodium chloride solution. Duration of test was 175 hrs.
Membrane crosslinked by immersion for 15 min at 60 C in a solution of KpSpOn (0.035 M) an^ NaHSO_
(0.040 M). d d G 3
Membranes crosslinked as in c except for 10 min at 70 C. Both membranes treated simultaneously in
same bath.
-------
membranes prepared from formulation 32E (based on acetamide) showed ex-
cellent flux stability, under the test conditions employed, veil vithin
the program performance goal (absolute value of flux decline slope less
than 0.02k}. The 60°C membrane, when tested at 200 psi, showed, for all
practical purposes, no flux decline, and the value of 13,000 gal/ft2 for
the first year's total water productivity, calculated for this membrane
compares very favorably with values at 800 psi of 7000 to 11,000 gal/ft2
calculated for current cellulose diacetate and 1:1 blend brackish-water
membranes operating on 5000 mg/£ sodium chloride. The performance of
the crosslinked 32E membrane (70°c) at 600 psi was outstanding in terms
of combined high water productivity and high salt rejection.
The difference in flux stability observed between the crosslinked 70°C
membranes prepared from the 32B formulation (based on maleic acid) and
the 32E formulation (based on acetamide) was evidently due to the differ-
ence in their extent of crosslinking. The 32B membrane was the less
stable of the two, with a flux decline of -0.05^ compared with a slope
of -0.020 for the 32E membrane. The former had an acetone-insoluble
fraction of only 29$ compared with 65$ insolubility of the latter mem-
brane. These differences and their interrelationships were quite clear-
cut in view of the fact that both membranes were crosslinked under
identical conditions in the same bath. A possible explanation of the
inferior results with the 32B membrane is that chemical degradation took
place either due to residual maleic acid or during the crosslinking
treatment through acid-catalyzed hydrolysis of the methacrylate groups.
Such degradation of CAM-360 has been observed with solutions of the 32B
formulation after storage at room temperature for periods of a week or
more.
The results obtained with both crosslinked and uncrosslinked CAM-360
membranes point up the superiority of the acetamide-based formulation
over the maleic acid-based formulation in terms of osmotic properties
and ease of crosslinking of flat-sheet membranes, fabricated under con-
ditions applicable to extrusion of tubular membranes. The outstanding
performance obtained with the membranes prepared from this formulation
was very encouraging and indicated high potential for development of
practical CAM membranes for wastewater treatment. The results of tests
conducted with CAM-360 membranes against secondary sewage effluent are
discussed in the next section.
26
-------
V. TESTING OF HIGH-FLUX MEMBRANES WITH SEWAGE
The best flat-sheet membranes developed from each of the three polymer
types examined during the program vere evaluated against sewage effluents
obtained from the Pomona Water Reclamation Plant, Pomona, California.
The tests were conducted at 600 psi for periods of 5 to 17 days in 3-in.
flat-plate test cells, constantly recirculating the feed from a 5-gal
reservoir. Except for two cases where primary and carbon-treated
secondary effluents were employed as feeds, all of the tests were con-
ducted with untreated secondary effluent. The pH of the sewage feeds was
maintained at 5.00 +_ 0.25 by the addition of dilute sulfuric acid. Five
milliliters per gallon of 5$ sodium hypochlorite solution was also added
to each feed batch as a routine measure. Each batch of sewage effluent
used at the start of the tests and to replenish the feed during the tests
was obtained from the water reclamation plant the morning of the same day
that it was used. Variation of the feed composition during tests in which
the feed was replaced periodically with fresh effluent, then, reflected
day-to-day variations in effluent quality originating at the reclamation
plant. In each test, the feed and product were analyzed for electrical
conductivity, chloride ion, TDS (by residue on evaporation), COD (total
and "soluble ), nitrate ion, ammonium ion (as NHo) and total phosphate.
The electrical conductivity and chloride ion concentrations were measured
twice daily and the concentrations of the other constituents determined
in most cases daily. "Soluble" COD in the feed was determined on the
filtrate after passage of the feed through a 0.^5 micrometer (^tn) Millipore
filter. The reverse-osmosis test loop employed contained six test cells.
Three specimens of each membrane were tested in each case (except in Test
No. 6 where two test specimens were used for each membrane evaluated) and
the results reported were the averages of each set of three test circles.
Prior to testing with sewage, the membranes were tested (at 600 psi) with
1000 mg/4 sodium chloride solution, and in several cases with 1000 mg/4
sodium sulfate solution, to determine their intrinsic flux and rejection
properties. These values provided a reference point for judging rejec-
tion behavior and rate and extent of flux declines with the sewage feeds.
Preliminary tests with Pomona secondary effluent were conducted for 5 to
6-day periods using E-383-^0 cellulose acetate membranes cast from two
promising formulations (those designated 33A and 33B in Table l) and
annealed at 85 C. Specimens of these membranes had previously exhibited
excellent osmotic properties and flux stability when tested with 1000
mg/4 sodium chloride solution, as shown by the data in Table 2. The pur-
pose of these first tests with secondary effluent was to observe the rate
and extent of flux decline due to fouling and the rejection of organic
and inorganic sewage components. The results of these tests, designated
Test Nos. 1 and 2 are summarized in Tables 9 and 10. The composition of
the sewage feeds, determined during each day of the tests (except for the
weekends), are given in Table 11. In Test No. 1, the sewage feed was
recirculated through the test cells at a rate of 5 GPH and allowed to
concentrate over the duration of the test to the point at which its
electrical conductivity had increased to approximately twice the original
value. In Test No. 2 and all subsequent tests, the feed was recirculated
27
-------
TABLE 9
TESTING OF E-383-UO CELLULOSE ACETATE MEMBRANES WITH POM3NA SECONDARY EFFLUENT
TEST NUMBER 1
Time,
Days
Membrane
Start (2
1
2
5
6
Membrane
Start (2
1
2
5
6
Flux/
gfd
Electrical ..
Conductivity8
Ad - Initial Osmotic- Properties6: 73 gfd
hrs) 14-9
31
26— -
12
9
Bd - Initial Osmotic
hrs) Iv8
31
29 — -
13
10
87
89
29f 89
80
79
Properties6.: .57 gfd
89
92
32f 91
86
83
Rejection, %
. a
TDS Cl"
flux, 6h% rejection
.
93 75
91 70
1*7
*•
flux, 72$ rejection
-
9!* 78
93 75
60
CODb'c
-
98
>99
91
81
.
98
>99
77
86
+b
.
91
77
62
77
-
92
85
70
80
b
NO,"
.
33
55
62
&
.
33
55
l;0
86
Total
Phosphate13
_
97
98
95
9^
-
99
99
97
96
Determined at 600 pei with 5 GPH feed circulation velocity. Values reported are the averages of morning and afternoon samplings.
One sampling per day carried out. TDS determined as residue on evaporation. ' .
'°Based on feed filtered through 0.8 ^tm Millipore filter.
Formulations were those of 33A (Membrane A) and 33B (Membrane B) given in Table 1. Both membranes annealed at'85 C.
eDetermined at 660 psi with 1000 mg/A sodium chloride solution after 1 hr of testing,
f '
-Flux increased to this value when feed circulation velocity was increased to 10 GPH for 1 hr.
'values probably due to faulty analytical results.
-------
TABLE 10
ro
Test
Time,
Days a
Membrane
Start (1
1
2
k
5
Membrane
Start (1
1
2
k
5
TESTING OF E-383-l*0 CELLULOSE ACETATE MEMBRANES WITH POMONA SECONDARY EFFLUENT
TEST NUMBER 2
Rejection,
Flux,
gfd"3
Ae - Initial Osmotic
hr) 57
28
18
15
15
B - Initial Osmotic
hr) 1*5
27
19
17
ll*
Electrical b b
Conductivity TDS° Cl"
Properties :
92
92
90
88
90
Properties :
95
9k
93
93
93
83 gfd flux, 70$ rejection
77
89 74
-
73 73
93 66
57 gfd flux, 82$ rejection
85
91 81
-
86 75
95 73
CODC)d mk+
83 88
-
_100 77
83 91
80 85
-
93 89
79 9k
N03-C
31
-
50
35
75
-
57
56
Total
Phosphate13
99
-
95
97
99
-
87
99
Feed replaced with fresh secondary effluent each day except over weekend.
b - "
Determined at 600 psi with 10 GPH feed: circulation velocity; Values reported are the averages of morning and afternoon samplings.
C0ne' sampling per day carried-out., TDS determined by residue on evaporation.
Based on feed filtered through 0.1*5 |*m Mlllipore filter.
Formulations were those of 33A (Membrane A) and 33B (Membrane B) given in Table 1. Both membranes annealed at 85°C.
Determined at 600 psi with 1000 mg/4 sodium chloride solution after 1 hr of testing.
-------
TABLE 11
Test
Time,
Days
Test No. lc
Start (2 hrs)
1
2
5
6
Test No. 2d
Start (1 hr)
1
2
h
5
COMPOS
Electrical
Conductivity,
ft mhos cm"
1200
1500
1825
2200
2700
1230
1530
1770
1555
1150
ITION OF -SEWAGE FEEDS IN TEST NUMBERS 1 AND 2
Concentration, mg/jj
a
TDS Cl"
loUi 165
1131)- 190
220
-
8k
1265 138
_
190
897 100
"cODb
52 (#0
te (50)
50
66
81 (98)
-
35 (53)
52 (69)
NV N03-
12 3.0
15 M
13 5-8
26 U.8
12 7.1
-
13 2k
33 28
Total
Phosphate
20
50
53
51
36
-
35
38
Average of morning and afternoon readings.
filtered (unfiltered).
cSee Table 9.
3See Table 10.
-------
at the rate of 10 GPH and replaced periodically with fresh effluent to
keep the feed concentration relatively constant.
The membrane fluxes decreased rapidly during both tests with secondary
effluent. After only 1 to 2 hrs on the sewage feed, the fluxes had
declined to 67 to 8U$ of the initial values produced with the sodium
chloride solution; after 1 day they had fallen to 5^ to 3^$ and after
5 days to 30 to 12$ of the initial values. The rate of flux decline was
greater with the higher-flux A-series membranes in both tests, as seen
by comparing the data for each membrane in Tables 9 and 10, and the plots
of log flux versus log time for the A and B-series membranes in Test No.
2 shown in Figure 2. The increase in flux decline rate which appeared to
occur after the first 20 hrs of Test No. 2> shown in Figure 2, cannot be
adequately explained because of insufficient data at times less than 20
hrs. This behavior, observed in subsequent tests to varying degrees, is
suggestive of more than one type of fouling or a change in the fouling
process taking place during the test. Similar abrupt increases in flux
decline rate have been observed in long-term tests with salt solutions
and have been attributed to iron deposits. It is entirely possible that •
inorganic deposits formed from heavy metal compounds present in the sewage
effluent or introduced externally during the test were responsible for
this behavior. Evidence for this hypothesis was obtained in a later test
and is discussed later in this section.
The rejection behavior of the high-flux E-383-kQ membranes observed in
Test Nos. 1 and 2 was excellent. With the exception of a few erratic
analytical results, the rejection of various components achieved.during
the two tests is summarized as follows: electrical conductivity, 80 to
95$; TDS, 89 to 95$; chloride, 60 to 85$; "soluble" COD, 80 to „ 100$;
ammonium ion, 77 "to 9^i>» nitrate ion, 31 to 75$; and total phosphate, 95
to 99$. The B-series membranes, with higher initial rejection of sodium
chloride, showed higher rejection of sewage components than did the A-
series membranes. The drop in rejection seen at the end of Test No. 1
and not observed in Test No. 2 was expected and is attributed to boundary
layer effects due to the deposit built up on the membrane as well as the
increased TDS concentration of the feed and decreased flux due to foul-
ing. The lower feed circulation velocity employed in the first test as
well as the increase in TDS concentration during this test (see Table 11)
are factors which could have been responsible for an increased boundary
layer.
The nitrate rejections were quite variable, and similar behavior has been
observed in a previous program. ^ The value of 8$ nitrate rejection
obtained with both membranes during the last day of Test No. 1 appears
excessively low and is attributed to faulty analytical results. Cellu-
lose acetate membranes reject nitrate ion to a somewhat lesser degree
than chloride ion but the difference is exaggerated when nitrate ion is
present in low concentration. Such behavior is attributed to a mixed
ion effect, which has been observed to be quite marked in tests conducted
with mixed feeds containing sodium chloride and sodium sulfate, in which
the chloride ion rejection was reduced in the presence of sulfate. The
magnitude of the effect depends both upon the sulfate-to-chloride ratio
31
-------
(JO
ro
LOO
90
80
TO
60
50
ho
U 30
I
20
10
Test Conditions
600. psi
25°C
10 GPH Feed Circulation Rate
Pomona Secondary Effluent
Membrane A
I i i
I
i I
I I I
10
Test Time, Hrs
Figure 2. Flux Decline of E-383-^0 Membranes
During Test No. 2
100
200
-------
in the feed and the membrane annealing temperature. The results of a
study of this phenomenon are presented and discussed in the Appendix.
With the exception of the suspect analyses, the nitrate rejections ranged
between 33 and 62%. This is satisfactory for the feeds employed in this
program which rarely contained greater than 20 mg/Jt of nitrate. Higher
nitrate rejections (60 to 99$) were obtained, however, in subsequent
tests (Nos. 3 through 5) conducted with sewage effluents using similar
E-383-40 cellulose acetate membranes as well as ds 2.55 blend membranes,
as is discussed later in this section.
The high rates of flux decline observed in these first tests conducted
with secondary effluent point up the seriousness of the fouling problem.
That the flux decline was due mainly to fouling was indicated not only
by the low flux decline slopes found previously for these membranes in
tests with salt solutions, but more directly in the substantial restora-
tion of the fluxes of the fouled membranes of Test No. 2 by cleaning.
At the end of the test, a 2000 mg/jfc aqueous solution of Procter and
Gamble's enzymatic laundry presoak, Biz (pH 9«3), was circulated at
ambient pressure and 50 to 6o°C through the test cells for 15 tnin at a
velocity of 10 GPH. Deionized water was then pumped through the cells
at 600 psi and room temperature for an additional 15 min and the mem-
branes tested at 600 psi with 1000 mg/^ sodium chloride solution. As
seen by the data summarized in Table 12, the fluxes were restored to
within 70 to 8h% of the original values, with no decrease in rejection
from that originally observed (except for one circle of membrane B which
apparently was damaged). The membranes appeared visually clean after
removal from the test cells. By contrast, the membranes at the end of
Test No. 1, where no cleaning was employed, were covered with a substan-
tial yellow-brown gelatinous deposit. The use of Biz in cleaning membranes
has also been investigated by others and reported by Aerojet under another
program.14
Additional specimens of the A and B-series E-383-^0 cellulose acetate
membranes were tested for a period of 8 days with carbon-treated secon-
dary effluent from the Pomona Water Reclamation Plant to observe flux and
rejection behavior with this higher-quality effluent. In this test (Test
No. 3, see Table 13), the sewage feed was replaced daily with fresh efflu-
ent to maintain feed composition relatively constant throughout the test.
The feed compositions are given in Table lU. During the first 20 to 30
hrs of Test No. 3 "the rate of flux decline was less than that observed in
the test conducted under similar conditions with untreated secondary efflu-
ent (Test No. 2). The rate and extent of flux decline observed after 5
days of testing in each case, however, was similar as shown by the percent
of initial flux after various test times tabulated below for the B-series
membranes with each effluent.
33
-------
TABLE 12
CLEANING 'OF FOULED MEMBRANES FROM TEST NUMBER
Membrane
Number
A - 1
A - 2
A - 3
(Average)
B - 1
B - 2
B - 3
(Average)
Initial0
Flux,
gfd
86
80
8k
(83)
52
60
60
(57)
NaCl
Rej .,..%
71
71
68
(70)
8k
82
82
(83)
Immediately
Before . Cleaning
Flux,
gfd
17
Ik
(15)
22
10
11
-------
TABLE 13
TESTING OF E-383-!*0 CELLULOSE ACETATE MEMBRANES WITH FOMDNA CARBON-TREATED SECONDARY EFFLUENT
TEST NUMBER 3
*5
30
25
23
19
17
17
Cleaned with Biz8 and tested at
76£ rejection (10756 of initial).
90
91
91
92
91
90
90
90
600 psi with
Membrane Be - Initial Osmotic Properties : 68 gfd
Start
1
3
It
5
6
7
8
(1 hr) 65
52
35
27
25
20
19
19
Cleaned with Biz8 and tested at
835» rejection (109& of initial)
92
93
93
9k
92
92
90
91
600 psi with
TDSC
flux, 7i%
_
95
-
91
95
92
93
1000 rng/H
flux, 76$
-
97
-
91
96
93
91*
-
1000 rag/*
b
Cl"
rejection
6k
67
62
58
58
60
66
65
sodium chloride:
rejection
67
70
68
62
58
66
68
69
sodium chloride:
CODC'd
80
5U
57
51
81
60
70
57 gfd flux
1*2
53
72
1*1*
80
67
57
62 gfd flux
NH4+C
85
87
83
88
89
83
88
(84$ Initial),
91
95
91*
-
91
88
91
(91$ of Initial)
c
NO,"
>75
>75
>75
-
1*9
50
59
>75
>75
>75
-
63
55
63
/
Total
Phosphate0
99
98
97
98
98
98
98
98
95
99
99
99
99
98
Feed replaced dally with-fresh effluent.
Determined at 600 psi with 10 GPH feed circulation velocity; 3 circles of each membrane tested and averaged. Values reported
are the averages of morning and afternoon samplings. Sevage feed pH maintained at 5.00 + 0.25.
One sampling per day carried out. TDS determined as residue on evaporation.
dBased on feed filtered through 0.1*5/im Milllpore filter.
eFormulations were those of 33A and 33B given in Table 1. Both membranes annealed at 85°C.
fDetermlned at 600 psi with 1000 mg/^ sodium chloride solution after 1 hr of testing.
flirty-minute wash with 2000 mg/X Biz solution (pH 9.3 to 9.5) at room temperature, 600 psi, 12 GPH.
-------
TABLE
FEED COMPOSITION - TEST MJ
Test Electrical &
Time, Conductivity,
Days jnnho . cm
Start (1 hr)
1
3
k
5
6
7
8
1025
1175
1^30
1300
1175
1350
1250
MEER 3 WITH POMONA CARBON-TREATED SECONDARY EFFLUENT
Concentration, mg/j2
a - + -
TDS Cl" CODb mk N°3
9k - - -
808 91 21 (21) 2k O.k
103 31 (32) 19 OA
838 86 33 (33) 2 o.h
813 7^ hi (56) 25 o.k
820 87 31 (31) 20 3-5
86^ 118 31 (35) 21 17
115 28 (31) 21 23
Total
Phosphate
-
32
k2
50
38
38
36
36
Average of morning and afternoon samples.
Filtered (xmfiltered).
-------
Flux Retention ($ of Initial Flux)
Carbon-Treated
Test Time Secondary Effluent Secondary Effluent
with Sewage (Test No. 2) (Test Mo. 3)
Start 81 96
1 hr 79 96
5 days 25 37
8 days - 28
This similarity is also seen by comparison of the log flux-log time plots
given in Figure 3« An increase in the rate of flux decline was observed
after about 30 hrs in Test No. 3 with carbon-treated secondary effluent
at least as marked at that seen during Test No. 2 with untreated secon-
dary effluent. This behavior with carbon-treated effluent was rather
surprising, in view of its higher quality compared with untreated secon-
dary effluent. These results together with results.,obtained in tests , ^
conducted with carbon-treated secondary effluent under a previous program,
are highly suggestive of extraneous matter, such as airborne dust, in the
feed.
The rejection behavior of the membranes observed during the te,st with
carbon-treated secondary effluent (Test No. 3) was similar to that ob-
served with untreated secondary effluent* (Test No. 2) except for the lower
"soluble" OOD rejection (generally 50 to 80$ vs. 80 to 100$ as shown in
Tables 10 and 13). This difference is illustrated by the fact that higher
product COD-values were obtained1in Test No. 3 than in Test Nos. 1 and 2.
The lower "soluble" COD rejection may have resulted from1,the presence of a
highly permeable organic or oxidizable inorganic component present in the
carbon-treated secondary effluent.
Cleaning of the fouled membranes at the end of Test No. 3 was conducted
by circulating a 2000 rag/4 solution of Biz through the test cells for 30
min at 600 psi and room temperature at a flow rate of 12 GPH. The mem-
brane fluxes were restored to Q^% (membrane A) and 91% (membrane B) of
their initial values with increases in salt rejection over that originally
observed for each membrane (see Table 13).
In view of the effectiveness of Biz cleaning in restoring the flux of
fouled membranes, observed after Test Nos. 2 and 3 with secondary and
carbon-treated secondary effluents, respectively, a test was -carried out
with more frequent cleaning. Specimens of Membranes A and B, annealed
at 85°C, were again employed and the test (Test No. ^) was conducted at
600 psi with Pomona secondary sewage effluent. The cells were cleaned
intermittently by treatment at 600 psi and 12 GPH circulation rate for
30 min with a 2000 rag/£ solution of Biz at room temperature. The highly
promising results of this test are summarized in Tables 15A and l^B, for
membranes A and B, respectively. The composition of the sewage feeds
used are summarized in Table l6. During the first 9^ hours of the test,
when the Biz cleaning was carried out every 2h hours, the flux of Membrane
37
-------
u>
CD
100
90
80
70
60
59
ho
-Initial Fluxes on 1000 mg/j0 Nad Solution
Carbon-treated Secondary Effluent
(Test No. 3)
7*
I I
I I I I I I I
100
10
Test Time, Hrs
Figure 3. Flux Decline of B-Series E-383-^0 Membranes with Different Effluents
200
-------
TABLE 15AT
TESTING OF E-383-UO'MEMBRAHES WITH SEWAGE UTILIZING PERIODIC' SLZ CLEANING
TEST NUMBER It- - MEMBRANE A
Date
3/17/70
3/17/70
3/17/70
3/17/70
3/18/70
3/18/70
3/18/70
3/18/70
3/19/70
3/19/70
3/19/70
3/19/70
3/20/70
3/20/70
3/20/70
3/21/70
3/22/70
3/23/70
3/23/70
3/23/70
3/23/70
3/23/70
3/24/70
3/24/70
3/24/70
3/24/70
3/24/70
3/25/70
3/25/70
Time
1100
12OO
1300
1600
0900
1130
1330
1600
0900
1000
1200
1600
0900
12OO
1600
1000
1000
0900
1100
1200
1600
2200
0900
1200
1300
l600
2200
0900
1200
Flux,'
gfd
73
68
64
56
73
64
74
51
68
65
56
51
66
67
48
43
42
61
58
31
21
13
49
33
21
17
1 Electrical
Conductivity
94 gfd flux, 72%
-
92
93
93
89
91
91
90
89
90
89
92
90
91
91
90
90
87
89
90
88
88
6l gfd flux, 79*
88
88
86
86
40 gfd flux, 85*
Rejection, %
i.
Cl COD
rejection
-
68
67
66 100
63 73
-
76
69 96
94
73
73
6k 83
-
-
82
-
81
-
79
-
-
6l
rejection
-
-
.
59
rejection
NH+ NO' Total
4 3 Phosphate Comments
1000 mg/& Nad
Start vith
secondary. effluent
.
.
92 80 97
86 79 97 Biz Cleaning0
.
Add fresh secondary
effluent
90 91 98
87 76 97 Biz Cleaning0
- - Add fresl?'"«"-"r"1ary-
effluent
-
93 72 98
- - Biz Cleaning
-
96 84 98 Add fresh secondary
effluent
.
90 95 98
Biz Cleaning
87 45 99 Start vith primary
effluent
.
.
88 67 >99
Biz vash , then
1000 mg/4 Mad
- - - Add fresh primary ef
-
.
85 67 98
Biz vash , then
1000 me/* Nad
BDetermined at 600 psi vith 10 GPH feed circulation',velocity. Values reported
are the averages of 3 test circles. Sevage feed pH maintained at 5.00 +:0.25.
Based on feed filtered through 0.1*5 urn Millipore filter.
°2000 mg/2 aqueous solution of Biz circulated at 12 QFH, room temperature and 600 psi
for 30 min, then svitch to previous feed.
washed for 30 min as in 0 above, then reduced pressure to ambient for 1, min and
continue vith 15 min vash at 600 psi as before.
e¥ashed as in d above except for only 15 min before pressure cycling.
39
-------
TABLE 15B
TESTING OF E-383-ltO MEMBRANES WITH SEWAGE UTILIZING PERIOIHC BIZ CLEANING
TEST NUMBER k - MEMBIANE B
Rejection, j£*
Date
3/17/70
3/17/70
3/17/70
3/17/70
3/18/70
3/18/70
3/18/70
3/18/70
3/19/70
3/19/70
3/19/70
3/19/70
3/20/70
3/20/70
3/20/70
3/21/70
3/22/70
3/23/70
3/23/70
3/23/70
3/23/70
3/23/70
3/2U/70
3/2U/70
3M/70
3/2U/70
3/2U/70
3/25/70
3/25/70
Tine
1100
1200
1300
1600
0900
1130
1330
1600
0900
1000
1200
1600
0900
1200
1600
1000
1000
0900
1100
1200
1600
2200
0900
1200
1300
1600
2200
0900
1200
-Flux,
gfd
50
M
k6
14
51
.50
51
1*2
51
1*9
">5
1*2
1*6
U7
1*1
38
37
1*6
1*1*
30
20
ll*
1*1*
30
18
ll*
Electrical
Conductivity
62 gfd nux, 87*
.
96
95
96
9*
95
95
95
9U
91*
91*
96
95
96
96
95
95
9*
95
95
95
9!*
1*6 gfd nux, 92*
9!*
9"
9U
93
35 gfd nux, 93*
0."
rejection
.
79
»
79
68
-
86
oU
-
8U
78
73
-
-
-
-
-
-
.
-
-
.
rejection
_
.
-
-
rejection
CODb
.
•
-
97
82
-
-
95
81
-
-
89
-
-
88
.
85
.
82
.
.
66
_
.
.
61
«m+ «m" Total
™V "°3 Phosphate
.
.
.
92 >98 99
93 >98 99
.
.
90 >96 99
91 >95 99
.
.
96 91 99
-
-
93 >97 99
-
91 95 >99
.
86 1*2 >99
-
-
91 85 >99
• — _
-
-
92 75 >99
Comments
1000 ng/i IfeCl
Start vith
secondary effluent
Biz Cleaning0
Add fresh secondary
effluent
Biz Cleaning6
Add fresh secondary
effluent
Biz Cleaning0
Add fresh secondary
effluent
Biz Cleaning0
Start vith primary
efnuent
d
Biz wash , then
1000 ng/* NaCl
Add fresh primary
effluent
Biz vashe, then
1000 ng/4 Had
*Detemdned at 600 psl with 10 GPH feed circulation velocity. Values reported are the
averages of 3 test circles. Sevage feed pH maintained at 5.00 +_ 0.25.
bBased on feed filtered through 0.1*5 urn fttllipore filter.
C2000 mg/* aqueous solution of Biz circulated at 12 GPH, room temperature and 600 psl for 30 nln,
then svitch to previous feed.
"washed for 30 ""in as in c above, then reduced pressure to ambient for 1 nin and continue, vith
15 min wash at 600 psl as before.
6Washed as in d above except for only 15 Din before pressure cycling.•
1*0
-------
TABLE 16
Feed
Secondary Effluent
Secondary Effluent
Secondary Effluent
Secondary Effluent
^Secondary Effluent
Secondary Effluent
Secondary Effluent
Secondary Effluent
Secondary Effluent
Secondary Effluent
Secondary Effluent
Secondary Effluent
Secondary Effluent
Primary Effluent
Primary Effluent
Primary Effluent
Primary Effluent
COMPOSITION OF PRIMARY AND SECONDARY ]
Electrical
Conductivity,
Date Time u mho cm"
3/17/70
3/17/70
3/18/70
3/18/70
3/18/70
3/19/70
3/19/70
3/19/70
3/20/70
3/21/70
3/22/70
3/23/70
3/23/70
3/23/70
3/24/70
3/24/70
3/25/70
1200
1600
0900
1130
1600
0900
1000
1600
0900
1000
1000
0900
1100
1200
0900
1600
0900
1360
1380
l44o
1625
1350
1460
1200
1300
1500
i4oo
1630
1700
1500
1350
1550
1550
1800
EFFLUENT FEEDS IN TEST NUMBER 4
Concentration, mg/
-------
A was maintained at 48 to 73 gfd (66 to 100$ of the initial flux on secon-
dary effluent) and that of Membrane B at 4l to 51 gfd (82 to 102$ of the
initial flux on the sewage). Of equal riote was the low rate of flux
decline observed with the secondary effluent between Biz washings, this
being comparable or lower than that observed in the test with carbon-
treated secondary effluent (see Table 13). Between each of the first
three Biz cleanings the flux of the A membrane decreased by 23 to 30$ and
that of the B membrane by 11 to 18$, compared with 26$ and 20$ decreases
in flux of the A and B membranes, respectively, during the first 2k hours
of testing with carbon-treated effluent. During the 4th and,5th day of
the test, when no Biz cleaning .was employed, the fluxes' of the A,and B
membranes decreased by only 12 and 10$, respectively, compared with
decreases of 33$ observed with each membrane during a corresponding
48-hour period in Test No. 3 with carbon-treated effluent. The effective-
ness of the periodic cleaning in maintaining high, flux levels^with secon-
dary effluent is illustrated graphically in Figure 4 by log flux-log time
plots for the B-series membranes in Test Nos. 2 and 4. Several plausib'le,
but speculative, explanations can be offered for the relatively low rates
of flux decline between Biz .cleanings, in,Test No. 4. First, it'is possi-
ble that the Biz treatment imparted some antifouling properties to the
membrane through adsorption of enzymes on its surface. Second, the quality
of the secondary effluent used during this test was unusually high. Third,
contamination of the feed by extraneous matter, such as airborne dust from
the laboratory ventilation system, which may have contributed to the flux
decline in the previous tests, may have been less in Test No. 4.
Although the rejection behavior of both membranes during Test No. 4 with
secondary effluent was excellent, Membrane B was again superior. The
B-series membranes, which gave initially lower flux and higher rejection
of sodium chloride than the A-series membranes, consistently exhibited
better rejection behavior as well as lower flux decline when tested against
sewage effluents'. The rates of, flux ..decline of the A and B-series mem-
branes in Test No. 4 during the first 24 hours of the test, between clean-
ings and over the duration of the test with secondary effluent are com-
pared graphically in Figure 5. Again, as in the previous tests, the rate
of flux decline was greater^with the'membrane producing higher initial.
flux. Slopes of -O.o8l for the' A-membrane (initial flux on sodium chloride
of 94 gfd} and -0.054 for the B-membrane (initial flux on sodium chloride
of 62 gfd) on Pomona secondary effluent are obtained for the lines defined
by the initial flux values on sodium chloride for each membrane and the
flux values obtained on the secondary effluent immediately after each
cleaning. These slopes yield extrapolated year-end post-cleaning flux
values of 44 and 38 gfd, respectively, for the A and-B-membranes. Still
more favorable flux decline slopes of- -0.071 and -0.033, respectively,
were calculated for the A and B-membranes using all of the points shown
in Figure 5« Even these, results, "-however, do'not reflect the full poten-
tial of these membranes, since similar membranes exhibited intrinsic flux
decline slopes of only -0.04 and -0.02, respectively, in long-term tests
with salt solution. Therefore, still greater productivity might be real-
ized by further improvement of the cleaning procedure, feed.pretreatment,
etc.
42
-------
100
90
80
70
60
50
30-min Biz Wash
I
30
20
Test No. k
Test No. 2
Test Conditions
600 psi
25°C
10 GPH-Peed-Circulation Rate
Pomona Secondary Effluent
10
1 I I i i I
'10
Test Time, Hrs
Figure 1|.. Effect of Periodic Biz Cleaning on Performance of B-Series
E-383-^0 Membranes with Secondary Effluent
100
200
-------
30-min Biz Wash
100
90
80
TO
60
50
30
20
Test Conditions
600 psi
25°C
10 GPH Feed Circulation Rate
Pomona Secondary Effluent
10
I I I
till
10
Test Time, Hrs
Figure 5. Effect of Membrane Initial Flux on Flux Decline During
Test Number h with Pomona Secondary Effluent
100
200
-------
During the last 2 days of Test' No. k, Pomona primary sewage effluent was
used as feed to observe membrane performance and the efficacy of Biz
cleaning with this lower quality effluent. The data obtained in this
experiment are shown in Tables 15A and 15B. As expected, the rate of
flux decline was considerably greater with the primary effluent than
that observed with secondary effluent (72 to 79$ decline for Membrane
A and 70$ decline for Membrane B in 21 hrs between cleanings as compared
with 23 to 30$ and 11 to 18$ declines, respectively, during corresponding
2k— hr periods with secondary effluent); and the cleaning procedure was
less effective with the primary than with the secondary:effluent. In the
first cleaning cycle with primary effluent, fluxes were not restored
appreciably when circulating the Biz solution under pressure but were
restored immediately after reducing the pressure to ambient for several
minutes. The fluxes were not completely restored to the original peak
values after the second cleaning attempt. At this point the test was
terminated and the membranes removed for examination. All of the membrane
circles were found to contain an opaque bluish-white surface deposit, the
color apparently caused by the bluing agent present in Biz. The amount
of the deposit appeared by inspection to vary inversely with the final
flux produced by each circle. Of equal significance was the observation
of the residue of a second type of deposit, more characteristic of the
organic matter deposited from sewage. This transparent, gelatinous,
yellow-brown material was present in small quantity on top of the bluish-
white surface deposit on one or two of the membranes. The presence of
this second type of deposit did not relate to final membrane flux and
since it was present in • small quantity on only certain membranes it
appears to be readily removable by the Biz cleaning treatment.
A sample of the bluish-white deposit was collected, dried, pyrolyzed in
a furnace at 1000 F and the residue submitted for flame emission analysis
to determine major metallic constituents. The results of the analysis
are summarized below.
Element %3 Based on Oxides
Fe 3-7
Al 36.3
Mg 1.5
Si 2.3
Na 25.6
No phosphorus was detected in the flame emission spectrum, and a quali-
tative test for sulfate was negative, eliminating the possibility that
the deposit had been comprised substantially of phosphates or calcium
sulfate. Precipitation of polyvalent metal phosphates during the Biz
treatment had been suspected, in particular, since the phosphate content
of Biz is' 18.7$ based on-elemental phosphorus.
On the basis of the above analyses, and its relatively dense physical
form, the original blue-white membrane deposit appears to have been
-------
inorganic in nature, and comprised of a refractory metal oxide mixture.
Its origin appears to have been extraneous having been introduced into
the primary effluent in the form of airborne dust either at the Pomona
Water Reclamation Plant or in the laboratory during the test. The
latter alternative seems to be a' good possibility^ the prime suspect
being fine particles of insulation material (e.g., mineral wool) from
the air ducts of the laboratory ventilation system entering the
uncovered feed tanks. Such contamination may also have been responsi-
ble for the increased flux decline with time in Test Nos. 1, 2, and 3-
Accordingly, the feed tanks were kept covered during the remaining tests
with sewage described below.
The results obtained in the tests of the E-383-^-0 cellulose acetate mem-
branes with various sewage effluents were very encouraging with respect
to rejection behavior and the utility of membrane cleaning in moderating
the flux decline due to fouling. At the.same time, they were instructive
as to the extent of the fouling problem in reducing,the productivity of
membranes having high intrinsic flux and flux stability. It was there-
fore of interest to evaluate and compare the performance of the other two
membrane types, namely the ds 2.55 blend and crosslinked CAM-360, in
similar tests with sewage effluent. The two remaining tests, Nos. 5 and
6 described below, were conducted for this purpose, using untreated secon-
dary effluent as the feed.
Specimens of ds 2.55 blend membranes cast from the formulations designated
35A and 35C in Table 4 and annealed at 85°C were tested for a period of
IT days at 600 psi with Pomona secondary effluent. The results obtained
in this test, designated Test No. 5, are summarized in Table 17.The
feed compositions are given below.
Test pS,C,S.C.?L Concentration, mg/l
C(
Soluble
1*0
36
35
)D
Total
40
^
^
K
19
^
16
NO"
17
29
l*l*
Total
Phosphate
33
36
38
Time,
Days (A mho cm"
Start (1 hr) lk6o
6 11*50
12 1690
17 1^00 - -
The test data indicate relatively low rates of flux decline (for secon-
dary effluent) throughout much of the test period, as well as generally
very high rejection of all components analyzed. The low COD rejections
obtained on the 6th and 12th days of the test were inconsistent with the
high rejections of other components observed on those days as well as
the high COD rejections obtained at the beginning of the test. This
discrepancy could have been caused by faulty analyses, sample contamina-
tion or the presence of highly permeable oxidizable components in the
feeds used on those days.
-------
TABLE 17
Time,
Days
Membrane 35AC
Start (1 hr)
1
2
5e
6
12
ll*6
17
Membrane 35CC
Start (1 hr)
1
2
5e
6
12
l>*e
17
Flux,a
en
- Initial
51
53
1*5
1*3
1*0
38
37
32
After
TESTING OF DS 2.
Electrical
Conductivity
Osmotic Properties ' :
9U
95
91.
95
95
91*
95
95
cleaning vith Blzf: 57
- Initial Osmotic Properties*' d:
1*6
1*2
36
33
28
26
21*
21
After
90
96
95
95
95
91*
95
92
cleaning with Blzf; 1*7
55 BLEND MEMBRANES WITH POMONA SECONDARY EFFUJENT
TEST NUMBER 5
Rejection, j&*
COD
Soluble* Total
68 gfd flux,
100
-
-
-
51
56
-
-
gfd flux, 89*
57 gfd flux,
100
.
.
-
1*7
148
gfd flux, 81*jt
TTJt rejection
100
-
-
.
68
72
-
-
sodium chloride
80% rejection
100
.
.
-
65
66
-
-
sodium chloride
v
96
-
-
-
91
87
-
-
rejection.*' d
93
-
.
-
91
92
-
-
rejection.8' d
?£.
99
-
-
.
99
99
-
-
99
.
.
.
99
99
-
-
Total
phosphate
99
-
-
.
99
99
-
-
99
.
.
-
99
99
-
-
Determined at 600 psl. Data reported are averages of 3 test specimens.
Based on feed filtered through 0.1*5 |un Hllllpore filter.
"Formulation 35A (parts ,by vt) - cellulose triacetate A-1*32-130B, 6; cellulose acetate E-398-3, ll*j dioxane, 1*5;
acetone, 1*0; malelc acid, 6; acetamlde, 10. Formulation 35C - same except used lU parts proplonamlde instead
of acetamlde. Membranes annealed at 85°C.
'^Determined with 1000 mg/X sodium chloride solution at 600 psl after 1-hr test period.
eFeed replaced with fresh effluent.
Carried out at end of test by circulating a 2000 mg/4 aqueous solution of Biz at room temperature, 600 psi and
10 GPH.
-------
Cleaning of the membranes with Biz at the end of Test No. 5 increased
the fluxes to more than 80$ of their initial values determined with
1000 mg/4 sodium chloride solution and revealed an improvement in salt
rejection. The flux decline behavior of the membranes during the test
is shown graphically by log flux-log time plots in Figure 6. The rate
of flux decline on the secondary effluent was similar for the two mem-
branes, and was approximately linear after the first 2k hours of oper-
ation. ' Slopes of -0.030 to -0.032 of the lines drawn'between the flux
values determined with 1000 mg/^ sodium chloride solution at the
beginning of the test and at the end of the test after Biz cleaning
are even better than the corresponding value of -0.05^ obtained in Test
No. k for an E-383-4-0 cellulose acetate membrane of comparable initial
flux (B-membrane). These values correspond to less than 25$ loss in
flux after the first year of operation. As in the case of the £-383-^0
membranes, the intrinsic flux decline slopes of these membranes (-0.011
to -0.013) with salt solution indicate that improved control of fouling
will yield still greater flux stability in the treatment of sewage.
A final test (Test No. 6) was conducted with the best examples of each
of the three promising high-flux membrane types examined in the program
in the same test apparatus using a common "secondary effluent feed, to
obtain a more direct comparison of their performance on sewage. The
test was conducted with 2 circles of each membrane type at 600 psi for
a period of 6 days and yielded the results summarized in Table 18 and
presented graphically in Figure 7. For the CAM membrane, formulation
32E in Table 6 was cast at room temperature, annealed at 70°C and cross-
linked at that temperature for 10 min as previously described. The
E-383-40 membrane was cast from the formulation designated 33A in Table
1 and annealed at 85°C; and the blend membrane was cast from the formu-
lation designated 35A in Table k and annealed at 85°C. Each of the mem-
branes was tested for 1 hr with both sodium chloride and sodium sulfate
before testing with the sewage feed. The sewage feed was replaced every
other day with fresh effluent and the flux and electrical conductivity
of the feed and product were measured twice daily.
All three membranes at the beginning of Test No. 6 gave high sodium
sulfate rejections (97 to 99$). The CAM membrane exhibited the best
rejection characteristics of the three - 92$ of sodium chloride, 99$ of
sodium sulfate and 95'to 97$ of IDS in sewage. Its initial flux on
secondary effluent was 57 gfd falling to 29 gfd after 6 days on the
sewage. The E-383-4-0 membrane exhibited the highest initial flux (85
gfd) on salt solution but the greatest flux decline on secondary efflu-
ent of the three (see Figure 7) producing after 6 days on the sewage
only 22 gfd flux. Although its rejection performance was greater than
the minimum requirement, the E-383-40 membrane, giving the lowest rejec-
tions of sodium chloride (70$) and sodium sulfate (97$), also gave the
lowest rejection of TDS (87 to 92$) with the sewage. The blend membrane
produced the lowest initial flux (50 gfd) on salt solutions and exhibited
a flux decline rate on the sewage comparable to that of the crosslinked
CAM membrane, giving after 6 days a flux of 20 gfd. The blend membrane
ranked intermediate in rejection performance among the three membranes
-------
100
90
80
70
60
50
30-min Biz Wash
30
20
Membrane 35A
z:
Membrane 35C
Test Conditions
600 psi
25°C
10 GPH Peed Circulation Rate
Pomona Secondary Effluent
10
J 1 I I I I
J ' I I I I I
100
10
Test Time, Hrs
Figure 6. . Flux Decline of DS 2.55 Blend Membranes During Test Number 5
200
300
-------
IABEE 18
COMPARATIVE TEST OF THREE MEMBRANE TYPES
WITH POM3NA SECONDARY EFFLUENT
TEST BOMBER 6
Crossllnked
CAM-3608
Flux,
gfd
TBS
ReJ.. *
E-383-l(0
Cellulose
Flux,
gfd
Acetateb
IDS
ReJ.. *
DS 2.55
Flux,
gfd
Blend0
IDS
ReJ.. %
Initial Osmotic
Properties4' e
1000 ing/*
1000 mg/Jl
Secondary
Test
Time,
Rrs
1
3
20
27
44
51
116
123
11)0
Had
*2so4
Effluent4' f
Electrical
Conductivity
of Feed,
|» mho cm"1
1525
1525
1600
1700
1700
1550
1700
1500
1500
69
66
57
56
1(2
44
38
41
34
32
29
92
99
95
96
97
97
97
97
97
96
97
85
83
71
66
1*3
43
36
36
25
2l(
22
70
97
87
88
90
90
89
92
91
90
92
1(7
51
41
40
33
3^
30
31
23
22
20
79
98
91
92
94
94
94
95
96
94
96
Cleaned with Biz,8
Tested with 1000 mg// NaCld
1(3 gfd
95* ReJ.
32 gfd
85* ReJ.
28 gfd
93* Re
aCast from 32E formulation (Table 6), annealed and crossllnked at 70°C.
bCast from 33A formulation (Table 1), annealed at 85°C.
°Cast from 35A formulation (Table 4), annealed at 85°C.
'resting conducted at 600 psl with 10 GPH recirculation velocity. Two circles of each membrane tested and results averaged.
Determined after 1-hr test periods.
Sewage feed pH maintained at 5.00 ± 0.2 and replaced every other day with fresh effluent.
8Aqueous solution of Biz (2000 mg/jt circulated for 30 mln at room temperature, 600 psl arid 10 GPH).
-------
100
90
80
TO
60
50
1*0
8
a 30
I
20
10
7,
DS 2.55 BLEND
Test Conditions
600 psi
25°C
10 OPH Feed Circulation Rate
Pomona Secondary Effluent
I I till
. I I
I L
I ill
li
10 100
Test Time, HTB
Figure 7. Flux Decline of the Three Membrane Types During Test Number 6
200
-------
with 91 to 96$ TDS rejection on the sewage.
Cleaning of the three membranes at the end of Test No. 6 with Biz was
not as effective as in previous tests with secondary effluent, as shown
by the data in Table 18. The fluxes were restored to only 38 to 62% of
the initial values on salt solution, and the membranes still contained
a gelatinous deposit. The CAM and blend membranes were cleaned to a
greater degree, and the deposit seemed, on inspection, to adhere less
strongly to them than the E-383-^0 membrane. The reasons for the incom-
plete cleaning achieved in this test with the Biz treatment are not
clear, but the nature of the deposit formed from the secondary effluent
used appears to have been the root of the problem. More frequent clean-
ing during this test would probably have been more effective.
The crosslinked CAM-360 membrane gave the best performance, in every
respect, of the three membrane types evaluated in Test No. 6. CAM mem-
branes appear to be very attractive candidates for further development
in the renovation of wastewater. The greater rate of fouling and lesser
ease of cleaning observed with the E-383-^-0 cellulose acetate membranes
could be attributed to both their lower ds and higher initial flux,
compared with the CAM and blend membranes. Clarification of these points
will require further work.
-------
VI. MEMBRANES RESISTANT TO COLLOID FOULING
METHODS OF ATTACHMENT OF ENZYMES TO CELLULOSIC MEMBRANES
Several coupling reactions were available for attaching enzymes to
reactive polymers. Two types of reactive polymers which offered ver-
satile intermediates were investigated, namely, carboxyl-substituted
and arylamine-substituted cellulose acetates. The carboxyl groups
could be converted to an activated ester and coupled with enzyme amino
groups, while arylamino groups could be converted either to reactive
isothiocyanates or diazonium salts, which would couple with.enzyme amino
or phenolic groups, respectively.
An activated ester group which appeared to fit .the requirements of water-
stability and mild reaction conditions was reported by Anderson and co-
workers, 5 who utilized N-hydroxysuccinimide esters, of amino acids for
peptide synthesis. Amide formation between an amine and the activated
ester occurs in aqueous solution at pH .8.0, conditions which are ideal
for enzyme attachment. Accordingly, cellulose acetate hydrogen succinate
(CAHS) was prepared by treating 212 g of cellulose acetate (Eastman Type
E-400-25, acetyl ds 2.38) with 24 g of succinic;anhydride in It of pyri-
dine. The product was purified by precipitation from 2i> of acetone in 7^
of 50$ aqueous methanol; the succinate ds values determined for two batches
were 0.17 and 0.15. The CAHS was converted to its N-hydroxysuccinimide
ester (CAHS-NHS) by treatment of a.stirred solution of 10 g pf CAHS (6.5
meq of COOH) in 250 ml of acetone with 0.75 g (6.5 ramole)-"'of N-hydroxy-
succinimide and 1.5 g (7-^- mmole) of dicyclohexylcarbodiimide at room tem-
perature for 16 hours. The NHS content was estimated by reaction of the
polymer with tryptamine. An ultraviolet absorption spectrum of the
product indicated that complete esterification of the hydrogen succinate
groups had occurred. The reaction scheme for the preparation of this
reactive support polymer is outlined below.
^OAc
*- Gel]
\ / pyridine
XOH ^
OAc
CellT ' + HONX I dicyclohexylcarbodiimide
" _ acetone
C02H
.OAc
i
0 0
53
-------
Aminoaryl-substituted support polymers were prepared by_acylation of cel-
lulose acetate with a nitroaryl acid chloride and reduction of the nitro
group to an amine. Sixty grams of cellulose acetate Type E-400-25 was
acylated with 18 g of p-nitrobenzoyl chloride in 500 ml of dioxane and
25 ml of pyridine at 100° for 5 hours. The p-nitrobenzoyl ;(ENB) ds of
the resulting polymer, found by ultraviolet spectrophotometry, was 0.28.
An earlier batch, prepared using an impure sample of p-nitrobenzoyl
chloride, had a 'PNB ds of 0.08.
Reduction.of 10'g of>the lower ds material with 2,g oft sodium hydrosulfite
in 150 ml of water at 25 C produced cellulose acetate p-aminobenzoate
(CAPAB) with a PAB:ds;of.0.05.- The product from a'similar reduction of
the higher ds .material was not sufficiently soluble in dioxane (or other
suitable solvents) to permit determination of the PAB ds by ultraviolet
spectrophotometry. Acylation of 60 g of E-400-25 cellulose acetate with
22 g of 4-(,V-nitrophenyl)butyryl chloride in 500 ml of dioxane and 25 ml
of pyridine. produced"cellulose acetate 4-(4'-nitrophenyl)butyrate (CAPNPB)
with a PNPB.
-------
ATTACHMENT OF ENZYMES TO BULK POLYMERIC SUPPORTS
Before preparing membranes having attached enzymes, the methods of attach-
ment were investigated by coupling of the enzymes with the bulk support
polymers. It has been generally noted, in studies of insoluble enzymes,
that a greater reduction in activity (compared with the native enzyme) g
is observed with macromolecular substrates than with monomeric substrates.
This effect is generally ascribed to steric hindrance by the polymeric
support. During investigation of methods of attachment, the resulting
insoluble enzymes were assayed by kinetic procedures utilizing a monomeric
substrate; this allowed a rapid evaluation of the efficacy of the attach-
ment procedures.
After samples of native trypsin, chymotrypsin, and papain were assayed
for activity against monomeric substrates, it appeared that trypsin could
be assayed most conveniently and with the greatest sensitivity. Trypsin
was therefore used in the majority of insoluble enzyme preparations, and
the activity was measured by ultraviolet spectroscopic determination of
the rate of hydrolysis of benzoyl arginine ethyl ester (BAEE).1" The
protein content of the insoluble enzyme preparations was determined by a
modification of Babeeb1s 2,4,6-trinitrobenzenesulfonic acid (TNBS) method
for monitoring protein hydrolysis.IT The insoluble enzymes were hydrolyzed
in 6 N hydrochloric acid, and an aliquot was neutralized and treated with
TNBS at pH 9.0. After acidification, the concentration of the colored
reaction product was determined spectrophotometrically. Calibration
curves for this analytical method were prepared using known concentrations
of each enzyme.
When 50 mg of trypsin was stirred with 500 mg of CAHS-NHS in 10 ml of 1%
aqueous sodium bicarbonate solution at 25°C for k hours, an insoluble
enzyme containing 3«2$ protein was obtained. The activity of this mate-
rial, as measured by BAEE assay, was 6% of the activity of a like amount
of native trypsin. When the reaction temperature was reduced to 0 to U°C
the protein content was almost the same, 2.8$ but the activity of the
insoluble enzyme was 13-1T$ of that of native trypsin.
Since a proteolytic enzyme such as trypsin loses activity in solution due
to self-digestion, it was thought that even greater retention of enzymatic
activity might be obtained by inhibiting the enzyme during the attachment
reaction. When attachment was carried out at 0 to U°C in the presence
of 0.1 M benzoyLarginine, which inhibits BAEE hydrolysis by trypsin, the
apparent protein content of the insoluble enzyme was k.Qfjo. Based on the
observed h.0% protein content, the enzymatic activity measured by BAEE
was 7$ of that of native trypsin. (The TNBS reagent used to determine
protein content measures the concentration of free amino groups. If
benzoyl arginine reacted with the CAHS-NHS, arginine would be liberated
upon acid hydrolysis, and an erroneously high apparent protein content
would be observed.) Because of these poor results, no further investi-
gation of the use of inhibitors during attachment was carried out.
The preparation of insolubiiized chymotrypsin and papain was carried out
by reaction of the native enzymes at room temperature with bulk CAHS-NHS.
55
-------
These preparations were attempted before the discovery, with trypsin,
that lower reaction temperatures (i.e., 0 to U°C) resulted in greater
retention of enzymatic activity after attachment. The insoluble CAHS-
chymotrypsin obtained was found to contain ^.5$ protein; its activity
was 2.8% of the activity of the native enzyme, as measured by hydrolysis
of tyrosine ethyl ester. 16 The sample of native papain used was found to
contain only 24$ protein, and the sample of insoluble CAHS-papain obtained
contained only 0.8$ protein. In view of the low purity and activity toward
BAEE hydrolysis of the native papain, no attempt was made to assay the
activity of the insoluble enzyme. Although a purified native papain would
probably give improved results, further efforts in the preparation of an
enzymatic membrane were limited to trypsin, as described below.
Although the lack of solubility of the aminoaryl derivatives of cellulose
acetate in the usual membrane casting solvents may present some difficulty
in preparing membranes from these polymers, an insoluble enzyme was pre-
pared by diazonium coupling of CAPAPB to trypsin. The protein content
of this material was 2.1$; the activity, compared with native trypsin,
was nil (probably less than 0.05$). This lack of activity may have
resulted from the inactivation of trypsin by the polymeric diazonium
salt, a low enzymatic activity having been reported for insoluble trypsin
prepared by coupling with a diazotized copolymer of p-amino-DL-phenyl-
alanine and L-leucine.° Diazonium coupling, therefore, is apparently not
a feasible method for preparing trypsin membranes.
Attachment of an enzyme to an insoluble carrier frequently results in a
shift in the pH optimum (the pH at which maximal activity occurs). The
shift may be to either more positive or more negative values, depending
on the electrostatic charge on the carrier. Since hydrolysis of N-hydroxy-
succinimide ester groups which were not involved in binding the enzyme
would result in a negative charge on the carrier, it was of interest to
determine whether a shift in optimum pH had occurred. The variation with
pH in activity of the insoluble trypsin, relative to native trypsin at
pH 8.0, is shown by the following data.
pH Activity
7.0 7$
8.0 17$
9.0
The optimum pH for insoluble CAH5- trypsin is apparently between 7 and 9,
not significantly different from that of the native enzyme. 1°
To provide a measure of the effectiveness of insoluble trypsin in hydroliz-
ing proteinaceous colloids, a sample of the insoluble enzyme ( CABS- trypsin)
was assayed with casein as substrate, using the TUBS procedure of Habeeb.17
The sample of insoluble trypsin which had shown an activity toward BAEE
hydrolysis of 13 to 17$ of that of the native trypsin was 2.9$ as active
as the native enzyme when assayed with casein. This lower activity toward
a macromolecular substrate is in agreement with literature reports on other
-------
insoluble enzjrme preparations.
PREPARATION OF ENZIME-COUPLED MEMBRANES
A membrane intended for enzyme coupling was prepared by machine casting
at room temperature from a solution containing (by weight) CAHS-NHS, 10
parts; propionamide, 7 parts; maleic acid, 3 parts; dioxane, 20 parts;
acetone, 20 parts. This membrane was gelled at 0°C after less than 5
sec drying, time and stored under methanoi at 5°C until ready for use.
Reaction of a sample of the CAHS-NHS membrane with tryptamine in the
same manner described previously for the bulk polymer, indicated that
j4$> of the NHS groups remained intact after casting. A second portion
of the membrane was annealed in water for 3 win at 7^°C; tryptamine
analysis of this sample indicated that 43$ of the NHS groups originally
present in the polymer remained after both,casting and annealing. These
experiments, together with the successful attempts at enzyme coupling
described below, demonstrated that although the very reactive CAHS-NBS
polymer was susceptible, to hydrolysis, membranes could be fabricated from
it under normal conditions which contained sufficient reactive sites for
enzyme attachment.
2 -
An enzyme-coupled membrane was prepared by immersing a 27.5 c^ specimen
of the unannealed CAHS-NHS membrane in a solution of 50 mg of trypsin in
5 ml of 2$ aqueous sodium bicarbonate for 70 hrs at 4°C. .Following this
treatment, the membrane was thoroughly washed, in succession, with 0.001
N hydrochloric acid, 0.01 M borate buffer (pH 9«P) and water. The result-
ing membrane was found to contain -1.1$ protein. Apportion of the mem-
brane which had been left in contact with the trypsin solution for 10
days before washing was found to have a protein content of 1.4$, indi-
cating that most of the enzyme attachment was complete after 70 hrs.
A sample of the enzymatic membrane prepared by reaction with trypsin for
70 hrs was cut into 1 to 2-mm squares, and its enzymatic activity assayed
with BAEE. The activity of enzyme attached to the membrane toward BAEE
hydrolysis was found to be 2.5$ of that of the native trypsin. After 4.
days in aqueous suspension (pH ca. 6, ,25°C) at room temperature, the
enzymatic activity was 1.7$ compared to native trypsin (68$ of its ini-
tial activity). Bar-Eli and Katchalski-^ have examined the stability
of trypsin solutions as a function of pH. They found that a solution of
trypsin containing 25 g enzyme per ml lost 60$ of its activity in 25
hours at pH 6 and 25°C. Their result indicates that native trypsin would
retain only 2 to 3$ of its activity under conditions where the enzyme
membrane retained 68$ of its initial activity. It is clear from this
experiment that attachment to the CAHS-NHS membrane stabilized the trypsin.
toward loss of activity, which presumably occurs by self-catalyzed hydrol-
ysis. Such a stabilizing effect has been observed for other insoluble
enzyme preparations, including other trypsin derivatives. These results
indicate that enzymes having greater stability than trypsin should be
investigated for attachment to membranes,, in order to.insure maximum
retention of-activity of enzymatic membranes during long periods of
reverse-osmosis operation..
57
-------
The trypsin membrane described above was prepared with both sides of the
membrane exposed to the trypsin solution. For prevention of colloid
fouling, only enzymes on the active layer side will be effective. Fur-
thermore, the substructure of the enzyme is much more porous than the
active layer, and enzymes could possibly have penetrated and become
attached within the substructure. Interior enzymes would contribute to
the observed protein content, but might be less reactive than surface
enzymes since the substrate would have to diffuse into the membrane and
products would have to diffuse out. To examine an enzyme membrane with
essentially only surface enzymes a portion of membrane was masked with
Mylar film and waterproof tape to allow only the active layer to contact
the trypsin solution. The resulting membrane contained 0.22$ protein;
its activity relative to native trypsin was 1.8$, slightly less than the
membrane with trypsin on both sides. The1 lower activity of the membrane-
bound enzyme relative to the enzyme attached to bulk polymer under the
same conditions can probably be ascribed to the effect of surface area.
Manecke has reported^-9 that increasing the surface area of synthetic poly-
mers used as enzyme carriers increases the amount of enzyme bound, and
also increases the reaction rate of the insoluble enzyme with its sub-
strate. Thus, trypsin attached to the membrane, especially on the active
layer side with its smaller surface area, is somewhat less reactive than
trypsin attached to the fibrous bulk polymer.
Verification that trypsin had become attached to both sides of the earlier
membrane arid only one side of the latter one, was obtained by staining
both membranes, with Pink RL dye, a protein-specific stain. Excess dye
was removed with ethanol and sections of the two membranes were examined
at ^20X magnification under reflected light. The pink dye was concentrated
at both -surfaces of the first membrane, with only a faint color in the
interior. In the second membrane (active layer side only), however, the
dye was concentrated at the active layer surface, with essentially no
color at;the other surface and very little color in the interior.
A demonstration of hydrolysis by the enzyme-coupled membrane of a colloidal
protein/ such as casein, was not practical. Based on the BAEE activity of
the enzymatic membranes and the bulk enzymes, hydrolysis of casein by either
membrane would -not be measurable by the assay procedures available. It has
been demonstrated (vide supra) that bulk CABS-trypsin is capable of effect-
ing hydrolysis of a model proteinaceous colloid, namely casein. Moreover,
the trypsin membranes have been shown to possess enzymatic "activity by'
hydrolysis of"BAEE. The colloid concentrations to be found in secondary
effluent probably do not exceed 5 rog/^j^ a concentration which is one or
two orders of magnitude below the sensitivity of the methods used to assay
casein hydrolysis. -While the enzymatic activity on the active layer is
small, a small activity may well be sufficient to prevent fouling by waste-
water colloids at these low concentrations. Tests of the enzymatic mem-
brane with sewage are required to answer this question.
MEMBRANES FROM CELLULOSE ACETATE HYDROGEN SUCCINATE
Prior to preparing enzyme membranes, preliminary membrane formulation
-------
studies were conducted with cellulose acetate hydrogen succinate (CAHS).
These were not extensive enough to optimize the formulation, but did
provide a formulation adequate for the highly swollen membranes used in
preparing trypsin membranes. Some reverse osmosis data were also
obtained on these membranes which may indicate another method of prevent-
ing fouling by colloids.
The best reverse osmosis performance with CAHS membranes was attained
with a formulation containing (by weight): CABS, 20 parts; propionamide,
14 parts; maleic acid, 6 parts; dioxane, k$ parts; and acetone, ko parts.
Membranes were cast from this solution at room temperature with less than
5 sec drying time and gelled at 0°C. After annealing at fk°C this mem-
brane exhibited a flux of 13 gfd with 89$ rejection when tested at 600
psi with 1000 mg/Jt sodium chloride feed solution. After two hours on
secondary effluent the flux was 46 gfd with 72$ conductivity rejection.
This increase in flux was accompanied by a change in the recirculating
feed from an initial pH of 7.0 to pH 8.2. After 3 hours the pH was
adjusted with 6 N sulfuric acid to pH 4.9. During the next 18 hours the
pH rose to 6.8, with a flux of 27 gfd at 78$ rejection. After 2k hours
the membrane was subjected to a Biz wash. After the Biz wash the flux
was increased only to 32 gfd with 32$ rejection, declining to 18 gfd and
21$ rejection over a period of three days. Visual examination after the
test showed very little of the slimy deposits generally accumulated in
that length of time.
The large changes in flux and rejection observed during the above test
can probably be ascribed to the effect of the CAHS carboxyl groups. The
charge density on the membrane would be expected to increase with increas-
ing pH in the pH range encountered above, which would alter the permeabil-
ity to water and ionic species. The increase in pH which occurred during
the test is also probably due to the presence of carboxyl groups.
The highly alkaline conditions of the Biz wash apparently resulted in
degradative damage to the active layer, as evidenced by the large decrease
in conductivity rejection.
To ascertain whether the negatively charged CAHS membrane might have inher-
ent antifouling characteristics another membrane from a similar formulation
(using 10 parts of acetamide instead of propionamide) was tested for 16
days on Pomona secondary effluent, maintaining the feed pH at 4.7 to 5.5
by periodic addition of dilute sulfuric acid. Initially the flux was 15.5
gfd, with 80$ conductivity rejection. After 6 days the flux was unchanged,
and the rejection was 87$. The flux dropped to 13.5 gfd after 8 days and
to 6 gfd after 16 days. The complete absence of flux decline over the
first 6 days, however, suggests that the CAHS membrane may have inherent
fouling resistance. The colloids involved in fouling may be negatively
charged at the pH maintained during this test, and would thus be repelled
by a membrane of like charge. This possibility should be pursued further.
59
-------
REFERENCES
1. D. L. Hoernschemeyer, e_t al., Research and Development of New and
Improved Cellulose Ester Membranes, Aerojet-General Corporation,
Report No. 1319-F, a final report to the Office of Saline Water,
USDI, Contract No. 14-01-0001-2205, January 1970.
2. C. W. Saltonstall, Jr., et al., A Study of Membranes for Desalina-
tion by Reverse Osmosis, Aerojet-General Corporation, Report No.
3292, a final report to the Office of Saline Water, USDI, Contract
No. 14-01-0001-338, October 1966; Office of Saline Water Research
and Development Progress Report No. 232.
3. U. Merten, et^al., Reverse Osmosis Membrane Research, General Atomic
Division of General Dynamics, Report No. GACD 7192, a quarterly
report to the Office of Saline Water, USDI, Contract No. 14-01-
0001-767, May 1967.
4. P. A. Cantor, et al., Research and-Development of New and Improved
Cellulose Ester Membranes, Aerojet-General Corporation, Report No.
3343-F, a final report to the Office of Saline Water, USDI, Contract
No. 14-01-0001-1732, January 1969; Office of Saline Water Research
and Development Progress Report No. 434.
5. R. B. Dean, et al., Environ. Sci. Tech., !_, 147 (1967).
6. I. H. Silman and E. Katchalski, Ann. Rev. Biochem., 35, 873 (1966).
7. A. White, P. Handler, E. L. Smith, and D. Stetten, Jr., "Principles
of Biochemistry," 2nd Ed., McGraw-Hill, New York, 1959, p. 152.
8. R. Axen, J. Porath, and S. Ernback, Nature, 214, 1302 (1967).
9. R. Goldman, et al.^ Biochemistry, 7, 486 (1968).
10. Development of Technology for Continuous Production of Low-Cost
Reverse-Osmosis Modules, Aerojet-General Corporation, Report No.
F1301-01, a final report to the Office of Saline Water, USDI,
Contract No. DI 14-01-0001-2146, November 1969.
11. Reverse Osmosis as a Treatment of Wastewater, Aerojet-General
Corporation, Report No. 2962, a final report to the U. S. Public
Health Service, Contract No. 86-63-227, January 1965.
12. P. A. Cantor, et_ al., Development and Scale-up of New and Improved
Cellulose Ester Membranes, Aerojet-General Corporation, Report No.
4865-01-F, a final report to the Office of Saline Water, USDI,
Contract No. 14-01-0001-1767, May 1969.
60
-------
13. W. M. King, et al.j High Retention Tubular Membranes for Reverse
Osmosis, Aerojet-General Corporation, Report No. 1372-8, a monthly
report to the Office" of Saline Water, USDI, Contract No. 1^-01-
0001-222^, February 1970.
14. Reverse Osmosis Renovation of Municipal Wastevater, Aerojet-General
Corporation, a final report to the Federal Water Pollution Control
Administration, USDI, Contract No. 1^-12-184, December 1969.
15. . .G. W. Anderson, J. E. Zimmerman, and F. M. Callahan, J. Am. Chem.
Soc., 86, 1839
16. W. Rick, in H. Bergmeyer, Ed., "Methods of Enzymatic Analysis,"
Academic Press, New York, 1962, pp. 800-823.
IT. A.F.S.A. Habeeb, Arch. Biochem. Biophys., 119, 264 (1967).
18. A. Bar-Eli and E. Katchalski, J. Biol. Chem., 238, 1960 (1963).
19. G. Manecke, Biochem. J., 107, 2 P (1968).
61
-------
APPENDIX
EFFECT OF SULFATE ON CHLORIDE REJECTION
IN MIXED FEEDS
CONTENTS
Table Page
A-l EFFECT OF FEED COMPOSITION ON REJECTION BEHAVIOR
OF CAM MEMBRANE 6k
A-2 EFFECT OF ANNEALING TEMPERATURE ON REJECTION
CHARACTERISTICS OF BLEND MEMBRANES 65
62
-------
APPENDIX
EFFECT OF SULFATE ON CHLORIDE REJECTION IN MIXED FEEDS
In brackish-water tests carried out several years ago, abnormally low
nitrate rejection was observed in feed waters containing sodium nitrate
in the presence of a large excess of calcium sulfate. It was considered
of practical interest, during this program, to determine the extent to
which this effect applied to sodium chloride rejection in mixed feeds
containing sodium chloride and sodium sulfate since such feeds were used
In evaluation of membranes in the program.
A crosslinked QAM-SoO membrane, cast at -1Q°C, and annealed at 80°C,"was
tested at 200 psi against a series of feeds, all of which had the same
osmotic pressure, in which the ratio of sodium=sulfate to sodium chloride
was varied. Included in the series were feed solutions containing either
solute alone. The results, summarized in Table A-l, indicate a strong
effect of sulfate on chloride rejection, the chlori'de rejection decreasing
with increased sulfate-to-ohloride ratio in the feed. The chloride rejec-
tion was observed to decrease from a value of 85$ with the feed containing
sodium chloride alone to 56$ with the feed containing the highest ratio
of sulfate to chloride. The sulfate rejection also was affected by varia-
tion in feed composition, increasing with increased sulfate-to-chloride
ratio,
A second series of tests was carried out to determine to what extent the
mixed ion rejection effect was a function of membrane annealing tempera-
ture, To this end, membranes were cast from a 1:1 blend (ds 2,63) for-
mulation, annealed at various temperatures and tested at 150 psi succes-
sively against the following feed solutions:
Feed No. 1 - 250 mg/4 NaCl, 750 mg/^ NagSO^
Feed No. 2 - 1000 mg/C NaCl
Feed No. 3 - 1000 mg/i, Na2SOj^
The results of these tests,_summarized in Table A-2, indicate, that the
extent to which.sodium^chloride rejection decreases in the presence of
sodium sulfate .4s greiater the "lower the annealing temperature for a
given membrane formulation../. With, the, unannealed membrane,, sodium
chloride rejection fell from.a value,of 52$ with pure sodium chloride
solution (feed No, 2) to 11$ with the mixed feed (feed No, l), a decrease
of 78%. The percentage decrease fell consistently as the annealing tem-
perature was increased, so that with the 70° C membrane, sodium chloride
rejection was reduced from 93$ with the sodium chloride feed to 85$ with
the mixed feed, a drop of only 10$,
-------
TABLE A-l
EFFECT OF FEED COMPOSITION ON REJECTION
BEHAVIOR OF CAM MEMBRANE
Feed Composition,
U
o
6ko
1500
2500
3500
Flux,
gfd
32.7
36.8
39.7
38.4
35.8
NaCl
Rejection,
85.2
7^.0
70.0
55-7
Rejection,
90.6
9^.2
97.3
96.1
Measured at 200 psi after 2-hour test period using membrane annealed
at 80°C. The casting formulation was as follows (amounts of ingredi-
ents in parts "by weight): CAM-360, 10; acetone, 4o; maleic acid, I1*-;
water, 10. Crosslinking was carried out after annealing by immersion
of the membranes for 3 min at 90°C in a solution containing 0.037 M
of potassium persulfate and 0.038 M of sodium bisulfite.
-------
TABLE A-2
EFFECT OF ANNEALING TEMPERATURE ON REJECTION
CHARACTERISTICS OF BLEND MEMBRANES
a
Annealing
Temp., °C
25
(unannealed)
50
60
70
Feedb
1
2
3
1
2
3
1
2
3
1
2
3
Flux,'
gfd
35.8
36.4
36.2
22.3
23.2
22.5
1^.2
13 .^
13.7
8.7
8.1
8.3
NaCl
Rejection,
11.3
52.0
38A
68.8
68.1
83.8
93.1
Rejection,
%
99-9
98.1
99.7
98.6
99.9
98.3
99.7
98.0
% Reduction of
NaCl Rejection
by Na2SO^
78.3
19-3
10.0
^mbrane formulation (parts by vt) - E-398-3, 10; A-432-130B, 10; dioxane, 55; acetone, 35;
methanol, 9; maleic acid, 3.
Feed No. 1 - 250 mg/t> sodium chloride, 750 mg/4"sodium,sulfate
Feed No. 2 - 1000 mg/A sodium chloride
Feed No. 3 - 1000 mg/6 sodium sulfate
"Determined at 150 psi after 2-hour test periods.
-------
The reduction of chloride and nitrate ion rejections in the presence
of other ions, such as sulfate, does not appear to be of serious con-
sequence in the treatment of waste-water by reverse osmosis because
the rejections obtained are still high enough to produce product water
with acceptably low concentrations of these ions* The .rejection of
organic components, furthermore, should not be affected by the presence
of ionic components, since the former are, in almost all cases, non-
ionic in nature.
66
-------
« 1 Accession Number
2
Subject Field St Group
05D
SELECTED WATER RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
Organization
Envirogenics Company, a division of Aerojet-General Corporation,
El Monte, California
Title
NEW TECHNOLOGY FOR TREATMENT OF WASTEWATER BY REVERSE OSMOSIS
10
Fisher, B. S.
Lowell, J. R., Jr.
16 Protect DesigMlion p^ program # 17020 DUD.
Contract # llt-12-553
2] I Note
Citmtlon
Descriptors (Stfrrml First)
^Reverse osmosis, *Semipermeable membranes, *Sewagor treatment, *Wastewater treatment,
*Tertiary treatment, Demineralization, Membranes, Membrane processes, Sewage
effluents, Water purification, Water treatment, Desalination processes
25 identifier* (Slurred Finn *Membrane fouling, *Proteolytic enzymes, *Enzymatic membranes,
*Wastewater renovation, Enzymes, Membrane cleaning, Enzymatic detergent.
27! Abstract stable, high-flux membranes were sought for use in the renovation of wastewater
—by reverse osmosis. Cellulose ester membranes were formulated to produce fluxes
greater than 60 gal/ft -day which would not decrease by more than 20$ after the first
year of operation, and reject at least 60$ of sodium chloride and 95$ of sodium sulfate
when tested at 600 psi with 1000 ppm feed solutions. The target osmotic performance was
achieved with each of three membrane types: A cellulose diacetate of moderately-low
acetyl content, a cellulose triacetate-diacetate blend, and crosslinked cellulose ace-
tate methacrylate. The intrinsic flux stabilities of these membranes extrapolated to
flux losses of only 12 to 18$ after the first year of operation. The fluxes of these
high-performance membranes declined rapidly in bench-scale tests with secondary sewage
effluent but were restored to within 80 to 90$ of the initial values after cleaning with
an enzymatic laundry presoak (Biz). Daily cleaning by this technique maintained the
fluxes at a nearly constant level over a 5-day test period. The rejection of sewage com-
ponents by the high-flux membranes was excellent. Techniques were explored for attach-
ment of proteolytic enzymes to cellulose acetate membranes to render them resistant to
colloid fouling. The proteolytic enzyme trypsin was chemically attached to the active
layer surface of a membrane prepared from the N-hydroxysuccinimide ester of cellulose
acetate hydrogen succinate. The resulting enzymatic membrane displayed hydrolytic
activity.
B. S. Fisher
institutionHSnvirogenics Company, a division ol
Aerojet-General Corporation
------- |