WATER POLLUTION CONTROL RESEARCH SERIES • ORD-17O4OEFOO6/7Q
MEMBRANE MATERIALS
FOR
WASTEWATER RECLAMATION
BY
REVERSE OSMOSIS
U.S. DEPARTMENT OF THE INTERIOR • FEDERAL WATER QUALITY ADMINISTRATION
-------
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, develop-
ment, and demonstration activities in the Federal Water
Quality Administration, in the U. S. Department of the
Interior, through inhouse research and grants and con-
tracts 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, Department of the Interior, Federal Water
Quality Administration, Room 1108, Washington, D. C. 20242.
-------
MEMBRANE MATERIALS FOR WASTE WATER RECLAMATION
BY REVERSE OSMOSIS
by
A. S. Douglas
M. Tagaml
C. E. Mllstead
Gulf General Atomic
San Diego, California 92112
for Che
FEDERAL WATER QUALITY ADMINISTRATION
DEPARTMENT OF THE INTERIOR
Program #17040 EFO
Contract #14-12-452
FUQA Project Officer, C. A. Brunner
Advanced Waste Treatment Research Laboratory
Cincinnati, Ohio
June, 1970
-------
FWQA Review Notice
This report has been reviewed by the Federal
Water Quality Administration and approved for
publication. Approval does not signify that
the contents necessarily reflect the views
and policies of the Federal Water Quality
Administration.
For sale by tbe Superintendent of Documents, U.S. Government Printing Office
Washington, D.C. 20402- Price 65 cents
-------
ABSTRACT
An experimental program was carried out to evaluate potential reverse
osmosis membranes for the tertiary treatment of secondary sewage effluent.
The evaluation program consisted of both direct osmosis and reverse osmosis
tests on various membranes using both single solutes and secondary effluent.
The types of membranes tested were polyurethane latices, cellulose diacetate,
cellulose 2.5-acetate, polyvinylpyrrolidone (PVP)-polyisocyanate interpolymers,
and polyelectrolytes.
The cellulose diacetate, polyurethane latices, and PVP membranes were
not suitable for waste-water treatment. Although rejection of most solutes
by PVP membranes was quite good for dense, homogeneous membranes, high-flux,
asymmetric membranes had poor rejection characteristics. Very-high-flux
membranes were prepared from polyacrylic acid cast onto cellulose nitrate,
cellulose nitrate-cellulose acetate, and polysulfone porous supports. While
2
NaCl rejection of over 80% was obtained at fluxes of less than 20 gal/ft -day
at 600 psi, increased flux was obtained only at the expense of good rejection.
The best combination of high flux and rejection was obtained with asymmetric
cellulose 2.5-acetate membranes heat-treated in water at 55° to 70°C.
This report was submitted in fulfillment of Contract 14-12-452 between the
Federal Water Quality Administration and Gulf General Atomic.
iii
-------
CONTENTS
Abstract ill
Section 1 Conclusions and Recommendations 1
Section 2 Introduction 3
Section 3 Experimental .... 7
Section 4 Results and Discussion 23
Section 5 Acknowledgment 63
Section 6 References 65
FIGURES
1. Direct osmosis equipment for monitoring electrolyte and water
transport 15
2. Time lag experiment with NaCl 17
3. Flow diagram of reverse osmosis system 20
4. Flux-rejection data for 300,000-MW PAA membranes on CN/CA
and polysulfone porous supports . 27
5. Flux-rejection data for 10 -MW PAA membranes on CN/CA and
polysulfone porous supports 35
6. Flux-rejection data for 10 -MW PAA membranes on CN porous support . 36
TABLES
1. Transport properties of PAA membranes prepared from 300,000-MW PAA 24
2. Transport properties of asymmetric cellulose 2.5-acetate membranes 28
3. Transport properties of PAA membranes prepared from 10 -MW PAA
on CN/CA porous support 29
4. Transport properties of PAA membranes prepared from 10 -MW PAA
on polysulfone porous support 30
5. Transport properties of PAA membranes prepared from 10 -MW PAA
on CN porous support 32
6. Transport properties of 10 -MW PAA-CN and cellulose 2.5-acetate
membranes 38.
7. Transport properties of 10 -MW PAA-polysulfone and cellulose
2.5-acetate membranes 39
8. Restoration of PAA membranes 40
9. Rejection of filtered secondary effluent by 300,000-MW
PAA-CN/CA membranes 42
iv
-------
10. Rejection of filtered secondary effluent by 300,000-MW
PAA-CN/CA and cellulose 2.5-acetate membranes 43
11. Rejection of secondary effluent by 10 -MW PAA-CN and
cellulose 2.5-acetate membranes 44
12. Rejection of filtered secondary effluent by 10 -MA
PAA-CN and cellulose 2.5-acetate membranes 45
13. Rejection of filtered secondary effluent by 10 -MW
PAA-polysulfone and cellulose 2.5-acetate membranes 47
14. Transport properties of polyelectrolyte membranes cast on CN
porous support 51
15. Permeability of PVP-PMPI and cellulose 2.5-acetate membranes
to water and various solutes 52
16. Effect of THF on properties of PVP-PMPI membranes 56
17. Effect of solution age and composition on the properties
of PVP-PMPI membranes 56
18. Effect of pyridine on PVP-PMPI membrane properties 57
19. Effect of heating on PVP-PMPI membrane properties 57
20. Effect of TEP on PVP-PMPI membrane properties 59
21. Effect of amyl acetate on PVP-PMPI membrane properties 60
22. Transport properties of cellulose diacetate 60
-------
SECTION 1
CONCLUSIONS AND RECOMMENDATIONS
The most apparent conclusion to be drawn from this program is that, of the
membranes surveyed, asymmetric cellulose 2.5-acetate, heat-treated in water
at 55° to 708C, was the best overall membrane for waste-water reclamation
with regard to both flux and rejection characteristics.
Polyacrylic acid (PAA) membranes were prepared with transport properties
that varied over a wide range, but in the flux range of interest for large-
scale applications, the rejection needs to be improved. In general, heating
the membranes in air at 80° to 110°C reduced the water flow and increased
salt rejection. The reproducibility among PAA membranes was poor, and severe
flux decline was in evidence over the first two days of operation. Attempts
to restore membrane properties in situ to increase the flux characteristics
produced only transient effects; within a few hours the flux had begun to
decrease again. Efforts to form a dynamic PAA membrane on the surface of a
cast PAA membrane resulted in no beneficial effects.
PVP-polyisocyanate interpolymer membranes had good rejection properties when
the mole ratio of isocyanate to PVP was 0.9 or higher. However, high-flux
asymmetric membranes of this material generally had poor rejection ability,
and the rejection was not improved by heating the membranes.
Neither the cellulose diacetate nor the polyurethane latex membranes studied
in this program were suitable for treatment of secondary effluent. These
membranes generally showed poor transport characteristics.
Of the new membrane materials studied, only PAA shows any promise for
commercial use and may be worth additional investigation. The best of these
membranes had flux and rejection properties that were comparable to those of
asymmetric cellulose 2.5-acetate membranes. The lack of reproducibility
among nominally identical membranes is a severe problem and presumably is
-------
related to the casting technique and/or support properties. It is believed
that the preparation of larger sheets of membrane on a continuous basis would
improve this situation.
-------
SECTION 2
INTRODUCTION
With increasing emphasis being placed on the problem of environmental
pollution, new and improved methods for waste-water purification are in great
demand. Treatment methods are being sought not only to purify domestic
and industrial wastes to maintain or improve quality of the receiving waters,
but also to meet the need for multiple water reuse that is expected to become
vital in some areas in the 1980s. Municipal waste treatment facilities
currently in operation generally utilize primary and secondary treatment
steps. Primary treatment usually consists of settling to remove suspended
solids, while secondary treatment involves the biochemical degradation of
organic compounds. The secondary effluent may then be chlorinated and is
finally discharged. Generally, a 90% (1) biochemical oxygen demand (BOD)
removal is effected from an initial value of about 150 mg/liter in the raw
sewage (2). The effluent also contains a significant amount of dissolved
inorganic salts (usually about 200 to 300 mg/liter higher than the local
water supply). In order to make this water suitable for reuse, both the
organic compounds and inorganic salts must be substantially reduced.
Several schemes for lowering the level of impurities present in sewage
effluent are currently under investigation. One process being studied uses
activated carbon to remove the organic material followed by electrodialysis
to remove the dissolved salts. Reverse osmosis has been proposed as another
possible technique for tertiary treatment. One significant advantage of this
process might be that both the organic material and the inorganic material
could be removed in a single step. Reverse osmosis is currently undergoing
field evaluation both at Pomona, California, and Lebanon, Ohio, under FWQA
sponsorship. In both of these evaluations, cellulose acetate-based membranes
limited to cellulose 2.5-acetate membranes made by the Loeb-Sourirajan tech-
nique (3) are being used.
-------
The purpose of the present laboratory study was to develop new membrane
materials that would be suitable for use with secondary effluent to provide
a product water with 95% of the organic matter removed as well as 70% to
80% of the dissolved electrolytes. This product might be suitable for use
as potable water.
Several different membrane materials were evaluated for use in waste-water
treatment in this study, i.e., polyelectrolytes, PVP-polyisocyanate inter-
polymers, cellulose diacetate, and latex materials. For successful applica-
tion, membranes must be effectively very thin (generally less than 1 p) to
allow a high water flux. (The Loeb-Sourirajan asymmetric cellulose acetate
membrane has this characteristic.) Thus, the problem of membrane development
is two-fold: (1) finding a membrane material with good transport properties,
and (2) preparing an effectively thin film. Several methods have been develop'
for preparing a membrane with these properties:
1. The technique of Loeb and Sourirajan (3) has received wide
application for preparing asymmetric membranes of cellulose 2.5-
acetate. In this procedure, the polymer is dissolved in a mixture
of a very volatile good solvent (acetone) and a poor solvent of low
volatility (magnesium perchlorate in water). The resulting solution
is cast on a glass plate (or steel belt); evaporation occurs for a
short time; and the membrane is immersed in water to remove the
remaining solvent and cause solidification of the structure. The
resulting membrane has a thin, dense skin on the air-dried surface,
and the rest of the structure is a porous open network. All the
resistance to flow appears to be in the thin skin.
2. A composite membrane is currently under development at Gulf General
Atomic (A)(5). This method allows the deposition of a thin
layer of cellulose acetate (typically about 1000 A thick) on the
surface of a finely porous support. Rejections of greater than
99.5% NaCl have been obtained with membranes of this type.
-------
3. Membranes with good transport properties have also been prepared
by employing a variation of the Oak Ridge National Laboratory (ORNL)
technique of dynamic membrane formation (6), i.e., depositing
a small amount of a salt-rejecting material into pores of a finely
porous support.
In the present study, high-flux membranes with substantial rejection of
both organic and inorganic materials were prepared from various polyelectro-
lytes, with PAA membranes receiving the most attention. The earlier work
at ORNL with dynamic PAA membranes had suggested that membranes cast on a
suitable porous support may have good transport properties. Subsequent
work at Gulf General Atomic (5) indicated that cast PAA membranes
had properties that were in some ways superior to the dynamically formed
membranes. Extensive testing of PVP-polyisocyanate interpolymers was
conducted using thick homogeneous films. This system was reported to have
good hydrolytic stability and should therefore have a significantly increased
lifetime compared with that of cellulose 2.5-acetate, especially in mildly
basic solution such as secondary effluent. Riley and co-workers (7) had
shown that a series of hydrophilic materials could be prepared from this
system with a wide range of water and salt permeabilities. Cellulose dia-
cetate membranes were prepared and tested, since earlier measurements by
Lonsdale and co-workers (8) had shown that the water permeability was
a factor of six higher than that of cellulose 2.5-acetate generally used
in reverse osmosis systems. The study of latex-based membranes was initiated
as a result of an observation made in this laboratory (9) that a membrane
prepared from Wyandotte E-207A latex provided more than 100 gal/ft2-day
(gfd) and 30% rejection using 1% NaCl feed at 1500 psi.
Membrane testing was performed using direct osmosis, reverse osmosis, and
a "desorption rate" technique (10). Desorption rate and direct osmosis
tests were used with homogeneous membranes to evaluate potential membrane
materials without having to prepare membranes with sufficient strength to
withstand the operating pressure of reverse osmosis. In particular, the
desorption rate technique was quite useful, because accurate values of solute
-------
diffusivity and permeability could be obtained even if a membrane contained
imperfections. Membranes that showed favorable properties in desorption rate
and direct osmosis measurements were then tested in reverse osmosis using
single solute feeds, and, in some cases, filtered secondary effluent.
-------
SECTION 3
EXPERIMENTAL
MEMBRANE PREPARATION
CELLULOSE 2.5-ACETATE
Homogeneous cellulose 2.5-acetate membranes were prepared from Eastman
E 398-10 cellulose acetate (15 wt-% in dioxane) by casting onto a glass
plate with a doctor knife. The membrane was enclosed in a Plexiglas box
to retard the rate of solvent removal, and the solvent was allowed to evaporate
overnight. Asymmetric membranes were made of the same material by the
method of Sourirajan and Govindan (11) from a mixture of cellulose
acetate, acetone, water, and magnesium perchlorate. The membrane was cast
on a glass plate at approximately 5°C and, after 3 to 4 min, was immersed
into ice water for about 1 hr. The membrane properties were then altered
by a heat treatment in water ranging from 55° to 80°C, depending on the flux
and rejection desired.
POLYELECTROLYTES
Membranes were prepared from several polyelectrolytes and evaluated in
reverse osmosis tests. These materials included:
1. American Cyanamid Company Cyanamer A-370 (modified sodium
polyacrylate)
2. American Cyanamid Company Cyanamer P-26 (acrylamide
copolymer)
*Mention of commercial products- does not imply endorsement by the Federal
Water Quality Administration.
-------
3. Stein-Hall Polyhall-295 (high-molecular-weight (MW) anionic
polyacrylamide)
4. Stein-Hall Polyhall-402 (high-MW nonionic polyacrylamide)
5. General Aniline and Film (polyvinyl methyl ether-maleic acid
copolymer)
6. Rohm and Haas Acrysol A-l and A-5 (PAA; 50,000 and 300,000
MW, respectively)
7. PAA, 106 MW (synthesized in this laboratory) (5)
The polyelectrolyte membranes were prepared on porous supports by both
dipping and casting from several concentrations in water solutions, ranging
from 0.04 to 25 wt-%. The dipping procedure consisted of preparing a dilute
(2% or less) solution of the polyelectrolyte in water and immersing a porous
support (see below) taped to a glass plate. The plate was withdrawn at a
controlled rate, then allowed to drain and air-dry. In some cases, membranes
were prepared by dipping several times, with air-drying between immersions.
Membranes were also cast from more concentrated (1% to 25%) polyelectrolyte
solutions onto similar porous supports and air-dried before use. The solution
was applied to the support by spreading with a glass rod. Thickness measure-
ments of eight membranes cast from 4.8 wt-%, 10 -MW PAA showed an average
thickness of the PAA film ranging from 2.4 to 5.6 y after air-drying and
heat-treating at 110°C. These measurements were made using a Federal thick-
ness gauge (Model Pll). No attempt was made to determine to what extent
the PAA penetrated the pores of the support. The uniformity of the membrane
was not believed to be critical, since much of the polyelectrolyte film
was washed off during testing in reverse osmosis. The effects on these mem-
branes of heat-treating in air were also studied.
-------
PVP-POLYISOCYANATE INTERPOLYMERS
Homogeneous PVP-polyisocyanate interpolymer membranes were prepared by
casting from a 7 to 15 wt-Z solution in chloroform. The PVP used in these
studies was General Aniline and Film K-90. Two isocyanates were studied:
polymethylene polyphenyl isocyanate, or PMPI (Upjohn Company PAPI); and
methylenebis-(4-phenyl isocyanate), or MDI (Matheson Coleman and Bell).
Studies performed in this laboratory showed that the isocyanates polymerize
or self-crosslink in the presence of water, forming an insoluble interpolymer
network with the PVP. In the case of the trifunctional PMPI, a three-dimen-
sional network is formed, whereas the difunctional MDI is restricted to the
formation of a linear polymer lattice. In some cases, the PMPI was used as
a stabilized material with the isocyanate groups blocked with e-caprolactam
(Upjohn Company Isonate 123 P). This material is destabilized at temperatures
above about 150°C, and the isocyanate groups are made availabl" to react with
water vapor to form a three-dimensional structure.
Membranes were prepared from solutions containing isocyanate-to-PVP molar
ratios in the range from 0.7 to 1.0. When blocked PMPI was used as the iso-
cyanate, which was the case for all homogeneous films, the membranes, after
drying in air, were heated slowly to 150°C and held at that temperature for
1 hr. The PVP-MDI membranes were cast under anhydrous conditions, and after
the solvent evaporated, water vapor was admitted at a slow rate to cause
reaction without rapid evolution of CO., which would result in the formation
of bubbles in the membrane. Details of these procedures are given in (7).
Asymmetric membranes were prepared by adapting the procedures of Loeb
and Sourirajan for cellulose 2.5-acetate membranes. Low-volatility
solvents were sought which would remain in the film when the chloroform
evaporated and which were relatively poor solvents themselves, so that
after the chloroform was gone the polymer would precipitate as a fairly
-------
porous open structure. Both short drying times and overnight drying were
used in an attempt to vary the film properties.
The unblocked PMPI was added to a solution of PVP in chloroform so that the
ratio of moles of PVP to equivalents of PMPI was 1:1. The weight ratio was
0.84 g PVP:1.0 g PMPI. The solid PVP and liquid PMPI were miscible in
chloroform at room temperature. A 10 wt-% stock solution was prepared to
which nonvolatile components could be added. The solution was kept anhydrous
to prevent the reaction of the isocyanate with water. To this stock solution
several miscible organic liquids were added* including tetrahydrofuran (THF),
pyridine, THF-pyridine mixtures, triethyl phosphate, and amyl acetate.
Solutions were prepared volumetrically. For example, a solution of 30% THF
was prepared by adding 30 volumes of THF to 70 volumes of the stock solution.
Following the Loeb-Sourirajan technique, the membranes were cast in the room
atmosphere at approximately 70°F and 50% relative humidity. After a drying
period ranging from 5 min to 24 hr, they were immersed in water at room
temperature. Simultaneously with the drying, the isocyanate reacted with
water vapor in the atmosphere to form the interpolymer network. The resulting
films were not uniform, but, in general, had a brownish air-dried surface
and an opaque-white bottom surface. The opacity indicates the presence of
light-scattering centers, suggesting the presence of porosity. Further
evidence of porosity was obtained from dye tests. The bottom surface absorbed
a water-soluble dye painted on it; the air-dried surface rejected the dye.
CELLULOSE DIACETATE
Homogeneous cellulose diacetate membranes were prepared from Eastman E 360
cellulose diacetate dissolved in reagent-grade pyridine. The membranes
were cast onto- a glass plate and were allowed to evaporate to dryness in a
closed box.
10
-------
LATEX
Several different latex materials were evaluated:
1. Wyandotte Chemical Company E-207A (nonionic polyurethane)
2. Wyandotte Chemical Company X-1017 (anionic polyurethane) (very
small particle size)
3. GAF Polectron 450 (PVP-styrene copolymer)
4. General Latex RA-150-7
Membranes were prepared from these latices by several different techniques.
One technique was to dilute them with water to the desired solids content
and then cast directly onto a cellulose nitrate-cellulose acetate porous
support. The water was evaporated to dryness, and permeability properties
were then measured in reverse osmosis tests. The effect of heat-treating
in air at several temperatures for various times was also examined. In a
second technique, the latex was formed into a solution by the addition of
ethanol, and this clear solution was cast onto the support. The solvent
was evaporated to dryness and testing proceeded as before. In some cases,
metal salts such as magnesium and calcium chlorides were added to the latex
in an effort to reduce the effect of the stabilizing surfactant and permit
a more continuous film to form.
v
POROUS SUPPORTS
Three types of support materials were prepared for use with polyelectrolyte
membranes: cellulose nitrate-cellulose acetate (CN/CA), polysulfone, and
cellulose nitrate (CN).
11
-------
CN/CA
The procedure for preparing this support has been described previously
(4). The casting solution consists of CN (du Pont DBA 35 E) and CA
(Eastman 383-40) in the weight ratio of 86:14 dissolved in a solvent system
consisting of acetone, ethanol, n-butanol, water, Rohm and Haas Triton
X-100, and.glycerin. The support is cast onto a glass plate and dried
in air at approximately 100% relative humidity. The resulting membrane is
finely porous with a bulk porosity of about 60%. The CN/CA porous supports
showed a water flux of approximately 700 to 3500 gfd when tested with dis-
tilled water in a low-pressure system at 40 psi.
Polysulfone
Polysulfone porous supports were prepared by casting a film of polysulfone
(Union Carbide 3500) in dimethyl formamide (approximately 6 to 20 wt-%)
onto a glass plate and rapidly immersing the film in water at approximately
5°C. Although at 40 psi the water flux of the wet membrane is comparable
to that of CN/CA, the air-dried support shows a considerable reduction in
flux, and in some cases no flux at all. This behavior is attributed to the
hydrophobic character of polysulfone (i.e., the lack of ability to readily
absorb water) and not to a reduction in porosity upon drying. Such flux
reduction is generally not a problem in reverse osmosis measurements because
of the higher applied pressure. (Polyelectrolyte membranes with initial
water fluxes greater than 300 gfd at 600 psi have been prepared using this
support material.)
CN
The preparation of finely porous CN membranes has been described by Baddour,
Vieth, and Douglas (12). The method consisted of dissolving CN in a
solvent to which certain soluble inorganic salts were added. After casting
the membrane, the solvent was allowed to evaporate and the salt was completely
12
-------
removed by leaching in water to produce a porous membrane. Porous membranes
prepared in this manner using CaCl~ or ZnCl., were found to be unsatisfactory
as a support for polyelectrolyte membranes because of excessive shrinkage
upon drying.
A CN support with minimal shrinkage upon drying was prepared by incorporating
anhydrous CrCl- with CN in acetone. A coordination complex trichlorotri-
acetone chromium (III), [Cr(C.,H,0)oCl,], was prepared according to the
procedure of Taylor and Milstead (13) by extracting CrCl3 and a trace
of a reducing agent (zinc powder) in a Soxhlet extractor with acetone. At
the conclusion of the extraction, CN was dissolved in the CrCl3~acetone
solution to form a casting solution of 6 wt-% CN and 16 wt-% CrCl3. The
membrane was cast on a glass plate and sealed in a closed box to allow slow
evaporation of the acetone. The film was then exposed to high-humidity air
until the conversion of the CrCl3~acetone complex to CrCl-'Sl^O was complete.
The formation of the hydrated CrCl- in the film was accompanied by a color
•* /
change from violet to a bright green. The CrCl^'dH^O was washed from the
membrane with water, yielding a porous CN membrane with properties similar
to those of the CN/CA support. Dry CN support material was prepared which
had water fluxes ranging from 100 to 2700 gfd when tested at 40 psi.
METHODS OF MEASUREMENT
The permeabilities of the dense homogeneous membranes to water and solutes
were determined by both direct osmosis and reverse osmosis, and in some
cases the distribution coefficient and diffusivity of solutes were indepen-
dently measured in "desorption rate" experiments. In addition, diffusivity
and distribution coefficients for some membranes were obtained from time-
lag measurements in direct osmosis.
DIRECT OSMOSIS
For the direct osmosis tests, a two-compartment cell, which had a membrane
area of 262 cm^, was used; the volume on each side of the cell was 2.7
liters. Internal stirring was provided by a glass paddle connected through
13
-------
a rotary seal to an external motor. The cell was immersed in a 25°C constant
temperature bath. For a given run, the homogeneous membrane was first equil-
ibrated with distilled water on both sides. The experiment was started by
introducing a known amount of concentrated solution of the solute (chosen
to give a desired concentration in the cell to provide a high osmotic pres-
sure difference and therefore a high water flux) at the top of one side of
the cell. Mixing was complete within about 15 sec. When the.solute was an
electrolyte, a conductivity probe was placed in the distilled water on the
other side of the cell. The conductivity of the solution was monitored
with a Beckman Instruments conductivity cell (Models CEL-VY001 and -002,
cell constants 0.01 and 0.02 cm"1, respectively) and a Solu Meter (Model
RA5) connected to a 10-mV recorder. The sensitivity of the measurements
was 0.23 ppm NaCl/mV using a conductivity cell of 0.01 cm"1 cell constant.
A diagram of the apparatus with the conductivity probe in place is shown
in.Fig. 1. The rate of solute transport was calculated from the known
calibration of the conductivity probe.
For the nonelectrolytes, continuous analyses were performed with a Waters
Associates Model R-4 recording differential refractometer. The dilute solu-
tion was continuously pumped through small-diameter polyethylene tubing to
the refractometer and back into the cell by means of a Cole-Parmer Masterflex
peristaltic pump (Model 7020V-13). Distilled water flowed by gravity through
the reference side of the differential refractometer. The electrical output
of the instrument was connected to a 50-mV recorder. At a sample flow of
11.0 cc/min and a reference distilled water flow of 0.8 cc/min, most solutes
gave a reading of 0.5 to 2 mV for each milligram of solute per liter. The
calibration for each solute was determined by a five-point standardization
curve in the concentration range of interest; in this range, the refractive
index is a linear function of concentration. The concentration of solute
employed was 10%, and the experiment was terminated before the dilute side
concentration was above about 100 mg/liter, so that the concentration gradient
was essentially constant for each run. The water permeability was measured
from the osmotic flow of water through the membrane by carefully filling Che
cell to exclude air bubbles at the start and measuring the osmotic water over-
flow from the concentrated side of the cell as a function of time.
14
-------
STIRRERS
CONDUCTIVITY
CELL
GRADUATED
CYLINDER
FOR WATER
COLLECTION
CONDUCTIVITY
METER
_ TO STRIP
- CHART
RECORDER
DIRECT OSMOSIS CELL
Fig. 1. Direct osmosis equipment for monitoring electrolyte and water transport
-------
The permeability of the membrane to water O^c.,, the product of diffusivity
and concentration) was calculated from the measured rate of water flow in
the direct osmosis experiments using the following equation (14) :
D, c, v"-
J
J
1 RT Ax »
2
where J, « water flow (g/cm -sec)
D. « water diffusivity (cm2/sec)
c.. » concentration of water in membrane (g/cc)
v^ = partial molal volume of water (cc/mole)
R = gas constant [cc-atm/(°K)(mole)]
T » absolute temperature (°K)
Air = osmotic pressure difference across the membrane (atm)
Ax = membrane thickness (cm)
The permeability of the membrane to the solute, D2K, was calculated from
the change in concentration of the dilute solution with time. A typical
concentration-versus-time curve is shown in Fig. 2. The straight-line
portion is described by an integrated form on Pick's law (15):
J2 - -°2K Ax" •
2
where J2 =• rate of solute transport (g/cm -sec)
o
D- = solute diffusivity (cm /sec)
K = solute distribution coefficient (g solute/cc membrane)/
(g solute/cc solution)
Ap« ** difference in solute concentration across the membrane
Therefore, D^K can be calculated from the flux and the concentration
difference across the membrane.
Extrapolation of the linear portion of the curve at long times to the base
line gives the "time lag" for the membrane and allows one to calculate the
solute diffusivity. The time lag, T, is related to the diffusivity by the
equation (16)
16
-------
1.0
ce.ce.
LUUJ 0.80
ri 0.60
CO LU
i-« O
O
CO
Z HH
i-» CO
0.40
5 z£
S3 0.20H
o u.
oo
l_) O
-------
(3)
Then, the ratio of the permeability to the diffusivity is the solute dis-
tribution coefficient.
DESORPTION RATE
For the desorption rate experiments, 7/8-in.-diam disks of homogeneous
membrane were equilibrated in a concentrated solution of the solute of
interest. The disks were quickly rinsed in distilled water to remove any
adhering droplets of concentrated solution and blotted dry. The membranes
were then immersed in a known amount of distilled water, and, during desorp-
tion of solute, the increase in solute concentration in the distilled water
was measured using the conductivity probe assembly or the refractometer
described for the direct osmosis procedure. Details of these procedures
are given in (10). The rate at which the desorption takes place is used
to calculate the diffusivity, and the total amount of solute removal provides
a measure of the distribution coefficient.
For diffusion into or out of a semi-infinite slab of thickness Ax, the rela-
tive concentration of the solute in the slab at time t is given by (15)
f _ t = 4 Z-f i 2 e*P I -<2V + *>' 5- I . (*)
f O TT V=0 (2V + 1)
where cf is the final concentration in the slab (i.e., the concentration at
t ->-«>), CQ is the initial concentration, c is the concentration at time t,
and D» is the diffusion coefficient. The higher-order terms of the series
22
decrease rapidly, and at t = 0.1 Ax /ir D2, the second term is less than 5%
of the first; for times longer than this, only the first term is important
and Eq. 4 reduces to
(5)
18
-------
Thus, a plot of In [(cf - ct>/(cf - CQ)] versus time is a straight line
[for (cf - ct)/(cf - CQ) < 0.7] with a slope of -ir2D2/Ax2, from which DZ
can be obtained, and an intercept of 8/ir2 - 0.81.
REVERSE OSMOSIS
The reverse osmosis experiments were performed in a flow system (Fig. 3)
that consisted essentially of a solution reservoir (or sump) maintained
at atmospheric pressure, a high-pressure pump, an accumulator, a back-
pressure regulator, and two parallel sets of three reverse osmosis cells,
as well as gauges and other regulators. The volume of solution in the system
was typically 54 liters. The flow system components on the high-pressure
side were all fabricated from Monel, Type 316 stainless steel, or similar
corrosion-resisting materials. The solution reservoir and interconnecting
tubing on the low-pressure side were constructed of plastic. A 5-y car-
tridge filter was installed between the reservoir and the pump. High-feed-
solution flow was achieved with a Manton-Gaulin positive displacement triplex
pump (Model MP-3) which has a rated capacity of 2 gptn. This pump permitted
the high-pressure solution to be circulated at a high linear velocity (greater
than 150 cm/sec) over the surface of the membrane, where boundary layer phen-
omena are known to be unimportant at the water fluxes observed in this work
(17). Pressure fluctuations were damped to less than ±5 psi with the
accumulator, and pressure drift during the measurements was essentially nil.
Temperature was controlled at 25° ± 1°C.
The test cell contained a rectangular membrane approximately 1 in. by 3 In.
2
(20-cm area). The detailed design of the cell is described in (18).
Testing was carried out at 1500 psi with the PVP membranes, at 800 psi with
the latex membranes, and at 600 psi with the polyelectrolyte membranes. Water
flow rates through the membrane were measured by collecting the output of a
cell In a graduated cylinder for a known period of time. Solute concentra-
tions in feed and product water were measured from conductivity for electro-
lytes and by use of a Beckman Instruments Carbonaceous Analyzer for non-
electrolytes and secondary effluent.
19
-------
Fig. 3. Flow diagram of reverse osmosis system
20
-------
In order to facilitate comparison of the reverse osmosis results and the
direct osmosis measurements for the homogeneous films, the permeability
of the membrane to water was calculated from the reverse osmosis results
by the equation
where Ap is the applied hydraulic pressure and the other items are as
previously defined. The direct osmosis and the reverse osmosis results
for solute flux are directly comparable as long as transport is by a solu-
tion-diffusion mechanism only (14). In most cases permeability to
solute can be calculated from Eq. 2. When solute rejection is poor, however,
a slight revision is required. For that case, Lonsdale, Herten, and Tagami
(19) have shown that the solute flux in reverse osmosis is given by
-D2KAp2 D2Kp2v"2Ap
J2 = Ax RTAx
where the prime refers to upstream concentration. The first term represents
the contribution of the concentration difference to the flux, and the second
term represents the contribution due to the pressure difference. The latter
occurs as a result of the changing thermodynamic activity of the solute with
pressure. Equation 7 can be written as
~D2Kp2
J,
2 Ax
For sodium chloride, v_ is approximately 18 cc/mole (20), and the right-
hand term within the brackets at 100-atm operating pressure is 0.073.
For a highly rejecting membrane, Ap2/p2 ~ 1, and the right-hand term in the
brackets can be neglected, leaving a simplified form of Pick's law. If the
rejection is not very good, Eq. 8 must be used to calculate solute permeability;
otherwise, an apparent pressure dependence of the permeability will be observed.
21
-------
The rejection of a solute Is calculated by the equation
(9)
P2 - P2
where the double prime indicates the product water. The product water
concentration is given by
P2 = j + j = j~ *
2 Jl + J2 Jx
where the approximation holds because in the cases of interest here,
J- » J2. Substituting into Eq. 9 yields
R - 1 - TrV . (ID
P2 J1
Substituting the equations for J1 and J. (Eqs. 6 and 8) results in the
following equation:
D KRTAp. D Kv,Ap
R = I ±* *"
Because Ap_/p' is the rejection, this equation can be simplified to give
- ATT) - D_Kv0Ap
- ATT) + DjKRT
When rejection is high, this equation reduces to
D, C./V., (Ap - ATT)
R - =
(13)
v"(Ap - ATT)
The required permeabilities were measured in direct osmosis experiments,
and the choice of an operating pressure permitted direct calculation of
the predicted rejection. This value was then compared with the value measured
in reverse osmosis. Values for partial molal volumes and osmotic pressures
were obtained from the literature (21) .
22
-------
SECTION 4
RESULTS AND DISCUSSION
POLYELECTROLYTE MEMBRANES
Polyelectrolyte membranes were evaluated only in reverse osmosis tests.
Direct osmosis and desorption rate experiments could not be conducted since
it was not possible to prepare a free-standing membrane. All membranes
were tested at 600 psi and 25° ± 1°C with 0.1% NaCl. Unless otherwise
indicated, all data were obtained after 16 to 20 hr, i.e., overnight tests,
when the initial rapid changes in membrane properties had subsided. Asym-
metric cellulose 2.5-acetate membranes were tested under the same conditions
to serve as control membranes during polyelectrolyte membrane evaluation.
Several factors were investigated during this program: (1) type of poly-
electrolyte, (2) molecular weight, (3) porous support, (4) method of mem-
brane preparation, and (5) posteasting treatment conditions. Membranes were
prepared on three types of porous support: CN/CA, polysulfone, and CN. The
surface pore size of the CN and CN/CA supports was estimated from electron
microscopy to be in the range 1000 to 3000 A. No pores were detectable in
the air-dried polysulfone.
An extensive investigation of membranes prepared from PAA was conducted,
since this material seemed to offer considerable promise for preparing
high-flux membranes suitable for waste-water purification.
Polyacrylic acid membranes were prepared from three different molecular-
weight materials: 50,000, 300,000, and approximately 10 MW. Membranes
prepared from 50,000-MW PAA gave essentially no NaCl rejection, apparently
because this material was not retained by the porous support, and its use
was therefore discontinued. Table 1 shows flux and rejection properties of
membranes prepared from 300,000-MW PAA, both by casting and dipping. The
effect of support type and low-pressure water flux as well as postcasting
treatment is shown.
23
-------
TABLE 1
TRANSPORT PROPERTIES OF PAA MEMBRANES
PREPARED FROM 300,000-MW PAA
Support Properties
Type
CN/CA
f
Water
Flux
at 40 psi
(gfd)
3500
i
800
i
Membrane Preparation
Method
Cast
1
Dip
'
Cast
Dip
1
Wt %
PAA
25
'
2.0
1
r
25
t
2.0
i
•
Heat
Treatment
None
<
r
80°C, 1 hr
\
110°C, 1 hr
None
i
80°C, 1 hr
t
110° C, 1 hr
t
115°C, 1 hr
1
NaCl
Rejection
(%)
50
50
46
46
28
25
29
32
45
57
57
88
78
88
85
75
87
87
88
97
86
96
Water
Flux
(gfd)
36
37
46
43
55
60
56
60
28
23
21
16
16
17
23
22
17
4.6
3.5
0.8
2.2
1.1
24
-------
TABLE 1 (Continued)
Support Properties
Type
Polysulfone
CN
1
Water
Flux
at 40 psi
(gfd)
(a)
r
400
i
Membrane Preparation
Method
Cast
i
r
Dip
Cast
1
Wt %
PAA
25
i
i
2.0
i
i
25
1
Heat
Treatment
None
I
80°C, 1 hr
i
110° C, 1 hr
I
None
1
80°C, 1 hr
V
110°C, 1 hr
1
None
1
NaCl
Rejection
54
54
54
58
67
61
57
19
58
68
48
46
32
32
32
Water
Flux
(gfd)
59
56
39
40
38
38
42
36
38
33
50
46
96
100
101
(a)
No flow through air-dried support at 40 psi.
25
-------
These data indicate that both flux and rejection are dependent on the
porosity of the CN/CA porous support, as characterized by the water flux of
the support measured at 40 psi. A higher flux is accompanied by a lower
rejection for the support with a 3500-gfd water flux. Heating the air-dried
membranes at 80°C for 1 hr had little effect on either flux or rejection,
but heating at 110° to 115°C caused significant decrease in the flux and an
increase in the rejection. This difference in behavior is attributed to a
reaction between the acidic groups of the PAA and the hydroxyl groups of the
CN/CA. It is noted that heat treatment of the PAA-polysulfone membranes had
essentially no effect on their transport properties. The PAA-polysulfone
membranes show somewhat higher flux and rejection than do the PAA-CN/CA
membranes. The three PAA-CN membranes tested show fluxes a factor of two
to three greater than the other membranes but lower rejection.
These data are shown in Fig. 4 as a plot of flux versus rejection. For
comparison, data for a set of asymmetric cellulose 2.5-acetate membranes
heat-treated in water at various temperatures (Table 2) are also plotted.
It can readily be seen that the cellulose 2.5-acetate membranes are superior
to the PAA membranes; i.e., they show higher flux at the same rejection.
Tables 3, 4, and 5 show the transport properties of membranes prepared from
10 -MW PAA on CN/CA, polysulfone, and CN, respectively. The overall properties
of the PAA-CN/CA membranes (Table 3) are inferior to those of the other mem-
branes, and the fluxes are generally lower than for the 300,000-MW PAA mem-
branes. It is noted that reproducibility in the membrane properties is poor,
particularly among the cast membranes where the flux ranges from 1.4 to 65
gfd for nominally the same membranes. No apparent benefit was observed by
initially casting with a 1.2% 10 -MW solution and then dipping the air-dried
membrane in a 2.0% solution of lower-MW PAA.
Table 4 shows the properties of 10 -MW PAA prepared on high-flux polysulfone
supports. The support is characterized by using the low-pressure flux of the
vet material, since upon drying the support loses much of its flow at 40 psi
and does not rewet easily because of its hydrophobic nature. The problem of
reproducibility of these membranes is not as severe as noted for the PAA-CN/CA
26
-------
UJ
ot
0
10
20
30
kO
50
60
70
80
90
92
96
98
99
oo
X MODIFIED CA2 $
• PAA-CN/CA
O PAA-POLYSULFONE
HEAT TREATMENT
O NONE
O 115°C - 1 HR
D 110°C - 1 HR
A 80°C - 1 HR
I
I
20
60 80
WATER FLUX (GFD)
100
120
Fig. 4. Flux-rejection data for 300,000-MW PAA membranes
on CN/CA and polyfiulfone porous supports
27
-------
TABLE 2
TRANSPORT PROPERTIES OF ASYMMETRIC
CELLULOSE 2.5-ACETATE MEMBRANES
Heat Treatment
(°C, 5 mln)
80 .
80
75
75
75
70
70
70
60
55
55
55
55
Water
Flux
(gfd)
18
27
30
28
28
41
38
37
48
55
53
62
65
NaCl
Rej ection
(%)
97.7
97.5
94.9
93.5
92.3
89
88
86
81
70
76
72
68
28
-------
TABLE 3
TRANSPORT PROPERTIES OF PAA MEMBRANES PREPARED FROM
106-MW PAA ON CN/CA POROUS SUPPORT
Water Flux
of Support
at 40 psi
(gfd)
3500
i
1400
1
700
1
Membrane Preparation
Method
Cast
1
r
Cast (106)
+ dip
(50,000)
Cast (106)
+ dip
(300,000)
Dip
1
r
Wt %
PAA
1.2
i
1.2 + 2.0
0.2
r
Heat
Treatment
None
r
NaCl
Rejection
CO
44
39
84
84
90
94
36
33
29
29
48
43
66
64
68
70
79
81
82
81
Water
Flux
(gfd)
58
65
1.4
1.6
3.4
3.3
49
53
64
73
59
62
22
19
19
11
7.4
3.1
10
10
29
-------
TABLE 4
TRANSPORT PROPERTIES OF PAA MEMBRANES PREPARED FROM
106-MW PAA ON POLYSULFONE POROUS SUPPORT
Water Flux
of Wet Support
at 40 psi
(gf d) .
•U2900
1
^2000
i
^10,000
1
•V4000
1
Membrane Preparation
Method
Di
i
P
Cast
Wt %
PAA
0.5
I
0.2
p
0.2 in
1:1 EtOH
4.8
<
Heat
Treatment
Noi
•
ie
p
80°C, 1 hr
110°C, 1 hr
\
None
t
110°C, 1 hr
*
None
110°C, 1 hr
None
i
NaCl
Rejection
00
34
38
38
46
34
40
31
33
38
35
39
44
60
39
43
44
42
38
39
35
58
30
35
50
52
53
55
55
56
Water
Flux
(gfd)
11
23
40
94
76
80
120
115
74
91
58
103
53
153
135
54
64
92
85
74
94
100
97
18
20
22
11
12
8.5
30
-------
TABLE 4 (Continued)
Water Flux
of Wet Support
at 40 psi
(gfd)
0,2900
^7000
r
MOOO
1
Membrane Preparation
Method
Cast
r
Wt %
PAA
4.8
'
1.3
i
1.3
(2 coats)
1.3 + 2.1
r
2.6
1.3 .
Heat
Treatment
None
<
110° , 1 hr
1/2 hr
r
120°C,
1/2 hr
None
110°, 1 hr
None
r
NaCl
Rejection
42
60
51
40
70
70
59
56
42
45
60
56
42
52
60
51
53
47
42
58
—
18
0
0
35
56
A3
38
60
28
28
58
Water
Flux
(gfd)
97
69
74
113
19
33
90
99
131
134
87
73
133
104
112
150
136
150
173
110
0
173
440
457
131
62
82
98
53
26
24
14
31
-------
TABLE 5
TRANSPORT PROPERTIES OF PAA MEMBRANES PREPARED FROM
106-MW PAA ON CN POROUS SUPPORT
Water Flux
of Support
at 40 psl
(gf
-------
TABLE 5 (Continued)
Water Flux
of Support
at 40 pal
(gfd)
2000
\
\
850
2000
\
\
800
I
I
2000
7
i
50
2000
750
i
\
100
800
1
\
2700
290
i
I
1800
\
\
290
1800
Membrane Preparation
Method
Cast
Dip
Wt Z
PAA
3.1
1
I
2.6
2.1
t
1.3
1.0
I
0.5
•
0.2
0.04
(2 dips)
I
0.04
(4 dips)
Heat
Treatment
None
110'C,
30 inin
None
Nad
Rejection
(Z)
20
66
25
0
42
3|4
34
34
76
10
10
20
10
28
68
42
58
36
36
74
68
62
51
33
43
40
55
44
45
45
38
41
38
39
Water
Flux
(gfd)
13
4
26
113
239
96
99
99
26
500
530
338
550
9.5
18
117
69
92
90
4.7
13
17
24
139
40
14
30
54
28
69
129
125
164
147
33
-------
membranes. Most of the membranes showed rejections of approximately 35% to
60% with fluxes of approximately 70 to 110 gfd, although several of the
membranes gave fluxes of greater than 130 gfd with rejections of 40% to 50%.
Higher fluxes were generally associated with the polysulfone with lower low-
pressure fluxes. This relationship is believed to be due to the presence
of finer pores and less intrusion of the PAA into the pores of the support.
It might be expected that plugging of the pores would result from the use
of very-high-flux polysulfone if large pores were present. The best membranes
were prepared by casting onto polysulfone (2900 gfd) from a 4.8 wt-% PAA
solution followed by heat-treating at 110° to 115°C. Membranes of this quality
were not prepared from any other combination of support and casting conditions.
Heat treatment above 115°C resulted in a non-flowing membrane, and it is
suspected that cross-linking of the PAA occurred above this temperature.
Transport properties of 10 -MW PAA-CN membranes are shown in Table 5. The
reproducibility among these membranes is not as good as was observed for
the PAA-polysulfone membranes, and there does not seem to be any correlation
between low-pressure flux of the support and water flux of the membranes.
No beneficial effects were observed to result from heat treatment, and fluxes
were generally lower for these membranes than for the PAA-polysulfone mem-
branes.
In Fig. 5, the data from Tables 3 and 4 are plotted and compared with cellulose
2.5-acetate membranes. Although there is only a small region of overlap, it
appears that the PAA-polysulfone membranes are superior to the PAA-CN/CA
membranes. Several of the high-flux PAA-polysulfone membranes exhibit flux-
rejection properties comparable to those of the asymmetric cellulose 2.5-
acetate membranes.
The data from Table 5 are plotted in Fig. 6. Only three of the PAA-CN
membranes are comparable to the cellulose 2.5-acetate membranes. Membranes
prepared from greater than 4 wt-% PAA solutions generally exhibit higher
rejections and lower fluxes than membranes prepared from lower concentrations.
34
-------
0
10
20
30
50
60
70
80
90
92
96
98
99
MODIFIED CA2 5
• PAA-CN/CA
O PAA-POLYSULFONE
HEAT TREATMENT
O« NONE
D 80°C - 1 HR
A 1108C - 1 HR
O 115°-120°C - 30 MIM
20
l»0 60 80
WATER FLUX (GFD)
100
120
Fig. 5. Flux-rejection data for 10 -MW PAA membranes
on CN/CA and polysulfone porous supports
35
-------
0
>o
20
30
1(0
50
60
70
80
O-
o
90
92
96
-CO
98
— MODIFIED CA2 $
O >1» WT % PAA SOLUTION
• <3 WT X PAA SOLUTION
99
I
l
20
60 80
WATER FLUX (GFD)
100
120
140
Fig. 6. Flux-rejection data for 10 -MW PAA membranes
on CN porous support
36
-------
Several of the PAA-CN and PAA-polysulfone membranes were tested in long-
term tests to observe changes in flux and rejection with time. In addition,
these membranes were tested with Na-SO,, NH.NO,, and urea for comparison with
cellulose 2.5-acetate membranes. These data are given in Tables 6 and 7 for
PAA-CN and PAA-polysulfone, respectively. Three of the membranes listed in
Table 6 were tested for the number of hours shown in the elapsed time column;
the other membranes were Installed at the indicated time. In both tables,
a rapid flux decline is noted for the PAA membranes over the first two days
of testing; the flux then continued to decrease at a rate of approximately
10%/day over the remainder of the test period. At the end of the tests, the
water flux through the PAA membranes was generally comparable to that through
the cellulose 2.5-acetate membrane. Rejections of Na^SO. were all above 85%,
even for membranes rejecting only approximately 36% NaCl. The rejection of
NH.NOo were comparable to that of NaCl, with the exception of the cellulose
2.5-acetate membranes. For these membranes, the NH.NO, rejection was sub-
stantially less than for NaCl. All the membranes rejected urea poorly. This
is in agreement with earlier measurements using heat-treated cellulose 2.5-
acetate membranes heat-treated at 85°C; the rejection of urea was only 45%
with an NaCl rejection of 98.5% (10).
An attempt was made to restore the performance of the PAA membranes by
depressurizing and draining the reverse osmosis test loop and by adding
10 -MW PAA to the feed. The results of this test are shown in Table 8. It
can be seen that draining the system and flushing it with distilled water
resulted in an increase in rejection and about a 10% increase in flux. No
further benefit was evidenced by pressure pulsing the system. The enhance-
ment in membrane performance was only transient; within a few hours the
performance returned to the original level. (This behavior can be seen
more clearly from Table 6 by noting the flux increase in membranes 120-1,
-2, and -4 at 24 hr elapsed time when the system was depressurized to install
membranes 120-3 and 120-5.) The addition of 10 ppm 10 -MW PAA resulted in
a sharp decrease in rejection with about a 20% increase in flux. After
an additional 18 hr, the membrane properties had returned to their original
level. Thus, it appears that at least partial restoration of membrane per-
formance can be achieved by periodic depressurization.
37
-------
, TABLE 6
TRANSPORT PROPERTIES OF 10 -MW PAA-CN AND CELLULOSE 2.5-ACETATE MEMBRANES
TiM
(br)
O.S
6
22
24
27
46
47
51
52
56
57
119
121
125
126
143
Solute
0.11Z
NaCl
0.09Z
"•2s0*
0.09Z
NH4N03
0.10Z
NaCl
0.12Z
urea
0.10Z
NaCl
120-l(m>
Rejection
(Z)
31
36
36
33
36
36
89
88
40
44
37
31
2
4
39
33
Water
Flux
(gfd)
140
122
92
102
98
83
78
77
84
82
81
60
59
61
66
59
120-2 (a>
Rejection
(Z)
29
36
36
33
36
36
89
88
40
43
37
34
6
7
39
33
Water
Flux
(gfd)
137
122
90
100
96
82
77
74
83
80
80
59
59
60
65
57
120-3(b>
Rejection
(Z)
46
56
58
94.5
93.7
61
61
63
55
11
11
63
57
Water
Flux
(gfd)
112
89
69
63
52
60
55
56
42
44
42
45
42
120-4(b>
Rejection
(Z)
22
39
43
40
42
45
90
89
48
50
47
44
10
12 .
49
43
Water
Flux
(gfd)
332
171
117
132
125
100
90
82
93
87
87
60
61
62
65
58
120-5
Rejection
(Z)
72
72
>99
25 98.9
34
28
31
20
22
20
23
19
43
43
71
70
13
12
71
72
Water
Flux
(gfd)
70
61
56
54
56
57
56
51
52
54
55
52
OJ
00
(a)Caat from 1.3 wt Z lO^-MH PAA on CN (750 gfd at 40 p«i).
CN (2000 gfd at 40 pei).
CN (2000 gfd at 40 psi).
-------
TABLE 7
TRANSPORT PROPERTIES OF 10 -MW PAA-POLYSULFONE AND CELLULOSE 2.5-ACETATE MEMBRANES
Elapsed
Time
(hr)
0.5
17
25
41
42
45
46
50
51
68
69
73
74
137
Solute
0.12Z NaCl
0.09Z Na2SO^
0.09Z NH4N03
0.1Z NaCl
0.11Z urea
0.1Z NaCl
126-l(a)
Rejection
(Z)
35
43
44
47
86
85
55
56
55
55
5
7
60
61
Water
Flux
(gfd)
130
82
76
55
62
60
68
65
60
49
53
50
53
35
126-2 (a>
Rejection
26
38
39
42
86
85
51
56
55
53
5
7
58
59
Water
Flux
(gfd)
175
98
90
64
72
70
78
74
70
57
58
57
61
41
126-3(a)
Rejection
(Z)
49
60
61
60
95.7
95.5
64
66
70
68
5
7
72
71
Water
Flux
(gfd)
92
53
50
37
43
41
45
42
40
32
35
31
34
26
125_5(a,b)
Rejection
(Z)
65 .
63
61
59
95.7
96.8
64
64
67
66
7
8
70
68
Water
Flux
(gfd)
54
48
52
43
46
46
50
48
46
40
41
39
42
35
125-6 (a'b)
Rejection
(Z)
70
70
68
66
97.6
97.5
69
70
73
70
8
8
74
73
Water
Flux
(gfd)
47
39
42
35
39
39
41
41
39
33
37
32
35
29
CA
Rejection
(Z)
71
70
68
98.3
98.9
41
40
70
70
10
9
70
70
Water
Flux
(gfd)
57
55
48
50
51
52
52
52
47
50
51
51
46
U)
v£>
(b)
(c)
from 1.3 wt Z 10 -MW PAA -f 2.1 wt Z PAA on polysulfone (2500 gfd at 40 psi),
Previously tested for 124 hr with 0.1Z NaCl.
Heat-treated in water at 55°C for 5 min.
-------
TABLE 8
RESTORATION OF PAA MEMBRANES
Elapsed
Tine
(hr)
0.5
18
26
72
91
92.5
100
117.5
118.5
124
106-MW PAA-CN
NaCl
Rejection
(X)
46
50
52
62
62
47
53
58
Water
Flux
(gfd)
67
39
32
33
33
42
31
33
NaCl
Rejection
(X)
46
50
52
62
62
47
54
59
Water
Flux
(gfd)
77
40
31
33
33
41
31
33
NaCl
Rejection
CO
37
46
46
57
58
44
52
51
Water
Flux
(gfd)
75
32
26
28
28
33
26
27
NaCl
Rejection
(X)
53
44
46
51
52
41
48
54
Water
Flux
(gfd)
53
29
24
27
27
32
25
27
106-MW PAA-Polysulfone
NaCl
Rejection
(X)
27
35
54
59
60
67
66
49
63
65
Water
Flux
(gfd)
400
131
104
62
53
59
59
68
54
54
NaCl
Rejection
CO
61
56
64
67
66
74
72
56
69
70
Water
Flux
(gfd)
108
62
76
50
41
47
47
57
46
47
^'Drained and flushed system; depressurized system for 5 nin
Bepressurlzed system three times in rapid succession
*c^Added 10 ppm 106-MW PAA to system
-------
Several sets of PAA membranes were tested in reverse osmosis at 600 psi
using filtered secondary tffluent from the Escondido, California, activated
sludge sewage treatment plant. The effluent was filtered first through No.
40 Whatman filter paper and then through 5-y cartridge filters installed in
the test loop. Although the effluent obtained from the plant varied from
day to day, a typical analysis would show approximately 1500 ymho/cm con-
ductivity, approximately 85 rag/liter total carbon (TC), and 25 to 30 mg/liter
total organic carbon (TOC). TOG analyses were performed by acidifing the
sample to convert all carbonates and bicarbonates to carbon dioxide and purg-
ing the sample for 5 min with N. to expel the CO.,; thus, all inorganic
carbon was removed. No attempt was made to determine the organic composition
of the effluent. Over a period of 4 to 6 days in the test loop, both TC
and TOC concentrations decreased by about 50%; little change in conductivity
was noted. The decrease in TOC hampered analysis of the product water. The
sensitivity of the Beckman Carbonaceous Analyzer was about 1 mg carbon/liter,
and many of the product water samples were below this limit.
Flux and rejection data for tests performed with secondary effluent are
presented in Tables 9 through 13. These tables also contain flux and
rejection data for 0.1% NaCl taken immediately before draining and flushing
the system prior to adding the secondary effluent. Tables 12 and 13 addi-
tionally show data for NaCl taken 30 min and 24 hr after the end of the
effluent test. The system was drained, then flushed with water for 20 min,
and 0.1% NaCl solution was added.
Tables 9 and 10 show the rejection of secondary effluent by 300,000-MW PAA-
CN/CA membranes. The membranes listed in Table 9 show about a 20% increase
in flux after addition of the effluent (as a result of flushing the system),
followed by a substantial decrease over the next 24 hr. No further decrease
in flux is noted over the next 4 days. The data in Table 10 show an imme-
diate flux decline of about 30% with essentially no change for the next 2-1/2
days. It is noted that the cellulose 2.5-acetate membrane displayed no sig-
nificant change in flux or rejection during the test. This indicates that
the flux decline is a result of fouling of the PAA membranes, not a compac-
tion phenomenon as is frequently observed (4). In both tests the
41
-------
TABLE 9
REJECTION OF FILTERED SECONDARY EFFLUENT
BY 300,000-MW PAA-CN/CA MEMBRANES
Elapsed
Time
(hr)
___
0.5
6
23
50
120
Membrane
66-1
-2
-3
66-1
-2
-3
66-1
-2
-3
66-1
-2
-3
66-1
-2
-3
66-1
-2
-3
0.1% NaCl
Water
Flux
(gfd)
22
20
23
Rejection
81
76
80
Secondary Effluent
Water
Flux
(gfd)
25
26
27
26
27
27
15
14
15
14
13
14
16
15
16
Rejection (%)
Conductivity
74
67
72
71
67
71
74
70
73
72
70
70
71
70
71
Total
Carbon
72
66
72
74
66
73
68
70
65
74
72
73
Total
Organic
Carbon
83
87
94.1
84
90
94.1
76
88
94.1
96.0
>96.8(a)
>96.8(a>
(a)
Limit of sensitivity
42
-------
TABLE 10
REJECTION OF FILTERED SECONDARY EFFLUENT BY 300,000-MW PAA-CN/CA
AND CELLULOSE 2.5-ACETATE MEMBRANES
Elapsed
Time
(hr)
___
1
17
52
66
Membrane
70-l(a)
_2(a)
-3
CA
70-1
-2
-3
CA
70-1
-2
-3
CA
70-1
-2
-3
CA
70-1
-2
-3
CA
0.1% NaCl
Water
Flux Rejection
(gfd) (%)
19 73
14 85
18 84
38 89
Secondary Effluent
Water
Flux
(gfd)
13
9.7
12
38
11
8.1
9.7
33
13
8.7
9.7
34
15
8.1
9.2
38
Rejection (%)
Conductivity
75
84
83
93.7
72
87
84
94.3
61
84
82
92.9
50
85
83
92.8
Total
Carbon
74
80
82
97.5
72
84
83
96.8
73
88
92.7
97.5
56
86
92.5
97.5
Total
Organic
Carbon
84
84
95.5
>96.8
88
82
93.0
>96.8(c)
76
88
>96.8(c)
>96.8(c)
67
85
>96.8(c)
>96.8(c)
(a)
(b)
(c)
Heat-treated in air at 80°C for 1 hr.
Heat-treated in water at 70°C for 5 min.
Limit of sensitivity
43
-------
TABLE 11
REJECTION OF SECONDARY EFFLUENT BY 10 -MW PAA-CN
AND CELLULOSE 2.5-ACETATE MEMBRANES
Elapsed
Time
(hr)
1
18
Membrane
101-1
-2
-3
-4
-5
CA
101-1
-2
-3
-4
-5
CA
101-1
-2
-3
-4
-5
CA
0.1% NaCl
Water
Flux
(gfd)
56
54
56
62
70
45
Rejection
74
76
74
62
58
73
Secondary Effluent
Water
Flux
(gfd)
18
17
17
17
19
29
1.8
1.7
1.7
1.7
1.8
1.4
Rejection (%)
Conductivity
58
64
59
48
45
85
75
79
78
69
64
74
Total
Carbon
75
78
74
66
64
96.5
86
88
83
82
79
90
Total
Organic
Carbon
91.3
>95.695.695.6
>95.6(C)
>95.695.6(C)
>95.6(c)
>95.6(C)
(c)
>95.6U;
(a)
(b)
(c)
Effluent was prefiltered only through No. 40 Whatman filter paper.
Heat-treated in water at 60°C for 5 min.
Limit of sensitivity
44
-------
TABLE 12
REJECTION OF FILTERED SECONDARY EFFLUENT
BY 106-MW J?AA-CN AND CELLULOSE 2.5-ACETATE MEMBRANES
Elapsed
Time
(hr)
0.5
3
19
A3
Membrane
120-1
-2
-3
-4
-5
CA
120-1
-2
-3
-4
-5
CA
120-1
-2
-3
-4
-5
CA
120-1
-2
-3
-4
-5
CA
120-1
-2
-3
-4
-5
CA
0.1% NaCl
Water
Flux-
(gfd)
59
57
57
58
19
52
Rejection
(%)
33
33
42
43
78
72
Secondary Effluent
Water
Flux
(gfd)
48
48
30
47
12
48
35
35
22
32
9.3
35
18
18
13
16
7.2
18
16
16
12
15
7.2
15
Rejection (%)
Conductivity
22
22
38
30
60
77
30
30
46
36
67
79
38
38
54
44
72
80
38
38
54
47
74
81
Total
Carbon
39
39
54
48
74
85
56
56
65
61
75
85
58
58
65
63
75
85
Total
Organic
Carbon
55
52
63
61
78
72
63
63
71
69
75
75
71
63
71
71
83
75
-------
TABLE 12 (Continued)
Elapsed
Time
(hr)
45
67
139
140
162
Membrane
120-1
-2
-3
-4
-5
CA
120-1
-2
-3
-4
-5
CA
120-1
-2
-3
-4
-5
CA
120-1
-2
-3
-4
-5
CA
120-1
-2
-3
-4
-5
CA
0.1Z NaCl
Water
Flux
(gfd)
33
33
29
37
14
47
33
33
29
36
15
47
Rejection
45
45
67
54
84
65
39
39
60
48
79
69
Secondary Effluent
Water
Flux
(gfd)
16
16
12
15
7.2
15
17
17
13
15
7.8
14
18
18
14
17
8.6
14
Rejection (*)
Conductivity
38
38
54
47
74
81
38
38
55
47
73
81
35
35
54
43
70
78
Total
Carbon
56
56
67
62
78
81
56
56
66
61
72
89
WHeat-treated in water at 55°C for 1 hr
System was depressurized rapidly at 44.5 hr elapsed tine.
*c*Llmit of sensitivity
Total
Organic
Carbon
70
70
79
75
79
83
67
67
67
72
>90
-------
TABLE 13
, REJECTION OF FILTERED SECONDARY EFFLUENT
BY 10 -MW PAA-POLYSULFONE AND CELLULOSE 2.5-ACETATE MEMBRANES
Elapsed
Time
(hr)
0.5
3
19
Membrane
126-1
-2
-3
-5
-6
CA(a)
126-1
-2
-3
-5
-6
CA
126-1
-2
-3
-5
-6
CA
126-1
-2
-3
-5
-6
CA
0.1% NaCl
Water
Flux
(gfd)
35
41
26
35
29
46
Re j ection
61
59
71
68
73
70
Water
Flux
(gfd)
27
28
28
37
34
46
24
24
27
36
36
44
16
16
18
29
30
33
Secondary Effluent
Rejection (%)
Conductivity
52
49
57
' 53
65
81
54
50
59
59
64
82
57
52
60
60
66
82
Total
Carbort
64
65
66
68
74
90
68
67
71
75
77
91.9
Total
Organic
Carbon
78
73
73
89
89
89
>94(b)
88
93.9
93.9
>94
>94Cb>
47
-------
TABLE 13 (Continued)
Elapsed
Time
(hr)
43
67
68
92
Membrane
126-1
-2
-3
-5
-6
CA
126-1
-2
-3
-5
-6
CA
126-1
-2
-3
-5
-6
CA
126-1
-2
-3
-5
-6
CA
0.1% NaCl
Water
Flux
(gfd)
23
23
30
52
54
43
25
25
28
46
47
42
Rejection
(%)
74
71
74
77
80
68
67
64
66
69
74
67
Secondary Effluent
Water
Flux
(gfd)
13
13
15
26
27
27
12
12
14
24
26
24
Rejection (%)
Conductivity
56
53
59
61
66
82
54
51
57
60
64
81
Total
Carbon
69
66
70
70
72
88
57
58
60
64
67
82
Total
Organic
Carbon
>94(b>
93.9
88
88
70
88
76
79
76
70
70
88
(a)
(b)
Heat-treated in water at 55°C for 5 min.
Limit of sensitivity
48
-------
conductivity rejections were about the same as the TC rejections, with TOC
rejections being slightly higher.
Table 11 gives the results of a test using 10 -MW PAA-CN membranes. In
this test the effluent was prefiltered only through No. 40 Whatman paper
and the 5-p filter cartridges were eliminated. Immediate and severe flux
decline was observed for all membranes (while rejection improved for all
but the CA membrane), and the decline continued until 18 hr, when the test
was terminated. This behavior illustrates the necessity for prefiltering
the effluent through a good filter system to prevent membrane fouling.
These membranes showed significantly higher TC rejection than conductivity
rejection and almost complete TOC rejection.
Tables 12 and 13 show flux and rejection data for 106-MW PAA-CN and PAA-
polysulfone membranes, respectively. The membranes used in these tests
were initially tested for up to 143 hr with NaCl, Na2SO,, NH^NO,, and urea
(see Tables 6 and 7). An appreciable initial flux decline was observed
with the PAA-CN membranes (Table 12) that was not as pronounced with the
lower-flux PAA-polysulfone membranes (Table 13). This decline was not
observed with the cellulose 2.5-acetate control membranes. All membranes
showed flux decline for the first 2 days; the flux then remained relatively
constant for the duration of the test. Both sets of membranes gave higher
TC rejections than conductivity rejections, with somewhat higher TOC rejections.
It is noted that all rejections become somewhat poorer with the PAA-polysul-
fone and cellulose 2.5-acetate membranes after 43 hr. It is suspected that
these data are the result of a slight loss of sensitivity of the carbonaceous
analyzer, since no evidence of membrane deterioration was seen in the longer
test with PAA-CN membranes. Removal of the effluent from the system, flushing,
and the addition of 0.1% NaCl resulted in a flux increase of about a factor
of 2 for the PAA membranes and a factor of 2 to 3 for the cellulose 2.5-ace-
tate membranes. After an additional day, a slight decrease in rejection was
noted, with no substantial changes in flux.
49
-------
Membranes were also cast from five polyelectrolytes, other than PAA, and
evaluated in reverse osmosis tests. The transport properties of these
membranes are given in Table 14. The various concentrations of polyelectro-
lytes were chosen to provide a solution viscous enough for casting. With
the possible exception of Polyhall 295, a high-molecular-weight anionic
polyacrylamide, none of these polyelectrolytes showed any superiority over
PAA with respect to flux and rejection properties.
In summary, although a few PAA membranes were prepared with transport
properties comparable to those of asymmetric cellulose 2.5-acetate membranes,
the lack of reproducib?lity and severe flux decline shown by these membranes
may make them unsuitable for use in waste-water reclamation. While PAA is
not subject to hydrolysis and may therefore prove to have some lifetime
advantages over cellulose acetate, unless a method is found to improve the
rejection without affecting the flux, the asymmetric cellulose acetate
membrane will remain preferable because of its greater rejection capability.
PVP-POLYISOCYANATE INTERPOLYMERS
HOMOGENEOUS FILMS
Extensive characterization studies were carried out on two different PVP
films. One had a PMPIrPVP molar ratio of 0.7 and the other had a ratio of
0.9. Both of these films (designated PVP 0.7 and PVP 0.9) were prepared
from the blocked PMPI.
The permeabilities to each of the solutes studied In direct osmosis and
reverse osmosis tests are given in Table 15, and similar data for homogeneous
cellulose 2.5-acetate membranes are included for comparison. The tabulated
distribution coefficient values were obtained from desorption rate experiments.
Solute concentrations used in direct osmosis and desorption rate tests were
10% for arabinose and dextrose, and IX for all other solutes. All the
solute concentrations In reverse osmosis tests were approximately 0.1%, with
the exception of the NaCl concentration, which was 0.9%. The predicted
rejection for each solute in reverse osmosis at 1500 psi was calculated from
50
-------
TABLE 14
TRANSPORT PROPERTIES OF POLYELECTROLYTE MEMBRANES
CAST ON CN POROUS SUPPORT (850 GFD AT 40 PSI)
Polyelectrolyte
Cyanamer A-370
Cyanamer P-26
Polyhall 402
Polyhall 295
Polyviny line thy lether-
maleic acid (PVME-MA)
Percent
Solids
2.5
2.5
1.0
1.25
5.0
Water
Flux
(gfd)
28
17
18
31
92
85
62
68
61
66
NaCl
Rejection
(%)
24
24
19
19
18
18
41
39
13
13
51
-------
TABLE 15
PERMEABILITY OF PVP-PMPl AMD CELLULOSE 2.5-ACETAIE .MEMBRANES
TO WATER AMD VARIOUS SOLUTES
Solute
Desorption
Rate, K
(>^-\
(a-?)
Direct Osmosis
DjCjXlO6
(g/cm-sec)
D2Kxl08
(cm2/sec)
Calculated
Rejection
(*)
Reverse Osmosis (1500 psi)
DlV106
(g/cm-sec)
Measured
Rejection
(X)
PVP 0.7
MaCl
Dextrose
Arabinose
Glycine
Acetic acid
NH4C1
NaN03
0.16
0.26
0.17
1.0
2.9
2.9
9.5
3.0
3.9
5.6
3.0
25
21
70
88
85
79
20
45
50
1.9
2.0
2.6, 1.5
2.1
1.8
1.8
55
80
76, 82
12
5.7
56
PVP 0.9
NaCl
Arabinose
Glycine
Acetic acid
NH4C1
NaNO.
NH4N03
0.16
0.17
0.15
1.3
0.73
0.73
0.78
0.15
0.17
9.0
1.4
1.0
2.6
88
97
96
20
80
84
67
0.64, 0.57
0.71
0.62, 0.49
0.69, 0.63
0.65. 0.58
0.66, 0.59
0.74, 0.66
84, 84
92
92, 93
14, 15
86, 85
86, 86
68, 67
Cellulose 2.5-Acetate
NaCl
Arabinose
Glycine
Acetic acid
NH4C1
MaNO.
NH.NO,
4 3
0.035(a>
0.26(a)
0.016
99.5
0.23
0.23
0.20
0.26
0.24
0.24
0.25
99.2
94
95
-13
98
96
95
(a)
Reference 8
52
-------
the permeabilities observed in direct osmosis. Good agreement was obtained
for almost all solutes. The only major discrepancy was observed with
acetic acid, where the measured rejection was only 12% for PVP 0.7 and 14%
for PVP 0.9. These values were lower than the predicted rejections.
For each of the solutes except acetic acid, the permeability decreased at
least a factor of 10 with increased polyisocyanate content. This was
apparently the result of the less hydrophilic nature of the membranes and
the decreased mobility of the PVP chains. For the three solutes checked,
the distribution coefficient was essentially unchanged. The distribution
coefficient was much higher for acetic acid than for the other solutes, and
it increased with increasing isocyanate content. The distribution coefficient
for most of the solutes was much less than unity.
The good agreement between the predicted and measured values of rejection
for most solutes indicates that coupling of water and solute flow does
not occur in these cases and that the homogeneous films were imperfection-
free (19). Because water and solute flows in direct osmosis are in
opposite directions, whereas in reverse osmosis they are in the same direc-
tion, there should have been a noticeable difference in the apparent solute
permeability if flow coupling occurred or if imperfections were present.
The difference between the calculated and measured values of rejection for
acetic acid, however, suggests that acetic acid transport does not occur
only by a solution-diffusion mechanism but rather was coupled to the water
flow. The small enrichment (negative rejection) of acetic acid observed
with the cellulose 2.5-acetate membrane was also probably the result of flow
coupling, although no coupling coefficients were determined in this study.
For the organic materials, glycine and arabinose, the rejection by the PVP
0.9 membrane and the cellulose 2.5-acetate membrane was about the same.
The water permeability of the PVP 0.9 is about three times higher than
that of the cellulose 2.5-acetate.
53
-------
ASYMMETRIC MEMBRANES
Reverse osmosis results for membranes prepared from unblocked PMPI are
shown in Table 16. These membranes were prepared from a stock solution
(10 wt-% solids) having a 1:1 ratio of moles PVP:equivalents PMPI (0.84
g PVP/g PMPI) and air-dried for 24 hr. Two of the membranes were cast
from solutions containing 30% THF. These tests were conducted at 1500 psi
and 25°C using 1* NaCl as feed.
Clearly, the addition of THF increased the water flux substantially (five-
to ten-fold), and the fact that good salt rejection was maintained indicates
that the higher flows did not result from imperfections. The membranes
appeared to be skinned on the air-dried surface. The thickness of the skin
can be estimated if two assumptions are made. The first assumption is that
the flow resistance is all in the skin and that the flux is inversely
proportional to the skin thickness. The fact that the thickest membrane
exhibited the highest flow indicates that gross thickness per se is not
Important. The second assumption is that the flow of water can be divided
into two parts: a diffusive flux of essentially pure water, and a leak
through flaws in the structure which permits the flow of undiluted brine.
The percent solute rejection is nearly the same as the percent of the
total flow which is diffusive, and it is the diffusive flow which must be
used to estimate the thickness of the skin.
For the third membrane listed in Table 16, the diffusive flow of water
is 0.58 x 0.96 - 0.56 gfd. The thickness of the thin skin can then be
calculated to be (0.096/0.56) x 25 - 4.3 y. The calculated thickness of the
last membrane in the table is 2.4 y. These membranes were made under
nominally the same conditions, and this difference is random scatter.
While these calculations should not be considered quantitative, they do
illustrate the fact that THF addition produced asymmetric membranes with
skin thicknesses only a few percent of the gross thickness.
54
-------
Since the solution containing THF tended to react in the mixing flask before
casting due to the presence of traces of water, the effect of time after
solution preparation (solution age) was investigated. The level of solids
content and the percent THF in the solution were also determined. The effects
of these variables are shown in Table 17. In each case, the air-drying time
was 18 hr. The data are sufficiently scattered that no important effect of
solution age or composition is apparent. The skin thicknesses, estimated
using the technique described above, ranged from 3% to 30% of the total mem-
brane thickness.
The water fluxes in Table 17 are lower than desired, presumably because
on drying the dense skin became too thick. Since it was believed, based
on other studies (22), that pyridine accelerated the isocyanate-water
reaction, small quantities of pyridine were added to a 10 wt-% PVP-PMPI
solution in chloroform in an attempt to produce a coherent film in a much
shorter drying time. The THF was completely replaced with pyridine in
s
another variation in the solution, again to reduce the required drying time.
The data obtained with these membranes are given in Table 18.
With a trace of pyridine, coherent films with properties comparable to
those obtained with overnight drying were produced with only several
minutes air-dry time. In the absence of THF, coherent films were produced
with short air-dry time, but the rejections were all significantly lower.
In no case was high rejection combined with high water flux.
Aqueous annealing did not appear to improve membrane performance. Some
of the membranes described in Table 18 were heated for several hours in water
at 95°C. The flux and rejection data obtained with 0.1% NaCl feed are
shown in Table 19. Again, the data are widely scattered and none of the
membranes combined high flux with high salt rejection.
Two other additives were tested at several concentrations and drying
times with the 10 wt-% PVP-PMPI stock solution. Triethyl phosphate (TEP)
was used in concentrations from 5% to 20%, and the membranes were tested
both as-cast and after heating in water. The drying times chosen were the
55
-------
TABLE 16
EFFECT OF THF ON PROPERTIES OF PVP-PMPI MEMBRANES
Casting
Solvent
CHC13
CHC13
CHC13-30% THF
CHC13-30% THF
Thickness
(u)
25
25
34
73
Water Flux
(gfd)
0.096
0.096
0.58
1.09
Rejection
GO
99
99
96
93
TABLE 17
EFFECT OF SOLUTION AGE AND COMPOSITION ON THE PROPERTIES
OF PVP-PMPI MEMBRANES
Solids in
Stock
Solution
(wt %)
15
15
15
10
10
10
10
10
10
Percent
THF
Additive
40
40
40
30
30
30
30
30
30
Solution
Age
(hr)
1/4
1
6-1/2
5
5
9-3/4
9-3/4
21
21
Water
Flux
(gfd)
0.43
0.22
0.37
1.1
1.9
0.26
0.83
0.43
0.52
NaCl
Rejection
(%)
90
98
95
87
78
80
72
90
88
Membrane
Thickness
(V)
66
73
76
38
49
36
76
61
82
Calculated
Skin
Thickness
(V)
6.1
11
6.8
2.5
1.6
11
4.0
6.2
5.2
56
-------
TABLE 18
EFFECT OF PYRIDINE ON PVP-PMPI MEMBRANE PROPERTIES
Additive
THF+1%
pyridine
THF+1%
pyridine
THF+1%
pyridine
Pyridine
Pyridine
Pyridine
Pyridine
Pyridine
Pyridine
Pyridine
Pyridine
Pyridine
Pyridine
Percent
Additive
30
30
30
30
30
30
30
30
30
20
20
20
20
Drying
Time
(min)
5
20
20
10
10
15
15
7
7
7.5
7.5
10
10
Membrane
Thickness
(y)
(a)
32
71
61
50
36
34
77
66
89 ,
93
71
84
Water
Flux
(gfd)
(a)
0.12
0.32
18
19
1.9
1.3
8.7
9.4
24.0
4.3
1.4
1.9
NaCl
Rejection
ta/\
\/o )
(a)
96
96
10
10
40
50
10
10
4
12
43
29
(a)
Membrane was too weak to test.
TABLE 19
EFFECT OF HEATING ON PVP-PMPI MEMBRANE PROPERTIES
Percent
Pyridine
20
20
20
20
30
30
Drying Time
(min)
7.5
7.5
10
10
7
7
Heating Time
(hr)
2
5
2
5
2
5
%
Water
Flux
(gfd)
2.9
3.1
0.8
1.0
13
12
NaCl
Rejection
(%)
21
14
61
61
6
8
57
-------
minimum times required to obtain a coherent film. The data are summarized
in Table 20. Increased TEP content increased the water flux but decreased
the salt rejection. Heating did not improve the rejection, nor did it appear
to reduce water flux; however, the fluxes were all too low to be of interest.
Amyl acetate, which is quite compatible with the chloroform-PVP-PMPI system,
was also tried as an additive at 10% concentration to the 10 wt-% PVP-PMPI
stock solution. The results are given in Table 21. The very low fluxes and
poor rejections indicate that a porous structure did not form as the chloro-
form evaporated.
When the difunctional isocyanate, MDI, was substituted in order to produce
a linear, rather than a three-dimensional, polymer, the membrane had essen-
tially the same properties as membranes made from PMPI and was not responsive
to heat-treating.
In summary, with none of the techniques studied was it possible to make an
asymmetric membrane from a PVP-polyisocyanate based system with the same
rejection properties as the homogeneous membrane.
CELLULOSE DIACETATE
The permeability of a single homogeneous cellulose diacetate membrane to
water and several solutes was measured in direct osmosis experiments. The
permeability data were used with Eq. 13 to calculate a predicted solute
rejection under reverse osmosis operating conditions of 100-atm net pressure.
The measured permeabilities and calculated solute rejections are given in
Table 22. The water permeability (DiC., = 1.2 x 10 g/cm-sec) was measured
only with NaCl and was assumed to be similar for the other solutes. The
value for NaCl is in substantial agreement with that previously obtained by
Lonsdale, Merten, and Riley (8).
The investigation of cellulose diacetate was terminated because of the
low rejection calculated for the organic materials.
58
-------
TABLE 20
EFFECT OF TEP ON PVP-PMPI MEMBRANE PROPERTIES
Percent
TEP
5
5
10
10
20
20
10
10
20
20
Drying Time
(min)
15
15
10
10
10
10
10
10
10
10
Heating Time
at 958C
(hr)
0
0
0
0
0
0
2
5
2
5
Membrane
Thickness
(V)
84
89
109
133
140
155
110
131
171
126
Water
Flux
(gfd)
0.4
0.7
5.2
4.7
10
6.9
3.7
5.2
11
12
NaCl
Rejection
(%)
75
64
24
24
17
17
22
11
15
7.5
59
-------
TABLE 21
EFFECT OF AMYL ACETATE ON PVP-PMPI MEMBRANE PROPERTIES
Air-Dry
Time
(min)
10
10
10
17
17
Membrane
Thickness
(V)
104
110
110
90
90
Water
Flux
(gfd)
0.26
0.19
0.15
0.52
0.17
NaCl
Rejection
(%)
20
38
26
10
4
TABLE 22
TRANSPORT PROPERTIES OF CELLULOSE DIACETATE
Solute Composition
1% NaCl
1% NaN03
10% arabinose
1% acetic acid
D2K
(cm /sec)
1.9 x 10~8
3.7 x 10~8
1.5 x 10~8
2.4 x 10~8
1 x 10~7
Calculated
Rejection
at 100 atm
81
70
85
79
47
60
-------
LATEX MEMBRANES
A number of reverse osmosis measurements were made on membranes cast from
latices on porous CN/CA supports. The method of applying the latex, the
solids content of the latex solution, and postcasting treatments such as
heating in air were examined. The original observation (9) of very high
water fluxes (greater than 100 gfd) with moderate salt rejection was
not made for any of these variations. In only one case, in fact, was any
salt rejection observed. In that instance, General Latex RA-150-7 was
dissolved in dimethylformamide (1 ml of latex in 19 ml DMF). This solution
was then used to prepare films approximately 4 y thick supported on CN/CA
porous supports. When tested In reverse osmosis at 800 psi, they provided
25% to 30% NaCl rejection at low water fluxes (0.1 gfd). It seems likely
that the rejection observed earlier (9) with the latices was the result
of formation of a dynamic membrane from some contaminant in the test loop
that was not present when these experiments were performed.
61
-------
SECTION 5
ACKNOWLEDGMENT
The authors would like to acknowledge the Important contributions to this
program by H. K. Lonsdale, Mrs. G. Hightower, and D. E. Want.
62
-------
SECTION 6
REFERENCES
1. Weber, W. J., Jr., C. B. Hopkins, and R. Bloom, Jr., J. Water Pollution
Control Federation 42^, 83 (1970) .
2. Hunter, J. V., and H. Heukelekian, J. Water Pollution Control Federation
37, 1142 (1965)..
3. Loeb, S., and S. Sourirajan, Advan. Chem. Ser. 38, 117 (1963).
4. Lonsdale, H. K., R. L. Riley, L. D. LaGrange, C. R. Lyons, A. S.
Douglas, and U. Herten, "Research, on Improved Reverse Osmosis Membranes,"
Office of Saline Water Research and Development Progress Report No.
484, Gulf General Atomic Incorporated, 1969.
5. Lonsdale, H. K., R. L. Riley, C. E. Milstead, L. D. LaGrange,
A. S. Douglas, and S. B. Sachs, "Research on Improved Reverse Osmosis
Membranes," Final Report to the Office of Saline Water, U.S. Department
of the Interior, Contract 14-01-001-1778, Gulf General Atomic Incorporated,
March 27, 1970.
6. Sachs, S. B., W. H. Baldwin, and J. S. Johnson, Desalination 6,215
(1969) .
7. Riley, R. L., C. R. Lyons, and U. Merten, Gulf General Atomic Incorporated,
"Transport Properties of Polyvinylpyrrolidone-Isocyanate Interpolymer
Membranes," presented at the National Meeting of the American Institute
of Chemical Engineers, Technical Session on Tailored Polymeric Materials
for Advanced Separations Processes, November 17-20, 1969, Washington
D.C., to be published.
63
-------
8. Lonsdale, H. K., U. Merten, and R. L. Riley, J. Appl. Polymer Sci. ^,
1341 (1965).
9. Merten, U., H. K. Lonsdale, R. L. Riley, and K. D. Vos, "Reverse
Osmosis Membrane Research," Office of Saline Water Research and Develop-
ment Report No. 369, Gulf General Atomic Incorporated, March 1968.
10. Lonsdale, H. K., C. E. Milstead, B. P. Cross, and F. M. Graber,
"Study of Rejection of Various Solutes by Reverse Osmosis Membranes,"
Office of Saline Water Research and Development Report No. 447, Gulf
General Atomic Incorporated, March 1969.
11. Sourirajan, S., and T. S. Govindan, "Membrane Separation of Some
Inorganic Salts in Aqueous Solutions," in Proceedings of the First
International Symposium on Water Desalination, Washington, D.C.,
October 3-9, 1965, v. 1, U.S. Government Printing Office, Washington
D.C., 1967, p. 251.
12. Baddour, R. F., W. R. Vieth, and A. S. Douglas, J. Colloid. Interface
Sci. 22, 588 (1966).
13. Taylor, K., and C. E. Milstead, U.S. Patent 3,076,833, February 5,
1963.
14. Merten, U. (ed.), Desalination by Reverse Osmosis, The M.I.T. Press,
Cambridge, Massachusetts, 1966.
15. Jost, W., Diffusion in Solids, Liquids, Gases, Academic Press,
New York, 1960.
16. Crank, J., and G. S. Park, "Methods of Measurement," in Diffusion
in Polymers, J. Crank and G. S. Parks (eds.), Academic Press,
New York, 1968, Chapter 1.
17. Merten, U., H. K. Lonsdale, and R. L. Riley, Ind. Eng. Chem. Fundamentals
_3_, 210 (August 1964).
64
-------
18. Lonsdale, H. K., U. Herten, and J. C. Westmoreland, "Reverse Osmosis
for Water Desalination," Annual Report to the Office of Saline Water,
U.S. Department of the Interior, Contract 14-01-0001-250, General Dynamics
Corporation, General Atomic Division, April 15, 1963.
19. Lonsdale, H. K., U. Herten, and M. Tagaml, J. Appl. Polymer Sci. 11,
1807 (1967).
20. Trlbus, M., R. Aslmow, N. Richardson, C. Gastaldo, K. Elliot,
J. Chambers, and R. Evans, "Thermodynamlc and Economic Considerations
in the Preparation of Fresh Water from the Sea," University of
California, Los Angeles, School of Engineering Report No. 59-34,
1960.
21. Washburn, E. W. (ed.), International Critical Tables. McGraw-Hill Book
Company, New York, 1928.
22. Riley, R. L., Gulf General Atomic Incorporated, private communication,
October 1969.
65
-------
• 1 Xcce.vsion Number
ry Subject Field & Croup
05D
SELECTED WATER RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
g- 1 Organization
Gulf General Atomic
San Diego, California
6
ride
MEMBRANE MATERIALS FOR WASTE WATER RECLAMATION BY REVERSE OSMOSIS
Douglas, A. S.
Tagami, M.
Mllstead. C. E.
16
21
Project Designation
#17040 EFO
Note
221 Citation
Final report for FWQA Contract No. 14-12-452, 1970, 65 p.
23
* *
Reverse osmosis, Semipermeable membranes, Demineralization, Membrane process,
Osmosis, Tertiary treatment, Wastewater treatment, Water pollution control,
Water reuse.
/den
------- |