EPA-660/2-73-033
December 1973
Environmental Protection Technology Series
New Membranes For Reverse
Osmosis Treatment of Metal
Finishing Effluents
Office of Research and Development
U.S. Environmental Protection Agency
Washington, D.C. 20460
-------
RESEARCH REPORTING SERIES
Research reports of the office of Research and
Monitoring, Environmental Protection Agency, have
been grouped into five series. These five broad
categories were established to facilitate further
development and application of environmental
technology. Elimination of traditional grouping
was consciously planned to foster technology
transfer and a maximum interface in related
fields. The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
'4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL
PROTECTION TECHNOLOGY series. This series
describes research performed to develop and
demonstrate instrumentation, equipment . and
methodology to repair or prevent environmental
degradation from point and -non-point sources of
pollution* This work provides the new or improved
technology required for the control and treatment
of pollution sources to meet environmental quality
standards.
EPA REVIEW NOTICE
This report has teen reviewed by the Office of Research and
Development, EPA, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policies of the Environmental Protection Agency,
nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.
For sate by the Superintendent of Documents, U.S. QoTemment Printing Office, Washington, D.C. 20102 - Price $1.40
-------
EPA-660/2-73-033
December 1973
NEW MEMBRANES FOR REVERSE OSMOSIS
TREATMENT OF METAL FINISHING EFFLUENTS
by
L. T. Rozelle
C. V. Kopp, Jr.
K. E. Cobian
Roap/Task 21 AZH 17
Project 12010 DRH
Program Element 1BB036
Project Officer
Dr. Hugh B. Durham
Heavy Industrial Sources Branch
Grosse lie Laboratory
Grosse lie, Michigan 58138
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D. C. 20460
-------
ABSTRACT
An Important new membrane has been developed for the reverse osmosis
treatment of both highly alkaline and acidic (non-oxidizing) metal
finishing rinse waters. This membrane, designated NS-1, and originally
developed for seawater desalination, consists of the following: a
microporous support film (polysulfone) coated with polyethylenimine
which is cross-linked with tolylene 2,4-diisocyanate.
Simulated alkaline copper and zinc cyanide plating rinses at pH's of
11.8 and 12.9 were treated by NS-1 membranes during 500- and 340-hour
tests without deterioration of reverse osmosis properties. Water fluxes
above 10 gallons per square foot (of membrane) per day (gfd) were observed
with cyanide rejections between 95 and 99 percent. The NS-1 membrane
also treated simulated copper sulfate rinse waters effectively at pH 0.5
during 550-hour tests without deterioration of reverse osmosis properties
(fluxes above 10 gfd with 99.8 percent rejection of copper). The NS-1
membrane is the only known membrane that can perform well using both
acidic and alkaline feed solutions.
Preliminary engineering considerations indicated the feasibility of
applying the NS-1 membrane to reverse osmosis treatment and recycle of
nickel and zinc cyanide electroplating rinse waters.
Tubular development studies were performed on three ultrathin membranes
[cellulose acetate (E 398-10), 3-glucan acetate dimethylaminoethyl ether
(B-GADE), and cellulose acetate 0-propyl sulfonic acid (CADPSA)] which
showed promise as flat sheets during Phase I for treating slightly
acidic metal finishing wastewaters. Cellulose acetate (E 398-10) and
CADPSA were successfully cast in 1/2-inch-ID tubes but 8-GADE membranes
were not. The spiral wrap configuration may be most effective for the
8-GADE membranes.
This report was submitted in fulfillment of Project 12010 DKH under
the partial sponsorship of the U. S. Environmental Protection Agency.
ii
-------
CONTENTS
Page
ABSTRACT i i
LIST OF FIGURES iv
LIST OF TABLES v
ACKNOWLEDGMENTS vi
SECTIONS
I CONCLUSIONS 1
II RECOMMENDATIONS 3
III INTRODUCTION 5
IV EXPERIMENTAL , 9
V RESULTS 15
VI POSSIBLE APPLICATIONS OF REVERSE OSMOSIS
TO THE METAL FINISHING INDUSTRY 55
VII REFERENCES 61
VIII "" PUBLICATIONS 62
IX APPENDIX 63
ill
-------
FIGURES
Page
1 Cross Section of a Fiber Glass Reverse Osmosis
Tube with a Polysulfone Support Liner 12
2 Photograph of Tubular Polysulfone Support Liner
(Left), Abcor Fiber Glass Tube with End Fittings
(Center), and Fiber Glass Tube Enclosed in a
Tygon Sleeve (Right) 13
3 Water Flux Behavior Across Cellulose Acetate
(E 360-60) Membrane During Reverse Osmosis
Treatment of Metal Ion Solutions 20
4 Schematic Representation of NS-1 Membrane 36
5 Idealized Structure of Polyethylenimine
Crosslinked with m—Tolylene 2,4-Diisocyanate 37
6 Reverse Osmosis Performance of the NS-1 Membrane
with Simulated Copper Cyanide Rinse Water 40
7 Reverse Osmosis Performance of an NS-1 Membrane
with Simulated Zinc Cyanide Rinse Water 43
8 Illustration of Iron Oxide Fouling. Photograph of
Cut-away Sections of Reverse Osmosis Tubes After
Casting and After 340-hour Zinc Cyanide Test 46
9 Effect of Operating Temperature on the Water Flux
Behavior of NS-1 Reverse Osmosis Membranes 47
10 Reverse Osmosis Performance of an NS-1 Membrane
with Simulated Acid Copper Rinse Water 49
11 Reverse Osmosis Performance of an NS-1 Membrane
with Simulated Chromic Acid Rinse Water 52
12 Design Concept for Nickel Plating Waste Treatment
Utilizing Reverse Osmosis 56
13 Design Concept for Zinc-Cyanide Plating Waste
Treatment Utilizing Reverse Osmosis 57
iv
-------
TABLES
Page
1 Reverse Osmosis Performance of Ultrathin Cellulose
Acetate Membranes (E 360-60) in Copper, Nickel,
and Chromium Solutions 19
2 Effect of Water Phase Additive on Reverse
Osmosis Performances of CAOPSA 22
3 Effect of Degree of Substitution (DS) of the
0-Propyl Sulfonic Acid Groups on the Reverse
Osmosis Performance of CAOPSA 22
4 Reverse Osmosis Performance of Ultrathin B-GADE
Membranes on Copper, Nickel, and Chromium Solutions:
Effect of Degree of Substitution (DS) of
Diethylaminoethyl Ether (DE) Group 24
5 Reverse Osmosis Performance of Ultrathin High DS
B-GADE (270-13A) Membranes on Copper, Nickel, and
Chromium Solutions: Effect of Membrane Thickness 25
6 Tubular Reverse Osmosis Performance of Ultrathin
Cellulose Acetate Membranes (E 360-60) on Copper
Solutions at Two Draw Rates 28
7 Tubular Reverse Osmosis Performance of Ultrathin
Cellulose Acetate (E 398—10) Membranes on Copper
Solutions: Effect of Draw Rate 28
8 Tubular Reverse Osmosis Performance of Ultrathin
B-GADE Membranes on Copper Solutions 31
9 Tubular Reverse Osmosis Performance of Ultrathin
CAOPSA Membranes 33
10 Reverse Osmosis Performance of NS-1 Membrane on
Simulated Copper Cyanide Rinse (1/10 of Plating
Bath Concentration) Water 38
11 Comparison of Cyanide and Copper Rejections of
NS-1 and Other Reverse Osmosis Membranes 41
12 Reverse Osmosis Performance of NS-1 Membrane on
Actual Zinc Cyanide Plating Bath Diluted 1:10 42
13 Comparison of Cyanide Rejection for Membranes
with Zinc Cyanide Plating Rinse Waters 44
-------
TABLES (Continued)
Page
14 Comparison of NS-1 and Ultrathln Cellulose Acetate
(Eastman 398—10) Membranes for Copper Rejection and
Flux on Simulated Acid Copper Rinse Waters in Flat
Test Cells 48
15 Comparison of NS-1 and Cellulose Acetate (Eastman
Type BD-97) Membranes for Nickel Rejection and Flux
on Simulated Watts Nickel Rinse Waters in Flat Test
Cells 50
16 Reverse Osmosis Performance of GANTREZ AN-Poly-
(Vinyl Alcohol) Membranes Prepared from Solutions
of Various Polymer Concentrations 53
vi
-------
ACKNOWLEDGMENTS
The authors are appreciative of the following staff members: Mr. John
Cadotte for advice concerning the NS-1 membrane; Ms. Barbara Nelson
who guided the early portion of the program, and Mr. Robert Reid who
carried out much of the tubular cellulose acetate testing. The authors
are also grateful for the cooperation of Honeywell Inc. and Mr. Tom
Zenk, in particular, for inspection and discussion of their electro-
plating operation.
The program was sponsored by the Minnesota Pollution Control Agency as
grantee from the U. S. Environmental Protection Agency with Dr. Hugh
Durham, Project Officer. Co-sponsors for this program were: Ecodyne
Corporation, The Lindsay Division; Honeywell Inc.; M & T Chemicals, Inc.;
The Udylite Corporation; and North Star Research Institute. The support
and assistance of these organizations is hereby acknowledged.
This report was submitted in fulfillment of Project No. 12010 DRH under
the partial sponsorship of the U. S. Environmental Protection Agency.
vii
-------
SECTION I
CONCLUSIONS
This study has led to the following conclusions regarding the treatment
of metal finishing wastewaters by reverse osmosis:
1. An important new membrane that is a major breakthrough
in the reverse osmosis treatment of both highly
alkaline and acidic (non-oxidizing) rinse waters has
been developed. This membrane, derived from a non-
polysaccharide polymer, is designated NS-1 and was
originally developed by North Star under Office of
Saline Water funding for seawater desalination. This
membrane consists of the following: polyethylenimine
(PEI)-coated microporous support (polysulfone) reacted
with m-tolylene 2,4-diisocyanate (TDI). The NS-1
membrane has shown considerable promise as an effective
membrane for the reverse osmosis treatment of:
a. Cyanide rinse waters. Simulated alkaline copper
and zinc cyanide plating bath rinses at pH's of
11.8 and 12.9 were effectively treated by NS-1
membranes for periods of 500 and 340 hours under
pressures of 600 psi without membrane deterioration.
At the conclusion of the copper and zinc plating
bath tests the average cyanide rejections were
98.5 and 95.5 percent, respectively. The membrane
also rejected 99.8 and 99.9 percent of the copper
and zinc ions present in the feed, with an average
product flux of 11 gallons per square foot (of
menfcrane) per day (gfd).
b. Acidic non-oxidizing metal finishing wastes. The
NS-1 membrane was found to be effective for the
treatment of acidic non-oxidizing metal finishing
wastes containing divalent ions. During long-
term tests on copper sulfate simulated rinse
waters at a pH of 0.5 the NS-1 membrane exhibited:
no membrane degradation during 550-hour tests,
an average water flux of 9.1 gfd after 550 hours
of testing, and high rejection of copper ions
(99.8 percent) .
2. The very promising reverse osmosis properties observed
for the flat-cast ultrathin membranes during Phase I
could not be achieved in the tubular configuration
-------
within the scope of this program. The tubular develop-
ment studies were performed on three membranes which
showed promise in Phase I for treating acidic waste-
waters. The results of these studies are as follows:
a. Cellulose acetate (E398-10) ultrathin membranes
may be successfully cast in 1/2-inch-ID tubes,
but it was decided that a membrane of this polymer
did not represent a major improvement over
commercially available reverse osmosis membranes
for the treatment of metal finishing waste.
b. Ultrathin 3-glucan acetate dime thylaminoethy1
ether (3-GADE) membranes could not be cast in
1/2-inch-ID tubes.
c. Cellulose acetate 0-propyl sulfonic acid (CAOPSA)
ultrathin membranes were successfully cast in
1/2-inch-ID tubes by compensating for the
inelasticity present in the polymer. Adequate
flux and salt rejection (23 gfd and 92 percent)
were obtained from these tubes.
Rirther tubular casting development would be expected
to result in effective membranes of 3-GADE and CAOPSA.
However, the high flux and rejection properties of
these ultrathin membranes, observed in flat sheets,
could be realized in the spiral wrap configuration
for reverse osmosis treatment of metal finishing
wastewaters.
3. Preliminary design concepts were developed on the
application of reverse osmosis treatment of zinc
cyanide and Watts nickel, plating rinse waters using
the NS-1 membrane. The product water (water which
has permeated through the reverse osmosis membrane)
was recycled back into the rinse bath and the
concentrate recycled back to the plating bath. An
equation was derived that accurately estimates the
concentration values for the product water and
concentrate by knowing (1) the fraction of dissolved
material rejected by the membrane and (2) the fractional
reduction in feed volume exhibited by the reverse
osmosis module. The calculation shows that a 99+
percent recovery of metal salts and water is
feasible when a reverse osmosis system, utilizing
the NS-1 membrane, is applied to the two metal
plating operations.
-------
SECTION II
RECOMMENDATIONS
The overall purpose of this research program was to develop and evaluate
new membranes for improved metal finishing waste treatment by reverse
osmosis. The major tasks of this second phase were (1) the development
of those membranes found most promising in Phase I into tubular
configurations and (2) the evaluation and development of new second
generation reverse osmosis membranes.
An important new membrane, designated NS-1 and developed by North Star
was observed to treat both highly alkaline and acidic (non-oxidizing)
rinse waters effectively using the reverse osmosis process. This is
the only known membrane with both acid and base stability and it
constitutes a major breakthrough in reverse osmosis treatment of waste
streams. It is, thus, strongly recommended that the application of the
NS-1 membrane to reverse osmosis treatment of metal finishing waste
effluents be continued to the field-demonstration phase.
In order to arrive at the field-demonstration phase, the following
three major tasks are recommended;
• Determine a set of casting procedures for fabricating
tubular membranes with consistently optimum performance
for water-flux and metal ion rejection on metal finishing
waste streams (e.g.j 20 gfd flux combined with over
95 percent cyanide rejection for zinc cyanide rinse
waters) . There is a specific need to develop casting
processes for tubular NS-1 membranes in order to
achieve consistent membrane performance and quality
for waste treatment on a commercial scale. This
task should thus include (1) optimization of the
polysulfone casting procedure (&.g*3 determine and
fabricate its optimum thickness); (2) establishment
of optimum PEI-TDI coating sequences; (3) development
of optimum reaction variables for the formulation
of NS-1 membranes such as coating concentration and
cure temperatures.
-------
Long-term testing on simulated metal finishing rinse
waters in order to establish performance capabilities
and reliability of the NS-1 membrane in prolonged operation
under field conditions. Tests of at least 1000 hours
should be performed on alkaline and acid metal finishing
rinse waters using the optimum NS-1 membrane.
Determination of the rejections for typical plating bath
additives, including various inorganic and organic
compounds that are currently used. This task will
provide insight for the total design and performance
of the reverse osmosis system in a metal finishing
waste solution.
Engineering design studies and economic analysis of both
NS-1 membrane production and utilization in a reverse
osmosis system must be carried out to prepare for the
large-scale field-demonstration.
-------
SECTION III
INTRODUCTION
North Star Research Institute has completed the second phase of research
on the program "The Use of Reverse Osmosis for Treating Metal Finishing
Effluents". The Minnesota Pollution Control Agency, as grantee from the
U. S. Environmental Protection Agency, together with five private
organizations, sponsored the work as EPA Program No. 12010 DRH.
The program was designed to serve the needs of the metal finishing
industry through improved pollution control and efficient conservation
of valuable materials. During Phase I of this program the treatment
of metal finishing wastewaters by reverse osmosis was shown to be
feasible. A number of membranes, both commercially available and
improved new types, were demonstrated to be capable of treating various
metal finishing effluents. The second phase of this program, described
in this report, consisted of (1) fabrication and testing of membranes,
found most promising in Phase I, in tubular configurations and (2)
developing new second generation membranes for improved metal finishing
waste treatment by reverse osmosis.
Background
General
The metal finishing industry has an ever-growing problem in controlling
the effects of its wastewaters. The wastes that cause the problems in-
clude rinse waters from metal electroplating solutions and from acid
and alkaline cleaning and pickling solutions. These rinse waters, if
discharged into the environment without treatment, can pollute our
natural resources, inhibit or destroy natural biological activities,
and adversely affect materials of construction. Specific examples of
these detrimental effects include the toxicity of heavy metals and
(2)
cyanides to various forms of aquatic life, the deleterious effect of
copper and chromium on biological sewage treatment processes (because of
-------
their toxicity to the tnicroflora) , and the corrosive effects of acids
(4 5)
and bases on sewer lines and metal and concrete structures. '
The inactivation or removal of the undesirable constituents from
metal finishing wastewaters prior to their disposal is therefore
necessary to minimize their detrimental effects on the environment.
Several methods exist which accomplish this task with varying degrees
of success. The simplest method is the neutralization of an excessively
acid or alkaline waste. Inactivation and removal of the metal and
cyanide species can be accomplished by oxidation or reduction to a less
contaminating state, precipitation to permit removal, or ion exchange
for removal or recovery.
Various problems are encountered in the use of any one of these
techniques, a few of which include large space requirements, complicated
operating procedures, high cost, and insufficient removal of the
contaminating species. In addition, the objective of most of the
conventional methods of treatment is ultimate disposal or destruction of
undesirable constituents, with relatively little attention being given
to recovery of the contaminating species or the water.
Reverse osmosis can be used in combination with existing methods of
treatment to increase efficiency, or it can be used alone. In combina-
tion with existing methods, it can be used to treat water from a
continuous destruction process for recycling back to the plant opera-
tions, or it can increase the metal ion concentration prior to an ion
exchange treatment. If used alone on an individual plating line, it
can provide rinse water for recycling and the reclamation of metal
salts or other chemicals for reuse.
A discussion of several metal finishing operations, the source of the
waste streams and their content, and the reverse osmosis process is
given in the progress report covering the first phase of this program.
-------
Review of First Phase
The first phase of the program .consisted primarily of determining which
meni)ranes, from those commercially available and experimental, were most
promising for the reverse osmosis treatment of metal finishing waste-
waters. The most important membrane performance criteria were high
rejections of specific components, high water fluxes through the
membranes, and low decline in flux during operation.
Seventeen different membranes were evaluated for the separation of metal
ions, acids, bases, and cyanides from water. They included commercially
available asymmetric membranes (approximately 0.002 inch in thickness—
n f __r
500 x 10 angstroms) and ultrathin membranes (1 x 10 to 2 x 10 inch
3 3
in thickness—0.25 x 10 to 5 x 10 angstroms). Experimental results
using flat sheets of membranes showed that reverse osmosis is feasible
and effective in treating these effluents for both pollution control,
and metal ion and water recovery for possible recycle. Although no
single membrane was found effective for all effluents, membranes for
four different polymers showed considerable promise.
Simulated acidic nickel, iron, zinc, and copper plating bath rinses
were effectively treated by ultrathin membranes of three polymers:
cellulose acetate, cellulose acetate 0-propyl sulfonic acid, and
3-glucan acetate dimethylaminoethyl ether. Water fluxes were generally
above 30 gallons per square foot (of membrane) per day (gfd) at metal
ion rejections up to 99.9 percent.
Simulated chromic acid rinses were effectively treated by ultrathin
cellulose acetate 0-propyl sulfonic acid. This membrane exhibited a
water flux (at pH 2.5) of 27 gfd with 95 percent rejection of chromium.
Sulfonated polyphenylene oxide membranes were tested for 95 hours in a
highly alkaline (pH 11.4) copper cyanide solution. The rejection was
98 percent for copper and 93 percent for total cyanide, with a water
flux of 45 gfd.
-------
Preliminary engineering considerations on the application of reverse
osmosis to the treatment and recycle of rinse waters from an acidic
copper (sulfate) plating bath were also carried out.
Current Research Program
The objective of this research program may be divided into two parts,
(1) the development of those membranes found most promising in Phase I
into tubular configurations and (2) the evaluation and development of
new second generation membranes for improved metal-finishing waste
treatment by reverse osmosis.
• Tubular Ultrathin. Three ultrathin membranes were
considered most promising for second phase tubular
development because, during the first phase, they
were the most effective in treating slightly acidic
wastewaters containing copper, nickel, and chromium:
Cellulose acetate
Cellulose acetate 0-propylsulfonic acid (CAOPSA)
3-Glucan acetate dimethylaminoethy1 ether (3-GADE)
• Second Generation Membranes. The polysaccharide membranes
tested in Phase I exhibited poor chemical stability in
highly acidic or alkaline wastewaters. Consequently,
efforts were directed toward developing and evaluating
new, nonpolysaccharide, second generation membranes
which would be chemically stable under highly acidic
and basic environments. A single polymer membrane
that could be used to treat a wide variety of metal
finishing waste solutions would provide for greater
practicality of the reverse osmosis process for this
application.
8
-------
SECTION IV
EXPERIMENTAL
Polymers
Cellulose Acetate
Cellulose acetate resin was obtained from Eastman Kodak Company. The
398-10 designation indicates 39.8 percent acetyl groups, or approximately
2.5 acetyl groups per molecule (out of a possible three), and a
viscosity of 10 seconds. The 360-60 designation indicates 36.0 percent
acetyl groups, or approximately 2.0 acetyl groups per molecule, and
a viscosity of 60 seconds (molecular weight of 360-60 greater than that
of the 398-10).
Cellulose Acetate O-PropylsuTfonic Acid
The cellulose acetate 0-propyl sulfonic acid (CAOPSA) polymer was
synthesized from pure cellulose in the presence of acetic acid.
The preparation procedures were designed for the synthesis of cellulose
methyl sulfonate 0-propyl sulfonic acid (CMSOPSA). However, elemental
analysis has indicated the presence of a negligible amount of methyl
sulfonate groups and a large amount of acetate groups; thus, CAOPSA is
the proper structure. The degree of substitution of 0-PSA groups was
found to be 0.1 or less.
B-G1ucan Acetate Dimethylamlnoethyl Ether (B-GADE)
B-GADE was prepared from B-glucan (Pillsbury). The structure was
confirmed by elemental analysis. The degree of substitution for
dimethylaminoethyl ether was 0.12 and lower.
NS-1 Membranes
The NS-1 membrane is a composite comprised of a microporous polysulfone
support film coated with polyethylenimine (PEI) and crosslinked with
-------
tolylene 2,4-diisocyanate (TDI). The thickness of the polysulfone
support is approximately 1.8 mils and the quantity of the PEI-TDI on
the polysulfone film is equivalent to approximately a 5000-angstrom
thick coating. The membrane is fabricated by coating polyethylenimine
onto the microporous surface of the polysulfone followed by an interfacial
reaction with TDI for the crosslinking step. Heat curing is then
necessary to achieve the final rejection and water flux properties.
Membrane Casting
Flat-Cast Membranes
For rapid evaluation of new membrane materials, flat composite membranes
were used since they were more easily fabricated than their tubular
counterparts. Two new polymers were studied in the flat sheet form:
the NS-1 membrane and a copolymer of GANTREZ AN and poly (vinyl alcohol).
GANTBEZ AN is GAP Corporations chemical trade name for poly(methylvinyl
ether/maleic anhydride).
The GANTREZ AN-poly(vinyl alcohol) condensation copolymer was fabricated
in membrane form by immersion of the polysulfone support film into
solutions containing varying amounts of the two polymers. The support
containing the absorbed polymers was then heat-cured at temperatures
varying from 100° to 150°C. The membrane-support composite was cut to
size and used in subsequent flat-cell tests.
The preparation of the NS-1 membranes in flat sheets was based on the
coating process described above.
Tube-Cast Membranes
The casting procedure for the water-cast tubular ultrathin membrane
composites is summarized below.
The membrane-support composite for a 1/2-inch commercial
reverse osmosis tube was cast as follows: the polysulfone
support liner was prepared by filling a 0.555-inch-ID
10
-------
stainless steel tube with a 15-percent solution of Union
Carbide P3500 polysulfone in DMF. The tube was drained and
a 0.547-inch-diameter aluminum bob was passed through the
tube to provide a uniform film of casting solution on the
inside wall. The tube was then lowered mechanically in a
smooth continuous motion into a one-percent DMF solution,
which gelled the polysulfone. After removal from the DMF
solution, the tube was filled with a dilute aqueous hydrophilic
polymer solution to produce a smooth water drainage, and the
membrane casting solution floated on top of this water
solution. The tube was drained at a specific rate (draw
rate) through a control valve and flow-meter. The wet
polysulfone support-membrane composite was then pulled
from the stainless stell casting tube and air-dried. The
dried composite was attached to a wood dowel and pulled
into the commercial fiber glass tube. The ends of the
composite were sealed to the tube wall with an elastomeric
adhesive.
The NS-1 membrane was fabricated in tubular configurations by immersing
a tubular polysulfone support film in an aqueous two-percent PEI solution
for ten minutes. Upon removal from the PEI solution, the support liner
was immersed in a one-half-percent TDI-hexane solution for thirty
seconds. The polysulfone membrane-support composite was air-dried
and pulled into a polysulfone-coated 1/2-inch Abcor fiber glass tube.
The fiber glass tube, containing the membrane-support composite, was
heat-cured at 110°C. Thermolastic adhesive was then applied to the
ends of the liner to seal it to the walls of the fiber glass tube.
Reverse Osmosis System
The reverse osmosis test loop contained a 20-liter reservoir, a Model
251-144 Milton Roy Pump, an accumulator (surge-tank), heat exchanger,
high-pressure filter, needle valve for system pressure control, and a
Rotameter-type flow meter. A detailed schematic diagram and description
of the test loop may be found in Reference 1.
Figures 1 and 2 represent the tubular configuration. Figure 1 depicts
a cross section of a fiber glass reverse osmosis tube with a poly-
sulfone support liner. Figure 2 is a photograph of the polysulfone
liner, Abcor fiber glass tube with end fittings, and the fiber glass
tube enclosed in the Tygon product water collection sleeve.
11
-------
Elastomer Sealant
Polysulfone
Support Lining
Elastomer Impregnated
Area Of Polysulfone
Support Lining
Stainless Steel
Bushing
Polysulfone Filler
Fiber Glass Tube
Figure 1. Cross Section of a Fiber Glass Reverse
Osmosis Tube with a Polysulfone Support
Liner
12
-------
Figure 2. Photograph of Tubular Polysulfone Support
Liner (Left), Abcor Fiber Glass Tube with
End Fittings (Center), and Fiber Glass Tube
Enclosed in a Tygon Sleeve (Right)
13
-------
A detailed schematic diagram and description of the flat test cell is
presented in Reference 1.
Reverse Osmosis Testing
The conditions used to measure the reverse osmosis properties of the
tube-cast membrane-support composites and flat reverse osmosis cells
were:
Pressure 600 psig
Temperature .... 25°C (77°F)
Feed Flow 1650 to 3350 ml per minute
The metal ion concentration in the feedwater used in the polysaccharide
membrane studies was generally about 1000 mg per liter. The simulated
rinse waters, used in the nonpolysaccharide membrane studies, were
usually 1/10 the concentration of the plating bath solutions. (The
formulations for the plating baths used during these studies are
presented in Reference 1.)
Water flux measurements were carried out by measuring the rate of flow
of the purified water stream from an individual flat cell or tubular
reverse osmosis unit.
Rejection measurements were appropriately based either on atomic
absorption, total carbon analysis (discussion presented in copper
cyanide section) or conductivity determinations of the metal ion
content of the feedwater and of the purified water stream. The
rejection was calculated as the percent of the total ion content
in the feedwater returned by the membrane (see Reference 1) .
More detailed feedwater make-up and analyses are given in the
appropriate sections of the report.
14
-------
SECTION V
RESULTS
The results of the experimental program are divided into two separate
sections, 1) polysaccharide membranes and 2) nonpolysaccharide membranes.
Basically, the polysaccharide section consists of the tubular develop-
ment of the ultrathin membranes found most promising in Phase I. The
nonpolysaccharide section deals with new second generation membranes
evaluated during the second half of this program.
Polysaccharide Membranes
In the first phase of the program, only flat-cast membrane-supports
were considered. The optimum set of these variables for flat-cast
membranes would not necessarily be the same as for tube-cast membranes.
Thus, a brief study of these variables was necessary for each polymer
to determine its applicability to tubular reverse osmosis. Three
polymers were considered: cellulose acetate (two types were used,
E 398-10 and E 360-60); CAOPSA and 3-GADE. Tube-casting considerations,
described in more detail, and reverse osmosis testing using the 1-3/8-'
inch— and 1/2-inch-ID tubes for casting are given below.
Tube-Casting Considerations
The tube-casting procedures for cellulose acetate E 398-10 had been
developed in previous programs for optimum reverse osmosis performance.
Thus, the tube-casting development for polysaccharide membranes
concerned cellulose acetate E 360-60, CAOPSA, and 3-GADE.
Solvent. The solvent used for casting an ultrathin polymer membrane in
a tube must meet four basic requirements:
1. It must be a good solvent for the polymer, giving clear,
molecular-dispersed solutions.
2. The density of the casting solution should be less than
1.0 so that it does not sink through the aqueous phase.
15
-------
3. The solvent should have some solubility in water to aid
in its removal from the membrane after it is cast.
However, too high a solubility in water would lead to
gelation of the polymer in the pool of casting solution
before it can be formed into a membrane on the support
liner. A high solvent vapor pressure also aids in solvent
removal from the membrane by evaporation.
4. The solvent must be of a polar nature so that it will
spread uniformly on water to give a membrane of uniform
thickness.
Cy clone xan one was used as a solvent for all the flat-cast membranes
evaluated in the first phase. However, its low water solubility
(2.4 percent), high boiling point (157°C), and density of 0.948
eliminated it as a solvent for tube-casting.
Ethyl acetate is an example of an attractive tubular membrane casting
solvent because it has a solubility in water of 8.6 percent at 20°C, a
density of 0.90, and a boiling point of 77°C. Ethyl acetate was used
most extensively in the brief solvent study.
The results of the solvent investigation for each polymer are listed
below. In each case many solvent combinations were studied and only
the best combinations are given that could be determined within the
scheduling of the program.
Cellulose Acetate E 398-10. The best solvent system for cellulose
acetate E 398-10 was found to be one part acetic anhydride in 15 parts
ethyl acetate. The polymer is first dissolved in acetic anhydride
and then diluted with ethyl acetate.
Cellulose Acetate E 260-60. This cellulose acetate has a lower acetate
degree of substitution and higher molecular weight than the E 398-10
type; thus its solubilities are different. A clear solution was obtained
from the following solvent system: 4 parts acetic anhydride, 1 part
cyclohexanone, and 5 parts ethyl acetate.
&-GADE. A ratio of 15 to 1 ethyl acetate to acetic anhydride gave a
clear solution for casting.
16
-------
CAOPSA. It was difficult to find a suitable tube-casting solvent for
this polymer. It could be dissolved in acetic anhydride only with
prolonged heating. The solvent system consisted of a 1 to 1 ratio
of acetic anhydride to ethyl acetate.
Water Additive. Water additives (linear, high molecular weight, water
soluble polymers) are necessary to produce a smooth water drainage
during tubular casting, which results in uniform membrane thicknesses
along the tube. A cationic polyacrylamide (Reten 210, Hercules) was
chosen as the standard additive for tube-casting cellulose acetate
E 398-10 in earlier programs. *'
Annealing. An annealed membrane will exhibit higher rejections but
lower fluxes than non-annealed membranes. Annealing is accomplished
by either drying or by exposure to heated water at a specified tempera-
ture. Drying the membrane-support composite produces three effects:
1) annealing by dehydration; 2) removal of residual casting solvent; and
3) bonding of the membrane to the support film. The membrane-support
composite for the 1/2-inch-ID tube is air-dried and then mounted in a
1/2—inch-ID fiber glass tube. All of the flat cell tests were conducted
with membrane-support composites that were not allowed to dry. (This
difference between the first-phase flat-cast membrane program and the
present tube-cast membrane program caused some differences in reverse
osmosis results.)
Reverse Osmosis Testing of Tube (1-3/8-Inch-ID)-Cast Membranes
The objective of this part of the program was to determine tubular
casting conditions for the polymers that will produce membranes with
good reverse osmosis properties. The polymers were cast on polysulfone
supports in 1-3/8-inch-ID tubes, cut from the tubes, and measured in
the flat reverse osmosis test cell. This procedure afforded a general
screening of tubular casting variables in the shortest possible time
and was carried out for cellulose acetate E 398-10 before direct
casting in 1/2-inch-ID tubes. The three polymers evaluated in this
study were cellulose acetate (E 360-60), CAOPSA, and B-GADE.
17
-------
CeTJjjTpse. Acetate (E 360-60). Reverse osmosis performance of ultrathin
cellulose acetate (E 360-60) membranes cast into the 1-3/8-inch-ID tube
with three casting solution concentrations is given in Table 1. The
ions in the feed solution during the test were varied in the following
order: sodium chloride, copper, nickel, chromium (pH 5.1), chromium
(pH 2.0), and sodium chloride. Each feed solution, except the last, was
tested for 24 hours with a total reverse osmosis testing period of
somewhat over 120 hours. The last sodium chloride solution was used as
a check for membrane degradation and a single immediate reading was
compared with the original sodium chloride solution. The test conditions
are listed in Table 1.
Table 1 shows that, of the three casting solution concentrations, two-
percent cellulose acetate produced a membrane with the highest metal
ion rejections. All three membranes exhibited a water flux of 30 gfd
after 120 hours of testing. Less flux decline was observed for the
thicker membranes cast from the two-percent solution. The magnitude of
the membrane thicknesses was generally proportional to the casting
solution concentration.
The rejections for copper and nickel were less than those observed
during the first phase of this program for flat-cast membranes (99.9
percent). However, the intent of these initial tube-castings was
not to optimize, but to show that the membranes could be tube-cast
and exhibit good reverse osmosis properties. The cellulose acetate
membrane cast from the two-percent solution exhibited higher chromium
rejections at both pH's compared to flat-cast membranes. In all cases,
the sodium chloride rejections increased from the beginning to the end
of the test, indicating no membrane damage.
Typical water flux behavior for the cellulose acetate membrane during
the reverse osmosis test, using all the feed solutions, is plotted as
a function of time in Figure 3. Changing the feed solution had a
negligible effect on the water flux during the 120-hour test. As
normally observed in reverse osmosis, the flux declined in the initial
18
-------
Table 1. Reverse Osmosis Performance of Ultrathin Cellulose Acetate Membranes
(E 360-60) in Copper, Nickel, and Chromium Solutions
Conditions:
Tube Cast (1-3/8-inch-ID), 0.14 inch per second draw rate
Flat cell tested
Pressure 600 psig
Temperature .... 25°C
Flow rate 1650 ml per minute
Metal feedwater
concentration . . 1000 mg per liter
Concentration of
Cellulose Acetate
in Casting Solution
(percent)
0.5
1.0
2.0
Membrane
Thickness
<400A
Increasing
V
1
Water Flux
(gfd)
25 hrs
50
52
44
120 hrs
30
30
30
Rejection (percent)
Cu
71.5
83.4
93.6
Ni
71.5
83.3
95.2
Cr
(pH 5.1)
58.9
73.2
89.7
Cr
(pH 2.0)
23.7
38.8
70.1
Sodium Chloride
Rejection
(percent)
Initial
40.0
57.5
81.2
Final
61.2
74.2
86.5
-------
N>
O
70
60
-. 50
•*-
9
x 40
d
* 30
20
10
0
0
5Hr
Pressure
Releoi
Conditions >
Cast In I 3/8 Inch ID Tubt
(2% Solution)
Flat C«ll Tasted
Pressure - - 600 psi
Temperature --25°C
Flow Rate t - 1650 ml / min
I
0.1% NaCI I OOOmg/ICu +*
81% Rej , pH5 ' PH5
lOOOmg/ICr*6! lOOOmg/ICr
I
I
pH 5
I
pH 2
I
I
b
HIM
O
,1
I
I
I
I
0.1% NaCI
86% Rej
20
40 60
TIME (HOURS)
80
100
120
Figure 3. Water Flux Behavior Across Cellulose Acetate
(E 360-60) Membrane During Reverse Osmosis Treatment
of Metal Ion Solutions
-------
stages of the test. This decline is probably caused by some compaction
of the membrane under pressure and by iron oxide fouling in the system.
In this case (Figure 3) the flux was leveling-out at above 30 gfd in
120 hours.
CAOPSA. Reverse osmosis tests similar to the cellulose acetate tests
described above, but using tube-cast CAOPSA membranes were carried out
for 120 hours. However, the rejections were considerably lower than
observed for the flat-cast membranes during the first phase of the
program. For example, the highest observed rejection for chromium was
71 percent compared to 97 percent for the flat-cast membranes. The
water fluxes were all below 10 gfd compared to values of greater than
20 gfd for the flat-cast membranes.
The reason for these poor reverse osmosis properties was the cationic
polyacrylamide (Keten 210) solution used for tube casting. CAOPSA
exhibits an anionic charge and thus reacts with cationic solutions.
This problem was not encountered during the first year because no
additives to the water were necessary for flat-casting of these
membranes.
Table 2 gives the reverse osmosis performance of two tube-cast CAOPSA
membranes using a copper feed solution. One membrane was cast using the
cationic polyacrylamide solution, the other using an anionic poly-
acrylamide solution (Separan AP 30 — Dow Chemical). This 18-hour test
indicates that the anionic material will give tube-cast membranes with
rejections and fluxes similar to those observed last year with the
flat-cast membranes.
During the first year of this program it was found that the degree of
substitution (DS) of the 0-propyl sulfonic acid group of CAOPSA was an
important factor in obtaining optimum reverse osmosis properties.
The higher the DS, the higher the flux and the lower the rejection.
Table 3 gives the reverse osmosis performance of several CAOPSA membranes
in the DS range of 0.05 to 0.10. The membrane with a DS of 0.05 exhibited
the highest copper rejection (97 percent), but exhibited relatively low
21
-------
Table 2. Effect of Water Phase Additive on Reverse
Osmosis Performances of CAOPSA
Conditions:
Tube cast (1-3/8-inch-ID), 1 percent solution,
0.14 inch per second draw rate
Pressure 600 psig
Temperature . . . . 25°C
Flow rate 3350 ml per minute
Feed solution . . . 100 mg per liter Cu1
Length of test ... 18 hours
-H-
Water Phase Additive
During Casting
Re ten 210
(Cationic polyacrylamide)
Separan AP 30
(Anionic polyacrylamide)
Water Flux
(gfd)
2.5
20.2
Salt Rejection
(percent)
85.1
95.9
Table 3. Effect of Degree of Substitution (DS) of the
0-Propyl Sulfonic Acid Group on the Reverse
Osmosis Performance of CAOPSA
Conditions:
Tube cast (1-3/8-inch-ID), 1 percent solution,
0.25 inch per second draw rate
Pressure 600 psig
Temperature .... 25°C
Flow rate 3350 ml per minute ,\
Feed solution . . . 1000 mg per liter Cu
Length of test ... 24 hours
Water additive . . . Separan AP 30
Membrane No.
270-48A
270-49C
270-4 9B
2 70-49 A
DS of OPSA
Increasing
from
'u 0.05 to
* 0.1
Thickness
(A)
300
650
—
800
Water Flux
15.1
19.1
97.0
133.0
Salt Rejection
(percent)
96.8
90.3
49.4
35.2
22
-------
water flux (15 gfd). It is interesting that even though the membrane
thickness decreased with decreasing DS, the water flux decreased also.
DS is an extremely important factor in regulating the reverse osmosis
performance of CAOPSA.
3-GADE. The reverse osmosis performance of 3-GADE membranes of varying
OS's for the dimethylaminoethy1 ether group (DE) is given in Table 4.
In this series of tests sodium chloride, copper, nickel, and chromium
feed solutions were used; each were run for 24 hours. The tube-casting
and measurement conditions are given in Table 4.
Table 4 indicates that the 3-GADE with the higher DS (0.12) produced a
membrane with high reverse osmosis potential for copper and nickel. The
thickness of this membrane was greater than the other two; thus, the flux
could be greater, if cast thinner. The flux decline was quite high but
leveled off fairly fast. For the 3-GADE membranes, a higher flux decline
was observed when the chromium feed solutions were used; this is in
agreement with results from the flat-cast membranes. The flux decline
did not result in membrane degradation because sodium chloride rejection
after testing was equal to or greater than that before testing.
The rejections for copper were similar to those for the flat-cast
membranes (DS—0.12). For nickel, the rejection was lower than observed
earlier. Optimization of casting conditions would be expected to improve
the reverse osmosis performance.
The effect of 3-GADE membrane thickness on reverse osmosis performance
by varying the 3-GADE casting solution concentration and draw rate is
shown in Table 5. This test series was run for a total of 96 hours
using copper, nickel, and chromium feed solutions. Sodium chloride
feed solutions were not used before or after, in this case.
The data in Table 5 show that the rejections were significantly higher
for the two thicker membranes (270-50C—900A, 270-50D—600A).
Theoretically, the thickness should not affect the rejection. However,
extremely thin membranes are more apt to develop small holes or leaks.
23
-------
Table 4. Reverse Osmosis Performance of Ultrathin S-GADE Membranes on
Copper, Nickel, and Chromium Solutions: Effect of Degree of
Substitution (DS) of Diethylaminoethyl Ether (DE) Group
Conditions:
Tube cast (1-3/8-inch-ID), 1% solution, 0.14 inch per second draw rate
Flat cell tested
Pressure 600 psig
Temperature .... 25°C
Flow rate 1650 ml per minute
Metal feed water
concentration . . 1000 mg per liter
Membrane
Number
270-29B
270-29A
270-13A
DS of DE
Group
Increasing
DS to
\
0.
f
12
Membrane
Thickness
<400A
Increasing
\
f
Water Flux
(gfd)
25 hrs
38
41
68
120 hrs
26
26
28
Rejections (percent)
Cu
98.8
95.2
94.1
Ni
91.4
95.4
94.7
Cr
(pH 5.1)
80.3
85.3
68.0
Cr
(pH 2.0)
44.3
66.9
52.2
Sodium Chloride
Rejection
(ppyppTit^
Initial
72.0
84.8
92.2
Final
87.4
93.7
92.5
-------
Table 5. Reverse Osmosis Performance of Ultrathin High DS P-GADE
(270-13A) Membranes on Copper, Nickel, and Chromium
Solutions: Effect of Membrane Thickness
Conditions:
Tube Cast (1-3/8-inch-ID)
Flat cell tested
Pressure 600 psig
Temperature .... 25°C
Flow rate . . . . . 1650 ml per minute
Metal feed water
concentration . . 1000 mg per liter
Membrane
Number
270-50C
270-50D
270-50E
270-50F
e-GADE in
Casting
Solution
(percent)
2
2
2
1
Draw Rate
(in. /sec)
0.15
0.50
1.50
—
Membrane
Thickness
(A)
900
600
450
<400
Water Flux
(gfd)
24 hrs
32
36
57
57
96 hrs
22
26
40
34
Rejection (percent)
Cu
98.8
97.2
84.8
93.4
Ni
99.0
98.3
83.4
95.0
Cr
(PH 5.1)
92.9
91.2
63.4
84.0
Cr
(PH 2.0)
64.7
64.7
30.1
62.7
-------
This was observed particularly in the case of the membrane designated
270-50E (4501 in thickness). The 96-hour water fluxes of the two
thicker membranes are not as high as might be desired but, combined
with their high copper and nickel rejections, would be adequate for
ultimate application.
The thinner membranes (270-50E and 270-50F) were adversely affected by
the pH 2.0 chromium solution with lower rejections and considerably
higher water fluxes. The lower chromium rejections were also noted
for the thicker g-GADE membranes although the fluxes were not affected.
Long periods of exposure to pH 2.0 chromium solution would probably
degrade most 3-GADE membranes—as indicated by increased fluxes and
lower rejections.
Conclusions. All three polymers have produced ultrathin membranes
cast in 1-3/8-inch-ID tubes that exhibited promising reverse osmosis
properties. These polymers were ready for adaptation to the 1/2-inch-
ID tubes which will be used in the actual application of reverse
osmosis to metal finishing wastes.
Reverse Osmosis Testing of Tube (1/2-Inch-ID) -Cast Membranes
The objective of this task of the program was to carry out short-term
reverse osmosis tests in 1/2-inch-ID tubes containing cellulose acetate,
CAOPSA, and 6-GADE ultrathin membranes to determine casting conditions
for high performance reverse osmosis membranes. The solvent systems and
general casting conditions were determined from the 1-3/8-inch tube-
casting studies. In this task the membranes were cast directly and
measured for reverse osmosis performance in the 1/2-inch-ID fiber glass
tubes (Abcor, Inc.).
Cellulose Acetate. The cellulose acetate E 360-60 membranes did not
show outstanding reverse osmosis properties in the 1/2-inch-ID tubes.
Because of a relatively "heavy" solvent system (4:1:5 acetic anhydride:
cyclohexanone:ethyl acetate) it was more difficult to cast this polymer
26
-------
In the narrower diameter tube. This resulted in many leaky and
imperfect membranes. Because of these poor results, further develop-
ment work on this polymer was not pursued.
A sampling of the best reverse osmosis results for the E 360-60 membrane
is given in Table 6. The copper ion rejection of 94.5 percent was the
highest attainable under the casting conditions developed during the
1-3/8-inch-ID tube-casting studies. This rejection value, coupled with
wide variability in the rejections of many membranes cast under the same
conditions, indicated that imperfect membranes were being formed.
Further casting development would be necessary to cast membranes of
cellulose acetate E 360-60 with consistently high reverse osmosis
performance.
Another reason for abandoning the 1/2-inch-ID tube-casting development
for the E 360-60 membrane was the excellent reverse osmosis performance
of 1/2-inch-ID tube-cast cellulose acetate E 398-10 membrane that was
observed during concurrent programs in water desalination. The solvent,
15:1 ethyl acetate:acetic anhydride, was ideally suited for 1/2-inch-ID
tube-casting. This membrane was considered during the first phase of
flat-cell testing, but was not particularly outstanding in reverse
osmosis performance. However, during the interval between the end of
the first phase of this metal finishing waste program (August 1970) and
the second phase, the E 398-10 cellulose acetate membrane was shown to
be particularly adaptable to 1/2-inch-ID tubes, giving high fluxes
(>20 gfd) and exhibiting excellent sodium chloride rejections (>97
(8)
percent) over a period of nine months on a brackish water stream.
The reverse osmosis performance of 1/2—inch-ID fiber glass tubes
containing ultrathin cellulose acetate E 398-10 membranes on copper feed
solutions is given in Table 7. Several draw rates were used at a
previously established casting solution concentration of two percent.
The thickness measurement for these 1/2-inch-ID, tube-cast membranes
was more difficult and was not carried out in these studies. However,
O
previous studies indicated a thickness variation of approximately 200A ^*
\,
(1.1 inch per second draw rate) to 600A (0.27 inch per second draw rate).
27
-------
Table 6^ Tubular Reverse Osmosis Performance of Ultrathin
Cellulose Acetate Membranes (£ 360-60) on Copper
Solutions at Two Draw Rates
Conditions:
Cast in 1/2-inch-ID tube (2% solution)
Tested in 1/2-inch-ID tube (2 feet in length)
Pressure ...... 600 psig
Temperature .... 25°C
Flow rate 3350 ml per minute
Copper feedwater
concentration . . 1000 mg per liter (pH 5)
Membrane
Number
2 70-42 A
270-43B
Draw Rate
(in. /sec)
0.24
0.33
Water Flux (gfd)
24 hrs
53
58
48 hrs
38
42
Cu
Rejection
(percent)
94.5
92.5
Table 7.
Conditions:
Tubular Reverse Osmosis Performance of Ultrathin
Cellulose Acetate (E 398-10) Membranes on Copper
Solutions: Effect of Draw Rate
Cast in 1/2-inch-ID tube (2Z solution)
Tested in 1/2-inch-ID tube (2 feet in length)
Pressure ...... 600 psig
Temperature . , . . 25°C
Flow rate ..... 3350 ml per minute
Copper feedwater
concentration . . 1000 mg per liter (pH 5)
Length of tests . . 24 hours
Membrane
Number
270-42C
270-43A
270-48B
270-48D
270-50A
270-50B
Draw Rate
(in. /sec)
0.27
0.33
0.70
1.00
1,10
1.10
Water Flux
(gfd)
25
30
23
38
34
33
Rejection
(percent)
99.9
99.0
99.5
98.4
96.3
98.8
28
-------
Table 7 shows that the highest copper ion rejections (99+) coupled
with adequate fluxes (>20 gfd) were observed for the "thicker" membranes
at draw rates below 0.7 inch per second. Because of the observed
membrane variability (Table 7, membrane 270-48B should be thinner and
exhibit a higher flux than membrane 270-43A) there is an obvious need
to optimize the tube-casting process.
Long—term reverse osmosis tests were performed on cellulose acetate
E 398-10 ultrathin membranes using several metal feed solutions over
500 hours and using a low pH copper solution (100 mg/1) for 800 hours.
During the 500-hour test the membrane exhibited rejections toward
copper (97 percent), nickel (95 percent), and chromium at pH 5.5 (96
percent). The flux remained above 40 gfd for these three feed solutions
over a period of 240 hours, indicating a very thin membrane. The lower
rejections (<99 percent) probably resulted from imperfections.
The second long—term test utilized two copper feed solutions: 1) 100
mg per liter copper ion at pH 2.8; 2) after 500 hours the copper ion
concentration was increased to 1000 mg per liter at pH 2.2 and testing
carried out for 300 more hours. At the end of 500 hours the flux was
above 30 gfd at a copper ion rejection of 96 percent. The change in
copper feed solution at 500 hours resulted in a lower flux (28 gfd)
I
and copper rejection (92 percent). The acid rejection during this
test was about 90 percent.
Although commercial tubes containing cellulose acetate E 398-10 ultrathin
membranes were being optimized for pilot testing at the end of the first
six months of Phase II, a membrane of this polymer did not represent a
major improvement over commercially available reverse osmosis membranes
for the treatment of metal finishing waste. Therefore, further
development of this membrane for this application was discontinued.
B-GADE. Ultrathin membranes of 3-GADE did not show immediate promise
as high-performing 1/2-inch-ID tubular reverse osmosis membranes. All
the tubular membranes exhibited very low copper rejections. Typical
29
-------
results are shown in Table 8. The low rejections for B-GADE were not
expected, since it had exhibited considerable promise in earlier
1-3/8-inch-ID tube-casting studies. Because of time considerations
on the programs, further attempts to cast 3-GADE in 1/2-inch tubes
were discontinued.
Cellulose Acetate 0-PrOpylsulfonlc Acid. During Phase I the cellulose
acetate 0-propylsulfonic acid (CAOPSA) membrane, in flat sheets,
exhibited the most promising overall reverse osmosis properties. In
addition excellent reverse osmosis results were obtained from this
membrane when cast in 1-3/8-inch-ID tubes.
At the beginning of the second phase considerable effort was spent trying
to fabricate CAOPSA membranes in 1/2-inch-ID two-foot Abcor fiber glass
tubes. All tubular-cast membranes exhibited very low salt rejections.
Two possible reasons for this poor performance in 1/2-inch tubes were
investigated:
1. Poor solubility of the CAOPSA polymer, resulting in
premature gelation during the tube-casting process.
[As the casting solution is drawn down inside the
1/2-inch-ID tube, solvent diffuses into the water
phase and water into the solvent phase. If the
polymer gels before the drawing is complete, a
potentially leaky membrane results.] Two approaches
were taken to alleviate the premature gelation problem:
a) varying the polymer synthesis procedure to increase
the solubility of the CAOPSA, and b) changes in the
solvent system. The original polymer synthesis
procedure was altered slightly to increase the solu-
bility of the CAOPSA, by distributing the 0-PSA groups
more equally throughout the polymer system. This CAOPSA
polymer possessed much improved solubility properties and
could easily be dissolved in a 7:1 ethyl acetate:acetic
anhydride solvent system. The new polymer, however,
exhibited approximately the same reverse osmosis
properties as unsubstituted cellulose acetate
(cellulose triacetate); i.e., adequate salt rejection
but water fluxes less than 10 gfd. Concerning the
second approach, attempts to alleviate the 1/2-inch
tube-casting problem by changing to an acetic
anhydride:any 1 alcohol solvent system produced results
similar to the previous acetic anhydride:ethyl
30
-------
Table 8. Tubular Reverse Osmosis Performance of Ultrathin
8-GADE Membranes on Copper Solutions
Conditions:
Cast in 1/2-inch-ID tubes
Tested in 1/2-inch-ID tubes (2 feet in length)
Pressure 600 psig
Temperature .... 25°C
Flow rate 3350 ml per minute
Copper feed
concentration . . 1000 mg per liter (pH 5)
Length of tests . . 24 hours
Polymer
3-GADE
(2% solution)
Membrane
Number
270-13A-2
270-13A-3
270-13A-1
270-13A-4
Draw Rate
(in. /sec)
0.17
0.31
0.45
0.62
Water Flux
(gfd)
59
97
100
93
Rejection
(percent)
22
22
11
30
-------
acetate solvent system (i.e., average salt rejections
of 55 percent and fluxes of 40 gfd). This new solvent
system should have decreased any water migration into
the organic phase and prevented premature gelling of
the polymer, since water is considerably less soluble
in amyl alcohol than in ethyl acetate.
These findings indicate that the poor membrane
performance, observed when CADPSA is cast in 1/2-inch-
ID tubes, is not a result of premature polymer gelling
due to water migration in the organic phase, as was
previously assumed.
2. Differences in testing procedures between membranes
cast in 1-3/8-inch tubes and 1/2-inch tubes. Initially
the CAOPSA polymer was cast in the 1-3/8-inch tubes
and tested in the flat reverse osmosis test cells with
excellent results. However, when CAOPSA was cast in
the 1/2-inch fiber glass tubes, the resultant membranes
were tested in the tubes. Using this difference in
testing procedure as the basic reason for the difference
in reverse osmosis performance of CAOPSA membranes cast
in 1-3/8-inch-ID and 1/2-inch-ID tubes, the following
is a possible explanation:
After an ultrathin membrane has been cast in
a 1/2-inch-ID tube, the system was then
pressurized, the polysulfone support and
membrane must conform to the sides of the
fiber glass tube. This expansion of membrane
and support does not occur in the flat test
cells. A certain degree of elasticity must
be present in both the support and membrane
to allov this expansion process to occur in
the tubes without damage to the membranes.
If these required elastic properties were not
present In the CAOPSA membrane, poor reverse
osmosis performance due to rupture in the
membrane would result.
A procedure was designed to eliminate the possible
elasticity problem encountered during fabrication
of the CAOPSA polymer In 1/2-inch-ID tubes. This was
accomplished by pressurizing the polysulfone support
liner In the fiber glass tube before casting the
ultrathin CAOPSA membrane onto it. Utilizing this
technique eliminated the requirement for the membrane
to expand to the sides of the fiber glass tube,
thus reducing the possibility of membrane rupture
during pressurization. The reverse osmosis results
obtained in two tubes with ultrathin CAOPSA membranes
cast by this procedure are presented in Table 9.
32
-------
Table 9. Tubular Reverse Osmosis Performance of
Ultrathin CAOPSA Membranes
Test Conditions:
Casting 1/2-inch-ID tubes
Tested in 1/2-inch-ID tube (2 feet in length)
Pressure 600 psig
Temperature . . . . 25°C
Flow rate 3350 ml per minute
Feed composition . . 1 percent NaCl
Tube
Number
50
70
Reverse Osmosis Results
Water Flux Through
The Membrane (gfd)
0.5 hour
24
29
24 hours
21
25
Salt Rejection
(percent)
0.5 hour
91
87
24 hours
94
90
In Table 9 the sodium chloride rejection was over
90 percent for both tubes after 24 hours. The mem-
brane would be expected to reject divalent ions,
such as Cu++, Zn""", or HI*"*", more efficiently than
sodium chloride. The results of this test indicate
that CAOPSA may be cast successfully in 1/2-inch
tubes by compensating for its lack of elasticity.
Testing of CAOPSA in tubes was terminated at this point because of
program time considerations.
Cond US ions. Three conclusions may be drawn from the preceeding studies.
(1) Cellulose acetate (E 398-10) ultrathin membranes may be success-
fully cast in 1/2-inch-ID tubes, but it was decided that a membrane of
this polymer did not represent a major improvement over commercially
available reverse osmosis membranes for the treatment of metal
finishing waste.
(2) Ultrathin 6-GADE membranes could not be cast successfully in 1/2-
inch-ID tubes. The results indicated that considerable development
33
-------
effort would be necessary to provide an effective tubular &-GADE reverse
osmosis membrane. However, this effort was not within the scope of
this program.
(3) CAOPSA ultrathin membranes were successfully cast in 1/2-inch-ID
tubes by compensating for the inelasticity present in the polymer.
Adequate flux and rejection were obtained from these tubes, but further
testing was precluded because of program time considerations.
General Conclusion of Tube-Casting Development
In general it was difficult to achieve the very promising reverse
osmosis properties for the new ultrathin membranes in the tubular
configurations as was observed for the flat-cast membranes during
Phase I. However, further tubular casting development would be expected
to result in effective membranes. In addition it is important to note
that the spiral wrap configuration for reverse osmosis would require
flat-casting of the membranes on a water surface followed by continuous
(9)
lamination to a polysulfone support. A similar process is currently
being carried out to produce artificial kidney membranes. Thus,
the excellent flat-cast reverse osmosis properties of the ultrathin
membranes may be realized more easily in the spiral wrap configuration.
Nonpolysaccharlde Membranes
In general polysaccharide membranes were found suitable for the reverse
osmosis treatment of metal finishing waste solutions only at pH's from
4 to 8. Many metal finishing effluents, however, are strongly acidic
or alkaline (i.e.* acid copper pH VL; zinc cyanide pH >11). Reverse
osmosis membranes comprised of nonpolysaccharide polymers would offer
greater chemical resistance to a wide variety of metal finishing waste
solutions. The NS-1 membrane originally developed for saline water
treatment showed promise for high chemical resistance during a concurrent
program funded by the Office of Saline Water. Thus, the NS-1 membrane
was applied to metal finishing waste solutions almost exclusively during
34
-------
the latter half of the program. A small effort was also directed
toward new polymers.
NS-1 Membrane
The NS-1 membrane [microporous polysulfone support film coated with
polyethylenimine (PEI) and crosslinked with tolylene 2,4-diisocyanate
(TEI)] is considered a major new-generation nonpolysaccharide membrane
for reverse osmosis. A schematic diagram of the NS-1 fabrication process
is given in Figure 4. The final coating on the polysulfone support film
consists of 1) amine-crosslinked PEI and 2) TDI crosslinked PEI (urea
linkages). An idealized structure of this coating is presented in
Figure 5. In this program the NS-1 membrane was evaluated for its
reverse osmosis performance on both highly basic (copper and zinc
cyanide) and acidic (acid copper chromic acid'and Watts nickel) rinses.
Cyanide Rinses. Reverse Osmosis Tests. For each test four two-foot
reverse osmosis tubes and one flat reverse osmosis test cell, containing
NS-1 membranes, were constructed. The simulated rinse water feed
solutions consisted of 1/10 the average concentration of common copper-
and zinc cyanide plating baths. The copper-cyanide rinse waters contained
1.9 grams per liter cooper and 3.1 grams per liter cyanide at a pH of
approximately 11.8. The simulated zinc-cyanide rinse water was prepared
by a 1:10 dilution of an actual alkaline zinc plating bath solution
obtained from Honeywell Inc. (approximately 1.3 grams per liter zinc
and 44.5 grains per liter sodium cyanide at a pH of 12.9).
Samples of the reverse osmosis product water were analyzed for cyanide
ions with the Beckman Model 917 Carbon Analyzer and metal ions with a
Varian Techtron AA-120 Atomic Absorption Spectrophotometer.
The results obtained from the four tubes and flat test cell during
the copper cyanide test are listed in Table 10. The average percent
copper and cyanide ion rejections after 500 hours were 99.9 and 98.7 with
an average flux of 10.1 gfd. In all cases the flux after 500 hours
35
-------
PEI IN WATER
TDI IN
HEXANE AND
HEAT CURE
SURFACE OF
POLYSULFONE
SUPPORT FILM
PEI COATING
CROSS LINKED
PEI-TDI
REACTED
ZONE
Figure 4. Schematic Representation of NS-1 Membrane
-------
10
•••J
;W
N
N^-*-
NH
CH
NH
^^.
II
o
•^N-»-+NH+-
NH
— NH
C=0
N
CHaCHaOROUP8 REPRESENTED
Figure 5. Idealized Structure of Polyethylenimine
Crosslinked with m-Tolylene 2,4-Diisocysmate
-------
Table 10. Reverse Osmosis Performance of NS-1 Membrane on Simulated
Copper Cyanide Rinse (1/10 of Plating Bath Concentration)
Water
Test Conditions:
Pressure 600 psig
Temperature . . . . 25"C
Flow rate 1650 ml/min
pH (Product Water) .11.8
Feed Composition:
CuCN . . 2.6 g/1
NaCN . . 4.4 g/1
NaOH . . 0.37 g/1
to
00
Tube Number
1
2
3
4
Flat Test
Cell
Reverse Osmosis Results
Water Flu
The Membr
2 hours
21.1
17.6
14.2
12.2
10.8
x Through
ane (gfd)
500 hours
11.9
11.6
10.3
9.1
8.2
Rejection (percent)
Copper Ion
2 hours
99.7
99.9
99.7
99.7
99.9
500 hours
99.9
99.9
99.8
99.8
99.9
Cyanide Ion
2 hours
97.4
98.7
98.8
98.7
99.1
500 hours
98.2
98.7
98.8
98.7
99.2
-------
had decreased to approximately 60 to 80 percent of the original values
while the rejection remained constant or increased. The tubes with
the highest initial fluxes decreased more than those with lower initial
fluxes. The flat test cell gave a higher rejection and lower flux than
the tubular membranes.
The data presented in Table 10 demonstrate effectiveness of the NS-1
membrane for the reverse osmosis treatment of alkaline copper cyanide
rinse waters. The NS-1 membrane was not observed to degrade under the
highly alkaline conditions of this test (the ion rejection remained
constant or increased). Also, if membrane deterioration had occurred,
the water flux through the membrane would have increased with time;
the opposite, however, was observed during the test.
Figure 6 is a typical plot of the water flux as a function of time in
the copper cyanide system. The flux decreased rapidly with time during
the first 100 hours of the reverse osmosis test after which the decline
leveled off. This decrease in flux through the membrane may be due to
fouling by a colloidal iron oxide which was observed on the surface of
the membrane after the 500-hour test.
An indication of how other membranes tested during Phase I compare with
NS-1 membranes in copper and cyanide ion rejection is given in Table 11.
A direct comparison, however, cannot be made between the rejection data
on NS-1 membranes and the other membranes since the tests were performed
under different test conditions and time periods. Both ultrathin
cellulose acetate 0-propylsulfonic acid and asymmetric cellulose acetate
degraded under the high pH environments. The sulfonated polyphenylene
oxide membrane remained stable for the 95-hour test duration. The NS-1
membrane remained stable for the 500-hour test duration at a higher pH
and exhibited significantly higher percent rejections of both copper
and cyanide ions than the latter membrane.
An actual zinc cyanide plating bath diluted 1:10 (pH 12.9) was utilized
during a 340-hour reverse osmosis test of the NS-1 membrane. The data
39
-------
O
LU
CC
LU
9
I
a:
LJ
I
100
a 99
98
TUBE * 2
TEST CONDITIONS:
PRESSURE
FEED
PH
.eoopsi
.1/10 COPPER PLATING
BATH
CONCENTRATION
.11.8
TEMPERATURE_25°C
o ©
200
300
TlME(hr)
400
500
Figure 6. Reverse Osmosis Performance of the NS-1 Membrane
with Simulated Copper Cyanide Rinse Water
-------
Table 11. Comparison of Cyanide and Copper Rejections of
NS-1 and Other Reverse Osmosis Membranes
Membrane
NS-1
Sulfonated poly-
phenylene oxide*
Ultrathin cellulose
acetate 0-propyl-
sulfonic acid*
Asymmetric
cellulose
acetate*
Cyanide
Rejection
(percent)
98.7
90.9
33.3
48.1
Copper
Rejection
(percent)
99.9
95.4
44.5
59.7
PH
11.8
10.4-11.4
10.4-11.4
10.4-11.4
Length of Test
(hours)
500
95
95
95
* Rejection data on membranes produced in Phase 1 given in
Reference 1.
listed in Table 12 were obtained from three tubes and a flat cell.
After the 340-hour test the average rejection of zinc and cyanide ions
was 99.8 and 95.5 percent with an average water flux of 11.3 gfd. These
results, similar to those obtained during simulated copper cyanide plating
bath reverse osmosis tests, indicate that the NS-1 membrane is effective
for the treatment of highly alkaline zinc cyanide waste waters.
Figure 7 contains typical plots of water flux and cyanide rejection as
a function of time for the zinc cyanide system. These results demonstrate
the absence of membrane degradation in the treatment of the highly
alkaline (pH 12.9) zinc cyanide plating bath rinse waters by NS-1
membranes.
The significance of this new membrane is illustrated in Table 13 which
is a comparison of the reverse osmosis performance of three commercially
available membranes with the NS-1 membrane. The first three membranes
listed represent commonly used reverse osmosis membranes.
Flux Decline. In both the copper and zinc cyanide tests the water flux
at the end of the tests was between 60 to 80 percent of the initial
41
-------
Table 12. Reverse Osmosis Performance of NS-1 Membrane on Actual
Zinc Cyanide Plating Bath Diluted 1:10
Test Conditions:
Pressure 600 psig
Temperature .... 25°C
Flow Rate ...... 1650 ml/min
Feed Composition:
Zn"^ . . 12.7 g/1
CN- . . 22.4 g/1
NaOH . . 75 g/1
pH . . . 12.9
Tube Number
30
33
35
Flat Test
Cell
Reverse Osmosis Results
Water Flux Through
The Membrane (gfd)
0,5 hour
16.8
22.0
14.8
13.7
340 hours
12.9
14.2
10.9
10.1
Rejection (percent)
Zinc Ion Cyanide Ion
0.5 hour
99.7
99.6
99.7
99.7
340 hours
99.8
99.7
99.8
99.8
0.5 hour
94.5
94.0
95.5
95.5
340 hours
95.5
94.5
96.0
96.0
-------
~ 97
I 96
LU
UL
LU
O
u>
95
94
TEST CONDITIONS
O
"•a
X
_J
a.
cc
UJ
i
28
26
24
22
20
18
16
14
12
10
8
6
4
2
o
FFFn
•>
rVs^__ TEMPERATURE —
_ — _ 1 i i ,
0 100 200 300 400
TIME (hr)
ouu psi
I / m n ATI i
I/IU BATH
CONCENTRATION
OF
Zn (CN)2
— 25°C
500
Figure 7. Beverse Osmosis Performance of an NS-1 Membrane
with Simulated Zinc Cyanide Rinse Water
-------
Table 13. Comparison of Cyanide Rejection for
Membranes* with Zinc Cyanide Plating
Rinse Waters
Membrane
Cyanide Rejection
(percent)
Cellulose Acetate
Eastman RO-97
Cellulose Acetate Butyrate
Universal Water Corporation
Nylon Hollow Fibers
DuPont "Permasep B-51
NS-1**
Membrane Deteriorated
Membrane Deteriorated
28.0
95.5
* Data on the cellulose acetates and hollow fibers given
by A. Golomb, Plating, April 1972. Test solution for
these membranes: zinc cyanide plating rinse at 1 percent
of bath concentration, pH 11.8 at 600 psi.
** NS-1 results obtained at 10 percent of bath concentration
pH 12.9 after 340-hour operation.
44
-------
readings. This flux decline may be the result of fouling on the membrane
surface by colloidal iron oxide which was observed after the 500- and
340-hour tests. Figure 8 is a photograph of the membranes in tubes
1) after casting and no testing and 2) after the testing. This iron
oxide apparently is produced by the chemical action of the highly basic
cyanide solutions on the stainless steel surfaces of the reverse osmosis
system. This problem may possibly be prevented by utilizing plastic
materials, such as poly(vinyl chloride), in the system whenever possible.
Currently, commercially available plastic tubing and fittings are being
utilized in high-pressure reverse osmosis operations. Chemically
removing the iron oxide with complexing agents such as citric acid,
oxalic acid or ethylenediaminetetracetic acid (EDTA) may also provide
a solution to this fouling problem.
Effect of Temperature. The effect of feed temperature on the reverse
osmosis performance of the NS-1 membrane was studied by varying the
temperature of the zinc cyanide test solution by 10°C intervals between
25° and 55°C. Samples of the water flux across the NS-1 membrane were
obtained after the system equilibrated for one hour. The results are
plotted in Figure 9 and indicate that flux increased linearly with
temperature in this temperature range. The increases in feed temperature
from 25° to 55°C caused no significant changes in the cyanide ion
rejection in the membrane. Thus if it were not feasible to reduce the
rinse water temperature from a hot plating bath prior to reverse osmosis
treatment the NS-1 membrane could perform effectively.
Acidic Rinses. The NS-1 membrane was found to be effective for the
treatment of highly acidic non-oxidizing metal finishing wastes.
Reverse osmosis tests were performed on simulated acid copper, Watts
nickel, and chromic acid rinse waters.
Acid Copper. The NS-1 membrane was found to be resistant to chemical
degradation under highly acidic conditions (pH 0.5). Ultrathin cellulose
acetate membranes did not demonstrate this attractive property. The
results of reverse osmosis tests on NS-1 and ultrathin cellulose
45
-------
Figure 8. Illustration of Iron Oxide Fouling.
Photograph of Cut-away Sections of
Reverse Osmosis Tubes (A) After
Casting and (B) After 340-Hour
Zinc Cyanide Test
46
-------
TEST CONDITIONS:
TEMPERATURE 25° - 55°
PRESSURE 600 psi
FLOW RATE I650mi/min
FEED 1/10 ZINC PLATING BATH
pH 12.9
36
34
32
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
CYANIDE ION REJECTION (%)
0 5 10 15 20 25 30 35 40 45 50 55 60 65
TEMPERATURE (C°)
Figure 9. Effect of Operating Temperature on the Water
Flux Behavior of NS-1 Reverse Osmosis Membranes
47
-------
acetate membranes using simulated acid copper plating bath rinse water
at pH 0.5 is presented in Table 14. After 210 hours of testing the
copper rejection in the cellulose acetate membrane had decreased
significantly with a corresponding increase in water flux; hydrolysis
of the cellulose acetate membranes was occurring. However, no degrada-
tion of the NS-1 membrane was observed after 550 hours.
Figure 10 is a typical plot of water flux and percent copper rejection
as a function of time for the acid copper system. The NS-1 membrane
exhibited a constant copper rejection at 99.8 percent with only a small
decrease in water flux (10.1 - 9.1 gfd) during the test. This lower flux
than usual was due to the high osmotic pressure of the feed solution.
These results demonstrate the absence of membrane degradation in the
treatment of highly acidic (pH 0.5) copper plating rinse waters by
NS-1 membranes.
Table 14. Comparison of NS-1 and Ultrathln Cellulose Acetate
(Eastman 398-10) Membranes for Copper Rejection and
Flux on Simulated Acid Copper Rinse Waters in Flat
Test Cells
Test Conditions:
Temperature 25 °C
Pressure 600 psig
Feed Composition:
(1/10 Bath Concentration) CuSO
45H2°
pH
20 g/1
0.5 (H,
Membrane
NS-1
Ultrathin
cellulose
acetate
(E 398-10)
Reverse Osmosis Results
Water Flux Thr
The Membrane (
0.5 hour
10.1
8.8
210 hours
9.4
13.7
ough
Rfd)
550 hours
9.1
—
Copper !
(
0.5 hour
99.8
99.6
Eon Rejection
percent)
210 hours
99.8
85.1
550 hours
99.8
—
48
-------
5* 100
*»
VO
O 99
Q
o
e—o-
UJ
cr
o:
UJ
t
o
o
en
X
_J
cr
UJ
98
97
TEST CONDITIONS:
PRESSURE
FEED
100
200 300
TIME (hr)
400
•600 psi
•CuS04-5H20-20g/l
12
10
8
6
4
2
n
TEMPERATURE 25 °C
•
•
•
m
— - ' 1 1 1 • '
500
600
Figure 10. Reverse Osmosis Performance of an NS-1 Membrane
With Simulated Acid Copper Rinse Water
-------
Watts Nickel. The results of tests on. simulated Watts nickel rinse
waters (pH 4.0) and a comparison of the NS-1 membrane reverse osmosis
performance with cellulose acetate membranes (Eastman Type RD-97) is
presented in Table 15. Both membranes exhibited similar initial reverse
osmosis results. The tests on the NS-1 membrane were discontinued after
210 hours because of time limitations. No deterioration of the NS-1
membrane was observed during the 210 hours of the test. Its performance
was slightly better than the standard cellulose acetate membrane;
however, it is apparent that no significant advantage would be obtained
using the NS-1 membrane instead of the cellulose acetate membrane
for Watts nickel baths.
Table 15. Comparison of NS-1 and Cellulose Acetate (Eastman
Type RO-97)* Membranes for Nickel Rejection and Flux
on Simulated Watts Nickel Rinse Waters in Flat Test
Cells
Test Conditions:
Temperature 25°C
Pressure 600 psig
Feed Composition:
(1/10 bath concentration) NiSO
4 6H2°
pH
33.75 g/1
7.50 g/1
3.75 g/1
4.0
Membrane
NS-1
Cellulose
acetate
(Eastman
RO-97)
Reverse Osmosis Results
Water Flux Thr<
The Membrane (j
[nitial
20.5
17.9
210 Hours
16.3
Jugh
fd)
10 Weeks
—
13.1
Nickel ion Rejection
(percent)
Initial
99.8
99.8
210 Hours
99.9
10 Weeks
—
98.9
* Data on cellulose acetate given by A. Golumb, Plating^ October 1970.
50
-------
Chromic Ac-id. The NS-1 membrane was ineffective in the reverse osmosis
treatment of chromic acid wastewaters at pH 1.5. The feed solution in
a 7.5 hour test contained 24.6 g/1 chromic acid and 0.247 g/1 H-SO or
1/10 the average plating bath concentration for both materials. Figure 11
illustrates the flux behavior during the test. At the end of 7.5 hours
the flux increased by a factor of five. An average chromic acid rejection
of 96 percent for the system was observed at the beginning of the test;
however, at the conclusion only a 13 percent rejection was recorded.
Apparently, the chromic acid oxidizes the PEI-TDI coating at the
nitrogen linkages breaking the chains and degrading the polymer. This
would cause a decrease in rejection accompanied by an increase in flux.
Thus chromic acid wastewaters that have not been adjusted to near
neutral pH's could not be effectively treated by NS-1 membranes.
GANTREZ AN-Poly(Vinyl Alcohol) Membranes
A small effort was carried out to determine new membrane polymers that
would be resistant to oxidizing acidic metal finishing wastes. The
most promising membrane found to date was a material formed by the
reaction of GANTBEZ AN [poly(methylvinylether/maleic anhydride)] with
poly(vinyl alcohol) on a polysulfone support. Water fluxes of over 8 gfd
at salt rejections of 90 percent were observed after a 48-hour test at
pH 1 for the feed solution.
•
Table 16 gives the results of the various membranes formed by immersing
the polysulfone support in solutions containing varying amounts of
GANTBEZ AN and poly(vinyl alcohol). The membrane formed by the reaction
of 2 percent GANTREZ AN and 0.5 percent PVA solution gave the most
promising results.
Conclusions
Five conclusions may be drawn from the data presented in this section:
1. The NS-1 membrane is an excellent candidate for treating
alkaline cyanide waste solutions. During the long-term
51
-------
U)
N>
70
65
60
^ 55
I SO
x
g
45
40
35
25
20
15
10
5
0
TEST CONDITIONS:
TEMPERATURE
PRESSURE-
FLOW RATE
FEED
PH.
25°C
eoopsi
1650 mi/min
24.7g/liter CHROMIC ACID
,0247g/liter SULFURIC ACID
1.5
CHROMIC ACID REJECTION
INITIAL__96%
FINAI 13 %
456
TIME (hours)
8
10
Figure 11. Reverse Osmosis Performance of an NS-1 Membrane
with Simulated Chromic Acid Rinse Water
-------
Table 16. Reverse Osmosis Performance of GANTREZ AN-
Poly(Vinyl Alcohol) Membranes Prepared from
Solutions of Various Polymer Concentrations*
Test Conditions:
Temperature 25° C
Pressure 600 psig
Feed 1 percent NaCl
pH 1.0
Preparation Solution
GANTREZ AN
(percent in
aqueous solution)
1.0
0.5
1.5
2.0
Poly (Vinyl Alcohol)
(percent in
aqueous solution
1.0
1.5
0.5
0.5
Reverse Osmosis Performance
Rejection (percent)
40
32
65
90
Flux (gfd)
1.5
0.8
3.2
8.0
Membranes were prepared by immersion of a polysulfone support film
into aqueous solutions containing the polymers.
tests on alkaline copper and zinc cyanide solutions
at pH's of 11.8 and 12.9, respectively, NS-1
exhibited:
a. No membrane degradation during both the
500- and 340-hour tests.
b. An average water flux of 10.1 and 11.8
after 500 and 340 hours of testing. An
average water flux decline of 20 to 40
percent for both tests may be attributed
to colloidal iron oxide coating on the
membrane. The coating apparently is
produced by the chemical action of the
highly basic cyanide solution on the
reverse osmosis system's stainless steel
surfaces.
c. High rejection of the solute species
(99.9 percent for copper and 98.7 percent
for cyanide in alkaline copper plating
bath and 99.8 percent for zinc and
53
-------
95.5 percent for alkaline plating bath
at the conclusion of the tests).
d. Stability under high temperatures (no
significant changes in cyanide rejection
occurred during operation at 55°C while
the flux increased approximately 84
percent over the 25°C value).
2. The NS-1 membrane is resistant to highly acidic acid
copper rinse waters (pH 0.5) and is probably the only
membrane capable of treating these effluents. The
membrane exhibited a constant copper ion rejection
at 99.8 percent with only a small decrease in water
flux (10.1 to 9.1 gfd) during a 550-hour test.
3. The NS-1 membrane is effective for the treatment of
Watts nickel rinse waters although not significantly
better than commercially available cellulose acetate
membranes.
4. The NS-1 membrane is ineffective in the reverse
osmosis treatment of chromic acid wastewaters at
pH 1.5.
5. A membrane which may have good resistance to chemical
degradation in acidic environments has been fabricated
from a polymer formed by the reaction of GANTBEZ,
poly(methylvinylether/maleic anhydride) with
poly(vinyl alcohol) on a microporous polysulfone
support.
54
-------
SECTION VI
POSSIBLE APPLICATIONS OF REVERSE OSMOSIS
TO THE METAL FINISHING INDUSTRY
During Phase I of this program reverse osmosis was shown to be ideally
suited for the treatment of metal finishing effluents. The process can
purify plating rinse waters and concentrate the plating ions to levels
where recycle to the plating bath is economically attractive. In cases
where the concentrated solution cannot be recycled to the plating tank,
the concentrate may be refined for the dissolved salts. However, in view
of the wide variety of conditions and operations found in the plating
industry, reverse osmosis treatment processes must be tailored for each
installation and plating process.
Before a reverse osmosis system can be applied to a metal finishing
operation, certain parameters must be established:
1. Type and concentration of dissolved material in the
rinse water.
2. The degree that various materials are rejected by the
membrane.
3. The dragout from plating tanks.
4. Maximum permissible concentration of dissolved material
in the last rinse.
5. Evaporation rate from the plating tank.
6. Feed flow rate in reverse osmosis module.
Using these parameters, calculations may be performed to establish the
most efficient designs for utilizing reverse osmosis in plating
operations. A detailed discussion of typical plating bath lines is
presented in Reference 1.
Figure 12 and 13 illustrate design concepts for reverse osmosis systems
utilizing NS-1 membranes for the treatment of Watts nickel and zinc
cyanide plating wastes. In these closed-loop systems the dissolved
material present in the rinse water, due to drag-out, is concentrated by
55
-------
Evaporation
Ul
1
, Dragout 50 gal/day
I
I
I
I
I
Nickel 12.5 oz/gal
Bath
i
120 gal/day
Concentration
Ni 5.17 oz/gal
1 |
! ! /
'
I
t I
Ni 0.26 oz/gal
Rinse 11
\
CM
R.O.
^^ f -— ^^ *HM MM •"••" «^^~ ^^ ___ '- ^^^
^ — ^x^ r
\ 1
XI -i 120 gpd
| j " Makeup
t 1
Ni 0.0068 oz/gal
Rinse #2
'
2280 gal/day
Permeability
Membrane — NS-1
Water Flux - 15 gfd at 600 psi
Nickel Rejection - 99.9%
Membrane Area — 160 ft2
Rinse pH «4
Ni 0.001 6 oz/gal
Figure 12. Design Concept for Nickel Plating Waste
Treatment Utilizing Reverse Osmosis
-------
Evaporation
t_n
.
Dragout 50 gal/day
r- -j r
1
Zinc 6 oz/gal
Cyanide 7.4 oz/gal
BATH
i
ra
5
-
1
1
1
t
CN=0. 178 oz/gal
Rinse #1
>.
Z "S
o *
~N CM
S
i
o
1
o
in
0
d
>
re
a
R.O.
95% Reduction
a
Perm
>
n
^
a
S
z
1
N
o
r^
o
d
Cone.
in Vol. of Feed
~" r^~"
i i *
1 1
l i
I !
CN = 0.0042 oz/gal
Rinse ffjL
>
to
1
CM
R.O.
\
— \ ' '
i i
t i
i i
1 i
CN = 0.00056 oz/gal
..» mnrn at
a
1
Perm
CN « 4.2 ppm
§
•= Membrane - NS-I
L/ry
+
120gpd
Make up
£ Water Flux - 1 5 gfd at 600 psi
8 Zinc Rejection - 99.8%
Cyanide Rejection — 95.5%
Membrane Area - 320 ft2
*
First Rinse pH * 12
Figure 13. Design Concept for Zinc-Cyanide Plating
Waste Treatment Utilizing Reverse Osmosis
-------
reverse osmosis and recycled to the plating rinse. These systems were
designed for 95 percent recovery; the rinse water (input feed) would
thus be reduced to 5 percent of its original volume. The final
concentration would be somewhat less than 20 times the original rinse
concentration, since a certain amount of the dissolved material present
in the rinse passes through the reverse osmosis membrane with the permeate
water. The amount of material passing through the membrane is dictated
by the ability of the membranes to reject the various dissolved materials.
Reve rs e Os mos i s Sys tern
The reverse osmosis system concentrations presented in Figures 12 and 13
are based on a 95 percent reduction in input feed volume and a 99.8
and 95.5 percent rejection of nickel and cyanide ions, which are the
values exhibited by the NS-1 membrane on these simulated plating rinses.
The calculations (derived in the Appendix) used to determine the
concentrations of the reverse osmosis product water and concentrate
required only the percent rejection values exhibited by the NS-1
membrane. To accomplish water recovery values presented in these
figures, the reverse osmosis systems must, have the following performance
parameters:
• Pressure 600 psi
• Membrane water flux . . 15 gfd
2
• Total membrane area . . 160 ft
• Rejection (percent)
Nickel 99.9%
Zinc 99.8%
Cyanide 95.5%
Watts Nickel
In the nickel plating system, Figure 12, 2,400 gallons per day of rinse
water would be concentrated to 120 gallons per day and recycled back
to the plating bath. The product water (2,280 gallons per day) would be
58
-------
utilized in the second rinse. With this system only 120 gallons per
day of DI water is required for make-up. In Figure 12 it was assumed
that the volume lost by evaporation and dragout in the plating bath
would be equal to the volume amount of rinse concentrate added to the
plating bath. However, this may not be the case in an actual plating
operation. In an actual operation, the volume of rinse concentrate
added to the plating bath would be dictated by total volume loss in
the plating bath due to dragout and evaporation. For example, if this
total volume loss was less than the final concentrate volume, the
concentrate volume could be further reduced by evaporation.
By utilizing the reverse osmosis system described in Figure 12, 620.4
ounces of nickel per day would be recycled back to the plating bath
and 2,280 gallons per day of water is recycled into the rinse system.
Assuning that the average cost of nickel is $1.75 per pound, a savings
of $67.85 per day for nickel alone could be realized with this operation.
Zinc Cyanide.
The zinc cyanide plating system (Figure 13) must employ a reverse
osmosis unit for the first and second rinses, since the cyanide ion
is not as efficiently rejected by the membrane as the nickel ion.
In this scheme the concentrated rinse water from the first rinse bath
is recycled to the plating bath, while the purified water is recycled
back to the first rinse. The concentrate from the second rinse is
cycled to the first rinse and its purified water is used in the third
rinse. This system also requires 120 gallons per day of DI water for
make-up. With this reverse osmosis system 369.6 ounces per day of
cyanide is recycled back to the plating bath with 2,280 gallons per
day of rinse water recycled back through the plating operation. The
same plating bath evaporation rate conditions discussed in the Watts
nickel bath section would apply in the zinc cyanide system. Assuming
that the average cost of zinc is $0.80 per pound, a savings of $30.02
per day for zinc alone could be attained with this operation.
59
-------
Operating Mode—Reverse Osmosis
The mode of operations used in both Figures 12 and 13 are single-pass
operations on the rinse waters using tubular units to support the
membrane. To maintain the volumetric flow rate and velocity in the
tubes during reverse osmosis treatment, the cross-sectional area or
the number of tubes would have to be reduced in downstream sections
of the reverse osmosis system.
The nvtn-tnmm amount of space that would be required by the reverse
osmosis tubes alone would be 14.7 cubic feet. This volume would
contain approximately 235 1/2-inch-ID by 60-inch tubes with a total
membrane surface of 160 square feet.
Results
Both the design concepts illustrated in Figures 12 and 13 are
represented as closed-loop systems, which would not only result in
considerable water and material savings but also eliminate many of
the pollution problems associated with plating operations.
The recovery of the materials and water for the two reverse osmosis
systems are given below:
Make—up deionized rinse water
Water recovered for reuse
Nickel recovered for reuse
Zinc recovered for reuse
Cyanide
With R. 0.
120gpd
99 + %
99 + %
99 + %
99 + %
Without R. 0.
2400gpd
0%
0%
0%
0%
60
-------
SECTION VII
REFERENCES
1. Nelson, B. R., Rozelle, L. T., Cadotte, J. E., and Scattergood,
E. M., Use of Reverse Osmosis for Treating Metal Finishing
Effluents, Final Report, EPA Program No. 12010 DRH. U. S.
Government Printing Office, Washington, D. C. (November, 1971).
2. Pickering, Q. H., and Henderson, C., "The Acute ToxLcity of Some
Heavy Metals to Different Species of Wanowater Fishes", Air and
Water Poll. Int. J., 10, 453 (1966).
3. Interaction of Heavy Metals and Biological Sewage Treatment
Processes, Public Health Services Publication No. 999-WP-22 (1965).
4. Shea, J. F., Reed, A. K., Tewksbury, T. L., and Smithson, G.R., Jr.,
A State of the Art Review an Metal Finishing Waste Treatment,
Federal Water Quality Administration, U. S. Department of the
Interior, Program No. 12010 EIE 11/68, Grant No. WPRD 201-01-68
(November 1968).
5. Dobb, E. H., "Metal Wastes, Contribution and Effect", Tech. Proc.
Amer. Electroplaters Soc., 53 (1958).
6. Rozelle, L. T., Cadotte, J. E., King, W. L., Senechal, A. J., and
Nelson, B. R., Development of Ultrathin Reverse Osmosis Membranes
for Desalination, Office of Saline Water Research and Development
Progress Report No. 659, U. S. Government Printing Office,
Washington, D. C. (June 1971).
7. Rozelle, L. T., Cadotte, J. E., and McClure, D. J., Development of
New Reverse Osmosis Membranes for Desalination, Office of Saline
Water Research and Development Report No. 531, U. S. Government
Printing Office, Washington, D. C. (June 1970).
8. Rozelle, L. T., Cadotte, J. E., Senechal, A. J., King, W. L., and
Nelson, B. R., "Tubular Ultrathin Membranes for Water Desalination",
Reverse Osmosis Membrane Research, H. K. Lonsdale and H. E. Podall,
ed., Plenum Press, New York, (1972).
9. U. S. Patent 3,551,244, "Method of Producing an Ultrathin Polymer
Film Laminate", Forester, R. H. and Francis, P. S., 1970.
10. Rozelle, L. T. and Petersen, R. J., Development of a. New Concept
in Membrane Structure for Application in Hemodialysis, National
Institute of Arthritis, Metabolic and Digestive Diseases, National
Technical Information Service, Springfield, Va., in press.
61
-------
SECTION VIII
PUBLICATIONS
Rozelle, L. T., Cadotte, J. E., Nelson, B. R., and Kopp, C. V.,
"Ultrathin Membranes for Treatment of Waste Effluents by Reverse
Osmosis", Journal of Applied Polymer Science, Applied Polymer Symposia,
in press; also presented at NASA symposium on Polymeric Materials for
unusual Service Conditions, Moffett Field, California, December 1972.
Rozelle, L. T., Cadotte, J. E., Kopp, C. V., and Cobian K. E., NS-1
Membranes: Potentially Effective New Membranes for Treatment of Washtfater
in Space Cabins, ASME Paper No. 73-ENAs-19, New York; also presented
at the Intersociety Conference on Environmental Systems, San Diego,
July 1973.
62
-------
SECTION IX
APPENDIX
Reverse Osmosis Process Calculations
The concentration values for the brine and product waters in Figures
12 and 13 (pages 56 and 57) were calculated from mass balance
determinations . The model and nomenclature for these calculations
are shown in Figure 1. In this model, the dissolved solids
concentration permeating through any finite area (A) of the membrane
was computed by:
Cp = (l-R)3Cp (1)
This equation expresses the concentration of dissolved solids permeating
through the membrane as a function of their concentration in the
feedwater (Cp). The fraction of dissolved solids rejected by the
membrane (R) and the concentration polarization factor (3) were
considered constant in the calculations.
As the feedwater passes through the reverse osmosis module, its
dissolved solids concentration (C ) continuously increases, due to water
permeating through the membrane. This change in feed concentration
consequently increases the amount of dissolved solids permeating
through the membrane with the product water. Therefore, both the
permeate and feed concentrations (C , C ) are a function of the feed
volume (Vp) . A clearer understanding of this process may be obtained
by considering reverse osmosis as a continuous method for concentrating
unit amounts of the input feed solution. Therefore, an approximation
for the function of feed volume (V ) may be obtained (for materials
which are efficiently rejected in the reverse osmosis system) by
assuming that the dissolved solids concentration in the feed at
a given time (CO is inversely proportional to the volume of feed
at the same time (V_) :
63
-------
Feed (F)
(Rinse Water)
REVERSE OSMOSIS MODULE
V
Product
water (P)
Brine (B)
(Concentration)
NOMENCLATURE
p
•»o
' F
'F
P
V =
Concentration of water permeating through
reverse osmosis membrane
Concentration of reverse osmosis module
product water
Concentration of feed before entering
module
Concentration of feed in reverse osmosis
module at a given time
Total mass of dissolved material in a
unit volume of product water
Volume of Feed before entering module
Volume of feed in module at a given time
Volume of product water (V°p-Vp)
Final volume of product water
Fraction of dissolved material rejected
Concentration polarization—-dimensionless
Figure 1. Reverse Osmosis Model-.and Nomenclature
for Mass Balance Calculations
64
-------
where C° and V°_ are the initial feed concentration and volume
r r
respectively.
A volume balance gives:
V° = V + V
F VF P (3)
where V is the volume of the product water.
Combining Equations 2 and 3:
C°F
C -
F
(4)
The feed concentration is slightly over-estimated by this approximation,
since a small amount of the dissolved material passes through the membrane
with the permeate water.
The total mass of dissolved material transported across the membrane
(M ) can be obtained by integrating C from 0 to V f :
M - f
P J.
"Vpfc dv
P P
(5)
65
-------
Combining Equations 1, 4, and 5:
rV CV
M -J *
-------
SELECTED WATER
RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
1. Report No.
3. Accession No.
w
4, Title
NEW MEMBRANES FOR REVERSE OSMOSIS TREATMENT
OF METAL FINISHING EFFLUENTS
5. Reporl Date
6.
Orfanization
7. Author(s)
Rozelle, Lee T., Kopp, C. V., Jr., Cobian, K. E.
North Star Research Institute
Under Contract To
Minnesota Pollution Control Agency
n^ : Environmental Protection Agency
il. Contract/Grant No.
12010 DRH
'.i • y ue of Report awd
. /wit Covered
15. Supplement^ Note*
Environmental Protection Agency Report Number
EPA-660/2-73-033, December 1973
16. Abstract
An important new membrane has been developed for the reverse osmosis treatment of both
highly alkaline and acidic (non-oxidizing) metal finishing rinse waters. This
membrane designated NS-1, and originally developed for seawater desalination, consists
of the following: a microporous support film (polysulfone) coated with polyethyleni-
mine which is cross-linked with tolylene 2,4-diisocyanate.
Simulated alkaline copper and zinc cyanide .plating rinses at pH's of 11.8 and 12.9
were treated by NS-1 membranes during 500- and 340-hour tests without deterioration
of reverse osmosis properties. Water fluxes above 10 gallons per square foot (of
membrane) per day (gfd) were observed with cyanide rejections between 95 and 99
percent. The NS-1 membrane also treated simulated copper sulfate rinse waters
effectively at pH 0.5 during 550-hour tests without deterioration of reverse osmosis
properties (fluxes above 10 gfd with 99.8 percent rejection of copper). The NS-1
membrane is the only known membrane that can perform well using both acidic and
alkaline feed solutions.
Preliminary engineering considerations indicated the feasibility of applying the
NS-1 membrane to reverse osmosis treatment and recycle of nickel and zinc cyanide
electroplating rinse waters.
a. Descriptors *Reverse osmosis, *Water pollution
i?b. identifiers ^Treatment of metal finishing wastewaters, *Polymer Membranes,
laboratory Study, Metal Salts, Electroplating Wastes
17c, COWRR Field & Group
18. Availability
19. Security Class.
(Report)
;0 Security C,..
(Page)
2). No: of
Fages
22. ?tke
Send To:
WATCH RESOURCES SCIENTIFIC INFORMATION CENTER
OS. DEPARTMENT OF THE INTERIOR
WASHINGTON. OJC. 20240
Lee T. Rozelle
Institution
North Star Research Institute
WRSIC 102 (REV. JUNE 1971)
G P O 488-935
------- |