EPA-600/2-76-223
October 1976
Environmental Protection Technology Series
DIALYSIS FOR CONCENTRATION AND
REMOVAL OF INDUSTRIAL WASTES
Robert S. Kerr Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Ada, Oklahoma 74820
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. 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.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-76-223
October 1976
DIALYSIS FOR CONCENTRATION AND REMOVAL
OF INDUSTRIAL WASTES
by
James K. Smith
Shyamkant V. Desai
R.E.C. Weaver
Ellas Klein
Gulf South Research Institute
New Orleans, Louisiana 70186
and
Louisiana State Department of Commerce and Industry
Grant 12020 EMI
Project Officer
L,. Frank Mayhue
Source Management Branch
Robert S. Kerr Environmental Research Laboratory
Ada, Oklahoma 74820
ROBERT S. KERR ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
ADA, OKLAHOMA 74820
-------
DISCLAIMER
This report has been reviewed by the Robert S. Kerr Environmental Research
Laboratory, U.S. Environmental Protection Agency, and approved for publication.
Approval does not signify that the contents necessarily reflect the views and
policies of the U.S. Environmental Protection Agency, nor does mention of trade
names or commercial products constitute endorsement or recommendation for use.
-------
FOREWORD
The level of organics in water supplies is increasing. Increased con-
centrations of organic pollutants in the Mississippi River have paralleled
closely the explosive development of a petrochemical industrial complex in the
1950s and 1960s. This expansion has resulted in the location of more than 60
major industries between Baton Rouge, Louisiana and the mouth of the river.
Most of these industries discharge their partially treated or untreated wastes
to the river. Industrial expansion is essential to the economy of our community
and nation, but just as critical is the effect that this expansion may have on
our environment. To control environmental pollution that can result as industry
expands, and to mitigate existing problems, it is necessary to establish treat-
ment procedures that will not place undue burden on industrial installations but
will assure a safe environment.
This investigation has studied one process: dialysis applied to treatment
of industrial effluent streams. Plant effluent streams were examined for poten-
tial treatment by dialysis according to the following criteria:
1. Pollution significance
2. Feasibility of treatment by dialysis
3. Industrial interest
The work plan for the project consists of three phases:
1. Selection of 10 waste streams potentially suitable for treat-
ment by dialysis, and a preliminary characterization of
potentially useful dialysis membranes applicable to waste
treatment.
2. A laboratory analysis of several waste-membrane combinations
employing actual waste to determine engineering and transport
data.
3. A semipilot continuous-flow evaluation of one of the optimum
waste-membrane combinations, employing a scaled-up dialyzer to
demonstrate technical and economic feasibility.
This report was submitted in fulfillment of Grant 12020 EMI by Gulf
South Research Institute for the State of Louisiana under the sponsor-
ship of the U.S. Environmental Protection Agency. Work was completed
as of July, 1972.
111
-------
CONTENTS
Page
Disclaimer ii
Foreword iii
List of Figures vii
List of Tables viii
Nomenclature ix
Acknowledgements xi
I. Introduction 1
II. Summary 7
III. Conclusions 8
IV. Recommendations 10
V. Contaminant Screening 11
A. The Apparatus 12
1. Rotary Batch Dialyzer 12
2. Pervaporation Test Cell 14
B. Membranes 16
1. Commercial Films 16
2. Films Prepared in the Laboratory by GSRI 17
C. The Permeability Measurements 17
D. Results 18
1. Dialysis with Conjugation 18
2. Dialysis with Complexing 26
a. Choice of Membrane 27
b. Selection of Complexing Agents and Solvent
Combinations 28
c. Results from Rotating Cell Screening Tests 28
Aluminum System 28
Copper and Lead System 32
3. Pervaporation 32
-------
CONTENTS, Continued
page
VI. Pilot Test of the Pervaporation Process 40
A. Equipment , 42
B. Results 43
1. Effect of Feed Flow Rate 43
2. Effect of Air Flow Rate 47
3. Effect of Other System Variables 47
VII. Projection to Commercial Operation 49
VIII. Appendices 55
APPENDIX A - Permeability Coefficients - Dialysis with Conjugation.. 55
APPENDIX B - Distribution Coefficients 59
APPENDIX C - Pervaporation Transport Analysis 60
1. Flux Formulations 60
2. Equilibrium Conditions at the Interfaces 61
3. Reduction of Flux Formulations to Usable Forms 62
4. Applications to Test Cell Configurations 63
APPENDIX D - Pervaporator Analysis 66
APPENDIX E - Membrane Area Requirement 69
IX. References 73
VI
-------
LIST OF FIGURES
Figure page
1 Dialysis schematic 2
2 Rotating dialysis cell, disassembled 13
3 Pervaporation test cell (Babb & Grimsrud) 15
4 Pervaporation test loop 16
5 The effect of BHP concentration in the organic
i i I
phase on the overall transport rate for Al 31
6 Pervaporation schematic 33
7 Vapor pressure vs temperature 36
8 Pervaporation unit scheme 42
9 Pervaporation unit gaskets 43
10 Pervaporation unit fluid flow pattern 43
11 Effect of feed flow rate on permeability 44
12 Effect of feed concentration on permeability 45
13 Effect of feed temperature on permeability 46
14 Effect of air flow rate on permeability 47
15 Flow scheme for proposed EDC recovery system 52
16 Dialysis scheme 55
17 Schematic of membrane transport 60
18 Schematic of stack-loaded pervaporator 66
19 Wilson plot 70
vii
-------
LIST OF TABLES
Number
1 Transport of Aniline and Water Through Polymeric Membrane ..19
2 Transport of Phenol Through Polymer Membranes 20
3 Diffusion Coefficients of Aniline in Polymeric Membranes 22
4 Diffusion Coefficients of Phenol in Polymeric Membranes 22
5 Single-Stage Extraction Efficiency 29
6 Solubilities and Vapor Pressure of Selected Organic
Solutes in Water 37
7 Permeability Coefficients - Nitrobenzene 38
8 Permeability Coefficients - Chloroform 38
9 Permeability Coefficients - Ethylene Bichloride 39
10 List of All Organic Compounds Found in the
Carrollton Water Plant (New Orleans) Finished Water 41
11 Economic Analysis for Hollow Fiber Unit 50
12 Economic Assessment 51
13 Comparative Cost of Processing 54
viii
-------
NOMENCLATURE
a = chemical activity
A = area
C = concnetration
C = average concentration in the dialyzer
conj = conjugate acid or base
C = charge concentration
d = downstream side of membrane
D = diffusion coefficient
f = feed (upstream) side of membrane
/ = Lewis fugacity
G = flow rate
H = Henry's law constant
i = solute i
J = Flux
j = solute j
K = equilibrium coefficient
eq M
k = mass transfer coefficient
k° = local mass transfer coefficient for membrane
L = thickness parameter
liq = liquid phase
1 = membrane length
M = membrane
m = membrane phase
2
N = molar flux (moles/cm sec)
N = Reynolds number
Re
org = organic solute
o = at 0 concentration, at initial state, or at length 0
p = specific permeability constant
P = partial pressure
P = vapor pressure in the pure state
prod = conjugated product
R = gas constant (lit-atm/mole-deg.)
S = solubility, or distribution coefficient
IX
-------
T = experimental temperature, C
T = glass transition temperature
O
t = time (sec)
U.. . = liquid phase mass transfer coefficient
U = change in liquid phase concentration
V = volume
v = feed velocity
vap = vapor phase
UliqA
ni = cTT^
liq
U, . A
llcl
G
vap
x? = solute mole fraction in liquid
z = solute mole fraction in membrane
Greek Letters
PS
al = L
RTpH
a2 = L
RTH
activity coefficient
density
Un . A
7^- 1 - K Y)
Gliq eq
-------
ACKNOWLEDGEMENTS
The financial support of the Louisiana State Science Foundation, supple-
menting the support from the U.S. Environmental Protection Agency grant for this
project, is acknowledged with sincere thanks.
Mr. William T. Hackett, Jr., Executive Director, Louisiana State Department
of Commerce and Industry and Mr. Vernon Strickland, the Project Administrator,
provided the necessary administrative assistance for the project.
Mr. Shyamkant V. Desai performed laboratory experiments, collected and
evaluated data and wrote an initial report.
Mr. James K. Smith and Dr. Elias Klein directed project activities and
helped prepare the final report.
Dr. Robert E.G. Weaver of the Department of Chemical Engineering, Tulane
University, New Orleans, did mathematical modeling for pervaporation process and
made many valuable suggestions in preparing the final report. Dr. Richard P.
Wendt developed transport equations for dialysis.
Vulcan Materials Company, Geismar, Louisana, cooperated in our efforts to
investigate typical industrial effluents for pervaporation. Mr. Charles Jones
of Vulcan was particularly helpful in several related discussions.
Support of the project by the U.S. Environmental Protection Agency and the
valuable suggestions by Mr. Frank Mayhue and Mr. James Horn, Project Officers,
are acknowledged with sincere thanks.
xi
-------
SECTION I
INTRODUCTION
The objective of this study was to explore the potential of dialysis
as a method of secondary water treatment. So far, industrial application of
dialysis has been directed primarily toward such operations as caustic recovery
in the rayon industry, separation of sulfuric acid from copper and other metals,
and separation of sugars from dextrins. In each of these cases dialysis proceeds
in response to a difference in chemical potentials of the solutes across the
membrane. Dialysis can therefore be classified as a passive process in which
the energy necessary to separate the constituents of a solution at a finite rate
comes from the free energy present in the system. Dialysis, then, is a process
whereby a solution containing a permeating species and separated by a membrane
from a solution at lower concentration loses solute through the membrane until
the activity of the solute is the same on both sides of the membrane.
A simple illustration of dialysis is shown in Figure 1. A solution contain-
ing two solutes i and j is separated from a pure solvent by a membrane M. The
solute i is able to pass through the membrane while solute ] cannot permeate the
membrane and is retained. At the beginning of dialysis the solution containing
the two solutes, shown in Compartment 1, is separated from the pure solvent,
shown in Compartment 2, by the membrane. As dialysis occurs, some of solute i
passes through the membrane into the pure solvent compartment. Because of the
chemical activity of the impermeable solutes in the solution, water tends to be
drawn into the solution Compartment 1. This movement of water is called osmosis.
Over a period of time the composition of solutions in the two compartments
change, as represented in Figure l(b). Solutes i and j both are now in more
dilute solutions than the original composite mixture.
The rate of transport of a solute through a membrane depends on several
factors: the chemical nature of membrane, the solute, the solvent, the tempera-
ture, differences in concentration, interaction of the membrane with solute and
solvent, etc. These factors are generally interrelated. In an ideal case, with
a given solute, solvent, temperature, and physical arrangement, the rate of
diffusion through the membrane can be described by Pick's law of diffusion :
1. Crank, J. Methods of Measurement. In: Diffusion in Polymers,
Crank, J. and G.S. Park (eds.). New York, Academic Press, 1968. p. 1.
-------
Compartment 1
Solutes
i and j
Compartment 2
Pure
water
w
(a) Beginning of dialysis.
•Dialysis membrane M
Dialysis membrane M
Compartment 1
J.
Solutes j
and i - Ai
Compartment 2
Water and
solute Ai
w
(b) Progress of dialysis after time t.
Figure 1. Dialysis schematic.
-------
Rate of transport per unit area = -pdc/dx
where p is the specific permeability coefficient, and dc/dx the concentration
gradient. At the steady or pseudo-steady state, the rate of transport is
directly proportional to the difference in concentration across the membrane.
It is also proportional to the diffusion coefficient. The permeability coeffi-
cient is a property of the membrane and the permeant, as defined by the relation:
where D is the actual diffusivity of solute within the membrane and S is the
partition coefficient for solute between the membrane phase and the solutions on
either side of the membrane. The membrane determines selectivity by allowing a
particular solute to permeate while rejecting others. The membrane also con-
tributes to the rate of transfer because of the interrelation between the mem-
brane structure and the diffusivity of the solute.
A variety of polymeric films or membranes can be used to separate and to
purify solutions by dialysis. Choice of the polymeric membrane depends on the
chemical nature of the mixture being separated and the stability of the polymer
under anticipated conditions of use. One of the solutes to be separated must be
permeable to the membrane at an acceptable rate without damaging the membrane.
The rate is inversely proportional to the membrane thickness, so it is desir-
able to employ membranes that are very thin. Thin membranes can be used without
concern about changing the value for S, as selectivity is not dependent on film
thickness.
The largest industrial applications of dialysis today are the recovery of
caustic soda from industrial waste and the refining of crude sodium hydroxide.
The press liquors from the viscose rayon process contain 16-17% caustic and some
hemicellulose. This solution is dialyzed to obtain a caustic sample which
contains 8 to 9% soda but less than 0.08% hemicellulose. The solution can then
be reused in the makeup of press liquor or in the viscose steeping process.
Waste caustic from the steeping step can also be dialyzed and reused in a
similar fashion. Mercerizing liquors, which are composed primarily of caustic
soda at a concentration of 20 to 30%, can be similarly treated. The removal of
-------
inorganic impurities, such as chlorides, alumina, silica, and ferric oxide, from
crude sodium hydroxide solutions can also be accomplished by dialysis.
Dialysis is employed in separation of sulfuric acid from copper and other
metal salts in contaminated copper-plating baths. This separation, accomplished
by a difference in diffusion rates of the various ions, results in a product
adequately pure for reuse. Another application of dialysis is in the separation
of sugars from dextrins in cornstarch conversion. Upon dialysis, dextrins, as
well as sugar, pass into the receiving liquid but at different rates. The
receiving liquid becomes enriched by the sugar—the faster permeating solute—
leaving the dextrin behind in the feed solution.
In the past, dialysis generally has been regarded as a slow and inefficient
process. Several factors have limited its development: (1) There has been a
small choice of membranes that could perform in the anticipated environment, and
these membranes lacked selectivity and uniformity of performance. (2) The
operating efficiency of the process has been dependent on the concentration of
the diffusible solute. (3) The process rate has been slow compared to other
chemical processes. (4) A reasonably large capital outlay has been required for
installation of equipment.
All the problems that have been associated with dialysis can now be improved
upon or resolved by utilization of up-to-date technology available in areas of
membranes and equipment, and by the utilization of up-to-date and precise mass
transport technology. To remove, by dialysis, a pollutant at low concentrations
in water, it is necessary to maintain the driving force at its maximum value.
This can be achieved in the following ways:
1. Organic acids or bases may be dialyzed through hydrophobic membranes.
The concentration of the acid or base in the receiving compartment is reduced to
minimum by formation of a nonpermeating or conjugate acid or base on the down-
stream side. Since the concentration of the pure acid or base will be higher in
the waste stream, the direction of permeation will be from the waste stream to
the recovery stream. Because the concentration of the solute acid will be close
to zero, the rate of transfer will be maximized for the system. The conjugate
acid and base can be concentrated in the receiving stream for recovery. This
method of dialysis is termed "dialysis with conjugation."
2. Dialysis of heavy metals in solution can be accomplished employing
hydrophilic membranes. To maintain the maximum driving force and to provide a
-------
means for concentration and recovery of the permeating species, an organic phase
containing a chelating or complexing agent can be employed on the downstream
side. As in dialysis with conjugation, the free metal ion is removed from the
downstream or receiving solution, thus enhancing the driving force. This
process of dialysis is called "dialysis with complexing."
3. Organic solutes can be separated from aqueous waste streams by dialysis
through hydrophobic membranes into a gas purge stream on the downstream side of
the membrane. The gas purge stream removes the permeating species and thus
maintains the driving force. The permeating species can then be recovered
downstream of the membrane. This process is termed "pervaporation." In each of
these cases, complete removal of solute from the upstream side and its collec-
tion on the downstream side is theoretically possible.
The present study to characterize these processes was carried out in three
phases.
A. SCREENING
A number of organic and inorganic solutes present in industrial waste
streams were examined for their susceptibility to separation and recovery by
dialysis. The transport rate of each solute was determined in one of the three
described dialysis processes. Different type membranes were employed in each of
the processes. Where more than one type of membrane was available for a par-
ticular process, all were included in the screening. The most promising
applications were then selected for further study.
B. CHARACTERIZATION
Process responses of the engineering variables for an ethylene dichloride
containing waste stream were studied in a pilot scale pervaporation process.
The selection of the pervaporation process and the EDC stream were made on the
basis of data gathered in the screening study, namely, with consideration of the
pollution significance of chlorinated hydrocarbons in waste water.
-------
C. DESIGN AND PROJECTION
The data from the characterization of the dialysis of the ethylene dichlo-
ride stream were used to project the technical and economic feasibility of a
full scale plant operation.
-------
SECTION II
SUMMARY
Conventional dialysis across a permselective membrane results in the
transport of the permeant to a lower concentration, maintained by simple
dilution, on the downstream side of the membrane. Conventional dialysis clearly
is not sufficient to remove a very dilute contaminant from a plant's total
effluence, unless a solvent completely free of the contaminant is used on the
downstream side; but it is possible, by using special techniques such as those
enumerated below, to remove certain classes of very dilute contaminants by
dialysis without requiring a continual fresh supply of downstream solvent.
Three novel techniques were developed and demonstrated as technically feasible
in the present study:
1. The use of acid and base conjugation, respectively, on the downstream
side of membranes permselective to aniline and phenol.
2. The use of a chelating complex for Al, Cu, and Pb ions.
3. The use of a pervaporation scheme in which a hydrophobic membrane
passes volatile contaminants such as nitrobenzene and ethylene
dichloride to a downstream vapor purge, which picks up the contami-
nants substantially free of water.
The pervaporation scheme was applied to a prototype EDC (ethylene dichlo-
ride) contaminated industrial waste stream in a pilot process configuration.
Economics indicated that while pervaporation is not a particularly inexpensive
route with established technology, the use of current improvements in membrane
technology could bring costs down into the range of routine process water
treatment costs (10C-50C per 1,000 gallons).
-------
SECTION III
CONCLUSIONS
An evaluation of dialysis as an advanced waste treatment process for the
removal of selected organics and metallic ions from industrial waste has been
performed by Gulf South Research Institute for the Louisiana Department of
Commerce and Industry. Dialysis on these industrial waste streams can be both
technically and economically feasible when removal of the membrane permeant is
supported by special schemes such as acid and base conjugation, complexing, or
pervaporation, rather than reliance upon simple dilution to maintain the
osmotic gradient. From the study the following conclusions and their close
affinities were drawn:
A. The economic removal of dilute contaminants from water by dialysis is
dependent upon the driving force of an essentially complete removal of contami-
nant downstream of the selective membrane. Methods for accomplishing this have
been developed and demonstrated in the present study. They include conjugation,
complexing, and the removal of the permeant as a vapor on the downstream side
of a semipermeable membrane, as follows:
1. Dilute aniline and phenol systems can be stripped of these contaminants
using acid and base conjugation downstream of the membrane. With polycarbonate-
-7 -7 2
co-silicone films, specific permeabilities of 3.9 x 10 and 2.06 x 10 cm /sec
were obtained for the aniline and phenol respectively. Corresponding values
of specific permeability with a styrene-co-butadiene film, Kraton 1101, were
—7 -7 2
2.5 x 10 and 0.71 x 10 cm /sec. Ethyl cellulose provided a good medium
-7 2
for the phenol service, with a specific permeability of 0.96 x 10 cm /sec.
2. An effective complexing sink for Al, Cu, and Pb ions permeating a
Cuprophan film exists in bis-(2 ethylhexyl) hydrogen phosphate. Unfortunately,
the solute transport rate to the organic receiving phase containing the complex-
—7 2
ing agent was only 0.56 x 10 cm /sec. In comparison to transport in a
. _-j 2
completely aqueous system (where a permeability of 2.58 x 10 cm /sec was
obtained) the resistance to the solute transport to organic receiving phase
containing the complexing agent was significantly high. It is believed that
Registered trademark of Enka Glanzstoff A.G.
-------
the major contribution to this resistance came from polarization on the down-
stream side of the membrane at membrane/receiver interface . Serious
attention would have to be given to minimizing this downstream resistance
between the aqueous phase and the organic phase before a really attractive
process could be claimed. This is the same kind of problem that has hampered
the development of liquid-liquid extraction. Because of the work performed in
this latter field, it was not deemed feasible to develop a solution to the
problem in the context of this study. However, in view of the notoriously
small transfer coefficients and tight fluid mechanical constraints experienced
in liquid-liquid extraction systems, dialysis through a selective membrane may
offer a viable alternative in this extractive process.
3. Pervaporation, based on a vapor-phase sink downstream of any of a
number of permselective hydrophobic films, provides an attractive approach for
the removal from industrial waste streams of chlorinated hydrocarbons such as
chloroform, ethylene dichloride, and other toxic chemicals such as nitro-
benzene. In general, the permeants in these applications will have reasonable
volatilities. The use of very thin membranes (25 microns or less), high liquid-
side velocities, and higher system temperatures is desirable for commercial
applications.
B. The pervaporation scheme was used in a larger scale test operation on
a stream representative of an ethylene dichloride process plant effluent contami-
nated largely with ethylene dichloride. This test was at such a scale that the
same parameters controlling a large on-site waste treatment unit were applicable,
but inordinate capital expenditure was not required. The test provides a
credible basis for projecting a treatment cost, based on existing technology,
of 90% EDC removal from a stream 0.8% (by weight) EDC in 3.5% HC1. Improvements
in membrane transport rates and reductions in membrane costs (such as are
feasible in the context of the developing hollow fiber technology) could drop
these processing costs to $0.10 - $0.50 per thousand gallons of treated water.
-------
SECTION IV
RECOMMENDATIONS
Field evaluations of two of the processes reported herein are recom-
mended. These field trials should also be supported by background studies
concerned with crucial engineering variables such as the following:
A. FOR "CONJUGATE DIALYSIS"
1. A design study to reduce fluid resistances on both sides of the sepa-
rating membrane.
2. Membrane life evaluation.
3. An evaluation of the use of pretreatment to reduce membrane fouling.
B. FOR PERVAPORATION
1. Development of hollow fiber configurations to provide the requisite
area in a simple assembly. A 90% reduction of the 0.8% EDC stream required
2
450 m of membrane area per gpm of waste flow. This can be accomplished in a
3
simple tube-and-shell assembly occupying a volume of 0.015 m /gpm of waste
flow.
2. Development of asymmetric fiber structures which could reduce the
required membrane areas by a factor of 50.
3. Evaluation of the use of stream pretreatment .to prevent fiber occlu-
sions.
4. Design of the EDC recovery system and design of air purge recycling.
C. FOR COMPLEXING
1. Field evaluation of the third process, complexing, is not recommended
at this time.
10
-------
SECTION V
CONTAMINANT SCREENING
In this phase of the project, ten different solutes which presented
potential concern as pollutants of the lower Mississippi River water in the
New Orleans-Baton Rouge area were screened to test the applicability of dialysis
in their concentration or removal from waste streams. These solutes are grouped
as:
a) Organic acids or bases: aniline, phenol, ethanolamine.
b) Toxic cations: Al"1"4"1", Cu4"4", Pb4"1".
c) Volatile refractory organics: nitrobenzene, chloroform, ethylene
dichloride, nitrotoluene.
More specifically, the solutions of aniline, phenol, and ethanolamine were
dialyzed through hydrophobic membranes aided by the formation of their respec-
tive conjugate acid or base on the downstream side in "Dialysis with Conjuga-
[ I i i I i i
tion." Solutions of Al , Cu , and Pb were dialyzed through hydrophilic
membranes using an organic receiving phase containing a complexing agent on the
downstream side in "Dialysis with Complexing." Nitrobenzene, chloroform and
ethylene dichloride solutions were "pervaporated" through hydrophobic membranes.
Our use of conceptual membrane design is based on a unit process technique.
The process seeks to sustain itself by taking advantage of the operative chemical
and/or physical process to overcome the singular process of dilution. That is,
the use of conjugation, complexing, or pervaporation is intended to remove the
solutes on the downstream side of the membrane so that the driving force (con-
centration differences) can be maximized. The removal and/or conversion of the
permeating solutes permit a concentration of the product.
Commercially available films including polyethylene (low and medium
density), Saran, Cuprophan, and laboratory cast films from Kraton 1101, poly-
carbonate/silicone and ethyl cellulose were used as membranes in the investiga-
tion. Permeability measurements for "Dialysis with Conjugation" and "Dialysis
2
with Complexing" were carried out in a rotating batch dialyzer . The permeability
measurements for pervaporation were carried out in a specially constructed Babb &
2. Wendt, R.P., R.J. Toups, J.K. Smith, N. Leger, and E. Klein. Measurements
of Membrane Permeabilities Using a Rotary Batch Dialyzer. Ind. & Eng. Chem.
10:406, 1971.
11
-------
Grimsrud cell (Figure 2), which had previously been used to determine transport
properties of hemodialysis membranes. In these previous studies the flow geo-
metry and mass transfer for this cell were defined.
A. THE APPARATUS
1• Rotary Batch Dialyzer
The cell shown in Figure 2 was used to evaluate the membrane compatibility
with the solution and its permeability constant. This apparatus is an attractive
one because of its ease of operation and construction. It was designed by Regan
3
and co-workers , for use in determining true solute permeability through a
related membrane. The cell consists of two hollowed-out discs which are clamped
together, with the membrane to be tested separating the two halves.
In dialysis there is a flow of permeable components into one side of the
membrane. At the other membrane surface there is flow of permeable components
out of the membrane and into the layer adjacent to the surface. This process
leads to permeant solute concentrations at the interfaces different from the
concentrations in the bulk solutions. These differences are great enough to
influence the observed mass transfer rate and are normally referred to as boun-
dary layer resistance or concentration polarization. This model has been
previously described
In an industrial application the feed solution is pumped past the membrane
surface at a high flow rate to reduce the concentration polarization. The
greater the feed solution velocity, or Reynolds number, the greater the shear at
the membrane surface and the lower the boundary resistance.
For the limited purpose of screening membrane performance, it is not neces-
sary to completely eliminate this boundary effect as long as it is held constant
and the magnitude of the effect is relatively small in comparison to the membrane
resistance. In the experimental screening of membranes these conditions were
achieved with the maximum contribution of the boundary never exceeding 30% of
the absolute membrane value. This constant error did not seriously compromise
3. Regan, T.M., W.G. Esmond, C. Strackfus, and A.M. Wolbarsht, Science.
162:1028, 1968.
12
-------
Figure 2. Rotating dialysis cell, disassembled,
-------
the evaluation of relative membrane performance. The error in the permeability
obviously becomes less as the membrane resistance becomes large compared to the
4
boundary resistance .
The two compartments of the rotary batch dialyzer were filled with 35 cc
each of the donor and receiving solutions, and the rotation was started. After
selected time intervals, varying from 15 minutes to 60 minutes, the contents of
each compartment were drained, neutralized to a common pH, and analyzed by U.V.
or atomic absorption. Dilutions were necessary to have concentration levels
suitable for analysis. Analyses were performed for solute concentration on both
sides of the cell as a check on the mass balance.
2. Pervaporation Test Cell
The pervaporation test cell is made of plexiglass and is shown in Figure 3.
Water was circulated through the liquid path (B) between the two membranes,
which were held in place by gaskets (E). Air circulated through channels (C)
counter current to the water flow around the outside of the membrane compart-
ment. The cell had an "effective" membrane area of 110 square centimeters. By
"effective" we mean that area which actually comes in contact with fluids.
4. Klein, E., J.K. Smith, R.P. Wendt, and S.V. Desai. Separation Science.
7(3):285-292, 1972.
14
-------
G A
A. Membranes
B. Liquid Channel
C. Air Channel
D. Liquid Ports
E. Gaskets
F. Air Ports
G. Air Channels
Figure 3. Pervaporation test cell (Babb & Grimsrud).
A pervaporation test loop, shown schematically in Figure 4, was assembled
to make quick determinations of membrane permeabilities for screening purposes.
This apparatus was easily assembled and proved very convenient to operate,
rapidly yielding very accurate data. At fixed intervals (varying from 15 min-
utes to 60 minutes) during the test recirculation runs, samples were drawn from
the feed charge for analysis of the solute concentration by gas chromatography.
Liquid flow rates of 0.5 to 1 liter/minute and an air flow rate of 12 liters/
minute resulted in a slight positive pressure on the liquid side. Air humidity
was not a significant factor during these runs since hydrophobic membranes were
used and all experiments were done at ambient temperatures.
15
-------
A. Feed Charge
B. Feed Pump
C. Test Cell
D. Liquid Flow Meter
E. Air Flow Meter
F. Trap
G. Vacuum Pump
Figure 4. Pervaporation test loop.
B. MEMBRANES
Membranes were either obtained commercially as films, or prepared by solvent
casting on clean glass plates in a controlled environment. (Commercially avail-
able polymers were used to make the casting solutions.) The following membranes
were used:
1. Commercial Films
a. Polyethylene with phthalate plasticizer.
b. Polyethylene (commercial bag film) without phthalate plasticizer.
c. Saran - A polyvinylidene chloride-polyvinyl chloride copolymer.
d. Cuprophan - A cellulosic hemodialysis membrane.
16
-------
2. Films Prepared in the Laboratory by GSRI
a. Kraton 1101 A styrene-butadiene block copolymer from
Shell Chemical Company, cast from methyl isobutyl ketone
solutions.
b. XD-7 A polycarbonate-polysilicone copolymer from
General Electric, cast from methylene chloride solutions.
c. Ethyl cellulose, with a high degree of ethyl substitution, cast from
methylene chloride solutions.
These films are not available commercially.
C. THE PERMEABILITY MEASUREMENTS
Membrane screening was based simply on the experimental determination of an
overall solute specific permeability as defined below. The derivations of this
equation for the computation of solute permeability constants from the recircu-
lating runs are given in Appendix A and Appendix C (pages 55, and 60, respec-
tively) .
The equation is given below:
V .
A
L
t
C
In °
- V . I; , O ,.. N
p = -T- — In — (1)
17
-------
3
where V = liquid volume (cm ) of feed or charge solution
3
C = charge concentration (g/cm )
2
A = membrane area (cm )
L = membrane thickness (cm)
t = time of run (sec)
3
C = concentration at time t (g/cm )
— 2
p = specific permeability constant (cm /sec)
The distribution coefficient S, as previously defined, was determined
between the aqueous solution and the membrane phase by use of a chromatographii
method which is also described in Appendix B, page 59.
Pure water permeability of the membranes at 25°C was determined using the
ASTM Method E 96-63T.
D. RESULTS
1. Dialysis with Conjugation
The following systems were studied to illustrate the use of conjugation as
a permeant sink:
Aniline Initial upstream aniline concentration: 2.98 wt %
Initial downstream H-SO, concentration: 5.33 wt %
2 4
Phenol Initial upstream phenol concentration: 0.94 wt %
Initial downstream NaOH concentration: 4.00 wt %
Ethanolamine Initial upstream ethanolamine concentration: 0.5 wt %
Initial downstream H?SO, concentration: 1 wt %
Specific permeability coefficients of aniline and phenol are listed in
Tables 1 and 2. The permeation of ethanolamine was too low to be detected. The
permeability of the aniline and phenol through the organic polymers was excel-
lent. The best results were obtained with the Kraton and XD-7 films. The
—7 2
observed permeabilities of 2 to 4 x 10 cm /sec compare favorably with commer-
—7 2
cial dialysis processes where, for example, a value of 6 x 10 cm /sec for the
2
permeability of glucose through uncoated cellophane has been observed . If a
18
-------
TABLE 1
TRANSPORT OF ANILINE AND WATER THROUGH POLYMERIC MEMBRANE
(Temperature 25°C)
Membrane
Designation
Thickness
L (microns)
Specific Permeability
p (cm /sec)
Water
Aniline
Distribution
Coefficient
Polyethylene A 20.0 1.16 x 10
Polyethylene B 13.6 2.32 x 10
Saran 11.9 (*)
Kraton 1101 35.7 3.8 x 10
XD-7 68.3 1.19 x 10
-10
-11
-8
-9
0.44 x 10
0.15 x 10
-7
-7
0.19 x 10
2.5 x 10~7
-7
3.9 x 10
-7
0.13
1.44
0.765
2.69
Distribution Coefficient:
S =
g of solute per unit volume of swollen polymer
g of solute per unit volume of solution
(A)
Transport was so low that the method was not applicable.
—12 2
Estimated to be less than 10 cm /sec
19
-------
TABLE 2
TRANSPORT OF PHENOL THROUGH POLYMER MEMBRANES
(Temp. 25°C)
Polymer
Phenol
„, . . Specific Permeability
Thickness v J
L(Microns) p (cm /sec) x 10
Distribution
Coefficient
S
Polyethylene A
(Medium Density)
20
0.5
Polyethylene B
(Low Density)
14
0.11
Saran
12
0.06
0.07
Kraton 1101
41
0.71
0.5
XD-7
68.1
2.06
0.5
Ethyl Cellulose
35
0.96
17.4
20
-------
correction factor for the error introduced by the boundary layer is applied to
the experimental values, these values will be even higher.
A diffusivity for aniline and phenol in the membrane matrix can be esti-
mated from the relationship:
P = DQS (2)
where p = specific permeability
2
D = diffusion coefficient (cm /sec)
and S = distribution coefficient
Tables 3 and 4 list such average diffusion coefficients for aniline and
phenol in the membranes tested.
It was observed that choice of solution temperature may be used to advant-
age in this process. The primary effect of a temperature increase upon membranes
operating under a pore-type mechanism is an increase in the permeation rate of
the solute in water-filled pores. For diffusive membranes, such as reported
here, however, the effect is both to increase the permeation rate in the mem-
brane and to alter the membrane structure.
The polymeric state of the membrane has a pronounced influence on the rate
and transport mechanism of a solute diffusing through its matrix. Below the
glass transition temperature the membrane is in a glassy state and the polymeric
chainr are restricted in their ability to move. The membrane may also
contain rigid voids which can trap permeant molecules, thereby contributing
little to the diffusive process. The restriction in the motion of the polymer
chain prohibits activated diffusion. Above the glass transition temperature,
however, chain mobility and diffusivity increase. This increased mobility and
diffusivity can be visualized as the membrane's becoming more and more rubbery.
The polymer films that appear most promising in this present study are all above
their glass transitions.
A review by Fujita presents temperature responses for a number of other
polymers. His conclusions appear to be applicable in the present work since
similar generalizations can be derived from the observed transport rates and the
5. Fujita, H. Organic Vapors above the Glass Transition Temperature. In:
Diffusion in Polymers, Crank, J. and G.S. Park (eds.). New York, Academic Press,
1968. p. 75-106.
21
-------
TABLE 3
DIFFUSION COEFFICIENTS OF ANILINE IN POLYMERIC MEMBRANES
Polymer Do(cm2/sec)
—8
Polyethylene B 11.5 x 10
Saran 1.32 x 10 8
Kraton 1101 32.7 x 10~8
XD-7 14.5 x 10~8
TABLE 4
DIFFUSION COEFFICIENTS OF PHENOL IN POLYMERIC MEMBRANES
2
_ , D (cm /sec)
Polymer o
Saran 8.6 x 10~8
Kraton 1101 14.2 x 10~8
XD-7 41.2 x 10~8
— ft
Ethyl Cellulose -55 x 10
22
-------
difference between experimental and glass transition temperatures
(T - T ). For example, the plasticized polyethylene, which can be expected to
&
have a lower T , has a greater permeability than the unplasticized polyethylene
O
membrane. Theory predicts that the higher the temperature above a glass tran-
sition, the greater the transport rate. Obviously a point is reached where the
integrity of the membrane can no longer be maintained or controlled. One solu-
tion to this problem is to use block copolymers, in which the polymer chain is
composed of distinct segments of at least two different polymers. One of these
segments is chosen to be a rubbery polymer above its glass transition and the
other is a crystalline polymer. The crystalline polymer provides the necessary
mechanical properties while transport occurs primarily through the rubbery
material. Examples of such polymers are the polysiloxane/ polycarbonate (XD-7)
and the polybutadiene/polystyrene (Kraton 1101). The polysiloxane copolymer,
whose blocks have a T of approximately -122°C, is the most highly permeable
&
membrane of a'.l. The butadiene copolymer, with blocks having a T of approxi-
o
mately -112 C, is the second most highly permeable membrane. From these data we
can expect that both Kraton and the polysiloxane copolymer will increase in
permeability up t" the temperature where mechanical structure of the membrane
begins to fail. he block copolymers thus apparently have an advantage of being
able to form large rubbery domains in contrast to cross-linked elastomers which
are relatively constrained by short chains between cross-links. Surprisingly,
the relatively stiff, crystalline ethyl cellulose shows phenol transport rates
higher than those of the butadiene-containing copolymer, Kraton 1101. Since the
apparent permeability values are the average values for the concentration range
of 0.1 molar phenol to infinite dilution, it is possible that the ethyl cellu-
lose membrane was plasticized by the phenol and that its morphology was signifi-
cantly altered in the presence of the permeant. Plasticization increases perme-
ability by reducing cohesiveness between polymer chains, which leads to increased
difftisivity. Alternatively, it is also possible that the 1 molar sodium
hydroxide—an effective swelling agent for the native cellulose—decreased the
ethyl cellulose crystallinity, leading to a more permeable structure.
The transport rates through the two polyethylene film samples are of the
order expected. The less-crystalline, lower density material allows more rapid
transport than the denser, more crystalline membrane. Saran, a copolymer of
vinyl chloride and vinylidene chloride, is a highly ordered polymer. It was
23
-------
included in the investigation since several of the solutes being screened were
chlorinated hydrocarbons. Because of chemical similarity it was expected that
these hydrocarbons might distribute favorably in the saran film and thus exhibit
an enhanced transport. The saran, however, is apparently unaffected by the
phenol and exhibits low permeability, as expected from the chemical nature of
this polymer.
The results achieved with aniline and phenol indicate that transport rates
of the same magnitude as achieved in commercially-useful dialysis processes are
possible, but without the dilutions so notoriously characteristic of ordinary
dialysis. The difference between the processes described here and the older
processes lies in the availability of new membrane materials which are selective
toward organic species coupled with the use of the artificially created chemical
potential gradient.
An example of these new membrane materials is the Kraton film, a copolymer
of butadiene/styrene. The organic solute material has a high permeability
through the mobile or rubbery phase, as discussed above. The ratio of the
butadiene to styrene, as well as the thickness of the membrane, can be varied.
The composition of the membrane and its amenability to fabrication in various
physical configurations make possible permeabilities of the same order of magni-
tude as those found in other commercial dialyzing processes and in reverse
osmosis. These permeabilities can be further enhanced by using solute conju-
gation in the receiving solution to maintain a maximum chemical potential
gradient.
In projecting the use of this process to the concentration of trace contami-
nants from plant waste streams for later disposal or recovery, consideration of
continuous processing will, of course, be required (see Section VII of this
report).
For concentration of dilute contaminants, according to the reaction org +
conj = prod, one would anticipate that the conjugate acid (or base) would be
maintained at a high excess concentration in the receiver compartment, and that
the following constraints would apply:
C , the feed concentration = constant (large V >» V )
org' * d
24
-------
C ., concentration of conjugate acid or base = constant
conj' J &
excess concentration
C ,, concentration of conjugated product = K C C
prod' j & r eq org con.j
receiver compartment always near equilibrium.
where:
C = concentration
f = upstream (feed) side of membrane
d = downstream side of membrane
org = organic solute
conj = conjugate acid or base
prod = conjugated product
K = equilibrium constant
eq
For these constraints it can be shown, by rearranging equation (17), Appen-
dix A, that the accumulation of the organic solute in the receiving compartment
of volume V will be given by the following:
org
= C
org
1 - exp (-
A P t
V,( 1 + C . K )
d conj eq
(3)
where P = membrane permeability, cm /sec
2
A = cell cross section, cm
t = contact time, sec
The concentration of the organic solute in the receiving phase will attempt
to increase until it begins to approach the feed concentration. If it were to
reach the feed concentration, the process would, of course, stop, since the
driving force, would be dissipated. The controlling factors are the membrane
permeability and the "sink" for the organic solute given by the product C
K .
eq
25
-------
This equilibrium product will predict the amount of the permeating solute
that will be present in the receiving phase. In the application of these
techniques the concentration of the permeating species is maintained at a very
low level by providing a large excess of the conjugating acid or base.
The total amount of organic solute accumulated in the dialysate is the
sum of the solute, C and its conjugate C . The relationship between
Cprod and Keq Can be used tO find the total quantity of solute transferred:
C = Cd (1 + K Cd .) (4)
org org eq conj v '
For reactions with large K values, the first term in the brackets reduces
d eq
to the product K C ., so that one can project that the maximum amount
which can be accumulated in the dialysis will approach the product of
d f Cdrod f
K C . C , which is equal to —*j x C (5)
eq conj org d org
org
The concentration ratio C ,/C is thus a measure of the "sink" avail-
prod org
able for the species to be transferred. The rate at which the process will
occur is a product of the membrane permeability times the concentration
difference of organic solute. The overall process can then be viewed as
a concentration process which can operate up to an amplification ratio estab-
lished by the equilibrium constant of the species to be transported. If recovery
of the product is desired, the dialysis driving force can be enhanced by removing
the conjugated product from the sink. Such recovery will, of course, involve
the neutralization of the conjugate acid or base. But recovery can be performed
in a solution more concentrated than the feed solution.
2. Dialysis with Complexing
The separation of transition metal ions such as alun inum, copper, and lead
has been studied, using dialysis with complexing. Permeant cations were com-
plexed on the downstream side of the membrane with soluble impermeable complexing
agents of relatively high molecular weight. The cations could thus be concen-
trated in the receiving phase for more efficient disposal or potential recovery.
Hydrophilic membranes and an organic "receiver" solvent were used in this mode
26
-------
of dialysis. The selection of these membranes and the solvent are discussed
below.
Dialysis with complexing has an advantage over simple solvent extraction in
that loss of the extracting solvent and/or complexing agent due to partial water
solubility and entrainment in the feed stream is eliminated. However, this must
be weighed against a potentially slower mass transfer rate common to the
dialysis membrane.
a. Choice of Membrane
The selection of Cuprophan for these studies was based on its unique proper-
ties: a high water content, good mechanical properties, low cost, and good
chemical stability. Since the transport of the metal ions will occur through
the water phase within the membrane, there will be minimal interaction between
the solute and the polymer. Therefore, membrane selection for dialysis with
complexing is based on different criteria than those that pertain to dialysis
with conjugation. Other membranes with similar properties could be considered;
however, this section of the investigation had as a goal the evaluation of the
complexing procedure. Since the properties of Cuprophan had been well documented
in other dialysis procedures , this membrane was selected in order to eliminate
the membrane as an independent variable in this evaluation. (Dialysis membranes
are routinely evaluated by GSRI under another program: Membranes and Materials
Evaluation, NIH NIAMDD Contract No. 72-2221). There are some experimental
membranes that may exhibit enhanced transport for the metal ions under study,
however, the chemical and physical stability of these membranes are not competi-
tive, and they are not available commercially. For these reasons they were not
included in this evaluation.
6. Klein, E., J.K. Smith, F.F. Holland, and R.E. Flagg. Membrane and
Materials Evaluation; Permeabilities, Physical and Mechanical Properties of
Hemodialysis Membranes—Bemberg Cuprophan Pt-150 Membrane. Gulf South Research
Institute. Annual Report AK-1-72-2221. Artificial Kidney-Chronic Uremia Pro-
gram, National Institutes of Health. July 1973. 24 pp.
27
-------
b. Selection of Complexing Agents and Solvent Combinations
I I i i I i I
A number of efficient agents for complexing Al , Cu and Pb are sug-
7 8
gested by Walsh . Based on reported information , and the requirement that the
systems should be commercially acceptable, some organic phosphorous compounds
were chosen for investigation in this study. Those selected were relatively
inexpensive, insoluble in water, and compatible with our test system. The
following compounds obtained from a local chemical company were tested:
1) Dibutyl Butylphosphonate
2) Tributyl Phosphorothionate
3) Triisooctyl Phosphorothionate
4) Bis (2-ethylhexyl) Hydrogen Phosphate
5) Bis (2-ethylhexyl) Phosphate
I i i
A contaminant found in Mississippi River water, Al , was used as a
representative metal cation of interest. A direct extraction test was run to
determine the extraction efficiency of each of the aforementioned compounds.
Results obtained appear in Table 5.
Based on this study, bis(2-ethylhexyl) hydrogen phosphate (BHP) was
selected for complexing the cations. This complexing agent has low water
solubility. It is a colorless, odorless liquid which is slightly lighter than
water (specific gravity 0.9). Its cost is about $1.00 per pound. Bis(2-
ethylhexyl) hydrogen phosphate dissolves in hexane and in heavy mineral oils,
acetone, alcohol, chloroform, etc. Hexane was chosen as the solvent for these
trials because, for practical purposes, it is insoluble in water and therefore
minimizes the transport of the receiver phase to the feed side.
c. Results from Rotating Cell Screening Tests
Aluminum System—Aluminum was chosen as a test solute because it is used
extensively as a coagulating and/or precipitating agent in waste treatment.
Its concentration in natural waters ranges from a few to several thousand
micrograms per liter. The toxicity of aluminum has not been well defined, but
concern has been expressed about the use of aluminum compounds in the food and
cosmetic industries. Data were obtained by dialyzing a 1% aluminum chloride
7. Walsh, A. Atomic Absorption Spectroscopy. ASTM Special Technical
Publication. No. STP 443,31. June 1968.
8. Marcus, Y. and A.S. Kertes. Ion Exchange and Solvent Extraction of
Metal Complexes. London, New York, Sidney and Toronto, Wiley-Interscience, 1969,
1037 pp.
28
-------
TABLE 5
SINGLE-STAGE EXTRACTION EFFICIENCY
Efficiency for Al
Complexing Agent Complexed at pH 3.5 (%)
Dibutyl Butylphosphonate 2
Tributyl Phosphorothionate 31
Triisooctyl Phosphorothionate 35
Bis (2-ethylhexyl) Hydrogen Phosphate 98
Bis (2-ethylhexyl) Phosphonate 43
29
-------
solution in water. The pH of the solution was maintained between 3-4 to
minimize hydrolysis of the Aid during the experiments, which lasted less
than 2 hours. The rotating cell was used at 90 rpm, employing 10%, 20% and
30% solutions of bis(2-ethylhexyl) hydrogen phosphate (BHP) in hexane to
determine the optimum concentration of the receiver phase. The results obtained
are shown in Figure 5.
Several factors can influence the diffusion rate in these experiments.
Three factors are present in all membrane processes: (1) the resistance to
transport through the aqueous liquid boundary next to the membrane; (2) resis-
tance to transport through the membrane; and (3) a boundary resistance through
the organic receiving phase. Figure 5 shows a reduction in the observed diffu-
sion rate with increasing concentration of BHP; this reduction reflects changes
in viscosity of the organic receiving phase. For the same rotational cell
velocity, a reduced diffusion rate leads to higher boundary resistances in the
organic receiving phase. In addition to the three boundary resistances mentioned
above, the system has an additional resistance to transport at the interface
between the organic and aqueous phases, in this case at the receiver side of the
membrane. To delineate the magnitude of this resistance, an experiment was
performed using distilled water as a receiver phase. The diffusion rate was
-7 2
found to be 2.58 x 10 cm /sec, approximately five times higher than that
obtained with the complexing agent present in the organic phase. These findings
indicate that the rate-determining step is the transition from the aqueous to
4
the organic phase. Previous studies with the rotating cell , utilizing turbu-
lence created at 90 rpm as in the present study, have shown that the transport
parameter is not likely to improve more than 30% with further increases in
turbulence.
The aluminum transport rates for this system are lower than those normally
anticipated in dialysis and thus will adversely affect the system's economic
attractiveness for industrial applications. However, conventional liquid/liquid
extraction processes are similarly burdened with the limiting factor in the
transfer from one liquid fraction to the other. Conventional liquid/liquid
extractions also suffer from hydraulic constraints and from some loss of solvent
and complexing agent due to water solubility and entrainment. The hydraulic
problems and loss of solvent and complexing agent would be minimized in the
dialysis mode because of the membrane interfacing the two solvents. Thus,
30
-------
v. r
0.6
If 0.5
Ul
-------
dialysis with complexing may prove profitable, compared to liquid/ liquid
extraction.
Copper and Lead System—Two additional pollutants found in the Mississippi
River, copper and lead, were also screened using this system. With 20% BHP in
-I.., I 1.1,
hexane and at 150 rpm, a 1% aqueous solution of Cu and Pb was dialyzed for 2
ii I
and 1/2 hours at conditions otherwise identical to those of the Al test. No
significant transport of copper was observed, indicating that a different
receiver phase was needed to achieve usable results. However, in the case of
I [ -, ty
Pb , the diffusion coefficient D was found to be 0.56 x 10~ cm /sec. This is
slightly higher than the corresponding Al transport because of increased
turbulence at the higher speed. It is our opinion, however, that the transport
rate is still low for most industrial applications.
3. Pervaporation
The pervaporation process is similar to classical dialysis in that the feed
mixture is placed on one side of a semipermeable membrane, and the product (the
permeant) is collected on the other side. The thermodynamic force behind the
permeation through the membrane is the difference in chemical potential of the
permeant on opposite sides of the membrane. This is accomplished in pervapora-
tion by removing the product from the membrane interface through evaporation. A
schematic of a pervaporation cell is shown in Figure 6. The cell interfaces the
liquid phase with a thin hydrophobic membrane; the permeant product is removed
from the "downstream" side of the membrane as a vapor. The pervaporation pro-
9
cess involves the following sequence of steps:
a. Absorption of permeating molecules at the liquid-membrane interface;
b. Diffusion of these molecules through the film;
c. Removal of these molecules from the downstream surface of the film into
a vacuum or gas receiving stream.
Pervaporation, as used in this report refers to an activated diffusion
process. Such diffusion occurs when there is a chemical interaction which
enhances the transport of some components over others in solution. Such mem-
branes are termed permselective. The proper selection of a permselective
9. Binning, R.C. and F.E. James. Petrol. Refiner. 37(5):214, 1958.
32
-------
Membrane
Liquid in Ai r out
I
Liq.
_l
^
X
X
X
^ 1
•>
Air
*
4^
, r
Liquid out Air in
Figure 6. Pervaporation schematic.
membrane and processing conditions such as temperature, fluid flow rates, etc.,
resulted in an effective separation. The versatility of this process is illus-
9
trated in the applications involving the separation of azeotropic mixtures , the
separation of hydrocarbons , and several other applications
In this pervaporation process, the solute (with a chemical activity a., in
the feed stream) distributes itself between the water and the membrane. The
activity of an organic solute at a level corresponding to its solubility limit
will approximate that of the pure liquid activity, a . We can write a distribu-
W. Li., N.N., R.B. Long, and F.J. Henley. Ind. & Eng. Chem. 57_(3):19,
1976.
11. Binning, R.C., R.J. Lee, J.F. Jenning, and E.G. Martin. Ind. & Eng.
Chem. 53^:45, 1961.
12. Binning, R.C. and J.R. Kelly. U.S. Patent 2,913,507. To American
Oil Company, November 17, 1954.
13. Choo, C.Y. Advances in Petroleum Chemistry. 6^:73, 1962.
14. Sanders, B.H. and C.Y. Choo. Petrol. Refiner. :39_(6) : 133-138, 1960.
33
-------
tion constant using the approximation
a™ = S a° (6)
where a is the solute activity in the organic membrane phase. On the gas side
of the membrane, the solute will attempt to reach its equilibrium partial pres-
sure, i.e., the partial pressure corresponding to a solution of the solute in a
fluid having the same activity coefficient as the membrane. This can be approxi-
mated by a Henry's law constant
H = am/P (7)
The transport equation has been derived for this case by Yasuda and is given
by:
dP/dt = ""
(1/RT) VL
_ 2
where p = Specific permeability constant (cm /sec)
2
A = Mass transfer area (cm )
3
V = Charge volume (cm )
L = Film thickness (cm)
P = Partial pressure of the solute on the upstream
side (atm)
P = Partial pressure of the solute on the downstream
side (atm)
P = vapor pressure in the pure state
R = Gas Law Constant, lit-atm/mole deg.
T = Temperature
t = time
The equation for our case can be simplified by letting / approach zero
through rapid gas sweeping. The result indicates that the transport rate will
15~. Yasuda, H. J. of Polym. Sci., Part A. ,5> 1967.
34
-------
be directly proportional both to membrane area, and to vapor pressure of the
pure solute (from the approximation that P at saturation equals P ).
Thus, compounds having an adequately high vapor pressure, for which one can
find the proper membrane, will be suitable for pervaporation separation.
With these criteria in mind, one can choose some likely candidate solute
from Table 6, which lists the pollutants detected in the finished water in the
New Orleans area. We can reasonably assume that membranes are available with a
favorable distribution constant. Thus, the most feasible transport can occur
where the highest vapor pressures are present. The level of the "pollution"
problem will be related to some extent to the solubility. Some examples of
solutes that may be worth considering are chloroform with a 0.82% solubility and
192 mm vapor pressure; ethylene dichloride, 0.81% solubility and 70 mm vapor
pressure; and trichlorethane with a 0.43% solubility and 125 mm vapor pressure
at 25 C. All of these solutes have a very strong dependence of vapor pressure
on temperature. (An increase of 20 C will more than double the vapor pressure
of ethylene dichloride, for example. The vapor pressure vs. temperature rela-
tionship is shown in Figure 7).
Nitrobenzene, chloroform, and ethylene dichloride were selected from this
list for screening purposes. Aqueous solutions of these solutes were pervapora-
ted at the following concentrations (by weight):
Nitrobenzene - 600 ppm
Chloroform - 8400 ppm
Ethylene dichloride 8000 ppm
These concentrations represent either solubilities approaching equilibrium
(chloroform, ethylene dichloride) or the concentration in the local waste
streams. In each run 400 ml liquid charge was used. Permeability constants
obtained from these runs are listed in Tables 7, 8, and 9.
The membranes listed in these tables are the same as those identified in
Section V.B of this report, with two additional membranes: (1) Pliolite (regis-
tered trade name of the Goodyear Chemical Co.) is a 50/50 butadiene/styrene
resin; (2) Adiprene L-315 (registered trade name of E.I. DuPont deNemours Co.)
is a polyurethane resin vulcanized with an aromatic diamine.
35
-------
600
400
200
40 80
Temperature ( C)
Figure 7. Vapor pressure vs temperature.
120
36
-------
TABLE 6
SOLUBILITIES AND VAPOR PRESSURE OF SELECTED ORGANIC SOLUTES
IN WATER
Compound
Solubility wt %
Vapor Pressure
Torr (mm Hg) at 25°C
Nitrobenzene
Dinitrotoluene
Chloro Nitrobenzene
Trichloroethane
Tetrachloroethylene
Dichloroethyl Ether
Ethylene Dichloride
Chloro Benzene
Dichloro Benzene
Chloroform
.20
.03
.003
.43
.29
1.1
.81
.04
.02
0.82
20
22
17
25
25
20
20
25
25
20
0.4
0.2
125
18
1
70
12
0.7
192
37
-------
TABLE 7
PERMEABILITY COEFFICIENTS - NITROBENZENE
Liquid Flow: 0.75 lit/min
Air Flow: 12.0 lit/min
Membrane
Thickness
(Microns)
p x 10
(cm sec)
Polyethylene (low density)
Saran
Kraton® 1101
®
Pliolite
®
Adiprene
Polyethylene (Plasticized)
14.7
19.4
25.0
38.0
59.0
24.4
6.85
6.42
9.6
10.7
.13
7.7
TABLE 8
PERMEABILITY COEFFICIENTS - CHLOROFORM
Liquid Flow: 0.75 lit/min
Air Flow: 12 lit/min
Membrane
Thickness
(Microns)
p x 10'
(cm /sec)
Plasticized Polyethylene
Polyethylene
Kraton" 1101
Saran
24.4
11.0
29.0
16
7.5
3.9
10.1
14.9
38
-------
TABLE 9
PERMEABILITY COEFFICIENT - ETHYLENE BICHLORIDE
Liquid Flow: 0.75 lit/min
Air Flow: 12 lit/min
Membrane
Thickness
(Microns)
p x 10
(cm /sec)
Plasticized Polyethylene
Polyethylene
®
Kraton
Saran
Polysiloxane (XD-7)
27
16
26
16
37
5.83
2.1
11.45
8.47
20.5
39
-------
SECTION VI
PILOT TEST OF THE PERVAPORATION PROCESS
The results of the contaminant screening experiments using dialysis with
conjugation and pervaporation showed good promise for industrial waste stream
refinement. Pervaporation, which could be used for the removal of halogenated
hydrocarbons and certain other toxic materials like nitrobenzene, is of imme-
diate interest from local water quality considerations. A list of 32 organic
compounds which have been identified in New Orleans' finished water supply is
given in Table 10. Halogenated hydrocarbons form a major segment of this list.
The list also includes nitrobenzene.
An ethylene dichloride (EDC) manufacturer supplied samples from a specific
waste stream originating in their ethylene dichloride process. This stream has
the following composition:
Ethylene dichloride: 0.8 - 0.9 wt %
Hydrochloric acid: 15 - 20 wt %
Water balance
(A modified process now under study is expected to decrease the HC1 content of
this waste stream to 3 - 3.5 wt %).
The objective in the ideal treatment process is total environmental con-
trol. Total environmental control means maximum water recycle/reuse and minimum
waste production, stressing acceptance of residuals into both raw materials and
marketable products. Such control utilizes stabilization techniques for the
remainder (minimum) so as to allow for its acceptance into the appropriate
environment.
A Kraton 1101 membrane was selected as the test vehicle on the basis of its
performance in screening experiments. The cost of this membrane, in addition
to its transport properties, was a factor in its selection. Other polymers like
polycarbonate/silicone offered higher solute transport but were ten to twenty
times more expensive (since they are still in the developmental stage) . The
polycarbonate/silicone polymer has recently been withdrawn from the commercial
market.
40
-------
TABLE 10
LIST OF ALL ORGANIC COMPOUNDS FOUND IN THE
CARROLLTON WATER PLANT (NEW ORLEANS) FINISHED WATER*
acetone
acetophenone
benzene
b r omob en z ene
bromochlorobenzene
bromoform
bromophenyl phenyl ether
(positional isomer?)
butyl benzene
a-camphanone
chlorobenzene
chloroethyl ether
chlorotnethyl ether
chloroform
chloronitrobenzene
chloropyridine
dibromobenzene
dichlorobenzene
(positional isomer)
1,2-dichloro-ethane
dichloroethyl ether
dimethoxy benzene
2,6-dinitro toluene
endo-2-camphanol
ethyl benzene
exo-2-camphanol
hexachlorobenzene
l-isoprophenyl-4-isopropyl
benzene (1,2 isomer)
isocyanic acid
methyl biphenyl
methyl chloride
nitrobenzene
o-methoxy phenol
p-methylphenol
tetrachloroethylene
toluene
1,1,2-trichloroethane
vinyl benzene
SOURCE: Industrial Pollution of the Lower Mississippi River in
Louisiana. U.S. Environmental Protection Agency, Region VI, Dallas,
Texas, Surveillance and Analysis Division. PF-611. April 1972. 146 pp.
41
-------
A. EQUIPMENT
The loading of a stacked pervaporation pilot unit is shown in Figure 8.
The assembly of the membranes, gaskets, turbulence promoters (a screen matrix
used to improve the mass transfer), and the polypropylene support plates is held
between two metallic frames by anchor bolts. Polyethylene gaskets (0.15 cm
thick) provide necessary liquid and air passages and also serve as fluid
distributors.
• i Plate
ii : i
j
Membrane
Gasket - T.P.
Membrane
Membrane
Gasket - T.P.
Membrane
Plate
Figure 8. Pervaporation unit membrane loading scheme.
Fluid distribution was accomplished by connecting the fluid ports to the
respective passages through grooves etched in the gaskets as shown in Figure 9.
Gaskets No. 1 and 3 distribute water along the membrane while gasket No. 2
distributes air. Two additional air passages are provided for air flow along
the grooved surfaces of the polypropylene plates. Figure 10 shows the counter-
current fluid flow pattern in the unit when holes corresponding to the gasket
holes are made on all except the bottom-most membrane and the gaskets are
inserted between them in the order shown. The bottom-most membrane has only two
openings (for the entrance and exit of the air stream).
The overall experimental setup was identical to that shown in Figure 4
except that the small test cell was replaced with the semipilot unit described
above.
42
-------
O
OE
a
1
so
»
o
Q?
o
o
^
2
N5
O
O
$D
IL J
^ . ^ fr-\
->-Air out
•< • Liquid
in
Figure 10. Pervaporation unit fluid flow pattern.
B. RESULTS
The following process variables were investigated for their effects on the
pervaporation rates: feed flow rate, feed temperature, feed concentration, and
air flow rate. Each variable was tested at three to four different levels
within the available range for its effect on p, the permeability. These results
thus provided a characterization of the design parameters.
1. Effect of Feed Flow Rate
Figure 11 represents the results of a change of feed flow rate from 250
ml/min to 1,000 ml/min. Permeability p at 1,000 ml/min increased more than 75%
over its base value at 250 ml/min. The increase in the liquid flow velocities
43
-------
apparently was tending to increase the solute mass transfer through an increased
turbulence. Another way to increase the turbulence is to reduce the liquid
channel height (consequently the cross section of the flow element) for the same
flow of liquid. Channel height is the height of the passage through which the
liquid is circulated. Normally, the width of this passage is constant for a
particular piece of equipment once it has been selected, but the height is
adjustable. The use of channel height adjustment to increase turbulence was
restricted, however, by pressure drop considerations. The fluid channel was
maintained between 0.07 and 0.20 cm to balance mass transfer and pressure drop
considerations.
o
CD
in
CM
la-
0.8
0.6
0.4
0.2
I
200 400 600
Feed Flow Rate ml/min
800
1000
Figure 11. Effect of feed flow rate on permeability.
44
-------
The permeability of the membrane is increased by an increase in flow rate.
This increased permeability is caused by reduction in the boundary layer forma-
tion due to the increased turbulence. Note that the relationship, as shown in
Figure 11, is not linear.
An analytical expression of the permeability and its flow rate dependence
can be established in a Wilson plot (see Appendix E, page 69). The linear
dependence found for p on Q " is similar to the Gilliland correlation , which
—0 8
relates the boundary layer mass transfer coefficient to Q . The Wilson plot
thus permits separation of the boundary layer mass transfer coefficient k .
(reflected in the slope) from the membrane diffusion coefficient D (from the
intercept). This procedure is simply the graphic representation of the series
resistance formulation:
Lk_.
liq
(9)
u
(D
tn
X
let
.75
.50
.25
0.2 0.4 0.6
Feed Concentration, wt%
0.8
Figure 12. Effect of feed concentration on permeability.
"l6T Wilson, H. Trans. Am. Soc. Mech. Engrs. 37.: 47, 1915.
17. Gilliland, E.R. Ind. & Eng. Chem. 30:506, 1938.
45
-------
Figure 12 shows the effect of ethylene dichloride concentration on the
membrane permeability. It should be mentioned here that only the ethylene
dichloride concentration was changed (from 0.8 wt % to 0.2 wt % in the feed);
the HC1 concentration remained in each case at 3.5%. No significant change was
observed in the permeability constant. This suggests that the parameters in
equation (8) are valid and that neither the diffusion coefficient D nor the
distribution constant S was affected by the change in the feed solution con-
centration over the concentration range studied.
0.8
CM
2 0.6
ID-
0.4
0.2
20
T
30
Feed Temperature C
40
Figure 13- Effect of feed temperature on permeability.
Figure 13 illustrates the effect of temperature on the membrane permea-
bility. An increase in p from 0.42 x 10 at room temperature (22 C) to
0.66 x 10 cm /sec at 40 C amounts to an almost 60% increase in the EDC trans-
port. The increase at higher temperature may be partly attributable to an
increase in the membrane diffusion rate but is mostly due to a decrease in the
feed viscosity (reducing the boundary layer resistance) and an increase in the
feed solute vapor pressure. It is frequently observed that increasing tempera-
ture causes a decrease in membrane selectivity. However, in the present appli-
cation, where the major effect was on the boundary layer, HC1 transport through
the film remained insignificant.
46
-------
2. Effect of Air Flow Rate
o
-------
in Section V.D.I., plasticization of the membrane lowers the glass transition
temperature, leading to enhanced solute permeability. Fujita has prepared a
detailed discussion of this phenomenon in the book Diffusion in Polymers,
edited by Crank and Park. Those membranes swollen by the pure solute, e.g.,
Kraton 1101 swollen by EDC, also exhibited better transport properties. These
changes can be used judiciously to obtain better transport properties. For
example, increasing the polybutadiene portion of the Kraton film would yield an
increased transport (though a poorer mechanical structure). The final film
selection thus must be based on a balance between physical and chemical
properties.
48
-------
SECTION VII
PROJECTION TO COMMERCIAL OPERATION
A design projection of full-scale operation has been assembled from the
results of the studies reviewed in Section VI. The actual industrial setting
under consideration concerns a feed of 20 gallons per minute containing 0.8 wt %
ethylene dichloride and 3.5 wt % HC1 in water (cf. the stream made available by
a chemical manufacturer page 40). Design specifications call for removal of at
least 90% of the chlorinated hydrocarbon from the waste stream. The HC1-
containing residual stream would be treated by conventional neutralization
methods.
A hollow fiber geometry was suggested for this application since hollow
fibers offer better mass transfer characteristics and membrane surface area at
•considerable economy. Due to the inherent parallel flow situation in fiber
modules, excessive pressure drops also are avoided. An effective mass transfer
2
area of 9100 m , to be provided in fibers with an inside diameter of 200 microns
and an outside diameter of 250 microns, proves necessary for 90% removal of EDC.
Membranes with higher permeabilities will have proportionately lower area
requirements. Calculations of membrane area requirements for these once-through
configurations, based on the experimentally obtained recirculating data, are
listed in Appendix D and Appendix E.
The economics of this hollow fiber design were evaluated at experimental
permeabilities corresponding to the 28-micron Kraton 1101 membrane (which was
actually studied in the laboratory) and for permeabilities one and two orders of
magnitude higher. The order-of-magnitude projection in permeability improvement
is considered well within the reach of existing membrane technology, since it is
possible to use thinner membranes (e.g. 5 microns vs. the 28-micron Kraton
membrane) and also to operate at higher temperatures (with a 60% increased
permeability observed with an 18 C temperature increase).
The flow scheme for the proposed process configuration is presented in
Figure 15. The economic analysis is assembled in Table 11. Summary results
for the several cases evaluated are compiled in Table 12.
49
-------
TABLE 11
ECONOMIC ANALYSIS FOR HOLLOW FIBER UNIT*
(Permeability equivalent of Kraton 1101 membrane)
Membrane area 9,100 m
Each plant requires its own economic analysis. Preliminary economic
estimations may, however, be based on the mass balance obtained by the
model discussed in Section VII. Some investment costs such as land cost,
etc., were not assessed since this operation would, in most cases, be
installed in existing plants where these costs have already been amortized,
Because of a small area requirement (6 sq. ft.), building costs also
were excluded. Estimation procedures recommended by Peters and Perry
were followed.
Investment:
— ty
Membranes and modules, initial cost $4/m $36,400
Pumps and piping 3,000
EDC recovery system installation 5,000
Total Investment $44,400
Operating Expenses (Annual);
Membrane maintenance $ 6,000
Manpower, 1/4 man @ $7,200 for 4-1/4 shifts
(vacation, holidays, etc. included) 8,700
Utilities, power $400 400
Total Operating Expense $15,100
Capital Cost;
Cost to pay investment in 5 years with profit
after 48% tax and 10% Depreciation $14,000
Capital and Operating Cost 29,100
EDC credit (80% recovery) @ 2-l/2(?/lb. 10,400
Cost of EDC removal $/l,000 gal water 2.16
Payout:
Payout is calculated on the basis of the annual return diminished by a 48%
tax rate, plus allowable depreciation (10-year basis). No provision was
made for interest, since these costs can be included in the risk analysis
calculations. For example, the capital requirement of $44,400 is paid
back by the sum of the depreciation $4,440 plus the net return per year
(100-48%) ($8,538), or:
,, , $44,400
Payout (5 years) = (0.52 x 8>538) + 4,440
This calculation allows a net 10% return on the investment per year.
Costs originally projected in 1972.
tPeters, M. Plant Design and Economics for Chemical Engineers. New York,
McGraw Hill, 1958.
**Perry, J.H. Chemical Engineers' Handbook. New York, McGraw Hill,
Section 26. 4th ed.
50
-------
TABLE 12
ECONOMIC ASSESSMENT
Feed: EDC Waste Stream
Processing Capacity: 20 gallons/minute
p = pR* p = 10 PK** p = 100 PK
Total Investment $44,440 $11,640 $8,360
Total Operating Expense 15,100 10,300 9,220
Capital Cost 14,000 3,680 2,640
Capital + Operating Cost 29,100 13,980 11,860
EDC Credit (80% Recovery) 10,400 10,400 10,400
Cost of EDC Removal, $/l,000 gal 2.16 0.422 0.17
* p Permeability constant
;
**p Permeability constant for lab cast Kraton 1101
51
-------
Stripped air recycle
Plant Feed Stream
Pervap
Unit
EDC loaded air
Compressor
T
-*- Aqueous effluent
I c~-
i Possible
IT"
by mass transfer
requirements.
Condenser
cooling at
,15 - 20°C
primary separa-
tions unit of
plant.
Figure 15- Flow scheme for proposed EDC recovery system.
The question of EDC recovery from the purging air stream is not a trivial
one and would not likely be justified purely in terms of the value of recovered
product. This necessary recovery operation, to prevent air pollution, in large
measure determines the acceptable air flow rate in the pervaporation system.
The recovery is aided by low flow rates and high effluent concentrations.
Transport across the membrane can be maximized by increasing the stripping gas
flow rate to maintain a low concentration in the stripping stream; thus a good
concentration driving force for transport across the membrane is maintained. In
Figure 14, the lowest flow rate tested appears to be adequate for maintaining
the concentration gradient in this particular test configuration. It is signi-
ficant that the air stream is free from the water pickup that would occur in a
straight air-stripping process. The pervaporation process avoids the latent
heat load associated with water content and also avoids a decrease in EDC
volatility which occurs in the presence of a water phase. The present calcula-
tions and economics were based on an air counterflow of 30 lit/rain (requiring a
52
-------
compression to 10-12 atm, given a cooling sink at 15-20°C). A formal optimiza-
tion on the air-flow specification was not undertaken since the pertinent cost
contribution to the overall capital requirements was modest (15%). A definitive
design for any given application would, of course, require further consideration.
The following observations are appropriate in reviewing the economic
evaluation:
a. The projected cost of pervaporation processing ($2.16/1,000 gallons
water) is comparable to the costs of processing by various other advanced waste
water treatment processes shown in Table 13. With reasonable improvements in
solute flux through the membranes, with larger scale operations, and with
parallel reduction in membrane costs, the processing cost by pervaporation can
be brought down toward costs associated with process water treatment techniques
($0.10 - 0.50/1,000 gallons)
b. The predominant cost items are the membrane investment and membrane
maintenance. These costs are based on the costs of small units (approximately
2
1m) now commercially available for other applications (e.g., hemodialysis).
The actual cost should be much less for the larger installations.
c. The technique of pervaporation is economically very attractive when
the recovered product has a high market value. Even for the "distress valued"
ethylene dichloride, a substantial portion (33%) of the total cost is recovered
in the credit of the recovered product. The product recovery is, therefore, a
consequential aspect of the pervaporation economics.
d. Membrane life will depend on the peculiarities of the application;
it will typically range between one and two years. A one-year life was
assumed for the present evaluation.
e. Overall efficiency of ethylene dichloride recovery was assumed
to be 80%. This allows for loss in the condensation system of as much as 10%
of the ethylene dichloride removed by the pervaporation. With a recycling air
purge, this loss can be significantly reduced.
53
-------
TABLE 13
COMPARATIVE COSTS OF PROCESSING
*
Cost per 1,000 gallons
Process (U.S. dollars)
Ultrafiltration $0.10 to $ 5.00
Reverse Osmosis .20 to 5.00
Centrifugation .30 to 10.00
Electrodialysis .20 to 5.00
Porter, M.C., and A.S. Michaels, Chem. Tech. 55, January 1971.
54
-------
APPENDIX - A
PERMEABILITY COEFFICIENTS
DIALYSIS WITH CONJUGATION
In Figure 16 is shown the schematic of the dialysis situation, ignoring
solution boundary layers for the present. The feed compartment contains water
and aniline solution. It is separated from the product (downstream) compartment
by a polymeric membrane which has a very low hydraulic conductivity; i.e.,
convective transport by a pressure gradient is very low. The product compart-
ment contains water, and an excess of acid over that required to establish the
equilibrium.
Membrane
cf
org
Feed
(upstream)
* cd
org
Cd .
conj
cd,rod
Product
(downstream)
Flux J
Figure 16. Dialysis scheme.
The concentrations are designated as follows:
C = concentration of organic solute in feed compartment
org
C = concentration of solute in product compartment
org
55
-------
C .= concentration of acid or base in product compartment
C ,= concentration of anilinium or phenolate in product compart-
ment.
The flux of aniline from the feed compartment to the product compartment is
given by:
d - cf )
L org org'
where J.. is the steady state flux, C and C are the bulk concentrations
1 J org _ org
under zero boundary layer conditions and p is a membrane permeability constant,
and L is the membrane thickness.
At equilibrium, the relationship between the three components in the pro-
duct compartment is given by:
K , (U)
» (Cd ,) (13)
v conj conj prod'
The zero superscripts denote that the values refer to the beginning of the
experiment. The rate equations, relating the changes in concentration in two
compartments having volumes Vf and Vd (separated by membrane of thickness L and
area A) , and containing an excess of acid in the product compartment, are given
by:
56
-------
(Cd - Cf ) (14)
org org'
nrd rrd rf ^ ^rd
dC. — (.C — C ) dC , /-ir\
org = -Ap org org prod (15)
dt V,L ~ dt
d
/rd f , d
TTT W ~^ / "" »• ^ -Jj.
V,L org org eq conj dt
We assume that equilibrium is attained when C is consumed by conju-
org
gation in the product compartment as rapidly as C diffuses into the com-
partment. With the boundary conditions cited, this leads to a discrete
expression for the permeability constant in terms of the measurable
experimental variables, as follows:
K C° .)L (Cf )
ea cony _ -, org
- J - ln — _fe
At (1 + K C . + V,/V.) (C1 - C )
eq conj f d org org
Since the equilibrium constant for aniline is known, the experimental data
allow the calculation of the permeability constant p. The form of equation (17)
is most useful for application to measurements with a rotating batch dialyzer.
The use of equation (17) requires a knowledge of the concentration of the free
aniline base in the product compartment. This was calculated from the analysis
for total aniline, and a knowledge of the acid concentration and the equilibrium
/ O
constant. The K as designated in equation (17) is 1.9 x 10 (cm /sec),
so that (Cdr ) is approximately equal to (C d> /(cconj) x 10 » which
is much smaller than C^r . If the product K (CCQ .) >» 1, and Vf - Vd>
then the coefficient for the logarithm term of equation (17) can be simpli-
fied to:
VJ. (Cf )°
P=vf- In °r* . (18)
At (Cf - Cd )
org org
57
-------
In our experiments (C ) was maintained much lower than (C ) , so
org f o ?rg
that the final expression contained only the term in ((C ) / (C1- )) .
org org
58
-------
APPENDIX - B
DISTRIBUTION COEFFICIENTS
To determine the distribution coefficient of aniline between an aqueous
solution and the membrane phase, a gas chromatographic method was used. A small
sample of the membrane was equilibrated in a solution of aniline in water over-
night. The membrane was blotted to remove surface liquids and then placed into
the injection port of a gas chromatograph equipped with a flame ionization
detector. The column packing used was a 10% OV-17 silicone on Chromosorb. The
18
method has been published by Dupuy and Fore, et al. . After the injection port
is closed, the carrier gas sweeps the membrane sample free of aniline. At the
completion of the run the membrane is retrieved and weighed on a microbalance.
With the same sensitivity and attenuation settings, a 2.0 microliter sample of
the equilibrating aniline solution is also analyzed. The ratios of peak areas
are then used to calculate the distribution coefficient with allowances incorpo-
rated for the relative densities of the solution and the polymer. The final
distribution coefficient is dimensionless; i.e., (g/cc) ../(g/cc) ,, where
subscripts pol and sol indicate polymer and solution, respectively.
18. Dupuy, H., and S- Fore. J. Amer. Oil Chetn. Soc. 4£:87a, 1971.
59
-------
APPENDIX - C
PERVAPORATION TRANSPORT ANALYSIS
We will consider the pervaporative membrane transport situation across the
membrane M, separating a liquid and a vapor phase. Transport is subject to
boundary layer resistances at 1-2 and at 3-4, as follows:
12 3 4
Liquid
Vapor
Figure 17. Schematic of membrane transport.
1. FLUX FORMULAS
For a given component i, one can formulate the transport through the three
phases (liquid, membrane, vapor) as follows:
m
-
L 2
H^-k.. (C}iq
i liq 1
k (cYap
vap 3
(19)
(20)
(21)
60
-------
2
where N = molar flux (moles/cm sec)
p = specific permeability coefficient
L = membrane thickness
m = membrane phase
liq = liquid phase
vap = vapor phase
k = mass transfer constant
C = concentration
Subscripts 1,2,3,4 refer to the boundaries as numbered in Figure 17
above. No convective flow is allowed across the membrane.
2. EQUILIBRIUM CONDITIONS AT THE INTERFACES
The fluxes given above will be equal at steady state and under conditions
where accumulations in the membrane and in the boundary layers can be considered
negligible. Additionally, one may accept the Lewis interface equilibrium:
liq jn jn vap
2 = J2 and J3 = T3
where / represents the Lewis fugacity.
This leads to the expression:
<23)
where y is the solute activity coefficient, P is the standard solute vapor
pressure, x« the solute mole fraction in the liquid, and z« the solute mole
fraction in the membrane. It is noted that for dilute solutions,
(24)
61
-------
Then
,liq jn liq
-,m
liq m
p
= S
(25)
where S is a distribution coefficient, and p's are densities.
Similarly at the vapor interface one obtains
nw
Y-p z = fvap _
Y3 o Z3 ^3
(26)
where P. is partial pressure. From this expression comes:
.vap
m
Y3Po
RTp
m
(27)
or a Henry's Law form:
rm m
c3 P
P.
i
= H
(28)
3. REDUCTION OF FLUX FORMULAS TO USABLE FORMS
Membrane concentrations in equation (19) can be replaced using equations
(25) and (26) to yield:
N =
i L
= k
k
liq
liq - liq
J*. V<-">
vap 3
62
(29)
-------
To eliminate concentration terms that are not tractable analytically, we
eliminate C and C*"1' algebraically to yield:
Ni = Uliq (Cl - ) (30)
where U.. . is the liquid phase mass transfer coefficient
.
liq vap
4. APPLICATIONS TO TEST CELL CONFIGURATIONS
The pervaporation test cell can be used to obtain coefficients in the
characterizing equation (30) as follows:
a. Assume that the membrane of area A is used to reduce the concentra
tion in a recirculating liquid sink of volume V, and initial concentration
C, (o) . The vapor side is maintained everywhere at a concentration C,
fresh supply. The mass transfer of solute is given by:
(35)
leading to
63
-------
- In
vap
(o) -
U. . A
liq
(36)
on integration.
b. Alternatively, for a long path length, the solute accumulation in
the vapor purge stream may not be ignored. For a vapor flow G, along a membrane
device of length 7- and cross-sectional area A', one can write the material
balance equation:
9 (GcYap)
4
TI
Uliq
A' -
A -
3 (V Cjap)
yap 4
3 —
When the vapor flow is fast relative to the liquid flow, then
(37)
dC
,vap
dl
• -u,
-------
out
(c
q
U A
6C4) exp -
To calculate the change in liquid phase solute concentration, we can now
write a material balance based on the change in vapor phase concentration, as
follows:
v 1 - r (r out r in>i
V dt ~ G (C4 - C4 }
(42)
dt
_G
VB
G (1 - exp
VA3
. Ag
Ag
exp
(43)
(44)
= U
(45)
which integrates to:
In
Cl -
in
in
U t
(46)
TT G
(47)
65
-------
APPENDIX - D
PERVAPORATOR ANALYSIS
The performance of a "once-through" pervaporator could be projected based
on the following transport analysis. In this analysis, flat sheet membrane
geometry, a countercurrent gas-liquid flow and constant fluid rates are assumed,
The analysis, as developed, is for dilute solutions.
Consider a section of stack-loaded pervaporator as shown in Figure 18.
vap
Figure 18. Schematic of stack-loaded pervaporator.
At steady state,
Liquid-side material balance:
fz ^iiqW + uiiqA (ciii ~ %C-P} = °
(48)
66
-------
where G = flow rate
C = concentration
A = area
U = mass transfer coefficient
1 = membrane length
liq = liquid phase
vap = vapor phase
K = equilibrium constant
eq
o = length 0
Gas-side material balance:
~T (G C ) + U.. A (C,. - K C ) = 0 (49)
at, vap vap liq liq eq vap
Equations (48) and (49) are a set of linear differential equations.
These can be solved simultaneously using Laplace Transforms and known
boundary conditions (Cn. , C ).
J liq' vap'
Rearranging equations (48) and (49) we have:
c
. \ dC,.
ML ) lii = „ (K c _c ) (50)
Ay dl liq eq vap liq
and
= n . (K C - C. )
'
(51)
Ay
-------
U1± A
Where n = -i-
- " G^~ (54)
liq
U, . A
"2 -iT33- <55)
vap
y-t
Y = 0^ = nf (56)
vap 1
Laplace Transformation of (52) and (53) would yield
C s - C° = n,K C - n, C,.
liq liq 1 eq vap 1 liq (57)
— o — —
P o — p — n v r •*
*-» t> v_* — iirt j\. L* ~~ n..,
vap 2 eq vap 2
In order to evaluate C-. in pure solution (solute mole fraction z = 1),
and Cva with no solute present (z = 0), equations (57) and (58) are
rearranged and subject to inverse transformation. Whence
•L [(l-KeqY) exp (-6E)]
Cliq [ l-KeqY exp (-?Z) ]
K [1 - exp (-?£)] 7
+ TT% - TTTTT (C + YC°. ) (59)
[1-K Y exp (-CZ.)] vap liq'
U A
where g = ^3- (1 - K v) (60)
liq
and C° = CZ + y(C° - c. ) (61)
vap vap ' liq liq
are obtained.
Equations (59) and (61) are used to generate concentration profiles in
liquid and gas phases along the length of the stack-loaded pervaporator .
68
-------
APPENDIX - E
MEMBRANE AREA REQUIREMENT
Plant Capacity : 20 gallons/minute
Feed Composition : 0.6 wt % ethylene dichloride
3.5 wt % hydrochloric acid in water
Specification : 90% removal of ethylene dichloride
Design Parameters : k° (mass transfer coefficient)
D (diffusion coefficient)
Design parameters k° and D are obtained from the permeability measure-
ments at different flow rates. The overall liquid side mass transfer co-
efficient is given by:
«.,,.„ = i r-^ (62)
+ ^7k
1 vap
pS
where a, = ^(63)
1 J-1
and a^ = RTHn/t (64)
where T is temperature, L is membrane thickness and S is solubility.
Since k is extremely high, equation (62) can be expressed as:
U.. --. -, (65)
iiq !_ + I
a. kn .
1 Iiq
but the specific permeability coefficient p and IL. are related in the
.form
U,. = p/L (66)
Uq
69
-------
Therefore, substituting the relationships (63) and (66) in (65) we get:
DS Lk
liq
(67)
However, liquid side mass transfer coefficient is typically related to
the feed velocity v (cm/sec) in the form:
k_. = k° N '
liq Re
based on Gilliland correlation , where N is the Reynold's number.
(68)
Re
Substituting (68) for k.. . we get
DS _. 0 „ 0.8
p Lk N_
K Re
(69)
_
and, therefore, a plot of p
—0 8
vs. N " (Wilson Plot) would give:
Slope =
Intercept =
Lk°
1
DS
(70)
(71)
Since membrane thickness L and solubility S are known, k° can easily be
computed.
2 2
X
V
T
10V1
Figure 19- Wilson plot.
T
8
10
70
-------
Figure 19 shows a Wilson Plot obtained from the experimental data. The
slope and intercept are respectively 2.03 x 10 sec/cm2 and 0.56 x 107 sec/cm2.
Since the film thickness L is 28 x 10~ cm and S is 50, k° and D are respectively
1.76 X 10~ and 3.6 X 10~ , as per equations (70) and (76).
The liquid phase concentration profile (see Appendix D) is given by:
I [(l-K£qY) exp (-q) ]
Cliq U-KY exp (-
K [1 - exp (- -
+ Tl=K~T exp (-a)] (Cvap + YC°llq> (72)
where K = RTH/S (73)
U . A
U-K Y> (74)
liq M
= Gliq/Gvap
Uliq= V^+VW (76)
• - *
Also:
C°
***-•
% Solute Removal = — - --- (78)
By use of equations (72) and (78) , it is possible to determine film area
required for a specified solute removal.
As an illustration consider that the module contains 100,000 fibers made
of the experimental Kraton 1101 film, each fiber with an inside diameter of
100 microns and an outside diameter of 250 microns. Then for 20 gallons per
minute feed flow rate,
NRe = 82.3
but k° = 1.76 x 10~6 (From Wilson Plot)
71
-------
Therefore, k^ = k° NRe°'8 = 6 x 10"5 cm/sec
-9 2
D = 3.6 x 10 cm /sec (Calculated from Wilson Plot)
Therefore, a,
Y
K
since
Therefore,
eq
R
7.2 x 10~5 since S = 50, L = 25 x 10~4 cm
2.55
RTH/S = 1.95 x 10~5
0.108, T = 298°K, H = 3.03 10
,-5
_c
= 1.21 x 10
Now,
C° = 0.008, C = 0
1 -"• ' vap
Thus if
solute removal is 90%.
1 = 145 meters, C from equation (72) is 0.008 and the
Thus,
—4 2
Area/meter fiber length = 6.28 x 10 m /m.
-4 5
Total Membrane Area = 145 x 6.28 x 10 x 1 x 10
= 9.1 x 103 m2.
72
-------
REFERENCES
1. Crank. J. Methods of Measurement. In: Diffusion in Polymers, Crank, J.,
and G.S. Park (eds.). New York, Academic Press, 1968. p. 1.
2. Wendt, R.P., R.J. Toups, J.K. Smith, N. Leger, and E. Klein. Measurements
of Membrane Permeabilities Using a Rotary Batch Dialyzer. Ind. & Eng. Chem.
10:406, 1971.
3. Regan, T.M., W.G. Esmond, C. Strackfus, and A.M. Wolbarsht. Science. 162:
1028, 1968.
4. Klein, E., J.K. Smith, R.P. Wendt, and S.V. Desai. Separation Science. 7_
(3):285-292, 1972.
5. Fujita, H. Organic Vapors above the Glass Transition Temperature. In:
Diffusion in Polymers., Crank, J. and G.S. Park (eds.). New York, Academic
Press, 1968. p. 75-106.
6. Klein, E., J.K. Smith, F.F. Holland, and R.E. Flagg. Membrane and Materials
Evaluation; Permeabilities, Physical and Mechanical Properties of Hemo-
dialysis Membranes—Bemberg Cuprophan PT-150 Membrane. Gulf South Research
Institute. Annual Report AK-1-72-2221. Artificial Kidney-Chronic Uremia
Program, National Institutes of Health. July 1973. 24 p.
7. Walsh, A. Atomic Absorption Spectroscopy. ASTM Special Technical Publica-
tion. No. STP 443,31. June 1968.
8. Marcus, Y. and A.S. Kertes. Ion Exchange and Solvent Extraction of Metal
Complexes. London, New York, Sidney and Toronto, Wiley-Interscience, 1969.
1037 p.
9. Binning, R.C. and F.E. James. Petrol. Refiner. 3T.(5) :214, 1958.
10. Li., N.N., R.B. Long, and F.J. Henley. Ind. & Eng. Chem. .57(3):19, 1965.
11. Binning, R.C., R.J. Lee, J.F. Jenning, and E.G. Martin. Ind. & Eng. Chem.
Chem. 53_:45, 1961.
12. Binning, R.C. and J.R. Kelly. U.S. Patent 2,913,507. To American Oil
Company, November 17, 1954.
13. Choo, C.Y. Advances in Petroleum Chemistry. 6^:73, 1962.
14. Sanders, B.H. and C.Y. Choo. Petrol. Refiner. 39^(6):133-138, 1960.
15. Yasuda, H. J. of Polym. Sci., Part A. _5, 1967.
16. Wilson, H. Trans. Am. Soc. Mech. Engrs. 3J:47, 1915.
73
-------
17. Gilliland, E.R. Ind. & Eng. Chem. 30^:506, 1938.
18. Dupuy, H., and S. Fore. J. Amer. Oil Chem. Soc. 48:87a, 1971.
74
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-76-223
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Dialysis for Concentration and Removal of Industrial
Wastes
5. REPORT DATE
October 1976 (Issuing Date_
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
James K. Smith, Shyamkant V. Desai, R.E.C. Weaver
and Elias Klein
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Gulf South Research Institute
P.O. Box 26500
New Orleans, Louisiana 70186
10. PROGRAM ELEMENT NO.
1BB610
11jj§ptKDHBQC/GRANT NO.
12020 EMI
12. SPONSORING AGENCY NAME AND ADDRESS
Robert S. Kerr, Environmental Research Lab. - Ada, OK
Office of Research and Development
U.S. Environmental Protection Agency
Ada, Oklahoma 74820
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/15
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This project evaluates dialysis for its potential for treatment/recovery of a
number of organics and inorganics found in industrial wastes along the Lower
Mississippi River.
The feasiblity of three membrane techniques was developed.
1. The use of acid and base conjugation on the downstream side of membranes perm-
selective to aniline and phenol.
2. The use of a chelating complex for Al, Cu, and Pb ions.
3. The use of a pervaporation scheme in which a hydrophobic membrane passed
volatile contaminants such as nitrobenzene and ehtylene dichloride to a downstream
vapor purge..
Rotating batch and mimi-plant plate dialyzers were used. A design projection of
full-scale operation was made for pervaporation of ethylene dichloride.
Theoretical and practical aspects of the diffusion phenomena were discussed.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
COSATI Field/Group
Membranes, Dialysis, Organic solvents,
Copper chlorides, Aluminum chloride,
Lead halides, Phenols, Anilines, Hydrogen
chloride, Chloroethanes, Cost effective-
ness
Ethylene dichloride,
EPA Research
Conjugation
Complexing
Pervaporation
Lower Mississippi River
7A & C
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)'
Unclassified
21. NO. OF PAGES
87
20. SECURITY CLASS (This page)
unclassified
22. PRICE
EPA Form 2220-1 (9-73)
75
* us. GovBM0r nmnw ofret IJTT—757-0 56 / 5494
------- |