WATER AND ION FLOW THROUGH
IMPERFECT OSMOTIC MEMBRANES
ERNEST JOSEPH BRETON, JR.
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
The work reported in this dissertation was made
possible by the support of others. Financial support for
the study was afforded by the Office of Saline Water of
the United States Department of the Interior. This Office
has been very cooperative in every aspect of our association.
The congenial relationship which exists between
graduate students and the Chemistry Department faculty
creates a very favorable atmosphere for graduate research.
Within the department many individuals have been exceedinrly
helpful. B. J. Otte, Morris Mixson, and J. H. Mans have
gone out of their way to furnish necessary supplies. Calvin
Workinger has been of inestimable help in building the ap-
paratus that was required for this work. He displayed a
great degree of resourcefulness and ingenuity in making
the apparatuses for measuring semipermeability and resistance.
Association with my faculty adviser, C. E. Reid, has
been very stimulating and harmonious. I am especially rrate-
ful for the very wide degree of research freedom that he has
given me. He has proved to be an authoritative source of
information and help in the areas of chemistry, chemical
physics, and mathematics. In three years of close
association, I never did perceive the limit to his knowledge
in these fields. He has also stimulated me to broaden my
interest in the arts--especially music.
Much of the analysis work involved in obtaining the
data for Tables 1 and 2 was done by Hugh Wise.
Several very helpful suggestions in the des.in of
the semipermeability apparatus were made by G. R. Kulkarni.
Mary Joy Breton has been a tremendous help in pre-
parlng this dissertation. In addition to typing several
drafts and the final copy, she assisted in editing the
TABLE OF CONTENTS
A"KY.' I'.LFDGMENTS . .
LIST OF TABLES . .
LIST OF FIGURES. . .
. '"'- C'rUCTION . .
* S 4
I DISCOVERY . .
Review of the Literature.
Semipermeability Tests. .
a 4 4 *
II T!TORY FOR THE SEMIPERMTEAILITY OF C"LLULOSE
ACETATE .. . .
Review of Existing Theories .
Fronosed Mechenism .
Literature Support for Theory .
III E(rP EIrflNAL SUPPORT OF THEORY BASED UPON
:..tiIl!P:. ABILITY DETERMINATIOiNS .
Effect of Pressure on Semipermeability. .
Effect of Crystallinity on Semipermeability
IV ELECTRICAL MEASURK' iE,.; S TO SUPPOhT THEORY .
Dielectric Measurements of Cellulose
Acetate. . .
Electrical Resistance Across Cellulose
V MECHANISM FOR LOSS OF rE.Bf'P.:'. SEMIPEfI1 -
ABILITY . .
Loss of Semipermeability. .
Reactions Between Solute and membrane .
Mechanism of Membrane Failure .
Structural Chnnres. .
. . ii
TABLE OF COIJTENTS (Continued)
VI RECOMMENDATIONS FOR AN IMPROVED inlEinRA1:E 75
VII SU3I A RY .. . 78
APPENDIX . . 81
Analytical Procedures. ... 82
Preparation of Films . 82
Determination of Dielectric Constants. ,. 84
Determination of Membrane Resistances. .. 86
BIBLIOGRAPHY .. . 91
BIOGRAPHICAL ITEMS . . 95
LIST OF TAPL:S
1. Semipermeability of Various Membranes to
Aqueous Sodium Chloride Solutions 9
2. Semipermeability of Cellulose Acetate to
Aqueous Solutions of Various Electrolytes .. 11
3. Effect of Semipermeability on Birefringence. 72
LIST OF FIGURES
1. Apparatus for determining the semipermeability
of imperfect osmotic membranes. . 7
2. Effect of pressure on the semipermeability of
40% cellulose acetate membranes . 26
3. Effect of acetyl content on salt rejection 28
4. Effect of acetyl content on flow rate. 28
5. Effect of pressure on the dielectric constant
of cellulose acetate (du Pont CA-43). 31
6. Resistance versus pressure across 40% cellulose
acetate membrane. . 34
7. Resistance versus pressure for sodium chloride
across cellulose acetate membranes (du Pont
CA-h3) of known salt rejection. . 36
8. Comparison of semipermeability and dielectric
constant of cellulose acetate membranes 38
Q. Resistance versus pressure for cellophane
(du Pont PT-300). . 40
10. Resistance versus pressure for cellulose
acetate membrane (du Pont 88 CA-43) with
a known salt rejection of over 90. .... 42
11, Effect of temperature on membrane resistance
of cellophane (du Pont PT-300) .. 43
12. Resistance versus pressure for 4C 0 cellulose
acetate film. No previous cycles 4
13. Resistance versus pressure across cellulose
acetate (du Pont CA-43) 47
1L. Resistance versus pressure across 40 cellulose
acetate ..... . 51
LIST OF FIGURES (Continued)
15. Effect of method of removin, solvent on the
durability of !7.2$ cellulose acetate
membrane . . 4
16, Effect of method of removing solvent on the
variation of salt rejection with volume. 55
17. Effect of method of removing solvent on the
variation of flow rate with time 56
18. Effect of acetyl content on durability of
cellulose acetate. .. 60
19. Effect of solute on durability of 37.""
cellulose acetate membranes. 62
20. Effect of pH on durability of 37.2'' cellulose
acetate. .. . 64
21. Unit cell of native cellulose . 69
22. Unit cell of swelled cellulose. . 6
23. Birefringence of a du Pont CA-143 membrane
that had a hiah salt rejection (over O0"). 73
2h. Birefrin7ence of a du Pont CA-l3 membrane
that had failed (5.h' salt rejection). 73
25. Apparatus for measuring membrane resistances. 87
26, Wiring diagram for determinlnr membrane re-
sistances.. . 88
As the population of the world continues to in-
crease, greater and greater demands are being made upon
the available fresh water resources. Consequently, there
are many heavily populated regions of the world in which
fresh water is becoming increasingly scarce. The problem
is especially acute in southwestern United States and in
the Mediterranean area. Fresh water needs could be more
than adeouately met if an economical process could be
found for purifying saline waters. In order to develop
a suitable process, the Congress of the United States has
set up the Office of Saline Water under the Department of
the Interior. This Office is considering every conceivable
means for removing salt from saline waters. One promising
method is based upon the principle of osmosis. At the
University of Florida, the Department of the Interior is
supporting a program to explore this particular method
for purifying sea water.
The phenomenon of osmosis has long been of scien-
tific interest. As far back as the eighteenth century it
was observed that when certain aqueous solutions were sepa-
rated from water by animal membranes, water spontaneously
passed through the membrane into the solution. The membrane
served the function of permitting the water to pass while
blocking the counter-flow of the solute. J. H. van't Hoff
(1) described these membranes as semipermeable.
Since then, osmosis has had a significant place in
textbooks on physical chemistry. It is frequently used to
illustrate the concept of reversibility. Consider an os-
motic cell in which the water is passing from the fresh
water side through a semipermeable membrane into an aqueous
salt solution. This flow of water through the membrane can
be retarded by applying pressure to the solution. The pres-
sure at which flow stops is the osmotic pressure. If the
pressure is decreased by an incremental amount, water will
again begin flowing into the solution. But if the pressure
is increased by an incremental amount, a reverse flow of
water will occur from the salt solution to the pure water
side. The term reversible is used to describe this process.
Recently it has been suggested (2) that this effect
of reverse osmosis might be practical for purifying saline
waters. Since the process would approximate reversible
conditions, the energy requirements should approach the
thermodynamic minimum required to separate salt from sea
However, no membrane was known that could be used
in such a process. In fact, no synthetic organic membrane
had yet been discovered that would function effectively as
a semipermeable membrane in salt water.
This, then, is the problem with which this disser-
tation is concerned--the semipermeability of synthetic or-
ganic membranes in strong electrolytes.
In Chapter I the discovery of a membrane (cellulose
acetate) which is semipermeable in strong electrolytes is
described. This discovery prompted much curiosity about
the mechanisms involved in the transfer of water and ions
across the membrane. A theory of these mechanisms is pro-
posed in Chapter II. In the two subsequent chapters, various
tests of the theory are described.
Although the discovery of cellulose acetate as an
effective semipermeable membrane was a big step in the de-
velopment of an osmotic membrane for the purification of
saline water, yet for various reasons this membrane would
not be practical. One of the reasons is its limited life.
After several weeks of continuous use, cellulose acetate
ceases to behave as a semipermeable membrane in strong
electrolytes. In order to extend the life of the membrane
it is necessary to gain an understanding of the process
involved in film failure. This work is presented in
In order to maintain continuity throughout the
text of the dissertation, experimental details have been
placed in the Appendix.
Review of the Literature
Most of the studies on osmosis reported in the
literature have dealt with systems in which the solutes
were relatively large organic molecules. Usually cello-
phane, collodion, and natural membranes were used. Very
little has been reported on osmosis of solutions of strong
electrolytes. This is due mainly to the difficulty which
has been encountered in finding a semipermeable membrane
for these systems.
Certain biological membranes are known to function
as osmotic membranes in strong electrolytes. Cooper (3),
for example, has reported that the membrane in a particular
species of seaweed found off the coast of India is semi-
permeable in sea water.
Of all the reported attempts to synthesize a perfect
osmotic membrane, the copper ferrocyanide membrane, first
discovered by M. Traube in 1864, comes closest to achieving
this goal. Using this membrane, Morse (4) perfected a
technique for measuring high osmotic pressures. Employing
a 0.5 N potassium chloride solution, this membrane was able
to hold an osmotic pressure of twenty atmospheres for twen-
ty days. However, the film failed rapidly in succeeding
experiments. Berkeley and Hartley (5), employing the same
membrane, measured the osmotic pressure of a wide range of
inorganic electrolytes. More recent work with this film
has been reported by Craig (6) and Mibashan (7).
The literature survey revealed that several syn-
thetic organic membranes have been evaluated. Trautmann
and Ambard (8) have checked the semipermeability of cello-
phane in solutions of uni-univalent and uni-bivalent salts.
When a 0.0017 molar solution was forced through cellophane,
they found that it prevented the passage of 30 per cent of
the uni-univalent salts and 75 per cent of the uni-bivalent
salts. Baumgarth (9) checked the effect of ultrasonics on
the diffusion of NaC1 through cellophane and observed that
its rate of diffusion was increased.
Dried collodion membranes have been found by North-
rop (10) to be only slightly permeable to salts of strong
acids and bases, yet highly permeable to water. Employing
various special treatments, Sollner (11), Elford (12), and
others have also worked with collodion membranes.
BurPess (13) made osmotic films from sodium algi-
nate that retarded the passage of the sodium ion.
An apparatus for ultrafiltration has been described
by Baudouin and Lewin (14) in which a cellulose acetate
membrane was used. However, no mention was made in the ab-
stract of filtering solutions of strong electrolytes. Re-
cently Terazawa (15) has proposed that in general crystal-
line semipermeable membranes should be impermeable to ions
of strong electrolytes. He classified copper ferrocyanide
as a crystalline membrane. Gels are another type of film
that may function as osmotic membranes. In this category
Hermans (16) has evaluated the osmotic properties of cellu-
lose xanthate gels in strong electrolytes.
In order to find a semipermeable membrane for
strong electrolytes, a trial-and-error approach was used--
although it was felt that the desired membrane should con-
tain some hydrophilic groups. Many types of thin films
were obtained from the leading chemical manufacturers in
the country. Some films were prepared in the laboratory.
The apparatus that was ultimately developed for measuring
the semipermeability is shown in Figure 1. It embodied the
following features: (1) Pressures up to 2000 p.s..i could
be obtained by means of compressed air or nitrogen. (2)
Stirring was accomplished by convection. One leg was
heated and the other cooled. (3) Monel was used throughout,
which reduced corrosion to a minimum. (4) The membrane was
supported by a ceramic disc which was recessed into the base
So Impres- ed__
Ja c ket
Heat i n Coil
380t /fo les
le rnb ra ne
Fip. 1.--Apparatus for determining the semi-
permeability of imperfect osmotic membranes.
plate and turned on a lathe to a flat surface. This design
permitted the testing of very thin films at high pressures.
To measure the semipermeability, a film was first
clamped into place as illustrated in Figure 1. The apparatus
was filled with salt solution and the pressure applied. Water
that passed through the membrane was collected at the bottom
and analyzed. From the concentration of salt in the appara-
tus and that in the water that had passed through the mem-
brane, the percentage of salt rejected was calculated. The
percentage of salt rejected is taken as a measure of semi-
The results of the screening program are recorded
in Table 1. As can be seen by the table, the semipermeability
of du Pont cellulose acetate far exceeded that of any other
film tested. Consequently, further study was confined for
the most part to cellulose acetate membranes. The semi-
permeability of du Pont CA-48 cellulose acetate to various
solutions is given in Table 2.
These results were significant. They indicated that
it is possible to synthesize an organic membrane that is
semipermeable in strong electrolytes. However, the cellu-
lose acetate membranes that were tested could not be used
as ultrafilters for purifying saline waters--first, because
the rate of filtration would be too slow and second, because
their life was limited to only a few weeks.
SEMIPIRMEIABILITY OF VARIOUS MEMBRANES TO
AQUEOUS SODIUM CHLORIDE SOLUTIONS
Pressure on Semipermeability,
Mb nMembrane % Rejection(a)
Tested Membrane RejectionRemarks
p.s.i. of Dissolved NaC1
(du Pont Elvanol
(Baked at 145C.)
(Baked at 1450C.)
50% Polyvinyl alco-
hol, 50f phenol-
75% Polyvinyl alco-
hol, 25% polytetra-
Amberplex A-I anion
membrane (Rohm and
Amberplex C-1 cation
membrane (Rohm and
TAELE 1 (Continued)
Pressure on Zemipermeability,
Membrane Hiembrane % Rejection(a)ak
Test ed Remarks
p.s.i. of Dissolved NaC1
by 1 membrane
Pont PT 300)
Saran (Dow Chemi-
(du Font 88 CA-43)
(du Pont 88 CA-b3)
(a)Percent rejection was based upon the change in
chloride concentration from influent to effluent.
(b)If after 48 to 72 hours, no water was collected,
an entry of no flow was made.
SENIPERrABILITY OF CELLULOSE ACETATE TO AQUEOUS
SOLUTIONS OF VARIOUS ELECTROLYTES
membrane Pressurelectro- Concen-(b) Rejection of Dis-
onMe- lye traction solved Electrolyte
brane by Membrane
Same film as
used for NH3
NagSO, 0.018 M
Pont CA-48 cellulose acetate membrane 0.88
mils thick was used for this series of tests.
(b)Analytical procedures are given in the Appen-
-- --- -- --
Before an improved membrane can be developed on a
systematic basis, some knowledge of the mechanisms in-
volved in the transfer of water and ions across this mem-
brane is needed.
THEORY FOR THE SEMIPERMEABILITY OF
Review of Existing Theories
Of the many theories proposed for the semiperme-
ability of membranes, none adequately explains the semi-
permeability of cellulose acetate. Several of the more
significant theories will be reviewed.
Sieve mechanism.--One of the earliest that was
proposed was the sieve mechanism (1). According to this
concept, a semipermeable membrane possesses pores inter-
mediate in size between the solvent and solute molecules.
Thus the solute molecules would be blocked and the smaller
solvent molecules would be allowed to pass. However,
several semipermeable membranes have been found in which
the pore size is larger than the solute molecules. Our
results indicate that the sieve mechanism could not account
for the high semipermeability of cellulose acetate. Refer-
ring to Table 1, nylon and saran which had much lower flow
rates than cellulose acetate (and presumably smaller pores)
exhibited no semipermeability under the conditions of the
experiment. Furthermore, the difference in size between
polymerized water molecules and sodium or chloride ions is
not sufficient to permit the sieve mechanism to apply.
Distillation mechanism.--Another theory is based up-
on the perfect semipermeability of the surface of a solu-
tion (1). If solute molecules (such as NaC1) are not vola-
tile, then only the volatile solvent molecules can pene-
trate the surface barrier. H. L. Callendar (1) considered
that solvent molecules distilled across a semipermeable
membrane. Of necessity the walls of the capillaries
through which the solvent distills would have to be non-
wettable to the solvent or the solution. This requirement
renders the theory inapplicable to cellulose acetate, be-
cause water has a high affinity for the free hydroxyl groups
in the film (17).
Ion exchange mechanism.--In recent years a consider-
able amount of research has been devoted to ion exchange
membranes. These membranes are permselective. That is,
they have a very low resistance to the passage of either
cations or anions but not both. This selectivity is due
to the high concentration of positive or negative charges
that are incorporated into the film (18, 19, 20). Conse-
quently, these membranes would be semipermeable if they
effectively blocked the passage of either the cation or
the anion of the solute and permitted solvent molecules to
It is unlikely that the high degree of salt rejec-
tion that cellulose acetate exhibits could be due to an ion
exchange mechanism. Cellulose acetate with approximately
40 per cent acetyl content has a low concentration of hy-
droxyl groups.. Thus, according to the theory of Teorell
and Meyer-Sievers (19), it would behave only slightly as
an ion exchange membrane.
On the basis of the ion exchange theory, the semi-
permeability of several ion exchange membranes was deter-
mined. None of them (i.e. tetraallylammonium bromide,
Amberplex A-1 and C-l in Table 1) exhibited a high degree
of semipermeability. McKelvey, in an unpublished report
to the National Colloid Symposium in 1056, indicated simi-
However, a series of experiments was conducted to
determine whether or not cellulose acetate actually does
behave as a permselective membrane. The results of these
experiments, which are reported in Chapter IV, indicate
that it does not.
Adsorption mechanism.--A theory of semipermeability
based upon the relative adsorption of solvent and solute
molecules by the membrane has been proposed (1). Semi-
permeability results when the solvent molecules are ad-
sorbed positively and the solute molecules negatively
adsorbed. This type of semipermeability is most pronounced
when the pore size is small and surface flow predominates
Solubility mechanism.--Closely related to the ad-
sorption theory is the solubility theory. This was first
proposed by M. L'Hermite in 1855 (1) and has today gained
wide acceptance (17). The theory proposes that semiperme-
ability is an outgrowth of the solubility of the solvent
and the insolubility of the solute in the membrane. Since
the solvent is soluble in the membrane, it can pass but
the solute cannot.
In their most Reneral sense, the solubility and
adsorption theories would certainly cover the observed be-
havior of cellulose acetate. However, this question is
then raised: Why is the solubility or adsorption of water
in cellulose acetate so much greater than that of strong
electrolytes? If it could be attributed merely to the
presence of hydrophilic groups, then cellophane and poly-
vinyl alcohol, which have a very high concentration of
hydroxyl groups, should have a greater degree of salt re-
jection than cellulose acetate does. But as indicated in
Table 1, they do not. Therefore, a clearer understanding
of the mechanisms involved in the transfer of water and ions
across cellulose acetate is needed.
To formulate a theory which would explain the semi-
permeability of cellulose acetate, it was necessary to find
out what unique combination of properties of cellulose ace-
tate caused this phenomenon. Fortunately a great amount of
work has been done on cellulose chemistry. This work is
well summarized in Volume V of High Polymers (17), and per-
tinent parts of it will be briefly described here before
the proposed mechanism is presented.
Cellulose type membranes are made up of a network
of lone polymers averaging 15,000 A in length and 5 to
20 A in width. These polymers have for the most part a
random orientation in the membrane. As an analogy, Soll-
ner (18) uses a pile of straw to describe their packing.
'.'hen polymer chains run parallel to each other in an or-
derly manner, crystalline regions are set up. In these
regions the polymers are held together by hydrogen bonds
and van der Waal forces and assume definite spaclnr-s. Qa
the other hand, when the chains lie in a disorderly manner,
the spaces between polymers are much greater. These are
called amorphous regions. The formation of these crystal-
line regions is a dynamic process. Mark, in High Polymers
(17), says that the individual chains are subject to
Brownian motion. During a given period within a given
volume, segments of polymer chains will become parallel to
each other and assume a high degree of order. Later these
crystalline regions will disappear as a result of Brownian
motion. Thus the system as a whole has an average degree
of order intermediate between complete crystallinity and
Compression of a membrane results in a complicated
rearrangement of the polymer network. During compression
the system goes through many transition states because of
the irregular shape of the polymers and the variety of
forces that come into play between individual polymer
chains. The consolidation of the three-dimensional lat-
tice thus becomes a complicated multi-step process of
settling down to an arrangement of highest compactness
This concept of the internal structure of membranes
applies to all types of cellulose films. Baker (22) has
shown, however, that cellulose acetate is unique in that
it possesses a very high degree of crystallinity. That
is, the internal forces are numerous enough and strong
enough to restrict Brownian motion and hold a large per-
centage of the polymer chains in highly ordered regions.
Of necessity the spaclnps in the amorphous regions are
much smaller than they would be if there were little or
no crystallinity. When water molecules are introduced
into this structure, they are concentrated in the amorphous
It is now postulated that these water molecules,
through hydrogen bonding, cross-link the polymer chains and
fill the voids in the amorphous regions with bound water.
The extent of cross-linking depends upon the size of the
voids. When they are large, as in the case of cellophane,
the structure is weakly cross-linked. When they are small,
as in the case of cellulose acetate, the cross-linking is
Furthermore, when there is a low degree of cross-
linking, the voids are only partially filled with bound
water leaving holes through which ions can pass. When the
network is highly cross-linked there is little or no free
space. All of the water is hydrogen bound to the polymer
network either directly or through other water molecules.
Brownian motion of the cellulose chains would thereby be
In a structure such as this, it is proposed that
two different mechanisms of diffusion occur. Ions and
molecules that cannot enter into hydrogen bonding are
transported across the membrane by hole-type diffusion.
That is, their diffusion depends upon the concentration
of holes in the membrane. Cross-linking cellulose polymers
with water reduces the probability of hole formation,
thereby reducing diffusion of this type.
Those ions and molecules that can combine with the
membrane through hydrogen bonding and can fit into the
bound water structure are presumed to be transported across
the membrane by alignment-type diffusion. Their diffusion
depends upon the establishment of hydrogen bonds with the
membrane. After these molecules combine with one side of
the membrane, they migrate across by transferring from one
hydrogen bonding site to another and are finally discharged
from the other side.
Based upon these concepts (i.e. bound water, hole-
and alignment-types of diffusion), it can now be deduced
that the semipermeability of cellulose acetate is lue to
this situation: that within the membrane strong elec-
trolytes diffuse very slowly compared to the rate of
diffusion of water. In cellulose acetate the high degree
of crystallinity reduces Brownian motion and pore size
sufficiently to permit bound water to fill the voids
which greatly reduces the probability of hole formation.
Hence the rate of diffusion of sodium chloride is very low.
If the membrane were compressed, more cross-linking would
occur, the bound water would become more stabilized, and
the rate of diffusion would become even less.
On the other hand, the diffusion rate of water
through cellulose acetate is relatively hih since it is
transported by a different mechanism. Comoression of the
membrane would not be expected to reduce appreciably align-
ment-type diffusion of water. Therefore, compression of
the structure should increase the semipermeability of the
The hydrogen bonding mechanism does not exclude the
sieve mechanism. The size of the holes that are continually
appearing and disappearing plays a secondary role. The rates
of diffusion of solutes that cannot enter into hydrogen
bonding depend upon pore size. However, in the case of
those ions and molecules that can enter into hydrogen
bonding, the permeability depends upon their ability to
fit into the ice-like structure of the membrane.
Literature Support for Theory
Cellulose acetate provides excellent conditions for
hydrogen bonding because there are large numbers of oxygen
atoms in each polymer. That water is bound to cellulose
acetate polymers is amply supported in the literature.
Gruner (23), Petitpas (24), and others (17) have discussed
chemical type bonding between water and cellulose polymers.
It has been reported by Davidson (25) that the density of
the sorbed water along the boundaries of crystalline regions
is 5 to 7 per cent higher than water at atmospheric pressure.
This corresponds to a pressure of 2000 kg./cm.2
If water were passing through cellulose acetate by
combining with the bound water within the membrane, its re-
duction in entropy should be comparable to the conversion of
liquid water to ice. Stamm and Loughborough (26) report
that the first water sorbed by cellulose acetate has a TA S
value equivalent to the heat of fusion of water indicating
that the water does have orientation (and hence bondinp)
comparable to that of ice. Neale (27) has also reported
that the bonding between cellulose and water is comparable
to that of ice.
The work of Baumgarth (9) on ultrasonics surgests
that strong electrolytes migrate in cellophane by hole-
type diffusion. He observed, that the rate of diffusion
of NaC1 in cellophane was increased two-fold by the action
of ultrasonics. The ultrasonic acoustical radiation in-
creased the vibration of the cellulose polymers and this
mil-ht be expected to increase the concentration of holes
in the polymer. Consequently, hole-type diffusion would
Henser (29) makes the point that the hydroscopicity
of cellulose derivatives depends upon the geometrical ar-
rangement of the cellulose molecules. This indicates that
any molecules that penetrated cellulose acetate membrane
by hydrogen bonding would have to be sterically compatible
with the water-cellulose acetate structure.
EXPERIMENTAL SUPPORT OF THEORY BASED UPON
Although there is abundant evidence to substantiate
indirectly the theory presented in the previous section,
little experimental work has yet been reported to support
it directly. Consequently, a series of experiments was
designed to test this concept. These experiments fall
into two broad categories. The first group was based
upon the determination of the semipermeability of cellu-
lose acetate under various conditions. They are reported
in this chapter. The second groun of experiments was
based upon the measurement of the electrical properties
of cellulose acetate under various conditions. These ex-
periments are reported in the following chapter.
The hydrogen bonding concept is compatible with the
results contained in Table 2. To facilitate an explanation,
the semipermeability will be expressed as:
Semlperme-,-., Rate of diffusion for solvent molecules (1)
ability Rate of diffusion for solute molecules
The relationship between the rate of diffusion of
salt molecules and the rates of diffusion of the individual
ions is not clear., However, the following general relation-
ship would be expected. An increase in the diffusion rate
of either one or both ions of a salt would result in an
increase in the diffusion rate of the salt.
In the foregoing expression, perfect semipermeability
corresponds to infinity, whereas, when semipermeability is
expressed in terms of per cent salt rejection, perfect semi-
permeability corresponds to 100 per cent salt rejection.
Inasmuch as neither the cations nor the anions of
NaC1 and NaBr can enter into hydrogen bonding, these salts
should be effectively rejected, and they are. The very high
rejection of ':gC12 and CaC1, may be attributed to the large
size of the hydrated magnesium and calcium ions which would
reduce their rate of diffusion.
Since the fluoride ion can enter into hydrogen
bonding, this ion should have a greater rate of diffusion
than the chloride ion. Sodium fluoride, therefore, should
have a lower semipermeability than NaC1, and this was found
to be true.
A very low salt rejection would be expected for
those solutes that can enter into the water-cellulose acetate
structure through hydrogen bonding. Ammonia is an example
of this type of solute. As indicated in Table 2, the semi-
permeability for NH3 was only 30 per cent--about a third of
that for NaCI.
Apparently the H3BO molecule can enter into the
water-cellulose acetate structure with considerable ease,
because its salt rejection was 46 per cent.
The mechanism presented in Chapter II is also com-
patible with the results presented in Table 1. Polyvinyl
alcohol and cellulose both have large concentrations of
oxygen atoms that permit cross-linking via hydrogen
bonding. However, the degree of crystallinity of these
two polymers is low; consequently, there is much Brownian
movement and the concentration of holes is relatively large.
For this reason these membranes have a low degree of semi-
For this same reason, cellulose acetate butyrate
and ethyl cellulose exhibit no semipermeability. Spence
(30) has shown that because of the plasticizing effect of
the butyrate group, cellulose acetate butyrate has a much
lower decree of crystallinity than cellulose acetate.
Effect of Pressure on Semipermeability
When the semipermeability of cellulose acetate
(expressed as per cent salt rejection) was determined at
various pressures, the relationship depicted in Figure 2
was obtained. An explanation for this -henomenon has al-
ready been suggested. As pressure is applied, the Brownian
movement is restricted and the "pile of straw" is compressed.
P r 0
-0 -4 o
Much more cross-linking by water then becomes possible and
the membrane becomes a more impenetrable barrier to sodium
Effect of Crystallinity on
Spence (30) has also shown that as the acetyl con-
tent of cellulose acetate is increased, the degree of
crystallinity likewise increases. Varying the acetyl con-
tent then becomes a means of varying the crystallinity.
Increasing the degree of crystallinity would have
a two-fold effect according to the hydrogen-bonding mecha-
nism. First, as explained in Chapter II, the semiperme-
ability should increase. Second, the flow rate of water
through the membrane should decrease since the percentage
of amorphous regions (through which the water passes) is
Variations of salt rejection and flow rate with
acetyl content have been determined. They are presented
in Figures 3 and 4 respectively. The procedure used in
preparing the membranes is given in the Appendix.
The role of crystallinity as depicted above is
borne out by the data. Up to an acetyl content of 40 per
cent there was a continual increase in the semipermeability
and a decrease in the penetration of water.
34 36 38 40 2 44
Acetyl content, per cent
Fig. 3 .--F"ffect of Acetyl Content on Flow Rate
34 36 38 40 42 44
Acetyl content, per cent
Fig. 3I..-Eff.ect of Acetyl Content on Flow Rate
ELECTRICAL MEASUREMENTS TO
In the previous chapter semipermeability-type ex-
periments were described which support the hydroren bonding
concept of semipermeability. However, these experiments
support only indirectly the crux of the theory--namely
that hydrogen bonding sets up an ice-like structure im-
penetrable to strong electrolytes. More direct evidence
was needed--first, to establish the existence of the bound
water regions and second, to show that two different types
of diffusion are involved in the transfer of water and
sodium chloride ions across cellulose acetate.
Dielectric Measurements of
Evidence for the existence of the water-cellulose
acetate structure could be obtained by showing that com-
pression of the membrane which sets up the structure and
increases the amount of bound water is accompanied by an
increase in hydrogen bonding. In order to detect an in-
crease of hydrogen bonding, some observable property which
depended upon hydrogen bonds was needed. The dielectric
constant meets this requirement. It has been established
that hydrogen bonding markedly increases the dielectric
constant of some liquids. Pauling (31) has explained this
effect on the basis that hydrogen bonding increases the
dipole moment which in turn increases the dielectric con-
Latimer (32) on the other hand explains this effect
on the basis of atomic polarization. He presents arguments
to show that the polarization of hydrpgen atoms involved in
hydrogen bonding causes a large contribution to the dielec-
tric constant. Furthermore, using Raman spectra, he shows
that the largest contribution to the dielectric constant is
made when the axis of the hydrogen bond is parallel to the
If Latimer's concept is correct, the dielectric
constant of hydrated cellulose acetate should increase with
pressure since the amount of bound water is also increasing.
If no additional hydrogen bonding were occurring when cellu-
lose acetate is compressed, the dielectric constant of hy-
drated cellulose acetate should remain constant.
The relationship between the increase in dielectric
constant and pressure for cellulose acetate is given in
Figure 5. Experimental details are included in the Appen-
dix. As shown, there was a decided increase in dielectric
constant with pressure. This large increase in dielectric
0 04 0
I o i
*H \ C O
I i I _
(amssoad oyxtqdsouq) 33 (eanssead ass osq-9) T9 V
;u3q\uo 01.1Q10TT UT es|veaouI
constant is due not only to the formation of hydrogen bonds,
but also to the large probability that they are formed in a
direction parallel to the electric field intensity.
It is felt that these results furnish direct evi-
dence of the existence of bound water regions in cellulose
Electrical Resistance Across
Hole-type and alignment-type diffusion in cellulose
acetate.--Electrical resistance measurements were used to
demonstrate that two different types of diffusion are occur-
ring in cellulose acetate. In electrolytic conduction the
current is carried by anions and cations in solution. Both
electrical resistance and rate of diffusion depend upon the
mechanical resistance that these ions encounter during mi-
gration. Therefore the electrical resistance across a mem-
brane can serve as a quantitative index of their diffusion
rates. Diffusion rate and electrical resistance, of course,
vary inversely with each other. In subsequent sections the
term membrane resistance refers to the electrical resistance
across the membrane.
The membrane resistance of a specific ion can be
determined by placing an appropriate ion exchange membrane
on each side of the semipermeable membrane. The procedure
is given in the Appendix. The membrane resistance of water
molecules was approximated by determining the resistance of
hydronium ions (HO ).
The electrical resistances for Na+, Cl-, NaC1, and
H0,O across cellulose acetate as a function of pressure is
plotted in Figure 6. At atmospheric pressure the resistances
of Na+, Cl-, and NaC1 are much higher than the resistance of
HO+ Further, as the membrane is compressed, the resistance
of H0+ remains relatively constant whereas the resistances
of Na+, Cl-, and NaCl rapidly increase.
Considered in terms of diffusion, this indicates
that at atmospheric pressure the diffusion rates of Na+ and
Cl" which cannot enter into hydrogen bonding were very low
and became even lower when the membrane was compressed. In
contrast, the diffusion rate of H0+ which can enter into
hydrogen bonding was relatively high at atmospheric pressure
and was only slightly reduced when the membrane was com-
It is considered that these results furnish strong
direct evidence to show that those ions and molecules that
cannot combine with cellulose acetate diffuse by a different
mechanism than those ions and molecules that can combine with
the membrane through hydrogen bonding.
Relationship between membrane resistance and semi-
permeability.--In Chapter ITT a qualitative expression was
aiven relating to semipermeability, the diffusion rate
Fig. 6 .--Resistance versus pressure
a 1 + +
0 400 800 1200 160
Fig. 6 .--Resistance versus pressure
g ^ /
across l^Of' cellulose acetate membrane.
of solvent molecules, and the diffusion rate of solute
molecules. Bearing in mind the inverse relationship be-
tween diffusion rate and membrane resistance, this expres-
sion may be written as:
Sem i perme- vMembrane resistance for solute molecules (2)
ability Membrane resistance for solvent molecules
That semipermeability does vary with the resistance
of the solute is borne out by a comparison of Figures 2 and
6. The rate of increase of semipermeability with pressure
siven in Figure 2 corresponds to the rate of increase in
resistance of the solute given in Figure 6. It can also be
deduced from Figure 6 that in the case of an ionized solute,
the semipermeability depends largely upon the ion with the
higher membrane resistance. Note that the increase in
semipermeability with pressure corresponds more closely to
the increase in Cl- resistance with pressure than with that
of either Na+ or NaCl.
To provide further evidence that semipermeability is
related to membrane resistance, the resistances of membranes
with high and low semipermeabilities were determined. It
can be seen from Figure 7, in which these runs are com-
pared, that the membrane with a high semipermeability also
had high electrical resistance, and the membrane with a
low semipermeability had a low resistance that did not in-
crease with pressure.
Over c00e salt rejection
S 0 o00 800 1200 1600
Pressure on membrane, p.s.i.
Fir. 7 .--Resistance versus pressure for
sodium chloride across cellulose acetate mem-
branes(du Pont CA-43) of known salt rejections.
Effect of bound water on hole-type diffusion.--In a
previous section dielectric constants were used to show
that compression of cellulose acetate sets up regions of
bound water. In the two preceding sections membrane re-
sistances were used to show that compression of cellulose
acetate causes the rate of hole-type diffusion to decrease.
A comparison of dielectric constants and membrane resistances
as a function of pressure illustrates that the formation of
bound water regions restricts hole-type diffusion. Such a
comparison is made in Figure 8. The lower curve in Figure
8 represents the increase in hydrogen bonding with pressure.
The close resemblance of this lower curve to the resistance
curve of C1- indicates very strongly that the hydrogen-
bound water-cellulose acetate structure is responsible for
the high membrane resistance or low diffusion rate of strong
The shift of the resistance versus pressure curve
to the left was probably the result of a hysteresis effect
as the pressure was reduced.
Semipermeability of cellophane.--It was mentioned
in Chapter III that the low degree of semiermeability of
cellophane is attributable to its low degree of crystal-
linity. A certain amount of energy is required to overcome
the Brownian motion of the cellulose polymers and to com-
press the system so that a cross-linked water-cellulose
8 Resistance of Cl
18- a a
o 16- -
I I-- ---
4 6 */- P
0 2000 00
S 4/ i C
T 12- -3 40J40
Pressure on membrane, p.s.i.
Fir. 8.--Comparison of semipermeability and
dielectric constant of cellulose acetate membranes.
4J 2 g7
0 2000 4000 6000
Pressure on membrane, p.s.i.
Fir. 8.--Comparlson of semlpermeability and
dielectric constant of cellulose acetate membranes.
structure can be set up. Crystallinity supplies energy re-
quired for the formation of the desired structure. Cellu-
lose acetate has a high degree of crystallinity, so only
little additional pressure is needed to set up the bound
water structure. With cellophane, however, much additional
pressure is needed to accomplish the same result.
To learn whether or not the semipermeability of
cellophane increases at higher pressures, the membrane re-
sistances of H+, Na Cl-, and NaC1 were determined up to
500 p.s.i. These results are presented in Figure 0. The
membrane resistance of H.O+ remained relatively constant.
The resistance of Cl~, however, increased rapidly. There-
fore the semipermeability of cellophane should increase
The semipermeability of cellophane was determined
up to 1500 p.s.i. The salt rejection of cellophane did
increase from about 8 per cent at 600 p.s.i. up to 14 per
cent at 1500 p.s.i. (33).
Effect of temperature.--Another type of experiment
was conducted to determine if hydrogen bonding was connected
with membrane resistance. It is recognized that hydrogen
bonds are very sensitive to temperature. At lower tempera-
tures the hvdroen bonding is greatly Increasei. For
example, at hC. water is more polymerized than at room
"/ J i..
o 2+ 4
Fig. 9.--Resistance versus pressure for cello-
phane (du Pont PT-300).
phane (du Pont PT-300).
Consequently, increasing the temperature should
decrease the semipermeability by breaking up bound water
regions and, conversely, decreasing the temperature should
increase the semipermeability. The former is illustrated
in Figure 10. The temperature was elevated by increasing
the passage of current through the membrane. By intro-
ducing four times as much heat into the membrane, the
change in electrical resistance between 0 and 60n p.s.i.
was decreased from 3600 ohms to 350 ohms.
To show the effect of decreasing the temperature,
resistance versus pressure determinations for cellophane
are given in Figure 11. The temperature was lowered by
packing the entire resistance cell in ice. Lowering the
temperature had two effects. The overall resistance was
increased and the rate of change of resistance with pressure
was also increased. This effect of temperature on membrane
resistance shows in another way that hydrogen bonding sets
up regions that restrict the diffusion of strong electrolytes.
Effect of ions in the water-cellulose acetate struc-
ture.--In the course of determining membrane resistances, an
unexpected phenomenon was observed. This is illustrated in
Figure 12. A cellulose acetate membrane was soaked in water,
placed in the resistance cell, and compressed to 7500 P.s.I.
After compression, the current was turned on and resistance
measurements were made periodically as the pressure was
2.8 ~ 12 x 10 calories developed per
a minute due to passage of current
50 x 10" calories developed
20 er minute due to passage of
0 400 800 1200 1600
Pressure on membrane, p.s.i.
Fig. 10.--Resistance versus pressure for
cellulose acetate membrane (du Pont 88 CA-f3)
with a known salt rejection of over 00'.
Cell packed in ice
2000 40C0 6000 8000
Pressure on membrane, p.s.,.
Fi. 11.--Zffect of temperature on membrane re-
sistance of cellophane (du Pont PT-300).
Cell at room temperature
o 18- CI
0 2000 4000 6000 8000
Pressure on membrane, p.s.i.
Pig. 12 .--Resistance versus pressure for
40% cellulose acetate film. No previous
lowered to 0 p.s.i. The upper curve was obtained. But
when the resistance measurements were made as the pressure
was increased, the lower curve was obtained.
It is felt that this effect may be caused by sodium
ions lodged in the membrane. To allow chloride ions to be
electrically forced through the membrane, electrical neu-
trality within the membrane must be maintained. This is
accomplished by the leakage of sodium ions through the
high resistance anion exchange membrane into the cellulose
acetate membrane. Once inside the cellulose acetate membrane,
the sodium ions are relatively immobile because of the very
high cation resistance of the anion exchange membranes. It
is considered likely that the presence of these sodium ions
disrupts the formation of bound water regions.
Now consider Fi1ure 12 in the light of this explana-
tion. When the membrane was first compressed to 7500 p.s.l.
no sodium ions were present so the bound water resIons
readily formed. The membrane resistance was then able to
reach 96,000 ohms. On the other hand, starting resistance
measurements at atmospheric pressure introduced large num-
bers of sodium ions into the structure before the bound
water regions were completely developed. These sodium ions
then disrupted the continued growth of the bound water
regions when the pressure was Increased. Consequently at
7500 p.s.i, the membrane resistance had increased to only
It appears that the water-cellulose acetate struc-
ture can be broken down at sufficient pressures. Referring
again to Pigure 12, it can be seen that the membrane re-
sistance begins to decrease at pressures above 4000 p.s.i.
In Figure 13 the effect is more pronounced. A du Pont CA-
43 cellulose acetate membrane was used. The breaks in the
Na+ and Cl- curves at 3000 p.s.i, are attributed to the
weakening of the water-cellulose acetate structure by
sodium and chloride ions. Beyond 3000 p.s.1. the weakened
structure begins to fail. On the other hand, the membrane
resistance of NaC1, which was determined last, continued to
increase up to 5500 p.s.i.
The difference between the resistance curves of
NaCl and its ions might be explained in the following way:
In the determination of the membrane resistance of NaC1,
both ions are free to migrate out of the membrane into the
solution. Since both ions are mobile, the formation of
the water-cellulose acetate structure would not be inhibited
or weakened. In the measurement of the resistances of C1l
or Na+, either Na+ or Cl" is immobilized within the membrane.
These immobile ions may then weaken the water-cellulose
It should be mentioned that the shapes of the re-
sistance versus pressure curves for du Pont cellulose ace-
tate varied widely from sample to sample. This difference
S300 2000 4000 6000
Pressure on membrane, p.s.I.
'ir. 13.--F.eslstance versus pressure across
cellulose acetate (du Pont CA-43).
between the resistance curves of NaC1 and its ions was in
general repeated. The erratic behavior of the commercial
films could be caused by the presence of plasticizers.
Good reproducibility was obtained with the cellulose
acetate membranes prepared in this laboratory--no plasti-
cizers were used.
Effect of pressure on transference numbers.--Origi-
nally the intention was to test the proposed mechanism of
semipermeability by determining the transference numbers
of various ions across cellulose acetate. Transference
numbers at elevated pressures were needed. The apDaratus
that was used for measuring membrane resistances was origi-
nally designed for measuring transference numbers. Employ-
ing the Hittorf method, the transference numbers for Na+ and
Cl1 across cellulose acetate were determined at pressures
up to 4000 p.s.i. No reproducibility or correlation of
results could be obtained. The difficulty was traced to
the heating effect of the current passing through the mem-
brane. To reduce the current sufficiently to eliminate
temperature effects would have made the experiment prohibi-
tively lonr. It was then decided to employ membrane re-
sistances for testing the theory.
It is still of interest to observe what effect
pressure has on the transference numbers of Na+ and Cl"
in cellophane and cellulose acetate. Basically, trans-
ference numbers depend upon the relative electrical
resistances of the ions in the system. For example, in an
aqueous sodium chloride solution, the electrical resistance
for sodium ions is greater than that of chloride ions.
When the solution is conducting electricity, the migration
of sodium ions accounts for only 38.5 per cent of the cur-
rent. The transference number of Na+ (tNa+) is then 0.385
and that of Cl" is 0.615. If the resistance for chloride
ions increased, its transference number would decrease.
Thus it seems that the resistance-pressure rela-
tionships which have been described previously should serve
as a basis for predicting (qualitatively) the variation of
transference numbers with pressure.
From these resistance versus pressure measurements
it appears that transference numbers within cellophane and
cellulose acetate are dependent among other things upon the
water-cellulose structure. To illustrate, in Figure 9 at
6000 p.s.i., the resistance of Cl- is very high compared to
Na+; therefore, the transference number of Cl" (tC1-) across
cellophane at this pressure is very much smaller than tNa+.
As the pressure on the membrane is decreased below 1000
p.s.i., the resistance of Cl- decreases very rapidly which
in turn results in a sharp increase in tC1-. Neale (34)
has found that at atmospheric pressure the transference
number of Cl- across cellophane is of the order of 0.4 in a
C.1 N NaC1 solution.
It might be concluded, then, that the formation of
the water-cellulose structure causes the transference num-
ber of Na to increase. The hydrated Na must be able to
fit into the water-cellulose structure with greater ease
than Cl can.
The transference number of sodium ions across
cellulose acetate that had not previously been compressed
is less than the transference number of chloride ions.
Note from Figure 14 that for this membrane the resistance
to Na is greater than that to Cl1. Yet after measuring
the membrane resistance for H0O+ up to 1600 p.s.i, the
situation is reversed. The transference number of Na+ is
now greater than that of Cl-. Referring back to Figure 6,
the resistance of Na is seen to be appreciably lower than
that of Cl". The effect of hydrogen ions is to set up the
water-cellulose acetate structure. This increase in tNa+
caused by the passage of H30+ through the membrane indicates
again that sodium ions can preferentially diffuse through
the water-cellulose-type structure.
The significance of this work is two-fold. First,
it shows that the semipermeability of cellulose acetate is
not caused by an ion exchange mechanism. Had the semiperme-
ability been due to an ion exchange reaction, there would
have been a very large difference in resistance between Na+
and C"-. Second, it illustrates the dependence of trans-
ference numbers on the internal structure of a membrane.
0 400 800 1200 1600
Pressure on membrane, p.s.,i
Fir. 14.--Resistance versus pressure across 60f
cellulose acetcte. Resistances of l'a and Cl" were deter-
mined across separate unused membranes.
MECHANISM FOR LOSS OF MEMBRANE
In Chapter I mention was made of the fact that
cellulose acetate lost its semipermeability after a
limited period of time. On the basis of the bound water
theory of semipermeability and the results of a series of
experiments, a mechanism for membrane failure is proposed.
The experiments that are first described illustrate
the nature of the film failure. The next set of experiments
shows that chemical reactions between the solute and mem-
brane are involved. And the final group of experiments
shows the effects of these chemical reactions on the
crystallinity and polymer orientation in the membrane.
Loss of Semipermeability
In order to study the nature of film failure, cellu-
lose acetate films containing 37 per cent combined acetyl
content were employed. By usinr cellulose acetate films
with a low acetyl content, the length of the durability
tests was reduced. The concentration of all solutes was
approximately 0.08 U NaCl. The apruratus and procedure
described in Chapter I were used.
Plots of semipermeability versus time and volume
of water passing through the membrane are presented in
Figures 15 and 16 respectively. In Figure 17 the relation-
ship between time and the rate of diffusion of water through
the membrane is given. Two methods of film preparation were
employed. Both membranes were cast from the same acetone
solution, and the solvent was allowed to evaporate at room
temperature. In one case, however, the remaining traces of
solvent were removed by forcing water through the film and
in the other this was accomplished by heating. These two
methods of film preparation are compared in Figures 15, 16,
Several observations can be made from the data
illustrated in these figures. First, the results indicate
that not all of the amorphous regions are available for
supporting a network of bound water. This is shown in
Figure 17. Removal of the solvent by water extraction
increased the diffusion rate of water seven-fold over
that of the heat-dried film.
This difference in diffusion rates mliht be ex-
plained in the following way: When the solvent is removed
by heating, the cellulose acetate polymers lying in non-
crystalline regions apparently become bound together by
intermolecular forces which prevent them from being used
for maintaining oriented water; whereas, when the solvent
O 0 0
0 CO 0
@0 0 0
t> -0 0
> 4 EH 0
0 > P 0
H 0 O
0 07 (D
o 0 0
-3 C\ H
0 ~0 rHj$:
o 0 0
r:1 Or-I C0
r-4 0O 4)41
0 Sb 4
0 0 r-4
43 > 'o 1f
0 W 43\ 00
EO 43 0
~~r 'U Trfl!IV
0 0 go
O .4 rt)0
H0 o 00
0 0 0 0
S3 a) C
\ 3a peas
> 0 Q
0 0 0 0
o I Cd "(D o 0
-0 0 4) 4)
0 *I 0
*8 cu a
0 4 0
0 0 0 0 0 0
is removed by water extraction, water has an opportunity to
occupy a much greater proportion of the amorphous reglons.
Some of the decrease in flow rate caused by heating could
be attributed to an increase in crystallinity; however, it
is felt that obstruction of the amorphous regions is the
primary cause for this loss in flow rate. Heating, as Mark
has pointed out (35), causes irreversible packing which
could block off some of the amorphous regions within the
Second, the water-cellulose structure apparently
persists until long after the solute break-through has
occurred. A break in the diffusion rate of water repre-
sents the beginning of extensive swelling of the structure.
For the heat-treated membrane, the salt break-through oc-
curred at 52 hours (Figure 15). Yet the break in diffusion
rate of water for this same membrane occurred at 110 hours
(Figure 17). The lag between solute and solvent break-
through has been observed in varying degrees. It was more
pronounced with heat-treated films.
Third, intermolecular forces stabilize water-
cellulose structure. It is to be expected that the film
prepared by removing the solvent through heating should be
more resistant to failure since additional intermolecular
forces between the polymers lying in the amorphous regions
are set up. That this is a factor is indicated by the
longer life of this film. Note from Figure 17 that the
heat-treated film lasted 100 hours before a break in the
flow rate, as compared to fifty hours for the water-
extracted film. Also, from Figure 15 it is seen that the
stabilized film tended to maintain salt rejection for a
little longer period of time. The two films reached 80
per cent salt rejection in approximately the same time.
The film containing the greater amount of bound water,
however, treated seven times as much water because the
flow rate was seven times higher.
Fourth, film failure seems to be associated with a
structural chance rather than the development of cracks.
From Figure 17 it is seen that after film failure, the rate
of diffusion of water climbs up to a limiting value and
levels off. This suggests that the completely failed film
has a definite structure. If large breaks or cracks de-
veloped in the membrane, the flow rate would be expected
to climb continuously with time. To investigate this
point further, the semipermeability of completely failed
films to fluorescein was checked. The fluorescein molecules
were effectively blocked by the membrane indicating that the
pores in the failed film were smaller than these molecules.
Reactions Between Solute and Membrane
Based upon the work that has been done on swelling
agents for cellulose, it appears that swelling is a factor
in the failure of cellulose acetate to act as a semiperme-
able membrane. Neale (17) has developed a theory for the
swelling of cellulose. He has shown that the hydroxyl
groups of cellulose may be regarded as weak monobasic acids.
These active hydrogens can then enter into exchange reac-
tions with cations such as lithium and sodium. Accompanying
the exchange reaction is a swelling effect.
When cellulose acetate undergoes swelling, the
probability of hole formation increases so the diffusion
rate of solute ions likewise increases. This relationship
between swelling and hole-type diffusion of strong elec-
trolytes is very important. Nicoll, Cox, and Conaway say
in High Polymers (17) that swelling is absolutely necessary
before alkali can penetrate into cellulose, and presumably
this applies to other electrolytes.
Effect of acetylation.--The swelling of cellulose
acetate depends upon the concentration of hydroxyl groups.
As the degree of acetylation is increased, the number of
active hydrogen atoms is reduced. Accordingly, membrane
durability should be dependent upon the degree of acetyla-
In Figure 18 the durability of a film of 37 per cent
acetyl content is compared to that of a film of 43 per cent
acetyl content. Probably several factors contribute to the
longer durability of cellulose triacetate (43 per cent
- 0 I
u' 'uoT^oae ^-[IS
acetyl content). One would be that cellulose triacetate is
more highly cross-linked by crystalline regions which would
make it more resistant to swelling. Another reason is that
the triacetate has fewer active hydrogens, so exchange
(hence swelling) reactions with sodium ions would be limited,
thereby permitting a longer period of semipermeability.
Effect of solute on durability.--If swelling is a
factor in film failure, the life of the film should depend
upon the type of solute in contact with it; i.e. solutes
that have a high reactivity with the hydrogen should shorten
film life. A series of tests was made to determine the
effect of various solutes on the durability of the film.
Avain, films of 37 per cent combined acetyl content were
used. The concentration of the solute in all runs was
approximately 0.08 N.
In Figure 19 the durability of cellulose acetate
in various solutions is illustrated. When dextrose was the
solute, the film showed no sipns of failure even after 300
hours of operation. When MgC1, was the solute, the film
also lasted 300 hours with no appreciable drop in its semi-
permeability. Neither of these solutes would be expected to
react with the acidic hydrogens in cellulose acetate. On
the other hand, when NaC1 was used as a solute, the film
began to fail at 40 hours. Swelling, therefore, seems to
be a factor in film failure, since the ions which cause
80 u0 0 0 0\ 0 0
rapid failure are swelling agents.
To obtain additional evidence to determine if the
solute causes film failure, a run was made using pure
water. After nine months of continual passage of water
at 600 p.s.i., a membrane containing 37 per cent acetyl
content exhibited no signs of failure. That is, there was
no break or increase in the diffusion rate of water through
If the reactivity of the acidic hydrogen atoms were
reduced, then the life of the film should be extended. This
can be done by lowering the pH. If the pH is lowered too
far, however, the ester groups within the membrane may be
hydrolyzed. As a compromise, a pH of 4 was selected. Hy-
drochloric acid was used to lower the pH. From the com-
parison of runs at high and low pH's included in Figure
19, it appears that the presence of hydrogen ions did
retard the NaCl break-throuph. On the other hand, the
rejection of chloride ions tended to level off at 60 per
cent at the lower pH. This could be caused by the higher
rate of diffusion of HC1.
A comparison of flow rates illustrates more clearly
the effect of reducing the reactivity of the active hydro-
gen. Figure 20 gives a comparison of the flow rates through
cellulose acetate at pH's of 4.0 and 5.5. The flow rate at
the lower pH remained essentially constant up to 180 hours,
I p I.o
1 I I
Fil. 20.--.ffect of pH on durability
of 37.2r' cellulose acetate.
indicating no extensive change in the internal structures
up to this time. Yet the NaC1 solution at the higher pH
caused a break in the flow rate after 87 hours. This dif-
ference, coupled with the fact that the flow rate at pH 4
was appreciably lower than at pH 5, indicates that the
hydronium ions stabilized the bound water regions, possibly
by reducing the reactivity of the active hydrogens in the
Mechanism of Membrane Failure
Although swelling is associated with membrane
failure, hydrolysis is unquestionably responsible for
extensive break-up of the water-cellulose acetate struc-
ture. Using infrared snectra, Kuppers (36) has shown that
accompanying the decrease in semipermeability of cellulose
acetate is an increase in the concentration of hydroxyl
groups. The increase in hydroxyl groups is at the expense
of carbonyl groups indicating that extensive hydrolysis is
The relationship between swelling and hydrolysis
has not been established. It is assumed that swelling
precedes hydrolysis. Initially, swelling results in the
penetration of the solute into the membrane. The solute
in turn causes hydrolysis. Since the product of the hv-
drolysis of cellulose acetate is more free hydroxyl groups,
Neale-type swelling continues. Concurrently, swelling and
hydrolysis probably continue until the film is converted
The structure of the failed film must be different
from other cellulose membranes since the diffusion rate of
water through the failed films was very much lower. Further-
more, the failed films were mechanically weak and difficult
The lac between the solute break-through and the
solvent break-through described earlier can possibly be
explained as follows: Solutes such as NaCl penetrate
cellulose acetate membranes by hole-type diffusion, whereas
water diffuses by alignment. When swelling and hydrolysis
commence, the probability of hole formation increases which
increases the diffusion rate of NaC1. Yet the original
structure is not altered enough to change the rate of
diffusion of H20. As hydrolysis proceeds, the particular
crystalline structure, which stabilizes the hydrogen-bound
water structure, is continuously being altered. Even
though the structure is continuously expanding, the bound
water regions continue to fill up the pores within the
membrane. A point is finally reached, however, at which
water within the pores ceases to be extensively associated
with the pore walls. Then the entropy of activation for
diffusion of water is greatly reduced which causes the rate
of diffusion to increase.
The structural arrangement of polymers has an
important bearing on the semipermeability of cellulose
acetate. Cellophane, it will be recalled, had a low de-
gree of semipermeability because the internal structure
was too open or expanded. It seemed of interest, therefore,
to establish a correlation between the changes in internal
structure of cellulose acetate and film failure. To do
this work X-ray diffraction and optical studies were used.
X-ray studies.--X-ray diffraction spectra were made
of two du Pont 88 CA-43 membranes. One had a salt rejec-
tion of ol,7 per cent and the other a salt rejection of
c.h per cent. No bands appeared for the film that had
94.7 per cent salt rejection. A broad band appeared at
approximately 4 A for the failed film. This latter X-
ray pattern was duplicated by using a failed film from an
The band which appeared at 4.4 A for the failed
film ties in very well with the existing data and concepts.
In connection with mercerization, a great deal of work has
been done on lattice extension due to swelling. In High
Polymers (17) the following concept is presented: Two
types of swelling are possible--inter- and intra-micellular
swelling. Intra-micellular swelling is caused by sodium
hydroxide, hydrazine, etc. attacking the crystalline regions
of the polymer. Inter-micellular swelling is brought about
by swelling between the micelles--the crystalline regions
remaining unaffected. In the case of native cellulose,
intramicellular swelling occurs in a direction normal to
the 101 plane of the unit cell. The 101 planes are held
together by hydrogen bonding. Looking down along the "b"
axis with the cellulose chains coming out of the page, the
unit cell of native cellulose is pictured in Figure 21.
When a swelling agent enters the unit cell, the hydrogen
bonds between the 101 planes are broken which causes an
increase in the 101 spacings. As shown in Figure 22, in-
creasing the 101 spacing flattens out the parallelogram.
This flattening effect decreases the 101 spacings to 1;.4 A.
Depending upon the swelling acent, the 101 spacings may
increase from 6 up to 16 A. For NaOH it is 7.4 A.
This mechanism for the swelling of native cellulose
appears to apnly to the swelling of cellulose acetate mem-
branes. The band which appeared at 4.4 A for the failed
cellulose acetate corresponds to spacinps between the 101
planes in mercerized cellulose (Figure 22).
The fact that only the failed film diffracted the
X-rays surgests that during failure the polymeric chains
oriented into a plane parallel to the surface of the film.
\. \ /
?ig. 21.--Unit Cell of Native Cellulose (17)
-S u r fa c e
-- #- K' r y
O- Fi l/ -" -"-
Fig. 22.--Unit Cell of Swelled Cellulose (17)
This rotated the 101 planes into diffracting position. In
Figure 22 the angle that the X-rays made with the surface
of the film is inserted to show why only the 101 plane
Microscopic studies.--In order to study the effects
of polymer orientation, the same two films used in the X-
ray studies were also studied under the polarizing micro-
scope. The polarizing microscope is extensively employed
in the field of mineralogy for determining the degree of
crystallinity of minerals. Drechsel (37), Spence (30),
McNally (38), and others have employed the technique for
studying the crystallinity of cellulose and its derivatives.
The polarizing microscope permits one to obtain conveniently
a quantitative index of the degree of crystallinity in a
membrane and to determine the direction of orientation of
When a beam of polarized light is passed through
most crystalline substances, it will be broken into two
parts--a fast ray and a slow ray. The difference in ve-
locity between these two rays (which is called the bire-
frinrence) is governed by the degree of crystallinity in
polymeric materials. The greater the degree of crystal-
linity, the greater the birefringence. The direction of
vibration of the slow ray in cellulose acetate corresponds
to the orientation of the polymer chains (37).
A Leitz Berek compensator was used with an American
Optical polarizing microscope to measure the birefringence.
Actually, the Berek compensator measures the retardation.
That is, it measures the amount the slow ray is retarded
behind the fast ray. This retardation produces interference
colors which appear in the microscope. The higher the order
of the color, the higher the retardation and, hence, the
higher the crystallinity. The retardation is used to cal-
culate the birefringence with the following formula (39):
n' n" = -
n' n" = difference in refractive index between
fast and slow rays called birefringence
D = retardation
d = thickness of sample in mx
First the birefringence was determined normal to
the surface of the film. The results are recorded in Table
These data indicate that both compression of the
membrane and hydrolysis cause orientation of the polymers.
Probably the number of unit cells was not increased, but
rather those that were present were changed from a random
orientation to an ordered arrangement.
The orientation in the film as received from the
EFFECT CF SEMIPERMEABILITY ON BIRFFPINGFNCE
88-CA-h3 dry As received (unused) P.00019
88-CA-h3 wet C..7 o. 0.0046
88-CA-i3 dry 9h. 7 0. 0C046
88-CA-43 wet 5.4 0.0022
88-CA-43 dry 5.h 0.0030
manufacturer was in the direction in which the film had been
An edge view of the film was obtained by cutting
strips of the films 108 microns wide with two razor blades
clamped together. By bending the strips, they could be
made to stand on edge on the microscope stage. We were
unable to measure the retardation with the Berek compen-
sator because of interference from light which did not pass
through the film. Attempts were made to block out the ex-
traneous light, but none were successful. However, the re-
tardation was estimated on the basis of the interference
colors. Ropers (40) gives a color chart that is very useful
for this purpose.
In Figures 23 and 24 the interference colors that
appeared for the high semipermeability and low semipermeabili-
ty cellulose acetate films are given. A magnification of
P Slow ray
S Corresponds to a birefringence of 0.0007
Fit. 23.--Birofringence of a du Font CA-43
membrane that had a high salt rejection (over 0,).
Sb~ Slow ray
Corresponds to a birefringence of 0.0074
Corresponds to a birefrinrence of 0.0060
Pij. 24.--Birefrinsence of a du Pont CA-43 membrane
that had failed (5.-' salt rejection).
X 860 was used. These observations indicate that film
failure is associated with extensive orientation of the
cellulose polymers into the plane of the membrane.
In view of previous work it is quite reasonable to
expect this orientation. McNally (38) and Sissons (41)
have shown that polymers will orient themselves to relieve
stresses. In the case of cellulose acetate when the films
bezin to swell and hydrolyze, the chains become free to
rotate. They then rotate into the plane of the film to
relieve the stresses.
Based on the interference colors obtained from an
edge view of a failed film, it appears that there is a
slightly greater degree of orientation at the surfaces of
the film than in the central regions. In the interior
there may not be as much freedom of rotation. Apr-arently
the orientation occurs at approximately the same rate
throughout the film rather than starting on the salt water
side and working its way through.
RECOMMENDATIONS FOR AN IMPROVED
In the last analysis, the use of semipermeable
membranes for purifying sea water will defend upon the
economics of the process. However, as a first step a
semipermeable membrane with a very high flow rate will
be needed. On the basis of the investigation described
in this dissertation, it now is possible to propose some
requirements for such a membrane.
According to this present study, the semiperme-
ability of cellulose acetate is caused by regions of bound
water within the membrane. The polymer network serves as
a skeleton for setting up the bound water regions. These
regions, it will be recalled, permit relatively rapid
diffusion of water molecules while greatly retarding
hole-type diffusion of sodium chloride. The problem, then,
becomes one of designing a skeleton that will maintain a
large amount of bound water.
To attain this objective, it apneers that several
factors in the design of the polymer.network will need to
be considered. First, the polymers that are used should
contain large numbers of sites for hydrogen bonding.
Second, sufficient cross-linking is needed to re-
strict the Brownian motion of the polymers that make up
the network. It is postulated that the bound water regions
can grow only when the kinetic energy of the polymers and
water molecules are reduced enough to permit extensive
association of the water with the polymer network. This
beinp the case, the optimum degree of cross-l1nkinr would
be related to the variables that affected the Brownian
movement of the system. Pressure was found to restrict
Brownian motion. Consequently less cross-linking would
be needed at high pressures, say 3000 p.s.i., than at low
pressures. Temperature is another variable that affects
Brownian motion. Hence, if saline water is being purified
at, say, 50C., less cross-linking is required than at
Third, the cells or channels within the polymer
skeleton should be of optimum size for setting up bound
water regions. The size is dependent upon how far a chain
of hydrogen-bound water molecules can span without rupturing.
At low temperatures--around the freezing point of water--
hydrogen bonds are strong; so, relatively larpe cells can
be used without increasing the probability of hole formation.
Fourth, the polymers should be cross-linked suf-
ficiently to render them insoluble in water. In all
probability the membrane will be insolubilized when it is
cross-linked enough to restrict Brownian motion.
Fifth, the polymer used for constructing the skele-
ton should be inert to chemical attack by sodium chloride
or any other solute in the system.
Cellulose meets in part the requisites for a semi-
permeable membrane--it has sites for hydrogen bonding and
sufficient cross-linking through crystallinity to set up
bound water in its pores. However, only a small effective
volume of the film is available for bound water. The rest
is taken up with crystalline regions. It seems that the
effective volume for bound water could be increased by
employing cross-linking agents smaller than the crystal-
line regions of cellulose acetate.
Cellulose acetate has the further limitation of
being subject to hydrolysis. Therefore, it is probable
that ultimately some other type of semipermeable membrane
will be developed for purifying saline water.
The primary objective of this investigation has
been to explain why cellulose acetate behaves as a semi-
permeable membrane in saline water. To explain this
phenomenon, two different mechanisms for the transfer of
water and ions through cellulose acetate membranes were
Those ions and molecules that cannot enter into
hydrogen bonding with the membrane are transferred by
hole-type diffusion. The rate of diffusion appears to be
governed by a water-cellulose acetate structure. The re-
action between water and the cellulose acetate polymers to
form bound water regions is induced by compressing the mem-
brane. As pressure is applied on the membrane, more bound
water is produced which causes the rate of hole-type dif-
fusion to decrease.
On the other hand, those ions and molecules that can
associate with the membrane through hydrogen bonding actually
combine with the membrane and are transported through it by
alignment-type diffusion. The formation of the water-
cellulose acetate structure does not appreciably diminish
the diffusion rate of water through the membrane.
Cellulose acetate begins to behave as an effective
semipermeable membrane in saline water when it is compressed
sufficiently to retard proatly the diffusion of NaCI.
Several types of experiments were conducted to sup-
port these hypotheses. The most important evidence was ob-
tained from resistance experiments. The resistance of
specific ions was measured across cellulose acetate at
various pressures. It was observed that the rate of dif-
fusion of those ions that cannot combine with the membrane
actually does decrease as the membrane is compressed. The
rate of diffusion of H.O+ that can enter into hydrogen
bonding with cellulose acetate is much hi-her and is not
appreciably reduced as the membrane is compressed. These
resistance-pressure relationships are correlated with the
semipermeability of the cellulose acetate.
On the basis of resistance-pressure relationships
for cellophane, it appears that at sufficient pressures
this membrane also should become semipermeable in strong
A secondary objective of this investigation has
been to explain why cellulose acetate ceases to behave as
a semipermeable membrane after a limited period of use.
It was found that film failure is due to a chemical
reaction between the solute and cellulose acetate. X-ray
diffraction and birefringence measurements were employed
to show that extensive structural chances are brought about
by this reaction between the solute and the membrane.
On the basis of this investigation it is considered
that an improved semipermeable membrane can be obtained
first, by using a polymer that is inert to chemical attack
by sodium chloride, and, second, by increasing the amount
of bound water regions within the membrane.
Preparation of Films
Determination of Dielectric
Determination of Membrane
Mohr titration for chloride ion
Mohr titration for chloride ion
'ohr titration for chloride ion
Mohr titration for bromide ion
Mohr titration for halides
Titration with standard HC1
Back titrated sulfate with barium chloride
using a versenate indicator
Complexed HBBO3 with mannitol and made an
acid base titration with KOH
During the early phase of this study, it was suprested
that the reduction in chlorides was due to their adsorption
rather than to the semipermeability of the membrane. To see
if this were true the per cent salt rejections were calcu-
lated on the basis of conductivity measurements. For the
most part the semloermeabilities calculated from conduc-
tivity measurements agreed with those calculated from chemi-
Preparation of Films
Films were made by dissolvin' cellulose acetate in
acetone and casting the solution on a suitable surface with
a doctor blade. The doctor blade could be rotated so that
films of solution 6 or 8 mils thick could be spread. The
solvent evaporated, leaving a film of cellulose acetate.
Several difficulties arose usinr this method. Be-
cause of the high humidity in the atmosphere, water would
condense on the surface of the film during evaporation of
the solvent. To avoid this difficulty the films were cast
in a controlled atmosphere. This was accomplished by con-
structini a dry box which would permit the manipulation of
the doctor blade by means of wires and also permit the
casting of the cellulose acetate solution without admitting
any outside air into the box. The removal of the solvent
was controlled by passing a stream of dry air through the
box. As a further precaution moisture was removed from the
acetone with anhydrous sodium sulfate.
Since the films that were used were only 5 to 7
microns thick, the handling of them was a serious problem.
Two methods were employed. A film could either be cast
directly on cellophane or cast on .lass and picked up by
laying a sheet of wet cellophane on top of it. The cello-
phane then acted as a backing when the films were placed
in the apparatus.
Above an acetyl content of 41 per cent, cellulose
acetate becomes Insoluble in acetone. Chloroform con-
taining 5 per cent ethanol was used to dissolve cellulose
acetates with high acetyl contents.
Determination of Dielectric Constants
The dielectric constant of du Pont CA-43 was
measured by placing the film between two brass plates ap-
proximately one inch in diameter and measurlnp the capacity
with a heterodyne beat apparatus. The instrument that was
used had an error of less than one ,,- (42). Pressure
was applied to the film with a 10,OCCt-pound hydraulic press.
The brass plates were insulated from the Dress.
As pressure was applied to the film, changes in di-
electric constant could have been due to four factors:
(1) an increase caused by structural changes within the
film, (2) a decrease caused by water squeezing out of the
film, (3) an increase caused by better contact between con-
denser plates and the membrane, and (4) a change resulting
from a change in the dielectric constant of water itself.
The change in dielectric constant due to structural
and dimensional changes was corrected for by measuring the
capacity of a dry film as a function of pressure. The nA
versus p.s.1. curve for dry CA-43 also serves as a correc-
tion for any increase in dielectric constant due to better
contact between the brass plates and the membrane. The
change in dielectric constant for dry CA-43 is included in
To insure that the change in dielectric constant
was not caused by a difference in water content, the film
was first subjected to the pressure at which the capaci-
tance was to be measured. After the squeezed-out water had
been removed, the condenser formed by the brass plates and
the film was wrapped with cellophane tape to prevent evapo-
ration losses. Then the pressure was reduced to atmospheric
pressure. The dielectric constant was determined at atmos-
pheric pressure and the pressure was arain increased for
the determination at the desired pressure. The difference
between the dielectric constant at atmospheric pressure and
at the desired pressure was used for the ordinate in Fi ures
5 and 8.
The following expression from Weissberger (43) gives
the relationship for calculating the dielectric constant
from the capacitance:
= d C
C = dielectric constant
C = capacitance in micromicrofarads
d = distance between plates in centimeters
A = area of plates in square centimeters
The dielectric constant determinations were made at
a frequency of 250 KC.
It is felt that the procedure employed pave the
necessary accuracy. The dielectric constant of the dried
sample of du Pont cellulose acetate was found to be 3.1
to 3.4 which lies within the range of dielectric constants
listed in Modern Plastica for cellulose acetate ("4). Be-
sides, the change in dielectric constant was needed instead
of the actual value.
Determination of Membrane Resistances
A drawing of the cell used for measuring resistances
is shown in Fir'ure 25. The circular carborundum whetstones
serve two purposes. First, they compress the film uniformly
and second, they act as electrolyte chambers. When saturated
they hold between 4 and 5 ml of solution apiece. With this
apparatus, compressive stresses of over 8000 p.s.i. can be
applied to a film. To eliminate evaporation losses, the
sides of the carborundum stones were covered with wide
elastic bands and wrapped with "Presstite" Insulation Tape.
This tape, which is pliable like putty, was pressed to cover
Figure 26 is a schematic of the electrical circuit
used for measuring the resistance across the cell. As indi-
cated in the figure, the voltage drop across the cell was
measured directly. The current was determined by measuring
the voltage drop across a known resistance and using Ohm's
law. The double pole, double throw switch was used for
Membrane Being Tested
Silver, Silver Chloride
Solid Brass Cylinder
Scale: 1" = It
Fig. 25.--Apparatus for measuring membrane
Fig. 26.--.Wiring diagram for determining
applying the potential drop across either the known re-
sistance or across the cell to the Beckman Model G pH
meter. The voltage was determined within an accuracy of
In the selection of electrodes, two factors were
considered. First, the products of the electrode reactions
had to be non-gaseous. Second, the products of the elec-
trode reactions should not introduce interfering ions into
To measure the membrane resistance of chloride ions,
discs of Amberplex anion permeable membranes in the chloride
form were placed on both sides of the cellulose membrane.
The cell was compressed to the maximum pressure at which
resistances were to be measured, and the excess solution
that squeezed out was removed. After wrapping the cell
with "Presstite" tape, resistance measurements were made
as the pressure was reduced. Two to three minutes were
allowed at each pressure before measuring the resistance.
A weighted lever was used to apply pressures up to 1600
p.s.I. For pressures up to 8000 p.s.l., a hydraulic
oress was used.
The same procedure was employed for measuring the
membrane resistance of Na+. However, an Amberplex cation
nermeable membrane in the sodium form was used on each side
of the cellulose membrane. The membrane resistance of H30+
was determined by using the cation permeable membrane in
the acid form.
To measure the membrane resistance of Na K1, no ion
exchange membranes were used.
The carborundum blocks were saturated with a C.018 N
NaCI solution for the Cl1, Na+, and 1,aCl determinations.
The blocks were saturated with 0.018 N HC1 for the H30+
The membrane resistance was obtained from the
membranee Total cell Resistance of cell without
resistance (resistance) the cellulose membrane
Cell resistance = Voltage across cell
Current passing through the cell
Current passing = Voltage across known resistance
through the cell Known resistance
The units of membrane resistance are ohms for the
total film area (7.1 square centimeters).
The resistance across the ion exchanPe membranes
is very low and does not vary significantly with pressure.
(1) Samuel Glasstone, "Textbook of Physical Chemistry,"
2nd ed., D. Van Nostrand Co., Jew York, 1'6.
(2) "Development of Synthetic Osmotic Membrane for Use in
Desaltina Saline Waters," Research Proposal for the
IT. S. Department of the Interior, February 19, 1953.
(3) P. F. Cooper and S. A. Pasha, J. Indian Botan. Soc.,
14, 237-55 (1935). C. A. 3, 2602 flr36).
(4) H. 1. Morse, "The Osmotic Pressure of Aqueous Solu-
tions," Publ. Carneire Inst. Washineton, No. 108
(1014). 222 p.
(5) Earl of Berkeley and E. G. J. Hartley, Proc. Roy Soc.
(London) A 92, i77 (l146).
(6) R. A. Craig and F. J. Hartunp, Trans. Faraday Soc.
8, 96t-9 (1052).
(7) A. Mibashan, Rec. Tray. chim. 70, 318-24 (1951).
c. A. 45, 60olo0 (lT5).
(8) S. Trautnann and L. Ambard, J. Chim. Phys., 4I,
220-5 (lc?2). C, A. L6, 8q35f I152).
(9) Franz Paumnarth, ".rzth Forsch" ., 525-30 (1L9);
"Chem. Zentr." (150) II, 1594. CA. 46, 10791d
(10) John H. Northrop, J. of Gen. Physiol., 11, 233 (1928).
(11) Karl Sollner and Harry P. Grepor, J. Am. Chem. Soc.,
61, 346 (1WL5).
(12) W. J. Elford, Brit. J. E.xr. Pathol., 10, 126 (1929).
(13) Laurie L. Burgess, J. Am, Chem. Soc., 56, 4l4-419 (1"31:).