Measurement of Diffusivity of Organic Liquids through
Polymer Membranes
A Simple and Inexpensive Laboratory Experiment
U. Shanthamurthy Aiial and Tejraj M. Aminabhavi'
Karnatak University. Dharwad, India 580 003
A common laboratory experiment to study the interaction
of polymer membranes with organic liquids is of great rele-
vance in chemical industries and allied areas in view of the
fact that the real tests of measuring the rate of absorption of
a liquid by large pieces of polymer samples are highly time-
consuming and often expensive. Thus, characterization of
liquid interaction with polymer membranes by using simple
laboratory tests using a small sample of polymer membrane
is certainlv
-~~~ ~
-
" of ereat value in understanding the actual behav-
ior of a large industrial rubber sample.
Theoreticallv. the contact of polvner membranes with
organic liquidscan be described by absorption and diffusion
ohenomena ( 1 ) . It is essential to know at what rate the liquid
biffuses inti the polymer matrix. Often simple laboratory
methods such as sorption experiments yield useful informa-
tion on the transport characteristics of polymer membranes.
Here we describe a simple and inexpensive laboratory ex-
periment to study the diffusion of organic liquids through
polymer membranes. This experiment can be performed by
most undergraduate physical chemistry students. This pro-
cedure helps to assess the resistivity of a polymer toward
well-known, commonly available organic liquids like ben-
zene, toluene, mesitylene, cblorobenzene, n-hexane, and
carbon tetrachloride. Interactions of these liquids with an
industrial polymer membrane, namely polyurethane, are be-
ing investigated.
Theory
Liquid absorption by polymer membranes is considered to
be a diffusion process. The diffusion of liquids and relax-
' Author to whom correspondence should be addressed.
82 Journal of Chemical Education
ation processes in a polymer above its glass transition tem-
perature (T,) are mainly governed by the segmental mobility
of the chains, which in turn are considered to be affected by
the total free volume and its distribution within the polymer
matrix (2). In rubbery polymers well above their T,, the
polymer chains adjust so rapidly to the presence of pene-
trant that they do not cause diffusion anomalies. For such
situations the diffusion process is generally described by
Fick's theorv and is often controlled by a concentration-
independent diffusivity. One of the easiest and simplest
~rocedurestoestablish the transport mode within apolymer
membrane is to analyze the sorption data and thereby esti-
mate the numerical value of the exponent n of eq 1.
MJM. = k t " (1)
Here, M,is the mass uptake at time t, M , is the equilibrium
value, and k is a constant that depends on the structural
characteristics of the polymer in addition to its interaction
with the solvent. The value of n determines the type of
transport mechanism (3.4). A value of n = 0.5 suggests the
Fickian mode, whereas for the uon-Fickian diffusion n is
--
unitv. Often. n mav v a n between 0.5 to 1. and this suggests
the Lomalo& diffusion pattern.
The solvent transport in a polymer membrane has been
described (2) bv Fick's second law of diffusion. which for a
constant diffnsrvity D is given as
where c is the concentration of the diffusing material (gI100
gof original sample) in the direction of x coordinate axis and
t is diffusion time (s). Equation 2 upon integration gives the
percentage fractional mass uptake Q(t) as

where h is membrane thickness. T h e first term of the inte-
grated infinite series can be used to determine the maximum
absorption; D of the penetrant within the polymer matrix for
values of Q(t) u p to 50% can he calculated from the slope ( 8 )
of the straight line of the plot of Q(t) versus t1I2and hy using
Treatment of Data
The sorption data are interpreted in terms of both percent
uptake Q(t) and percent increase in thickness h(t) of the
membrane versus square root of time, t112.~ h e s plots,
e as are
given in Figures 1and 2, exhibit linearity during early stages
of absorption [up to -50% of either Q(t) or h(t)]. From the
slopes of the linear portions of the curves of Q(t) or h(t)
versus tlR, diffusivities are calculated using eqs 4 and 6,
respectively, and these data, when compared (see Table l ) ,

I I I I
0 Benzene Chlorobenzene
Often attempts have been made to calculate the theoretical
sorption curves from eq 3 using D as obtained from eq 4. o Toluene 8 Carbon tetrachloridt
Another useful approach (5-7) for studying diffusivity is A Mesitylene A n-Hexane
through the dimensional response of the membrane: this
approach is based on the of "hygroelasti&ty",
where one is confronted with questions such as how much
the polymer swells and what is the magnitude of the internal
stress that develops in it. Thus, hygroelasticity can be treat-
ed by examining the relative dimensional change in relation
t o absorption, which may be expressed in a number of ways.
Expressing liquid absorption in terms of the relative weight
gain yields the definition of "coefficient of hygroe1asticity"p

*=- Ahlh,
AWIW,
where Ahlh, and AWIW, are. res~ectivelv. the relative
change in thkkness and weight' of &e polymer membrane.
The coefficient fi can be determined from the slope of the
curve relating thickness change to weight change. For many
polymer-solvent systems, p is independent of AWIW, and
may he regarded a s a constant throughout the swelling pro-
cess. Such considerations lead to the following relation for
diffusivity:
Figure 1. Variations of percent maos uptake 40 of solvem wRh square rwt of
time (t'I2). Symbols: 0, benzene; 0, toluene: A, mesitylene; 0 , chlwoben-
Zene; B, w b o n tetrachiaide; A, rrhexane,

where 9, is the slope obtained from the initial linear portions

o f t h e plot h(t) I-(Ahlh,)1001 versus 1' ".and h, is themaxi-
mum thickness: 0 Benzene Chlorobenzene
Several experiments on a wide variety of polymer-solvent Carbon tetrachloride
35
systems (8) tell us that eqs 4 and 6 give nearly identical
results on diffusivities through polymer membranes, and A Mesitylene A n-Hexane
this approach will be demonstrated in this paper.

Experimental Procedures
Any elastomer available as a rubhery material at room tempera-
ture may be used for experimentation. In this investigation,due to
the widespread use of polyurethane in industry and engineering (9),
we have employed a commercially availshle polyurethane. The base
polymer was a Vibrathane B600 (Uniroyal)cured with 4,4'-methyl-
ene-bis-o-chloroaniline (MOCA). A sample of uniform thickness
(0.250 cm) was used throughout; these were cut into circular pieces
of diameter 1.9 cm, weighing approximately 0.8 g. These should be
dried in vacuum oven at room temperature before experimentation.
The solvents, namely benzene, toluene, mesitylene, chloroben-
zene, n-hexane, and carbon tetrachloride, were obtained in their
highest purity and were double-distilled to ascertain extreme puri-
ty. Caution should be exercised in handling these chemicals. As
many of these are highly toxic, the experimentsshould be conduct-
ed under hood, with eztrerne core.
Sorption experiments were performed by placing the cut polymer
membranes in the respective liquids in screw-tight metal-capped
test hottles maintained at 25 OC (10.5 "C). At regular time intervals
the samples were removed from the test containers, blotted with
Kimwipes to remove the surface-adhered excess liquid. These were
then weighed (10.05 mg) under closed environments, the thiek-
nesses were measured (10.01 em) at several points by means of a Figure 2. Variatimofpercem increase in mickness Wnot ths memocans w m
micrometer and then placed hack into the containers. sqme rwt ol time 0' 4. Symbols have ths same meanmg as in Figne 1.

Volume 67 Number 1 January 1990 83

 Table 1. Solvent Properties and DlWuslvRy Data

DiffusioncDefficlent
Molar Exponent D x lo' (cm2/5) Maximum Maximum
volume (4 by wt. gain by hygmelasticily wt. gain thickness
Penetrant (cm3/mol) eq 1 eq 4 eq 6 M. (%) h- (%)

benzene 89.41 0.563 2.55 2.67 67.19 25.01

toluene 106.85 0.590 2.59 1.83 60.17 21.92
mesilylene 139.58 0.532 0.86 0.98 40.15 13.88
~hl~~obenzene 102.24 0.605 2.81 2.31 105.52 31.02
shexane 131.61 0.501 1.22 1.06 6.94 5.75
wrbon tebachl~~lde 97.09 0.553 1.15 1.29 105.70 21.60

Tabla 2. Thermodynarnlc Data and Swelllng Parameters

1.0 - Molswlar
Solubility weight
0.8 - parameter Swelllng Volume Interaction between
6. (wl/ coefficient fraction parameter mass-links
Penetrant ~m~)''~ a *P X &.t
0.6 -
.
T
benzene
toluene
9.2
8.9
0.769
0.698
0.541
0.565
0.354
0.405
596
703
2-
0.4 - -
-
expCrimPnf
c ~ ~ c u l a t e from
d
masilylene
chlarobenzene
Mexane
8.8
9.5
7.3
0.463
0.958
0.106
0.662
0.467
0.896
0.456
0.340
1.416
600
890
614
carbon tetra- 6.6 0.666 0.576 0.473 710
0.2 - rhlnrida

used t o estimate the Flory-Huggins-type polymer-solvent

interaction parameter x as
Figure 3. ComDarisan of meOretlcal and experimental sorptlon curves for
polyurethane-Mexane system at 25 T.

where 6, and 6, represent the solubility parameters of liquid

agree quite satisfactorily. The concentration-independentD
thus calculated is further used to generate the theoretical
curve from eq 3; a typical plot is shown in Figure 3 for n-
hexane. The ohsemed good agreement between theoretical
and exnerimental curves is sue~estiveof the reliahilitv of the
m e t h 2 of estimating D frogeither eq 4 or eq 6. F k h e r -
more, the exponent value n of eq 1, heing close to 0.5 but
varying u p to a maximum of 0.6, suggests that the diffusion
Drocess deviates sli,qhtly from the expected Fickian value
i n d could be classified as anomalous.
Several other useful ohsewations are ohtained from Fig-
ures 1 and 2. For instance, chlorobenzene shows highest
value for both Q(t) and h(t), whereas n-hexane shows the
lowest values. exhibitine almost a flat de~endenceof both
Q(f) and h(t)'on It ). ~ G p r i s i n ~forl ~carbon
, tetrachloride
the Q ( 0 is hieher than those of the remainine Denetrants: on
the other hand, i t has a smaller h(t) than do benzene, tolu-
ene, or chlorobenzene.
Further insight into polymer-solvent interactions may be
ohtained by the application of the Flory-Rehner model
(10.11) to the swelling data. This model assumes the depen-
dence of factors such as (1) molecular weiaht (M,) between
cross-links, (2) volume fraction (Qp) of the polymer in its
swollen state, and (3) polymer-solvent interaction parame-
ter (x) on the overall morphological behavior of the polymer.
In order to compute the M, values, we must obtain the
sorption data in terms of swelling coefficient (a)defined as
M-.-M, 1
a=-
M, PS
where M, is the mass of the swollen polymer (i.e., equilibri-
um saturation) and M, is the original mass of the polymer
membrane; p, is the liquid density. These data are further
and polymer, 6 is a lattice constant whose value is about 0.34,
and Vs is the molar volume of the solvent, the term RT has
the usual meaning. For equilibrium swelling of a cross-
linked network in a solvent, the M, can be calculated using
Flory and Rebner theory:
where Q,, volume fraction of the polymer in the swollen
membrane a t equilibrium with solvent, is calculated as
The results of this analysis are given in Table 2. The value of
x indicates the strength of polymer-solvent interaction.
Generally, lower values of x (i.e., well below the assigned
value of x = %) indicates that the particular penetrant would
be a good solvent for the polymer in the absence of the
restraining effects of hard segments of the polymer. In any
case. this method vields a somewhat satisfactorv Drocedure
to estimate M, vilues, which are found to v&from one
Denetrant to the other.
From the experimental result of Tables 1and 2 it is ohvi-
ous that high M, (106%) and h , (31%) as observed for the
polyuretha~e-chlo~oben~ene system resulted in lower val-
ues of x (0.340) and higher values of M,(-890) as compared
to the remaining liquids. This suggests that, out of all the
monocyclic aromatics chosen for this investigation, chloro-
(7)
benzene is the good solvent. Similarly, carbon tetrachloride
behaves ina manner similar to that ofchlorobenzene.On the
other hand, n-hexane, being a poor solvent for the polymer,
exhibited the smallest values for M , and h. and the highest
value for x (1.416).

84 Journal of Chemical Education

 Planning of Student Assignments
This experiment is somewhat time-consuming as it re-
quires careful preparation by the student for a sequence of
experimental operations. We find that four to five continu-
ous laboratory sessions of 4 hours each are required for a pair
of students 6 carry out duplicate determinations of D using
both the experimental techniques. The entire experiment
can be completed in three days. We require the students to
start from scratch-cutting the membrane to appropriate - - -
size, purifying and characterizing solvents, etc.
One of the virtues of this experiment is that i t provides
different experimental and mathematical approaches to the
same physical values. Effective educational use of the meth-
od and economy of the time can be achieved by assigning
different pairs of students to different experiments. For
example, one pair of students can be assigned to one type of
liquid-polymer pair and evaluate D from both eqs 4 and 6.
The second, third, fourth, fifth, and sixth pairs of students
can be assigned the same determinations but using different
liquids for the same polymer. Another variation suggested is
the determination of the same quantities over a ranee of
temperatures. I t is desirable t h a t t h e values of diffusiiities
be obtained for a t least three temperatures. Such studies
lead to the evaluation of ~rrhenius-typeactivation parame-
ters involved in the transport processes. The three most
convenient recommended temperatures for rubbery poly-
mers are 25,40, and 55 "C.
We find that followine such a nrocedure nroduces a
healthy and stimulating interchange bf ideas ambng various
groups of students. At the end of the work. a session mav he
Leldduring which results from various groups are presented
and compared. Thus, students will be exposed to "real-
world" and will develop skills necessary to bridge
from basic science to the solution of technological problems.
Afurther analvsis of the data in terms of suitihle theoretical
~ ~~ ~~
lmpllcations and Conclusions
The study of transport of organic liquids through polv-
mers is important for a variety of engineering applicaiio"s.
Contrary to gases, which diffuse through polymeric materi-
als with little interaction, the transpwt of liquids causes
polymers to swell; the extent of swelling depends upon ther-
mal condition, chemical nature, and deeree of cross-linkine
of the polymer in addition to the molar&ass of the
and of the liquid molecule. I t is obvious that eq 6 suggests a
new way of measuring D. This method is advantageous over
the weight-gain technique in that the evaporation losses, if
any, do not affect the final results. However, such evapora-
tion losses are critical in the weight-gain method and affect
the final results. The agreement of diffusivity values as oh-
tained by two entirely different methods is quite remark-
able. I t is therefore concluded that the dimensional chanee
technique provides a very convenient method for measuring
-
the diffusivity of a solvent throueh a polvmer memhrane.
Furthermoreithis technique is apilicabie to many polymer-
liquid systems whose p values are independent of the ab-
sorbed penetrant; also, the experiments can be performed
with great precision and do not require any sophisticated
instrumentation.
Acknowledgment
We thank the University Grants Commission, New Delhi.
for the award of teacher fellownhir, to USA under the Facul-
ty Improvement Program.
Literature Cited
1. Jmt, W.Diffurion in Salids,Liquidsund Gares;Aeademie: New York, 1952.
2. Crank, J. The Molhsmaficsaf Diffusion, 2nd ed.: Oxford University: ond don. 1975.
3. Peppsu,N.A.;Urdhal,K.G.Eur.Polym.J. 1988,24,13.
4. 8mith.M. J.:Peppaa,N.A.Polymer1985.26.569.
5. Cahn, D.: Marom, G. Polym. Ens Sci. 1978,18,IW1.
6. Cohn. D.; Msrom, G.Polymer 1979. W.501.
7. Cohn, D.; Merom, 0.Palym. En#.Sci. 1982,22,8A.
~~~~~~

8. Aithal. U.S. PhDTheais, Kamatak University, 1989.

models, nameiy Flory-Rehner theory, leads to a better un- 9. Sanders,J. H.: F-h, K. C. Polyurethanes Chemistry and Technology, PLI, Chem.;
Krioger: New York. 1978.
derstanding of the polyrner-solvent interactions and gives 10. Flow, P.J. ??inciplssofPolymor Chemlalry: Cornell University: Ithaea, NY, 1953.
an insight into swelling mechanism. 11. Flow P. J.; Rehner, J., Jr. J. Chom.Phys.1943.11.521.

Call for Papers-Semon National Undergraduate Research Symposium

The 12th Annual Semon Lecture and National Undergraduate Research Symposium will be held at Kent State
University an Monday, April 2, 1990. This event is named in honor of Waldo Semon, who was the pioneer at the BF
Goodrich Company of Akron, OH, in the development of PVC and, after retirement, senredas a Research Professor at Kent
State University. A major focus of this event, which is co-sponsored by Kent State University and Goodrich, will be the
SemonNationalUndergraduateResearch Symposium. Students at colleges and universities anywhere in the United States
are invited to submit a paper describing their undergraduate research work. This should be limited to 10 double-spaced,
typed pages with any number of figures, tables, and references. Six finalists will be invited to present short seminars
describing their work at the symposium. The student presenting the best paper will receive a $2,000 prize, with $200 going
to each of the five remaining speakers. Limited travel funds will also be available. Interested students should write for
further details to: Paul Sampson, Chair, Semon Lecture Committee, Department of Chemistry, Kent State University,
Kent, OH 44242, or call (216) 672-2032. The deadline for receipt of papers is March 1,1990.

Volume 67 Number 1 January 1990 85