a40 J . Phys. Chem.

1991, 95, 6040-6044

rate limitation by internal electron transfer would interrupt an
expected increase in klzat low ionic strength, provided the internal
reaction is less affected by variations in ionic strength.
Elucidation of Site of Reactivity with Electron Acceptors. The
preceding analysis demands evidence that the measured kI2values
correspond to reaction exclusively at the SO heme site on our
experimental time scale. Sulfite oxidase is a dimer of two identical
subunits, each of which contains a Mo atom present as a mol-
ybdopterin c ~ f a c t o r , ' ~the
, ~ ~site at which oxidation of sulfite
occurs.13 Each subunit also contains the above mentioned cyt
bs-like center where reduction of the physiological electron ac-
ceptor, cytochrome c, occurs subsequent to an internal charge
transfer.I3 Cleavage of the holoenzyme into Mo- and Fe-con-
taining domains by trypsin and isolation of the Mo fragment yields
a sulfite-reducible molybdoprotein which has lost the ability to
reduce cytochrome c, but which retains a level of ferricyanide
reductase activity which is virtually unchanged compared to that
observed for the h ~ l o e n z y m e . ~ ~ ~ ~ "
Against this established background, we tested whether various
metal complex electron acceptors catalytically oxidize sulfite in
the presence of the molybdoprotein fragment, using voltammetric
experiments conducted as described above. No catalytic currents
are observed in mixtures of M o fragment, sulfite, and mediator,
for [C0(4,FMe~-phen)~] 2+/3+, [Co(bpyI3]2+/3+, [Ru(NH3),] '+I3+,
or the anionic complex [ C o ( d p ~ ) ~ ] & (vide
/ ~ supra). The ability
of the sulfite-reduced fragment to reduce ferricyanide was con-
firmed both electrochemically and spectrophotometrically, thus
demonstrating that the Mo fragment sample was active. From
this we infer that the reaction site of these mediators (except
[Fe(CN)6]*'/3-) is the enzymic Fe(I1) site, as is the case for the
physiological acceptor, cytochrome c.13 It should also be noted
that no evidence was found for direct electron transfer between
the EPG electrode and the Mo fragment, either in the presence
or in the absence of sulfite.
Acknowledgment. This research was supported in part by
grants from the National Science Foundation and the North
Carolina Biotechnology Center.
Registry No. [C0(3,4,7,8-Me-phen),]~+, 47889-06-5; [Co(3,4,7,8-
Me,-phen),]'+, 86176-94-5; [C0(4,7-Me,-phen),]~+,47872-45-7; [Co-
(4,7-MeZ-phen),]'+,62791-75-7; [Co(5,6-Mez-phen),l2+,47872-55-9;
[Co(5,6-Mez-phen)#+, 62869-82-3; [Co(4-Me-phen),lZ+,8071 1- 1 I- 1;
[Co(4-Me-phen),13+, 807 1 I - 14-4; [Co(5-Me-phen),lZ+,47860-25-3;
[Co(S-Me-phen),lZ+,96504-30-2; [Co(phen)J2+, 16788-34-4; [Co-
(phen),])+. 18581-79-8; [ R u ( N H ~ ) ~ ] 19052-44-9;
~+, [RU(NH,)~]~+,
18943-33-4; [C0(5-NH~-phen)~]~+, 1 13442-71-0; [Co(S-NHZ-phen),l3+,
1 13461-17-9; [C~(terpy)~]*+, 18308-16-2; [C~(terpy)~]'+,19137-07-6;
F ~ y t c9007-43-6;
, [Co(bpy)J", 15878-95-2;[C0(bpy),l3+, 19052-39-2;
[ Fe(CN)613-, 13408-62-3; [ Fe(CN),]@, 13408-63-4; TMPD", 34527-
55-4; TMPLY', 100-22-1;SO,9029-38-3; molybdoheme, 126858-98-8;
sulfite. 14265-45-3.

Determination of Water Dlffusion Coefficients in Perfluorosulfonate Ionomeric

Membranes

Thomas A. Zawodzinski, Jr.,* Michal Neeman, Laurel 0. Sillerud, and Shimshon Cottesfeld
Los Alamos National Laboratory, Los Alamos, New Mexico 87545 (Received: September 17, 1990;
In Final Form: March 18, 1991)

Pulsed field gradient spin-echo 'HNMR measurements of 'Hintradiffusioncoefficients at 30 "C in hydrated Nafion membranes
are reported. The dependence of the 'Hself-diffusion coefficient on membrane water content was a central part of this
investigation. 'Hdiffusion coefficients ranged from 0.6 X 10-6 to 5.8 X IOd cmz/s for the range of membrane water content
2-14 water molecules per sulfonate. The membrane water content was controlled by isopiestic equilibration of the membrane
sample with water vapor above aqueous LiCl solutions of well-defined water activities. The dependence of membrane water
content on water activity enables us to estimate "chemical diffusion coefficients" from the intradiffusion coefficients measured
by NMR.

Introduction content and possible modes of water management in polymer

Fuel cells using perfluorosulfonate ionomers (e.g. Nafion) as
proton (H+) conducting electrolytes are attractive candidates for
use in electric vehicles.' In our laboratory, such fuel cells are
constructed by hot-pressing together a "sandwich" consisting of
an ionomeric membrane separating a pair of gas diffusion elec-
trodes impregnated with the recast ionomer.2 The only liquid
added to this system in order to operate it as a H2/02 fuel cell
is distilled water. Water is supplied to the fuel cell by humidifying
the gas feed stream. Adequate water content of the ionomeric
material is essential to maintain the conductivity of the polymer
electrolyte membrane. This has, in turn, a strong effect on cell
performance. However, excessive amounts of liquid water could
impede mass transport within the electrode structure. As part
of a larger effort directed at understanding the role of water
( I ) Lemons, R. J. Power Sources 1990, 29, 251.
(2) (a) Raistrick, 1. D. In Diaphragms, Separators, and Ion-Exchange
Membranes; Van Zee,J. W., White, R. E., Kinoshita, K., Burney, H.S.,Eds.;
The Electrochemical Society: Pennington, NJ, 1985; p 172. (b) Ticianelli,
E. A.;Derouin, C.R.; Srinivasan, S.J . Electrmnanl. Chem. 1988, 251, 275.
(c) Gottesfeld.S.; Raistrick, 1. D.; Srinivasan, S.;Springer, T. E.; Ticianelli.
E.; Detouin, C. R.;eeCry,J. 0.;Pafford, J.; Sherman, R. J. Presented at the
1988 Fuel Cell Seminar, Long Beach, CA.
0022-3654/9 1/2095-6040$02.50/0
electrolyte fuel cells, we are engaged in a variety of experimental
studies of the properties of water in perfluorosulfonate ionomers
with the goal of developing models of water distribution across
a fuel cell under typical operating conditions.
The diffusion coefficient of water in Nafion and related mem-
branes as a function of water content is a necessary input for the
analysis of the performance of cells based on these membranes.
Nafion-based polymer electrolyte fuel cells (PEFC) are usually
operated under conditions of partial hydration of the polymer.
Water is lost from the membrane during fabrication of the
membrane/electr& assembly and is apparently not fully replaced
by humidification of the membrane with water vapor carried into
the fuel cell with the reactant gases. The level of hydration could
also vary significantly with position in the membrane when the
cell is under current. A nonuniform water concentration profile
is set up in the membrane by an electroosmotic drag of water from
anode to cathode and by the production of water a t the cathode.
The water concentration gradient set by these two effects is ex-
pected to be made less steep due to water "back-diffusion" from
cathode to anode. Clarification of the water profile across the
ionomeric membrane in a cell under current thus requires
knowledge of water diffusion coefficients as a function of water
0 1991 American Chemical Society
Perfluorosulfonate Ionomeric Membranes
-
ELECTRO-OSMOTICDRAG
-
BACK DIFFUSION
ANODE
H2(g)+2H*+2e
t
MEMBRANE
CATHODE
02+4H*++2H20
Figure 1. Schematic illustration of water transport within a PEFC.
content. A schematic description of water currents in the mem-
brane of a fuel cell is given in Figure 1.
Diffusion coefficients for sorbed solvent and ions in Nafion have
been estimated by using several techniques. Ye0 and Eisenberg3
studied the sorption of water by a dry slab of Nafion and estimated
the diffusion coefficient of water in the membrane over the tem-
perature range 0-99 OC from the water uptake dynamics. Dif-
fusion coefficients from these measurements increased over the
range (1-10) X IO4 cm2/s with increasing temperature with a
reported activation energy of 4.5 kcal/mol. The method used to
estimate the diffusion coefficients by Ye0 and Eisenberg, taking
advantage of the t1I2dependence of the uptake in the initial portion
of the uptake curve, is not necessarily appropriate. The sorption
of water takes place in a membrane with varying water content,
whereas the diffusion coefficient is not known a priori to be
independent of the state of membrane hydration. As Crank has
pointed out for the case of variable diffusion coefficients: sorption
will depend on t112no matter what the relationship is between the
diffusion coefficient and the concentration of water in the film.
Yeager and SteckS reported diffusion coefficients of water in
Nafion 120 membranes containing various alkali-metal cations
determined by radiotracer measurements. In these studies, the
diffusion coefficient of water was measured only for fully hydrated
membranes. The water diffusion coefficient is dependent on the
cation present in the membrane, with the value of the diffusion
coefficient of water in Na+ form (fully hydrated) Nafion equal
to 2.65 X lv cm2/s at 25 OC and the values for K+and Cs+ forms
slightly smaller.
More recently, Verbrugge and co-workers have published studies
of the transport of methanol and of protons (referred to throughout
this manuscript as H+with no intent to imply specifics of spec-
iation) and other ions through Nafion membranes by radiotracer
techniques.6 These experiments were carried out with the
membrane exposed to a bulk aqueous acid solution. This situation
is not identical with that found in a PEFC in terms of either the
hydration state of the membrane or the presence of co-ions in the
membra ne.
We report here some results obtained by pulsed field gradient
spin-echo (PGSE) NMR measurements of IH intradiffusion
coefficients in Nafion membranes of various, well-controlled states
of hydra tion.
Experimental Section
Nafion 117 membranes were pretreated by boiling in dilute
H202solution, rinsing in boiling water, boiling in dilute sulfuric
acid solution, and rinsing in boiling water. Establishment and
maintenance of a well-defined water content are prerequisites for
measurement of transport properties in membranes of different
states of hydration. To control the membrane hydration state,
we have employed isopiestic equilibration of membrane samples
(3) Yeo, S.C.; Eisenberg, A. J . Appl. Polym. Sci. 1977, 21, 875.
(4) Crank, .I.
The Mathematics o/Dlffusion, 2nd ad.; Clarendon: Oxford,
1975; Chapter 9.
(5) Yeager, H.;Stcck, A. J . Electrochem. SOC.1981, 128, 1880.
(6) (a) Verbrugge. M.; Hill. R: J . Phvs. Chem. 1988. 92. 6778. (b)
Verbkgge, M. J . l?iectrochem. Soc. 1989, i34,417. (c) Hill, R.;Verbrukc,
M.J . Electrochem. Soc. 1990,137,886,893.
The Journal of Physical Chemistry, Vol. 95, No. 15, 1991 6041
1
Figure 2. Cell used for determination of membrane conductivity: (1)
Kel-F block; (2) thumbscrew; (3) open area to allow equilibration; (4)
membrane sample; (5) blackened Pt foil; (6) Pt ribbon lead.
with water vapor above aqueous LiCl solutions of known water
activity, following Pushpa et al.' The water content of a mem-
brane sample was set by suspending the membrane following the
above pretreatment over an aqueous LiCl solution in a sealed glass
jar kept in a temperature bath at 30 f 1 "C. Equilibrium, as
indicated by constant sample weight, was reached under these
conditions after 4 days. All membrane weighing steps were carried
out by rapidly transferring the membrane to a weighing bottle
and weighing by difference. The water content of a given mem-
brane was determined by weighing the equilibrated membrane
followed by completely drying the same membrane sample by
suspending it over P205in a sealed jar for several days and re-
weighing it. This last treatment leads to a membrane water
content of (practically) zero based on the following argument.
Drying the membrane under vacuum at room temperature for long
periods leads to a water content of X = 1 (A = the number of water
molecules per sulfonate), as demonstrated by Bunce et a1.* We
find that further drying at 105 OC under vacuum of a membrane
dried in the latter fashion results in a weight loss corresponding
to loss of one water per sulfonate. That all water is lost at 105
OC is consistent with the results reported by LaConti and co-
w o r k e r ~ . ~This is slightly at odds with data presented by Yeo
and Ei~enberg,~ which suggests a small amount (roughly 0.2 wt
%) of residual water present after drying at 105 OC under vacuum.
Drying by suspending the membrane over P205leads to weight
losses equivalent to those obtained via the 105 OC vacuum
treatment. Furthermore, both our vapor-phase isotherms (see
below) and maximum uptake data upon immersion in liquid water
are in quantitative agreement with those reported by others using
other extreme drying protocols. We believe that our low-tem-
perature drying method is to be preferred since exposure of
polymer samples in the acid form to elevated temperatures gen-
erally leads to some charring of the polymer, with unknown effects
on, for example, the ion-exchange capacity of the polymer.
Membrane conductivity was determined by using the cell
sketched in Figure 2. The "window" structure was employed to
allow membrane equilibration in situ. The resistivity of the
membrane was measured by using a pair of pressure-attached high
surface area Pt electrodes at 5 kHz. Both real and imaginary
components of the impedance were measured and compared to
ensure that the real Z-axis intercept was closely approximated
and that the measurement was free of the effects of lead induc-
tance. For determinations of conductivity of partially hydrated
membranes, the cell was suspended in a closed vessel above the
equilibration solution. Conductivity measurements on fully hy-
drated membrane samples were carried out with the cell immersed
in liquid water. Temperature was controlled in both cases by
(7) Pushpa, K.K.;Nandan, D.; Iyer, R. M. J. Chem. Soc., Faruday Trans.
I 1988,84, 2047.
(8) Bunce, N.; Sondheimer, S.; Fyfe, C. Macromolecules 1986, 19, 333.
(9) LaConti, A.; Fragala, A.; Boyack, J. In Electrode Materials and
Processesfor Energy Conversion and Storage; McIntyre, J. D. E., Srinivasan,
'.,. ....,.,
s Will F.. Us. .:
The Electrochemical Societv: Penninnton. NJ. 1977; D 354.
 The Journal of Physical Chemistry, Vol. 95, No. 15, 1991
ISOPIESTICSORPTION CURVE 30°C
14 I I I I I I I I I 90
- m-
12 - m -
- -
10 -
-
m -
- Gradient Pulses
8- -
- -
6-
m -
- -
4- m -
- m m -
2- -
- -
0 I l l I l l I I I
Figure 3. Isopiestic sorption curve for Nafion 117 membrane, T = 30
OC.
immersing the entire vessel in a water bath. The cell constant
was calculated from the spacing of the electrodes and the thickness
of the membrane sample. (The latter was measured with a
micrometer in each case.)
NMR experiments were carried out using a Bruker AM-400
NMR spectrometer using a Bruker microimaging probe equipped
with 50-mm gradient coils and a 5-mm Helmholtz coil insert.
Gradient pulses were shaped through a preemphasis network to
minimize the effect of eddy currents. The effect of any remaining
eddy currents was checked by measuring the signal intensity of
a sample of water as a function of time between the end of an
applied gradient pulse and the onset of acquisition. For times of
20 ms or longer, the signal intensity was constant. Experiments
were carried out with delays between gradient pulses and rf pulses
or acquisition of 20 ms or more. Samples were prepared by
equilibration and then placed in a homemade multicompartment
NMR tube within which equilibration could be maintained. A
small Kel-F container with the equilibration solution was screwed
onto the bottom of a chamber containing the membrane which
was then screwed into a positioning rod. Small bits of threaded
rad were used to join the various compartments. A hole drilled
through the center of the threaded connector between the cham-
bers containing membrane and solution allowed vapor phase
contact. The sample temperature was controlled by means of the
Bruker VTI 000 probe temperature controller and calibrated by
a Luxtron IOOOB fiber optic temperature probe.
The gradient was calibrated by observing the profile of a
calibrated water phantom in the presence of gradients covering
the range used experimentally. To extract the diffusion coefficient
from the raw experimental data, the observed signal intensity for
16 incremented field gradient strengths was fit to the predicted
decay curve using a nonlinear least-squares procedure from SAS
software (SAS Institute Inc.) on a micro VAX computer.
Results and Discussion
Equilibration of Nafion membranes with aqueous LiCl solutions
at 30 OC leads to the water sorption curve shown in Figure 3. The
shape of this curve is similar to that obtained by Pushpa et al.
at 25 OC. However, a somewhat lower water uptake was observed
by us over most of the water activity range. A particularly in-
teresting feature of this curve is the relatively small change in water
content over the range of water activity 0.15-0.75. At water
activities above 0.75, the water content of the membrane changes
much more rapidly with water activity.
Though measurements of diffusion coefficients by the PGSE
technique have been reported for polymer/penetrant systems,I0
to our knowledge none at all have been reported for ion-exchange
polymers. In the pulsed field gradient spin-echo (PGSE) NMR
experiment," an intradiffusion coefficient for species bearing the
detected nucleus is determined from the diffusional dephasing of
~
(10) Stilbs, P. Prog. Nucl. Mugn. Reson. Spectrosc. 1987, 19, 1 and ref-
erences therein.
(1 1) Stejskal, E. 0.;Tanner, J. E. J . Chem. Phys. 196§,42, 288.
Zawodzinski et al.
180
RFPulses n
1
(a)
Gradient Pulses
(b)
lp
-*-
LAA81p
Figure 4. (a) Pulsed field gradient spin-echo (PGSE) sequence. (b) Slow
diffusion enhanced modification of PGSE sequence (SDE-PGSE).
a gradient-encoded magnetization. Experimentally, a pair of
symmetrically spaced field gradient pulses is placed onto a normal
spin-echo sequence, as illustrated in Figure 4a. The predicted
dependence of signal attenuation on gradient strength is
where A(g) is the signal intensity observed with applied gradient
g, A(0) is the signal intensity observed in the absence of an applied
gradient, y is the nuclear gyromagnetic ratio, D is the intradif-
fusion coefficient, and 6 and A are as shown in Figure 4a. The
pulse spacing and duration can be altered to obtain information
on any restrictions to diffusion (e.g. pore walls) on the distance
scale determined by the experimental time scale and the diffusion
coefficient.12
A fundamental requirement of the PGSE technique is that
diffusional attenuation be significant relative to relaxation of the
observed nuclei. If relaxation is too rapid, determination of
diffusion coefficients via the PGSE experiment is difficult or
impossible to carry out. Boyle et al.13 have indicated, based on
their relaxation study of water in Nafion, that PGSE experiments
would be impractical for this system. However, the Nafion
samples used by Boyle et al. were, as they demonstrated, con-
taminated with iron, causing significantly shortened relaxation
times (TI on the order of 10 ms for 1100 EW Nafion). Other
workers have measured proton Tlsand T2swhich are significantly
longer at room temperature (roughly 100 and 50 ms, respec-
tively).14 We find in our experiments on well-purified, protonic
forms of Nafion that proton longitudinal relaxation times vary
over the approximate range 80-200 ms for water at various
concentrations in Nafion 117 membranes (less water in the
membrane leads to shorter relaxation times), while transverse
relaxation times are generally about half as long at 25 OC. There
is thus an ample time window in which to perform PGSE mea-
surements.
'H diffusion coefficients in pieces of hydrated Nafion 117
membrane were determined for samples with various water con-
tents, as established by the isopiestic technique described above.
A typical plot of the observed signal attenuation versus gradient
strength (for a sample containing 9 waters per sulfonate) is shown
in Figure 5 together with the best-fit curve predicted from eq 1.
The measured intradiffusion coefficients are summarized in Table
I as a function of water content. IH diffusion coefficients in Nafion
decrease with decreasing water content. No systematic difference
(12) Tanner, J. E.; Stejskal, E. 0. J . Chem. Phys. 1968,49, 1768.
(13) Boyle, N. G.; McBrierty, V. J.; Douglass, D. C. Mucromolecules
1983, 16, 75.
(14) (a) Slade, R. C. T.; Hardwick,'A.; Dickens, P. G. Solid Stare lonics
1983, 9/10, 1093. (b) Sivashinsky, N.; Tanny, G. B. J . Appl. Polym. Scl.
1981,26, 2625.
Perfluorosulfonate Ionomeric Membranes The Journal of Physical Chemistry, Vol. 95, No. 15, 1991 6043
I I I

EXPERIMENTAL DATA POINT -

-m- CALCULATED BEST FIT

0.8 -
-
* 0.4 -
0.3 -
0.2 -
0.1 -
0' I I I
I I I I 1 I
0 1 2 3 4 5 6 7 8 9
GRADIENT STRENGTH ( G k m ) MOLES OF WATER PER EQUIVALENT
Figure 5. Fit of PGSE data to cq I . Figure 6. Specific conductivityof Nafion 117 as a function of hydration
state.
TABLE I: 'H Intndiffusion Coefficients, 30 O C
water content D, water content D, 0 DH+(cond)
(HzO/SO,H) IO6 cm2/s (HzO/S03H) IO6 cm2/s D i H( N W
2
3
0.6 f 0.06' (SDEb)
1.2 0.08 (SDE)
6
9
*
3.7 0.15
4.4 & 0.07
1

4 2.1 f 0.06 (SDE) 14 5.8 f 0.21

*Errors reported here are the standard error of the nonlinear least-
squares fit to experimental data. bThe label SDE indicates that this
diffusion coefficient was determined by using the SDE-PGSE techni-
que.

TABLE 11: Effect of Gradient Strength on D Measured by the

SDE-PCSE Experiment
gradient strength,
G/cm2
D,
IO6 cmz/s
gradient strength,
G/cm2
D,
lo6 cm2/s
fL
s "i
l

0
-

m'
10 20
MOLES OF WATER PER EQUIVALENT
I

30

Figure 7. Comparison of H+mobility (from protonic conductivity) and

0 3.6 0.3 7.14 4.3
2.38
4.76
4.0
3.9
* 0.3
0.2
9.52 3.3
is observed between the diffusion coefficients measured for dif-
ferent values of A. In addition to suggesting that there are no
restrictions to diffusion apparent on the length scale probed by
this experiment is on the order of a few microns for these
experiments), this result enhances our confidence that eddy
currents are not significantly affecting the measurement.
A modification of the Stejskal-Tanner sequence, referred to
here as the slow diffusion enhanced (SDE) PGSE experiment,ls
was employed for two purposes: (i) determination of slow diffusion
coefficients and (ii) testing for the presence of two distinguishable
types of water, one rapidly diffusing and one slowly diffusing. This
pulse sequence is shown in Figure 4b. Additional gradient pulses
are inserted symmetrically before and after the gradient pulses
of the normal PGSE experiment so that the spins precess under
the influence of both gradients during time A. Neeman et aI.l5
have indicated the potential application of the SDE-PGSE
technique to enhance the signal attenuation in slowly diffusing
systems, in our case equilibrated Nafion samples with low water
contents. Diffusion coefficients for membranes with water contents
in the range 2-4 HzO/S03H were determined by using the
SDE-PGSE technique with a single value of gradient gr (except,
of course, for the determination of S(O), for which no gradients
were used).
' H diffusion coefficients were determined for a hydrated
membrane sample by using several incremented values of gr.
Under the influence of gr, any rapidly diffusing components are
more strongly attenuated. A decrease in the measured diffusion
coefficient with increasing gr indicates that several water envi-
ronments with different diffusional character exist in the sample.
As shown in Table 11, the measured diffusion coefficients are
independent of the value of gr, indicating that no dispersion in
the diffusion rates of 'H nuclei exist in the sample. Thus, we
conclude that no large "pockets" of bulk water (e.g. water-filled
( I S ) (a) Neeman, M.;Freyer, J.; Sillerud, L. J . Magn. Reson. 1990, 90,
303. (b) Neeman, M.;Jarrett, K.A.; Sillerud, L.; Freyer, J. P.Cancer Res.,
in press.
'H intradiffusion coefficient for water in Nafion 117 as a function of
* 0.2 extent of hydration.
0.4
bubbles in the membrane) exist under the conditions of these
experiments.
Having described the determination of 'H diffusion coefficients
in hydrated Nafion membranes, we must now consider what the
diffusion coefficients mean. In a hydrated Nafion membrane,
'H nuclei exchange rapidly between H20and H+ (we reiterate
that the symbol "H+"covers all forms, including aqueous com-
plexes, of H+ in the membrane) environments. The measured 'H
diffusion coefficient is a weighted average of the diffusion coef-
ficients for the separate environments. Thus, the diffusion
coefficient measured in this work is not exactly the intradiffusion
coefficient of water in these membranes. However, the identi-
fication of the 'H diffusion coefficient with that of water is rea-
sonable at high water contents since 'HzOis in large excess over
'H+ (28 'H present as HzO to 1 present as H+ when X = 14).
At low water contents, the identification of the 'H diffusion
coefficient with that of water is based on a more subtle argument
outlined below. The conductivity of Nafion 1 17 as a function of
membrane water content is shown in Figure 6. The mobility of
H+ can be estimated from the conductivity data using the
Nemst-Einstein equation. A plot of the H+ diffusion coefficient
(from ionic mobility) and the 'H diffusion coefficient (from NMR)
versus membrane water content is shown in Figure 7. The dif-
fusion coefficients are similar at low water contents, differing
increasingly as the water content increases. From this behavior,
we infer that transport of H+ by Grotthus hopping probably
becomes increasingly significant at high water contents, whereas
it seems to be negligible at low water contents. This is not un-
reasonable, since water in the polymer a t low water contents is
likely to be solvating H+and SO3- and becomes more bulklike
only at high water contents. Falk has shown that H-bonding
interactions between water molecules, of importance for proton
transport by hopping, are significantly weaker in Nafion even for
fully hydrated membranes.I6 Indeed, the ratio of H+ mobility
in liquid water to the diffusion coefficient of water molecules in
water is higher than the corresponding ratio (Figure 7) for H+
and water in the fully hydrated polymer. The N M R spectrum
(16) Falk, M.Can. J . Chem. 1M0, 58, 1495.
6044 J. Phys. Chem. 1991, 95,6044-6047

TABLE III: Chemical Diffusion Coefficients
water content Dh, water content D-,
(H,O/SOIH) 106 cm2/s (H20/S03H) lo6 cm2/s
2 1.3 6 2.0
3 4.2 9 1.7
4 2.3 14 1.5
indicates no immobile water at low membrane water contents since
no very broad components are observed in the spectrum. Thus,
we conclude that, in the extreme of low water content, H20and
H+probably diffuse by an identical mechanism, i.e. that the mobile
species under an electric field is solvated H+, of mobility identical
with that of H20.At such low water levels in the membrane, all
'Hnuclei would thus diffuse together and the measured diffusion
coefficient of 'His, again, a good estimate of the water diffusion
coefficient in the membrane. Thus, on the basis of the above
arguments for high and for low water contents, we tentatively
identify the 'Hdiffusion coefficient measured here as the intra-
diffusion coefficient of H20in the membrane over the entire range
of membrane water contents. We are, however, planning ex-
periments employing H2I7Oto provide definitive water diffusion
coefficients in order to further substantiate this point.
In a real membrane in which a water concentration gradient
exists, such as in an operating fuel cell, diffusion of water through
the membrane takes place in a chemical potential gradient. Flux
as described by Fick's law is generally relative to a concentration
gradient, Le.
flux = Dc,,,(dC/dx) (2)
Conversion of a measured intradiffusion coefficient to a "chemical"
diffusion coefficient is carried out using the equation
and C is the concentration of the diffusing species (Le. the
membrane water content). The "chemical" diffusion coefficients
are obtained by use of eq 3 and the isopiestic data shown above.
The curve shown in Figure 3 is transformed into a plot of In "a
vs In C,,,,,. This curve is next fit to a third-order polynomial,
and the derivative term in eq 3 is then directly obtained. The
calculated values of Dchcmshow a much smaller range of variation
with water content than do the self-diffusion coefficients (Table
111). This is caused mainly by the much stronger variation of
water activity with water content a t the lower water contents
studied, as can be seen in Figure 3. The large variations of the
activity coefficient of water at low water contents provide an
additional "drive" for water transport, compensating for the large
drop in the value of the intradiffusion coefficients.
This report demonstrates the capability of the NMR technique
to provide the desirable information of diffusion coefficients of
water ('H) in PEFC membranes maintained at a well-defined state
of partial hydration. More extensive work is currently in progress
in our laboratory to determine diffusion coefficients of water in
Nafion and other ionomeric membrane systems under a variety
of conditions, most importantly, elevated temperatures. PGSE
measurements on hydrated Nafion membranes employing "0as
the probe nucleus will be carried out to directly ascertain the
diffusion coefficient of water in the membrane. As suggested
above, the results described in this communication are currently
being applied in the development of a water transport model for
polymer electrolyte fuel cell systems.17
Acknowledgment. This work was supported by the U S . De-
partment of Energy, Office of Conservation and Renewable En-
ergy.
Registry NO. H20,7732-18-5.

Dchem = KC)^*
d In a
(3)
(17) (a) Springer, T. E.; Gottesfeld, S.;Radzinski, S.;Zawodzinski, T. A.
Book of Abstracts, 178th Meeting of the ElectrochemicalSociety, Seattle,
where Dchcmis the "chemical" diffusion coefficient, D* is the WA, Oct 1990; Abstract No. 118. (b) Springer, T. E.; Zawodzinski, T. A,;
intradiffusion coefficient, a is the activity of the diffusing species, Gottcsfeld, S . J . Electrochem. Soc., in press.

Radical Scavenging In the Sonolysls of Aqueous Solutlons of I-, B r , and NS'

Maritza GutiCrez, Arnim Henglein,*

Hahn-Meitner-Institut Berlin GmbH. Bereich S, 1000 Berlin 39, FRG

and Fernando Ibaiiez

Facultad de Quzmica. Pontifcia Universidad Catblica de Chile, Santiago, Chile (Received: December 28, 1990)

Iodide and bromide solutions are sonolyzed under pH conditions, where reactions of the products, i.e. hydrogen peroxide
and iodine (or bromine), do not occur. The total yield of the products as well as the hydrogen yield is independent of solute
+
concentration. The results are understood in terms of the competition of the OH OH and the OH solute reactions.+
A local concentration of 4 X M of the OH radicals in an interfacial region between the cavitation bubbles and the liquid
is derived from the data obtained. The sonolysis of azide solutions is also investigated. The main product is nitrogen, which
is formed in the reaction of N3- with OH radicals in the millimolar concentration range. At higher azide concentrations,
additional nitrogen is formed as hydrogen atoms are also scavenged. Ammonia and hydrazine are minor products of the
N3- sonolysis.

Introduction homogeneously in the solution but in "hot-spots" where cavitation

The kinetics of chemical reactions that are initiated by ,Itra- bubbles pulsate or collapse. During the adiabatic compression
sound are rather complex.'-' The reactions do not take place Phase in such gas bubbles, temperatures of several lo00 K are
reached and pressures of up to 100 bar. Molecules can be dis-

( I ) (a) Henglein, A. Ulrrasonlcs W87, 25, 6 . (b) Aduances in Sone (3) Suslick, K. S . ulrrasound, I r s Chemical, Physical, and Biological
chemistry; Mason, T.,Ed.;Jai Press: Vol. 3, in press. Eflects; VCH: Weinheim, 1988.
(2) Mason, 7. J.; Lorimer, J. P. Sonochemistry, T W ,Applicarions and (4) Carmichael. A. J.; Mossoba, M.; Riesz, P.; Christman, C. L. IEEE
Uses of Ulrrusound In Chemistry; Ellis Honvood Ltd.: Chichwter, 1988. Trans. 1986, UFFC-33, 148.

0022-3654/91/2095-6044$02.50/0 Q 1991 American Chemical Society