TECHNICAL PAPERS

ELECTROCHEMICAL SCIENCE AND TECHNOLOGY

Performance Modeling of the Ballard Mark IV

Solid Polymer Electrolyte Fuel Cell
I. Mechanistic Model Development

J. C. Amphlett, R. M. Baumert, R. F. Mann, B. A. Peppley, and P. R. Roberge*

Royal MilitaTy College of Canada, Kingston, Ontario, Canada K7K 5LO

T. J. Harris
Queen's University, Kingston, Ontario, Canada

ABSTRACT
A parametric model predicting the performance of a solid polymer electrolyte, proton exchange membrane (PEM) fuel
cell has been developed using a combination of mechanistic and empirical modeling techniques. This paper details the
mechanistic model development. Mass transport properties are considered in the mechanistic development via Stefan-
Maxwell equations. Thermodynamic equilibrium potentials are defined using the Nernst equation. Activation overvoltages
are defined via a Tafel equation, and internal resistances are defined via the Nernst-Planck equation, leading to a definition
of ohmic overvoltage via an Ohm's law equation. The mechanistic model cannot adequately model fuel cell performance,
since several simplifying approximations have been used in order to facilitate model development. Additionally, certain
properties likely to be observed in operational fuel cells, such as thermal gradients, have not been considered. Nonetheless,
the insights gained from the mechanistic assessment of fuel cell processes were found to give the resulting empirical model
a firmer theoretical basis than many of the models presently available in the literature. Correlation of the empirical model
to actual experimental data was very good.

A model mapping overvoltage in a proton exchange Stage 1 is discussed in this paper, stages 2 and 3 are dis-
membrane (PEM) fuel cell as a function of val~ious con- cussed elsewhere ~ and in the following paper. 9
tributing variables would be of great utility to researchers
and operators. Two common modeling approaches are: (i)
Physical Properties, Assumptions,
theoretically based mechanistic approaches and (it) empir- and Approximations
ically based analysis. Although, many mechanistic models The fuel cell for which the present model is proposed, and
can be found in the literature, ~-4they generally require the with which data for the empirical analysis were gathered,
knowledge of parameters which are not readily available, is a single ceil of the MkIV design produced by Ballard
such as, transfer coefficients, humidity levels, membrane, Power Systems Inc. of North Vancouver, BC. The cell con-
electrode, and active catalyst layer thicknesses. Empirical sists of a sandwich of Du Pont's Nation 117TMmembrane
m o d e l s y on the other hand, are generally accurate over a between two carbon fiber paper electrodes. The faces of the
small operating range. They fail to reflect the actual pro- electrodes contacting the membrane are impregnated with
cesses involved in fuel cell operation and thus are not ap- platinum catalyst and Teflon. The geometric surface area of
plicable over a broad range of conditions. An approach the cell is 50.56 cm 2. At the time of experimentation, the
combining both mechanistic and empirical modeling tech- cell had already logged in excess of 500 h of operation (non-
niques has the potential to couple both mechanistic valid- continuous), and some performance deterioration had been
ity and the inherent simplicity of the empirical approach. observed in that time (output voltage reduction of approxi-
The goal of the present work was to couple a mechanistic mately 0.1 V at 400 ASF for a fuel cell system operating at
evaluation of fuel cell processes with the application of 70~ using 30 psig H2 and 30 psig 02). Over the course of
empirical linear regression modeling techniques, in order experimentation (approximateIy 75 h), however, no further
to develop a performance model for a F E M fuel cell. A deterioration was observed, and experimental replicability
three-stage approach was followed. Initially,mechanistic was achieved. Further details on the MkIV cell can be
modeling techniques were employed to determine the char- found in various Ballard publications.~~
The envisaged electrode structure comprises two regions;
acteristic dependencies of fuel cell operation. This yielded
a porous, partially hydrophobic, carbon fiber-paper back-
simple algebraic expressions defining the thermodynamic
ing layer, and a very thin, hydrophobic layer of platinum
equilibrium potential, as well as overvoltages due to acti-
catalyst and Teflon binder lying between the electrode and
vation, ohmic resistance, and mass transport. In the second
the membrane. Of this second layer, only the uppermost
stage, numerical parametric equations for each of the
catalyst agglomerates (5 nm < radius < 50 nm) are believed
voltage loss terms were determined from experimental
to directly contact the membrane phaseJ '2 It is assumed
data. These equations reflected the form of the algebraic
that the walls of the macropores between carbon fibers are
expressions developed in stage I. The third stage involved
coated with hydrophobic Teflon, while the walls of the mi-
the statistical analysis of the resulting parametric model. cropores within carbon fibers will not be Teflon-coated.
* Electrochemical Society Active Member. The wet-proofed macropores provide a path for gas flow in
J. Electrochem. Soc., Vol. 142, No. 1, January 1995 9 The Electrochemical Society, Inc.
Downloaded on 2019-04-19 to IP 186.121.125.90 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 J. Electrochem. Soc., Vol. 142, No. 1, January 1995 9 The Electrochemical Society, Inc.

the cathode, while the micropores provide a path by which fore assumed in all mathematical calculations as a first
liquid product water can exit to the flow channels. A simi- approximation.
lar approach to water and gas transport in the electrode has
been suggested previously by Verbrugge and Bernardi2 Development of the Mechanistic Model
The hydrophobic nature of the Teflon binder is expected Determination o f m a s s t r a n s p o r t e f f e c t s . - - B e f o r e ther-
to cause the product water to bead. This will repel the re- modynamic and kinetic equations can be developed, the
suiting liquid microdroplets out of the catalyst layer and effective concentrations of hydrogen and oxygen at the cat-
into the nonhydrophobic micropores of the carbon fiber- alyst surfaces must be determined. Three distinct mass
paper backing, through which they will flow to the gas flow transport interfaces exist in both electrodes: (I) between
channels. Here they will be entrained in the gas flow and the gas flow channels and the porous electrodes, (II) be-
removed from the fuel cell. While the Teflon is expected to tween the gas phase in the electrodes and any water film
be very effective in removing product water from the which may cover the catalyst sites, and (III) between the
catalyst layer, the agglomerate theory of Ridge e t al. 2 sug- w at er film and the catalyst surface. To define effective re-
gests that a microfilm of w a t e r wilt still coat the catalyst actant partial pressures or concentrations at each of these
particles. interfaces, a mass transport analysis is required.
Considering the proposed physical characteristics, the reactant
L E f f e c t i v e p r e s s u r e s in t h e f l o w c h a n n e l s . - - T h e
expected transport processes at the cathode are: (i) diffu- partial pressures in the inlet flow channels will vary with
sion of oxygen (in a mixture of 02, H2Ovap, and possibly N~) both the humidification level of the inlet streams, and the
through the wet-proofed pores of the porous backing layer, consumption rates of oxygen and hydrogen. The partial
(it) diffusion of dissolved oxygen through a water film in pressure of water vapor in the flow channels can be defined
the catalyst layer, and (iii) capillary and surface tension by the saturation vapor pressure at the fuel cell operating
induced flow of water microdroplets through nonwet- temperature
proofed pores in the carbon fiber backing layer. It is as- -- sat
PH2o.cathoao - PH~O [1a]
sumed that the nonwetproofed pores are closed to gas flow
at all current densities, such that the effective porosity of PHuO.anode= 0.5 - PH2os't [lb]
the cathode (in terms of gas flow) is constant. The assump-
tion of a constant effective porosity has been employed in The partial pressures of oxygen and hydrogen in the hu-
all of the mechanistic models reviewed thus far. ~-4 midified inlet gas streams can be determined from the dry
gas compositions and the water partial pressure
The anode and cathode gases are saturated with water
vapor in a humidifier, prior to entry into the flow channels hum (P - (PH20,~node)~d~y [2a]
o f the cell. Water from a heated reservoir is circulated PH2,inlet -- p •H2,inlet

through the humidifier, where both heat and humidity are

transferred to the feed gases through a semipermeable hum (P - (PK20,o~thode) a~y
Po2,inlet = p Po2,1nlet [2b]
membrane. The humidification water is also passed
through channels in the fuel cell for heating purposes (the The outlet dry gas composition at the cathode can be
single cell requires heating to maintain operating tempera- calculated as a function of the dry inlet mole fractions, the
tures in excess of 30~ prior to returning to the reservoir. oxygen stoiehiometrie ratio ( S R ) (the ratio of the inlet flow
For modeling purposes, the reactant streams exiting the rate of oxygen to the rate at which it is consumed), and the
humidifier are assumed to be saturated with water vapor at total pressure
the temperature of the humidification water exiting the
in d~2/
fuel cell (the temperature drop of the humidification water in dry /?202
across the fuel cell is less than 2~ which will henceforth Xo2 SR
p~td~= i~a~., . 9P [3]
be termed the fuel cell operating temperature. As a simpli- (~nd~ _ XO2 + X ~ dry
fying approximation, isothermal properties are assumed Xo~ SR ]
for the flow streams in the flow channels, and it is assumed
An identical calculation can be used to determine the dry
that the product wat e r generated at the cathode will be in
gas outlet hydrogen partial pressure by substituting hydro-
the liquid state and in equilibrium with the water vapor in
gen and carbon dioxide mole fractions and incorporating
the cathode gas stream. In actual fuel cell systems, and
the stoichiometric ratio for hydrogen. Equations 2a and b
especially in multicell stacks, thermal gradients are likely
can then be used to determine the humidified outlet gas
to apply, due to the heats of reaction at the electrodes, and
partial pressures for the fuel and oxidant flows by substi-
the product water is likely to be a mixture of liquid and
tuting outlet dry gas partial pressures for inlet dry gas par-
vapor. With the available apparatus, however, it was not
tial pressures.
possible to characterize these gradients or the state of the
In a mixed flow stream (e.g., 02, N2, and H20 in the
product water.
cathode or H2, CO2, and H20 in the anode), consumption of
It is expected that a gradient of water concentration will oxygen or hydrogen at the electrode will cause concentra-
exist across the electrolyte membrane, with the level of tion gradients to develop along the length of the flow chan-
hydration increasing from the anode to the cathode. This is nels. The shape of these concentration profiles will depend
expected to lead to diffusion of water from the cathode to on various factors, such as reaction kinetics, flow channel
the anode. A constant boundary condition of full hydration geometry, flow distribution, and flow velocity. If first-order
is assumed for the membrane at the cathode/membrane kinetics are assumed at the electrodes, ~5 the oxygen and
interface (i.e., the membrane at the interface is fully wetted hydrogen partial pressures will decline exponentially with
and will have a constant water content), due to the produc- respect to the residence time
tion of water in the cathode catalyst layer.
o e -k,
Po2 = Po2 [4a]
Due to the wat er concentration gradient in the mem-
brane, the anode side of the membrane is not expected to be P~2 = P~2 e k~ [4b]
fully wetted during fuel cell operation. As a result some
absorption of wat er from the humidified fuel feed is ex- If plug flow behavior is assumed as a first approximation,
pected. This will progressively reduce the vapor pressure of the exponential decay will depend on the velocity of gas
water in the anode gas a]ong the length of the anode flow flow in the flow channels. On the cathode side, where the
channel. ~3'~4The only direct mathematical use of the anode consumed oxygen represents a small fraction of the tota]
stream water vapor pressure in the model is in the calcula- flow (stoichiometric ratios are generally >1.75 using air,
tion of the thermodynamic equilibrium potential, where i.e., 733 % excess flow due to the nitrogen present), velocity
small deviations in hydrogen partial pressure are of negli- will be approximately constant. The effective oxygen par-
gible consequence. An effective anode stream water vapor tial pressure can then be approximated using a log-mean
pressure of 50% of the saturation vapor pressure is there- average of the inlet and outlet oxygen partial pressures
Downloaded on 2019-04-19 to IP 186.121.125.90 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 J. Electrochem. Soc., Vol. 142, No. 1, January 1995 9 The Electrochemical Society, Inc.
hum hum
avg P02,in -- P02,out Combining Eq. 13 and 14 and integrating yields
_b~O2 : hum [5]
in ~ hum I XN~t2er~aCe=Xw [ 4 R ; ~ I" ~e~ ]
\PO2,out/
' ' 'DN2,O2J [15]
In the anode flow channels, on the other hand, the decline
in flow velocity due to consumption and absorption is a where Q is the thickness of the gas diffuser portion of the
much more significant fraction of the total velocity, since electrode (-0.25 ram), xw a=el is the log mean average mole
the excess flow of fuel is typically quite small (SR = 1.15 to fraction of nitrogen in the humidified gas at the interface of
1.3). Consequently, T will increase along the length of the the cathode and the cathode gas flow channel, and x ~ terrace
flow channels and the consumption profile, Eq. 4a, will be is the effective mole fraction of nitrogen at the gas/liquid
less curved. For this model, an arithmetic mean is more film interface of the cathode catalyst layer. The pressure
likely, then, to suffice as a good first approximation of the diffusivity product has been estimated by Springer et al., 4
effective hydrogen partial pressure using the Slattery and Bird correlation ~
hum hum
avg PH2,in + PHmom
PH2 = 2 [6] PDr~2,o2 - 0.0002745 \ ~ /
~ O 2 ~N2
If either of the dry inlet gases is undiluted (i.e., no N2 or CO2 (pcrit~crit~l/3trpcritrpcrit~5/12[ 1 1 ]1/2
in the inlet stream), no average is required as the inlet and
outlet partial pressures will be equal 9
where e represents a correction factor for the diffusivity to
II. Diffusion in the electrodes.--When the effective reac-
account for the void fraction, tortuosity, and varying pore
tant partial pressures in the flow channels are specified,
cross sections of the electrode 9 Values for e of between 0.25
diffusion of the multicomponent gas streams through the
and 0.35 are commonly observed in the literature for car-
porous carbon electrodes can be described using Stefan-
bon fiber-paper electrodes) -3 For this model 9 has been set
Maxwell equations
to 0.3 as a simplifying approximation.
The effective partial pressure of oxygen at the gas/liquid
Vxi = R Tf (xiNj,g - xjNi,g) i = 1,2 ..... n [7] interface of the catalyst layer can now be calculated using
Eq. 8
where D~]~ is an effective binary diffusivity for the i-j pair in
the porous medium. For the purposes of the present model, p ~ t ~ e = p 9 [ 1 - XH~o
~t - XN~
~terfaoe] [17]
only one-dimensional diffusion normal to the face of the
Combining Eq. 15, 16, and 17 with appropriate values for
electrodes is considered.
the known constants yields 8
Two separate cases must be considered at each electrode.
In the first, the inlet flows contain an inert diluent (i.e., interface [ sat charmel (0.291. _/)]
nitrogen in the cathode flow or carbon dioxide in the anode Po2 =P- 1--XH~o--XN2 -exp\~jj [18]
flow). In the second case the flows contain no diluent.
On the cathode side, the inlet stream in the first instance A similar development for the interracial hydrogen partial
will be a mixture of oxygen and nitrogen, saturated with pressure on the anode side yields 8
water vapor at the fuel cell operating temperature, i.e.
interf. . . . [ sat channel (0 9 ~]
PH~ -P" I--0.5"XH20--Xc02 .exp\~]j [19]
XO2 -}- XN2 -b XH20,vap = 1
or [8]
In the second case, the cathode feed stream contains only
Po~ + P ~ + PrimO.yap= P
water vapor and oxygen. Now Eq. 8 indicates that the drop
The derivative of Eq. 8 gives the relationship along the axis in oxygen partial pressure along the diffusion path will
of oxygen diffusion into the electrode (i.e., along the gas result in a corresponding drop in the total pressure. In ad-
diffusion path through the electrode), i.e. dition, Eq. 9 indicates that the gradient in water vapor
mole fraction is no longer zero (as in Eq. i0). There will
dxo~ dx,~ + dxtt~o,va p = 0 [9] therefore be a molecular flux of water vapor "out" just
dz + ~ az sufficient to balance the bulk motion of the gas due to the
Assuming constant pressure in the flow channels and as- diffusion of oxygen "in." The net flux of water vapor
suming that constant temperature maintains the partial through the cathode is, therefore, still zero.
pressure of water vapor constant at the saturation vapor Again assuming that the Stefan-Maxwell equation de-
pressure scribes the process, Eq. 7 leads to

dx~o,v~ = 0 [10]
dz dz - P - b-~2o,o2
Therefore the gas-phase flux of water normal to the Substitution and rearrangement of this equation yields
cathode face can be set to zero
d(xo2) R- T. I
-- eft dz [21]
NH~O,~= 0 [11] (1-Xo2) 4.F.P.DH2o,o2
As nitrogen is inert and cannot be removed from the diffu- Integration of this expression from the flow channel to the
sion channel, the net nitrogen gas-phase flux normal to the catalyst surface, again taking the product PDi~ as a con-
cathode face is also zero stant, gives
(1 - - X o ~interface
)]
NN~,~ = 0 [12] R" T:I. Q
in (I -Xo2 oh~e~ - ~ [22]
)J 4 . F 9 P . DH20,O2
The one-dimensional Stefan-Maxwell equation for the ni-
trogen component, as derived from Eq. 7, can be simplified Since xo2 + X~2o = 1
using assumptions in Eq. 11 to 12 to give
i~te~f. . . . halo1 [ R . T . I . ed ]
XH2O =X~2o 9exp [4 .~w. .P. . ". DH2o,o2J~ [23]
d x ~ _ RTes (xN2No2,g) [13]
dz PDN~,o~
As the partial pressure of w at er vapor in the cathode is
The gas-phase flux of oxygen can be related to current fixed at the saturation level, the partial pressure of oxygen
density by at the interface is
sat
I p~zterface= (1 in te rfa c e ( Ph~b
No~,~- 4F [14] - XH~O ) ' i~t~fa~e [24]
\X.~o /
Downloaded on 2019-04-19 to IP 186.121.125.90 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 4 J. Electrochem. Soc., Vol. 142, No. 1, January 1995 9 The Electrochemical Society, Inc.

Combining Eq. 23 and 24 yields w a t e r film t h i c k n e s s w a s 8.1 • 10 -6 cm, t h e c o n c e n t r a t i o n

of d i s s o l v e d o x y g e n a t t h e c a t a l y s t s u r f a c e w o u l d b e zero.
Since existing data shows that significant voltages
( - 0 . 3 7 V) c a n b e a c h i e v e d a t 6000 A S F (using t h e D o w
m e m b r a n e ) , 1~ t h e w a t e r f i l m t h i c k n e s s c a n b e n o g r e a t e r
e x p ~4 9F . P . D ~ o , o / " XH~O
than 8.1 • i0 -6 cm and is likely to be significantly less. The
E v a l u a t i o n of t h e S l a t t e r y - B i r d e q u a t i o n for t h e P D H20,02
~ maximum current available using a Nafion membrane is
constant, using the appropriate critical values a n d con- limited to approximately 1200 ASF by the internal resis-
stants for a water/oxygen mixture finally yields ~ tance of the membrane. For a given set of operating condi-
tions, if it is assumed that the water film thickness varies
linearly with current density for a given electrode, then the
Po2 ...... tie9
- (P~o) {(4.192. I)~
xp \ ~ - ~
1
] 9XH20
~l
- 1 ] [26]
maximum film thickness which would be seen in a Nafion
cell would be on the order of i0 7 cm. A film of this thick-
ness will not present a significant barrier to oxygen diffu-
sion to the catalyst sites. Thus the surface oxygen concen-
A s i m i l a r d e v e l o p m e n t for h y d r o g e n o n t h e a n o d e side
tration Co* can be approximated using the interfacial
yields t h e e q u a t i o n 8
int d e v e l o p e d i n Eq. 30.
c o n c e n t r a t i o n Co2

.... [ 1
The Thermodynamic Equilibrium Potential
PH2 - {(1.653 I)~ oh~ T h e o v e r a l l r e a c t i o n i n a h y d r o g e n / o x y g e n fuel cell c a n
exp \ T1.334 ] "XH~o b e w r i t t e n as 18

III. D i f f u s i o n t h r o u g h a w a t e r f i I m . - - I t is n e c e s s a r y to H2(g ) + 21 O2(g) --> H20(aq) [31]

a n a l y z e t h e t r a n s p o r t of d i s s o l v e d g a s e s t h r o u g h a n y w a t e r
film w h i c h m a y c o v e r c a t a l y s t sites i n t h e a n o d e or c a t h o d e T h e t h e r m o d y n a m i c p o t e n t i a l of t h i s r e a c t i o n is d e f i n e d b y
c a t a l y s t layer. A r e q u i r e d f i r s t step is t h e d e t e r m i n a t i o n of t h e N e r n s t e q u a t i o n 3'~8
t h e l i k e l y t h i c k n e s s of s u c h a film, a n d its effect o n m a s s
t r a n s p o r t rates. It is f i r s t a s s u m e d t h a t a n y w a t e r l a y e r o n RT
E = E ~ - ~ In [p*2(p*2) ~ [32]
t h e a n o d e side will b e e x t r e m e l y t h i n , as n o w a t e r is b e i n g
p r o d u c e d a t t h e a n o d e , a n d t h e e l e c t r o l y t e m e m b r a n e is
w h e r e E% r e p r e s e n t s a r e f e r e n c e p o t e n t i a l a t u n i t activity,
v e r y retentive of a n y w a t e r it contains. This film is not
a n d t h e p a r t i a l p r e s s u r e t e r m s c o r r e s p o n d to t h e c o n c e n t r a -
expected to present a significant barrier to hydrogen diffu-
t i o n s c~2 a n d c~. T h e s t a n d a r d s t a t e d e f i n e s a s t a n d a r d s t a t e
sion to the catalyst, and it can be assumed that any dis-
r e f e r e n c e p o t e n t i a l E3 (equal to 1.229 V a t 298.15 K a n d
solved hydrogen gas at the anode will be in equilibrium
I atm), a n d E ~ will v a r y f r o m t h e s t a n d a r d s t a t e r e f e r e n c e
with the bulk gas. The effective concentration can be de-
i n a c c o r d a n c e w i t h t e m p e r a t u r e TM
fined by a Henry's law equation of the form

cH* : p~ter~ace E~ = E3 + ( T - To) (\ ~S~

nF / [33]
[28]
1.09. 106 exp (7) w h e r e To is t h e s t a n d a r d s t a t e t e m p e r a t u r e (298.15 K). T h e
e n t r o p y c h a n g e of a g i v e n r e a c t i o n is a p p r o x i m a t e l y c o n -
The Henry's law constant in the right denominator of Eq. 28 s t a n t (the v a r i a t i o n i n specific h e a t w i t h t h e e x p e c t e d
has been determined from available solubility dataJ v c h a n g e s i n t e m p e r a t u r e is m i n i m a l ) a n d c a n b e set to t h e
Contrary to the anode case, the production of water at s t a n d a r d s t a t e value. T h u s t h e r e f e r e n c e p o t e n t i a l v a r i e s
the hathode makes the existence of a significantly thick directly with temperature via TM
water layer much more probable. However, data for a MkIV [34]
E ~ = ~i + ~T
fuel cell using a Dow experimental membrane indicates
that current densities in excess of 6000 ASF can be where
achieved with pure hydrogen and oxygen feeds at 3.83 atm
abs and 358 KJ ~ Even at 6000 ASF there is no indication of ~1 = 1 . 2 2 9 V - 298.15 9AS~
nF [35a]
a diffusion limit to the current. Thus the limiting cathodic
current density must be greater than 6000 ASE Diffusion of
AS~ [35b]
oxygen through a thin water film would be defined by ~ = nF
Fick's law
U s i n g l i t e r a t u r e v a l u e s for t h e s t a n d a r d - s t a t e e n t r o p y
No2 = -Do2
dco2
dz
[29] c h a n g e , t h e v a l u e of [~2i n t h i s e q u a t i o n c a n b e c a l c u l a t e d to
b e - 0 . 8 5 • 10 3 V/K. 18 E q u a t i o n 32 c a n n o w b e w r i t t e n as
w h e r e t h e c o n c e n t r a t i o n of d i s s o l v e d o x y g e n at t h e g a s / l i q - RT
u i d i n t e r f a c e c a n b e d e f i n e d b y a H e n r y ' s l a w e x p r e s s i o n of E = 1.229 - 0.85- 10-3(T - 298.15) + ~ In [p*~(p32) ~ [36]
the form
F u r t h e r e x p a n s i o n of t h i s e q u a t i o n yields
interface por2tedace [30]
Co~ E = 1.229 - 0.85 . 10 -3 9 (T - 298.15) + 4.3085
lO o p
.10 ~.T .[ln(p*2)+l.ln(p~)] [37]
T h e H e n r y ' s l a w c o n s t a n t in t h e d e n o m i n a t o r is a g a i n
d e r i v e d f r o m r e a d i l y a v a i l a b l e s o l u b i l i t y d a t a f T h e diffu-
s i v i t y of o x y g e n i n w a t e r a t 358 K is a p p r o x i m a t e l y 8 •
10 -~ cm2/s~% F o r a n u n d i l u t e d o x y g e n / w a t e r i n l e t s t r e a m a t
Fuel Cell Voltage Losses
358 K a n d 3.83 a t m abs., w i t h a c u r r e n t d e n s i t y of 6000 A S F The combined effect of thermodynamics, mass transport,
(6.46 A/cm2), Eq. 26 d e f i n e s t h e i n t e r f a c i a l o x y g e n p r e s s u r e kinetics, and ohmic resistance determines the output
to b e 2.15 a t m abs. T h u s Eq. 30 d e f i n e s t h e i n t e r f a c i a l oxy- voltage of the cell, as defined by
g e n c o n c e n t r a t i o n as 1.7 • 10 6 m o l / c m < E q u a t i o n 14 d e - V = E + ~]act+ ~]ohmlc [38]
f i n e s t h e o x y g e n f l u x f o r a c u r r e n t d e n s i t y of 6.46 A / e r a 2 as
b e i n g 1.67 • 10 -5 m o t / e r a ~- s. T h e o n e - d i m e n s i o n a l c o n c e n - where ~ represents an overvoltage, and act is short for
t r a t i o n g r a d i e n t i n t h e w a t e r f i l m n o r m a l to t h e c a t h o d e activation. The effects of mass transport will be seen in all
face 4 is c a l c u l a t e d for Eq. 29 as 0.21 m o l / c m . T h u s if t h e three terms on the right side of this equation.
Downloaded on 2019-04-19 to IP 186.121.125.90 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 d. Electrochem. Soc., Vol. 142, No. 1, January 1995 9 The Electrochemical Society, Inc. 5

Activation Overvoltage nearest to the cathode, with constant full hydration, will
In the solid polymer fuel cell models found in the litera- have the lowest concentration of protons, while the mem-
brane region nearest to the anode will have the highest.
ture to date, the kinetics of hydrogen oxidation on plat-
Given the assumption of constant water concentration at
inum have been considered to be very rapidJ -~ The TafeI
the membrane/cathode interface, the proton concentration
slope [slope of overvoltage vs. log(i) plot] for the reaction at
in Eq. 43 will be constant, as will the terms k ~ A, n, F, R,
standard state (298.15 K, 1 atm) on p l a t i n u m catalysts is
AFt, and cq. The only remaining variables are the concen-
approximately 0.03 V/decade, ~ and its exchange current
tration of oxygen at the reaction sites, the current, and the
density is approximately 50 ASFP As a result, activation
overvoltages remain low even at high currents. The cathode temperature. Grouping constants, Eq. 43 can be expressed
in parametric form as
reaction, by contrast, is rather slow with a standard state
exchange current density in the range of i0 -~ ASF and a
~cto~thodi~: ~I + ~2T+ ~3Tln (c~2) + B4TIn (i) [46]
Tafel slope in excess of 0.07 V/decade) ~ Thus the activation
overvoltage at 298.15 K and a current density of I00 ASF where the constant parametric coefficients are
would be 0.009 V for hydrogen oxidation and more than
0.35 V for oxygen reduction. Consequently, the voltage B~ = (-AFD
drop due to activation can be predominantly attributed to cqnF [47a]
the cathode reaction.
When mass transfer effects are negligible, reaction kinet-
ics can be described by the Butler-Volmer equation ~ R in [nFAk~247 ~ *r r [47b]

i = i0 [ e - ~ n F ~ / R T - - e (1 =)nFB,'RT] [39] R
~ =~:~(1 - ~o) [47c]
Considering the general electrode process
R
aA + he- ~- bB [40] ~4 = e~onF [47d]

The exchange current io can be expressed as I~ The activation overvoltage of the hydrogen oxidation re-
action has been defined as z8
io=nFAk~ [exp [ - A F ~ ) (CA*)~c~-~)(CB~)~(~ [41] RT
~qactanod~c= ~ F In ( ~ ) [48]

where n is the number of electrons involved per mole of As the rate controlling step at the anode is likely to be
electrolyzed component in the rate controlling step, a and b chemisorption of hydrogen, Berger ~8 suggests that the ex-
represent the appropriate reaction orders of the rate con- change current can be defined as
trolling step with respect to the reactant and product, A is
the cross-sectional surface area over which the reaction is to, ~odir = 4FAk o
occurring, k ~ is the intrinsic rate constant, and AF~ is the
standard free energy of activation for the reaction.
For large values of ~], the second exponential in Eq. 39 is
negligible, yielding the Tafel expression
where AFe~ is the standard state free energy of activation
for chemisorption and AF~ is the standard state free energy
2.3. R T . io
9lact = --~- log ~ [42] of chemisorption from the gas stateY Thus the activation
overvoltage at the anode can be defined as
Given the relatively high overvoltage of the cathode reac-
tion, its kinetics can be described in this manner. Thus the ~{ln[4FAkOexp/-(AF~-O-5AFc)~(c,,
cathodic activation overpotential can be expressed as
...... 9 ~= \ -RT ) ~
1
exp \ ~ ) j - ln(i)p [50]
~(]act cathodic
= R T [ln ( n F A k ~
(~cn F L
)

The variation in hydrogen concentration in the inlet fuel

o=>
feed for most low pressure fuel cell applications is expected
to be relatively small (x~ > 0.75, 2 atm < P < 4 atm), and the
resulting change in activation overvoltage will be less than
where ao represents the cathodic transfer coefficient. The 0.010 V. Thus for most empirical modeling applications the
concentration of protons in the electrolyte membrane, c*§ hydrogen concentration term in Eq. 50 can be approxi-
will be fixed by water content and the law of eleetroneu- mated as a constant.
trality, which requires that The total activation overvoltage for a fuel cell is the sum
of the anodic and cathodic overvoltages, in parametric
form
ZmCm+ ~ ZlCi = 0 [44]
i
1]act total = "l]aet anodic q" "qact cathodic
where z~cm is the total charge on the solid membrane and
= ~1 + ~ T + ~3T[ln (c~2)] + ~4T [ln (i)] [51]
Ez~c~ is the total charge on mobile species in the electrolyte.
Protons are the only ionic species, and sulfonic acid ions where the constant parametric coefficients are
are the only charged membrane species. As Zm and z~ are
both constants ( - 1 and +1, respectively) _- (-A~o~ (-~reo~
~1 \ aonF ] +\ 2F I [52a]
c~ = ci [45]
R
~2 = ~ In [nFAk ~ (~)] + 2@ [In (4FAk ~
As the sulfonic acid ions are fixed to the backbone of the
membrane's polymer chains, the n u m b e r of these ions in a [52b]
given portion of dry membrane will be fixed. The dry mem-
brane is compressible, however, such that different parts of R
=~ (1 - ao) [52c]
the membrane can hold different concentrations of water,
and thus different concentrations of sulfonic acid ions,
while still being wet throughout. The membrane region ~ = - (~nF+~) [52d]
Downloaded on 2019-04-19 to IP 186.121.125.90 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 6 J. Electrochem. Sac., Vol. 142, No. 1, January 1995 9 The Electrochemical Society, inc.

Ohmic Resistance Losses of hydration across the membrane. Thus it is not possible to
Ohmic polarization is expected to result from resistance separate the various phenomena of water transport in the
to proton transfer in the solid polymer membrane and re- membrane so as to make an informed assessment of how
sistance to electron transfer in the graphite collector plates the three proton transport processes interact. Instead, it is
and the electrodes. Resistance to electron transfer is ex- simply observed that the water content and distribution in
pected to be approximately constant over the operating the membrane will be in some way dependent upon the
range (328 K -< T -< 358 K), and the resultant voltage loss operating current and temperature. (Diffusivities and wa-
can be determined using an Ohm's law equation of the form ter partial pressures vary with temperature, while proton
fluxes, and therefore water fluxes, will vary with the cur-
electronic " R electronic
~ohmic -- - i [53] rent). Equation 60 would thus indicate that the internal
w h e r e R eiectr~ is the resistance to electron transfer. resistance of the fuel cell will also be in some way depen-
dent upon the current and the temperature.
Resistance to p r o t o n t r a n s f e r t h r o u g h the m e m b r a n e
In the absence of an explicit mechanistic model, linear
c a n n o t be considered to be constant. The f l u x of protons
regression techniques are used to determine a best fit para-
t h r o u g h the m e m b r a n e n o r m a l to the electrode faces can be
metric model for the internal resistance. A model consider-
described using the N e r n s t - P l a n c k e q u a t i o n
ing the main, interaction, and quadratic effects of current
and temperature would have the form
Ni = - z i f T Dici OcP
c~z - Di OCi+
az civ [541
R i~ternal = ~i + y2T + ~3i + %Ti + ~sT 2 + ~5i 2 [61]
where q~ represents the electrical potential. The relation-
ship between the flux of mobile ions and the current den- 0.46 /
sity can be defined as I'
/m
/I
I = F ~ z~N~ [55] //
i .,p, 0,42 //
S u b s t i t u t i n g this e q u a l i t y into Eq. 54 and rearranging, one o
obtains P~
| /
0.38
--~Z -- K --K \ ~ i i~ + 7 - - ZiCi V [56] >
t- /
O /
t,p
w h e r e K is the m e m b r a n e conductivity, d e f i n e d as .o
F2 "0 0.34 7"
K = ~ T DH+CH+ [57] P
n

All three c o m p o n e n t s of Eq. 56 will be non-zero. An elec- "0

trical p o t e n t i a l g r a d i e n t will exist to create m i g r a t i o n , a 0
0.30 /
p r o t o n c o n c e n t r a t i o n g r a d i e n t will exist to create diffusion
of protons from the anode to the cathode, and diffusion and
absorption of water in the membrane will lead to a convec-
/"111
tive proton flux. Integration yields 0.26 . . . . . .
0.26 0.30 o.,2 0.46
h~)protons protons I ~m R T I dz E x p e r i m e n t a l Data V a l u e (Volts)
~- ~ohmic ----J 0 F2DH+CH+
Fig. 1. Parily plot for activation overvoltage of experimental data
foe~ RTdcH§ ~e~ R T v as a function of model predictions for the Ballard Mark IV solid
Rc~+dz dz + Jo FDH+dz [58] polymer fuel cell.
where ~ is the thickness of the membrane.
The total drop in potential due to ohmic resistance will
result from the combined resistance to electron and proton
"~" 54

y- /
transport and can be defined via an Ohm's law relation
total electronic protons ~ internal E
c-
~qoh~io= ~Qohmlc + ~Qohmic = -i 9 [59]
O
where ~internal is the total internal resistance of the fuel cell. ,'T
Assuming that the current normal to the cathode face is LU 5O
uniform in the membrane, and using Eq. 58 and 59, one
obtains
/
Rinten~a 1 Relect" RT fl ~n, dz t-
= + F~ DH, CH§ o 4~ /
R T (e= dc~+ , R T (~=vdz ._o i
-- 9 ~ a z + [60] "0
Fi Jo CE+dZ ~ Jo P
0_
The diffusivity a n d c o n c e n t r a t i o n of protons, as well as 42
the c o n v e c t i v e p r o t o n velocity, will all be d e p e n d e n t u p o n "0
the w a t e r content and d i s t r i b u t i o n in the m e m b r a n e . While o i

m a n y papers dealing w i t h issues of w a t e r d i s t r i b u t i o n in

iJ
the m e m b r a n e h a v e been presented, ~3'14'2~ there is a very
l i m i t e d u n d e r s t a n d i n g of the i n t e r a c t i o n of m i g r a t i o n , dif-
fusion, and c o n v e c t i o n processes and their overall effect on
w a t e r c o n t e n t and distribution. The c o m p l e x i t y of w a t e r F_.xpedmental D a t a V a l u e (1 E-4 O h m s )
d i s t r i b u t i o n arises f r o m the i n t e r d e p e n d e n c e of all of the
p a r a m e t e r s involved. With t h e e x p e r i m e n t a l a p p a r a t u s Fig. 2. ParitypIot for fuel cell resistanceof experimental data as a
used in our lab, there is no m e a n s of d e t e r m i n i n g the flux of function of model predictionsfor the Ballard Mark IV solid polymer
w a t e r v a p o r into the m e m b r a n e via absorption, or the level fuel cell.
Downloaded on 2019-04-19 to IP 186.121.125.90 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 J. Electrochem. Soc., Vol. 142, No. 1, January 1995 9 The Electrochemical Society, Inc. 7

0.96 %ct = -0.9514 + 0.00312T

/
/ - 0.000187T [ln (i)] + 7.4 9 10-5T [ln (c32)] [62]

and
Ri~tr = 0.01605 - 3.5 9 10-5T + 8.0 9 10-~i [63]
o
"-1
0.85 When coupled with Eq. 37 defining the thermodynamic
equilibrium potential, these p a r a m e t r i c models very accu-
~> rately defined fuel cell performance over the experimental
[..-

o range (328 K -< T -< 358 K, 0.6 atm abs. -< p ~ -< 3.1 atm
0
abs, 30 A S F -< I <- 450 ASF) as shown by the very good
"a
,,= agreement between experimental and model-generated
/ data displayed in Fig. 1 through 6. Figures 1 to 3 show
o_ 0.75 ,"
0
p a r i t y plots of experimental data v s . model predictions for
"o activation overvoltage, fuel cell resistance, and total out-
o put voltage. Figures 4 to 6 plot the predicted and experi-
mentally determined response in output voltage for various
independent changes in the operating current, tempera-
ture, and oxygen p a r t i a l pressure.
0.65 , , Conclusions
o.65
The very good agreement between the empirical model
ExperimentaJ Data Value (Volts) and the experimental data indicates the value of develop-
ing a mechanistic basis for empirical modeling exercises.
Fig. 3. Parity plot for total output voltage of experimental data as This approach is of considerable benefit in expanding the
a function of model predictions for the Ballard Mark IV solid polymer
applicability and validity of the empirical model produced.
fuel cell.
In spite of the assumptions and approximations employed
in the mechanistic approach, the predictions of the model
seem to reflect the actual trends found in the experimental
1.2
- - Model Prediction 0 Experimental Data

1.1 1.2-
Model at 328K ......... Modal at 343K - - - - Modal at 358K
[] Data at 328K • Data at 343K 9 Data at 358K
1 1
q

6)

g o.9
>

0
G)
0.8
~- 0.9
0

0"~ 0.7 -~a 0.8t

0.7
0.6 f)_

0.6
0 160 ' 260 360 460
0"50 160 260 ' 360 ' 460 ' s o o Current Density (ASF)
Current Density (ASF) Fig. 5. Predicted and experimental response in output voltage as a
function of temperature for the Ballard Mark IV solid polymer fuel
Fig. 4. Predicted and experimental response in output voltage as a cell.
function of operating current for the Ballard Mark IV solid polymer
fuel cell.
1.4
-- Model (21% 02) .......... Model (46% 02) -- Model (100% 02)
Da'~a at 21% 0 2 9 Data at 46% 0 2 X Data at 100% 02'
where the ~/ values represent constant parametric coeffi-
cients. Equation 61 shows the most expansive form which 1.2
a second-order p a r a m e t r i c model would take if the resis-
tance m a p p e d linearly with current and temperature. Such
a relationship will not necessarily apply, however, and it is
possible that the relationship to either of these variables
might be logarithmic, exponential, or otherwise.
>-- 0.8-
Empirical Application of the Model
If all of the parameters in Eq. 37, 51, 52, 59, and 61 were 0 ...............

known, they could be combined as shown in Eq. 38 to de- 0.6.

fine fully the output voltage of the fuel cell. Unfortunately,
not all of the parameters are directly available, and it is
necessary to use empirical means to estimate values for 0.4
these u n k n o w n terms. Details of this approach as well as
o 5o 16o i go 260 2go abo aiD .6o .so
Current Density (ASF)
the statistical evaluation of the resulting model are pre-
sented e l s e w h e r e Y Fig. 6. Predicted and experimental response in output voltage as a
The parametric equations developed from the empirical function of the partial pressure of oxygen for the Ballard Mark IV
analysis were solid polymer electrolyte fuel cell.
Downloaded on 2019-04-19 to IP 186.121.125.90 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
 8 J. Electrochem. Soc., Vol. 142, No. 1, January 1995 9 The Electrochemical Society, Inc.

data. As a result of its mechanistic basis, the empirical equil, equilibrium

model can be extrapolated beyond the experimental range H2 hydrogen
with much more confidence than would otherwise be justi- HfO water
hum humidified gas conditions
fied. Yet its application loses none of the ease and empirical in gas conditions at the inlet to the fuel cell
validity for which empirical models are known. Thus it is interface conditions at the interface of the gas diffusion
concluded that the effort involved in coupling mechanistic cathode and the water film surrounding the catalyst
and empirical approaches to model the performance of a sites
PEM fuel cell is worthwhile, and yields a final model which N2 nitrogen
will be both mechanistically valid and easily applied. O2 oxygen
out gas conditions at the outlet of the fuel cell
Manuscript submitted July i, 1993; revised manuscript sat saturation conditions
received July 25, 1994. This was in part Paper 89 presented * effective value
at the Toronto, ON, Canada, Meeting of the Society, Oct.
11-16, 1992. REFERENCES
The Royal Military College of Canada assisted in meeting 1. R. P. Iczkowski and M. B. Cutlip, This Journal, 127,
the publication costs of this article. 1433 (1980).
2. S.J. Ridge, R. E. White, Y. Tsou, R. N. Beaver, and G. A.
LIST OF SYMBOLS Eisman, ibid., 136, 1902 (1989).
cl concentration of species i, mol/cm 3 3. M. W. Verbrugge and D. M. B ernardi, AIChE J., 37, 1151
D~ diffusivity of species i, cmf/s (1991).
effective binary diffusivity coefficient of the i-j gas 4. T. E. Springer, T. A. Zawodzinski, and S. Gottesfeld,
pair, cmf/s This Journal, 138, 2334 (1991).
E thermodynamic equilibrium potential, V 5. H. Ide, T. Yoshida, H. Ueda, and N. Horiuchi, in Pro-
E ~ reference potential, V ceedings of 24th Intersociety Energy Conversion
E3 standard state reference potential, V Conference, pp. 1517-1522, IECEC (1989).
e- electrons 6. J. C. Amphlett, M. Farahani, R. F. Mann, B. A. Peppley,
F Faraday's constant, equal to 96,487 C/equiv. and R R. Roberge, in Proceedings of 26th Intersociety
Hi Henry's law constant for species i, atm 9cm3/mol Energy Conversion Conference, pp. 624-630, IECEC
I current density, A/cm 2 (1991).
Id thickness of the gas diffuser, cm 7. A. Cisar, ibid., pp. 611-616.
l~ thickness of the water film, cm 8. R. M. Baumert, M.Sc. Thesis, Queen's University,
i current, A Kingston, ON, Canada (1993).
io exchange current, A 9. J. C. Amphlett, R. M. Baumert, T. J. Harris, R. F. Mann,
K proportionality constant B. A. Peppley, and R R. Roberge, This Journal, 142, 9
N~ superficial gas flux of species i, mol/[(cmf)(s)] (1995).
n n u m b e r of equivalents involved in a reaction 10. D. S. Watkins, K. Dircks, D. Epp, and C. De La Franier,
P pressure, arm Paper presented at 4th Annual Battery Conference,
pi partial pressure of component i, atm Long Beach, CA, 1989.
R gas constant, 8.3143 J/(mol)(K), or other appropriate 11. D. S. Watkins, K. Dircks, D. Epp, and J. Blair, Paper
units presented at 5th Annual Battery Conference, Long
Rext external electrical resistance, Beach, CA, 1990.
Rin, internal electrical resistance, 12. D. S. Watkins, K. Dircks, D. Epp, a n d J . Blair, Can. Pat.
AS~~ standard state entropy change, kJ/mol Appl. CA 2,052,221 (C1.HolM4/94) (May 20, 1992).
T temperature, in Kelvin unless otherwise stated 13. T. Nguyen, J. C. Hedstrom, and N. E. Vanderborgh,
t time, s Proc. EIectrochem. Soc., p. 39, 1989.
xl mole fraction of species i 14. P. C. Rieke and N. E. Vanderborgh, J. Mem. Sci., 32,313
z distance in the "horizontal direction" (direction of (1987).
diffusion of gases), cm 15. A. Damjanovic, D. B. Sepa, and M. V. Vojnovic, Elec-
z~ charge on species i trochim. Acta, 24, 887 (1979) as cited in S. J. Ridge,
R. E. White, Y. Tsou, R. N. Beaver, and G. A. Eisman,
Greek This Journal, 136, 1902 (1989).
c~ transfer coefficient 16. J. C. Slattery and R. B. Bird, AIChE J., 4, 137 (1958) as
9 diffusivity correction factor = 0.3 cited in T. E. Springer, T. A. Zawodzinski, and S.
~l overvoltage, V Gottesfeld, This Journal, 138, 2334 (1991).
~]act overvoltage due to activation, V 17. Chemical Engineers' Handbook, 4th ed., R. H. Perry,
~ohm~c overvoltage due to ohmic resistance, V C. H. Chilton, and S. D. Kirkpatrick, Editors, Mc-
p~ viscosity of water, g/(cm)(s) Graw-Hill, Inc., New York (1963).
18. C. Berger, Handbook of Fuel Cell Technology, Prentice
Superscripts and Subscripts Hall, Englewood Cliffs, NJ (1968).
act activation 19. J. T. Wang and R. F. Savinell, Electrochim. Acta, 37,
channel conditions at the interface of the cathode gas 2737 (1992).
flow channel and the cathode itself 20. T. E Fuller and J. Newman, This Journal, 139, 1332
CO2 carbon dioxide (1992).
crit critical values 21. N. E. Vanderborgh, J. R. Huff, and J. Hedstrom, in Pro-
dry dry gas conditions ceedings of 24th Intersociety Energy Conversion
eft effective (i.e., average) conditions Conference, p. 1637 (1989).

Downloaded on 2019-04-19 to IP 186.121.125.90 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).