Fuller (1993)
Fuller (1993)
Fuller (1993)
ABSTRACT
A mathematical model of transport in a solid-polymer-electrolytefuel cell is presented. A two-dimensional membrane-
electrode assembly is considered. Water management, thermal management, and utilization of fuel are examined in detail.
Because the equilibrium sorption of water between the gas phase and the polymer-electrolyte depends strongly on temper-
ature, water and thermal management are interrelated. The rate of heat removal is shown to be a critical parameter in the
operation of these fuel cells.
As interest in electric vehicles increases, solid-polymer- Vanderborgh4 characterize the membrane-electrode inter-
electrolyte fuel cells are receiving renewed attention. Dra- face and present a photomicrograph of the region. There
matic improvements in the performance of single labora- are three phases present: a solid phase consisting of the
tory-scale cells with low platinum loading have been made porous graphite electrode and platinum catalyst, the solid
over the last few years. 1'2Because several interrelated pro- polymer electrolyte, and a gas phase. Descriptions of gas-
cesses occur simultaneously, the development of a mathe- diffusion electrodes and corresponding models can be
matical model is critical to the design and scale-up of solid- found in the literature? -7
polymer-electrolyte fuel cells. In particular, understanding Nguyen et al. 8 raise mass- and heat-transfer design is-
thermal effects, which are inseparable from the manage- sues in polymer-electrolyte fuel cells, but by and large the
ment of water in the polymer, is necessary to optimize the subject has not been treated in the literature. Recently,
Kimble and White 9 presented a detailed model of an alka-
performance of these fuel cells.
line fuel cell; solid-polymer-electrolyte fuel-cell models
One method of fabricating solid-polymer-electrolyte fuel have been developed by Bernardi and Verbrugge 1~ and by
cells is to hot-press two gas-diffusion electrodes onto a thin Springer et al. 13 In contrast to these, our model uses con-
polymer membrane above its glass-transition tempera- centrated solution theory, treats the two-dimensional
ture? This is shown schematically in Fig. 1. Leddy and membrane-electrode assembly, and accounts for thermal
effects.
* Electrochemical Society Student Member. We consider a membrane-electrode assembly operating
** Electrochemical Society Active Member. at steady state on air and reformed methanol. The fuel and
Downloaded on 2016-05-10 to IP 130.203.136.75 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. 140, No. 5, May 1993 9 The Electrochemical Society, Inc~ 1219
air streams are heated and humidified before entering their Model
respective channels where they are consumed in electro-
The cell sandwich is composed of five regions (Fig. 1) and
chemical reactions. Hydrogen gas diffuses through the
three phases. The species considered in the electrolyte are:
porous electrode and is oxidized on platinum catalyst sites
at the anode i n a three-phase region containing polymer the polymer with covalently bound sulfonic acid groups,
electrolyte, gaseous reactants, and carbon matrix. Oxygen hydrogen ions, and water. Transport in the polymer elec-
passes through the gas-diffusion electrodes to the cathode. trolyte is described with concentrated solution theory. The
At the cathode, the hydrogen ions react with oxygen at fundamental equation is
similar catalyst sites to form water. Water also absorbs into n
or evaporates from the membrane depending on its partial dj = ~ Kij(vj - vi) [1]
pressure.
After developing the model, we divide the analysis into where di is the driving force for transport and Ki/s are fric-
three areas: (i) constant-temperature study of a unit cross tional coefficients. Equation 1 may be put in the form of the
section of the fuel cell, (it) water management in a mem- Stefan-Maxwell equation; for n species
brane-electrode assembly at constant temperature with
ciVlii = ~-~xiJ l -xjJi
eocurrent flow of air and fuel, and (iii) combined material [2]
RT 3~~i cDIj
and energy balance for the membrane-electrode assembly.
We focus on the following water-management issues: the where the molar flux is given by
net transport of water, which must be supplied or recycled
thereby reducing the efficiency of the cell from the anode Ji = ci(v~ - Vrof) [3]
to the cathode; dehydration of the anode leading to high In developing Eq. 2, we introduced the molecular weight of
ohmic losses since the conductivity of the polymer elec- the polymer. However, the fundamental transport equa-
trolyte is a strong function of the degree of hydration, and tions are independent of molecular weight, and this choice
flooding of the cathode.
is somewhat arbitrary. We assumed the molecular weight of
It is common practice to saturate the inlet gases with the polymer is equal to its equivalent weight. More details
water at a temperature higher than the operating tempera- of the membrane-transport model and on the measurement
ture of the cell (producing a mist). Since water is produced of their transport properties are given by Fuller and New-
at the cathode, it is not obvious that additional water need man. ]4-16
be supplied to the fuel cell. We hope to elucidate when and There are three species in each gas stream: nitrogen, oxy-
why additional water might be needed to enhance the per- gen, and water vapor in the cathode; carbon dioxide, hy-
formance. drogen, and water vapor in the anode. The Stefan-Maxwell
In addition, the utilization of hydrogen is an important equations also can be used to describe multicomponent dif-
factor in the optimization of these fuel cells. One could fusion of gases. 17For convenience, the reference velocity is
circumvent water and thermal management problems by set to zero (that of the nondiffusing gases, carbon dioxide
increasing the flow rates of the gas streams. This, however, and nitrogen, and the polymer).
affects the utilization of fuel, the net transport of water Equation 2 is combined with a material balance on spe-
across the cell, and the efficiency of the fuel cell. cies i
electrode
I
gas-dif fuslon I
1
reaction
zone
reaction gas-diffusion
~ z~x~ = 0 [7]
zone electrode i=l
I
I
I Within the reaction zones, we consider two chemical reac-
tions
at the anode H2 ~- 2H § + 2e-
Z
at the cathode O2 + 4H § + 4e- ~ 2H20
Fig. 1. Membrane and electrode assembly. The five regions of the
model are shown (not to scale). and H20(1) ~---H20~)
Downloaded on 2016-05-10 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
1220 J. Electrochem. Soc., Vol. 140, No. 5, May 1993 9 The Electrochemical Society, Inc.
representing the absorption and evaporation of water from Table I. Gas composition at the boundaries.
the membrane. Electron-transfer reactions are described
by Butler-Volmer kinetics, using the bulk concentration of Cathode-gas Anode-gas
each species in their respective phases.
The problem is two dimensional, but the large aspect xn~0 0.40 XHzo 0.40
ratio allows us to consider transport only in the y direction xN~ 0.48 Xco~ 0.15
Xo2 0.12 XH~ 0.45
in the cell sandwich (Fig. 1). Although this is not strictly
correct, the expected improvements in results do not war- Parameter Value
rant the additional complexities of a more detailed two-di-
mensional model. We solve for transport in the y direction T 353.0 K
at a given value of z and integrate down the membrane- L 0.0175 m
electrode assembly in the z direction. The gas outside the p 1.0 bar
gas-diffusion electrode is assumed to be of uniform compo- em 0.20
e~ 0.40
sition in the y direction. For large P~clet number, axial
diffusion is neglected. A material balance on species i in the
anode gas channel, for example, gives Appendix summarizes the thermodynamic and kinetic
data and the transport properties used in the simulation.
r i -- F ~ = Jily=L5 dz [8]
Results and Discussion
Unit cross s e c t i o n . - - First, we examine a unit cross sec-
References 18 and 19 discuss general energy balances for tion of the fuel cell. The temperature is fixed, and we
electrochemical systems. We assumed that temperature is a specify the mole fractions of the gaseous species at the ends
function of z only. The work done by the system is iV. The of the two gas-diffusion electrodes. The current-voltage
rate of external heat transfer to the system may be ex- curve for the conditions of Table I is shown in Fig. 2. For all
pressed by the simulations the thickness of the gas diffusion electrodes
q = -h(T- TA) [9] was 360 ~m, and the thickness of the reaction zone was
10 ~m. At low current density, ohmic losses are small, and
Application of the first law of thermodynamics to the fuel polarization at the cathode controls the cell potential. As
cell gives the current density increases, ohmic losses grow, and the
curve becomes roughly linear; in this region the slope is
-FTAH=f$ h(T- TA)dZ+ f0z V i d z [10] inversely proportional to the electrical conductivity. Be-
cause the water content, which determines the conductiv-
ity of the membrane, increases with current density, the
where AH is change in enthalpy per mole of gas and FT is the curve is slightly concave upward (the water profiles in
total molar flow rate of gas per unit width of the mem- Fig. 3 clarify this effect).
brane-electrode assembly. Thus, the energy released by the Ticianelli et aI. 3 presented experimentally measured
overall chemical reaction is split among: electrical work, curves. They observed what appear to be mass-transfer
change in enthalpy (and hence temperature) of the flowing limitations in experimental cells, a sharp drop in current
gas streams, and external heat transfer. density at low cell potentials. The model does not predict
At each end of the fuel-cell assembly, y.= 0 and y = L5, the mass-transfer limitations in the gas-diffusion electrodes
mole fractions of the gaseous species are determined from until current densities of about 5 A/cm 2. The model, how-
a material balance in the z direction of the gas channel (Eq.
ever, does not include adsorption and mass transfer of hy-
8). At a phase boundary, the fluxes of the species within that
drogen or oxygefi in the water or polymer film, which could
phase are set to zero. At L4, for example, the fluxes of
limit the current density. The mass-transfer limitations
electrolyte species (water, hydrogen ions, and membrane)
are zero. Where there is a discontinuity in the porosity, the may be the result of flooding; as the pores fill up with water,
concentration and superficial fluxes are continuous. There- the volume fraction of the gas decreases, and access to the
fore, at L3, the fluxes of gas species are set to zero, and the catalyst may be restricted.
concentration and superficial fluxes of electrolyte species
are matched. The solid carbon phase is assumed to be [9 i i i i i i
tempt a parametric study. Instead, using the best available 0 200 400 600 800 1000 1 200
data, we examine the complex relationship between the Current density, mA/cm 2
operating conditions and transport in the fuel cell. Where
possible, the parameters were obtained from our experi- Fi~. 2. Cell potential vs. current density for unit cell. Operating
mental results or from data available in the literature. The conditions are fisted in Table I.
Downloaded on 2016-05-10 to IP 130.203.136.75 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. 140, No. 5, May 1993 9 The Electrochemical Society, Inc. 1221
. . . . i . . . . i . . . . i . . . . i . . . . i
V = 0.72 V
11
13
i=1150 mA/cm 2
z = 12 era ~
12
10 II
'~ ~'~ 10
9
9 ~ 6
8 0 ~ 7
0.~
, , , , 1 , , , , I , , , , I ,
Figure 3 presents profiles of water content in the mem- concentration gradient established that counteracts the
brane at different current densities, k is the number of electro-osmotic drag of water and reduces the net transport
moles of water per sulfonic acid group. At the cathode, of water across the cell.
water is produced in stoichiometric proportion to the cur- The mole fractions of hydrogen, oxygen, and water as a
rent; and with increasing current, more water is dragged function of axial position are shown in Fig. 5. The local
from the anode. A concentration gradient builds up to current density is depicted by the solid line in Fig. 5. At
counteract this electro-osmotic drag. small z, there is a large net flux of water from the anode to
As the current density grows, one observes an increase in the cathode gas stream. Consequently, the mole fraction of
the water content at the cathode and dehydration at the
hydrogen is nearly constant for the first couple of centime-
anode. A/though the flux of water across the membrane is
ters. Initially the current density decreases; but as the
constant, because of variations in physical properties, the
membrane becomes more hydrated, the conductivity in-
water profiles are not straight. There is little resistance to
mass transfer in the gas phase or in the membrane; conse- creases substantially, and the local current density rises.
quently, even at high current densities, dehydration of the Figure 6 shows the composition of water in the two gas
anode is slight. streams. We have introduced the variables
Isothermal c a s e . - - N e x t , we consider a two-dimensional Fi xi [11]
membrane-electrode assembly (Fig. 1), with cocurrent flow Xi - Fmen - x ~
of air and fuel streams. The temperature is constant, but as
the gas streams flow down the channels, fuel and oxygen where inert refers to the nonreacting gas in each stream:
are consumed, and water is produced at the cathode. There- nitrogen for the cathode and carbon dioxide for the anode.
fore, with the electrode potential fixed, the local current
density and flux of water change. Here we account for these 9 ' ' ' ' ' " ' " ' " 4 1 5
Parameter Value
9 i , i . i . i . i , i , i . i . q , i , i .
0 1 2 3 4 5 6 10 11 12
7 s 9
T 353.0 K
L 0.0175 c m z, distance from entrance (cm
P 1.0 b a r
Em 0.20 Fig. 5. Mole fraction of hydrogen, oxygen, and water in the gas
0.40 channels. Local current densi~ is shown by the solid line; V= 0.72 %',
1.0 x 10 -2 m o l / m 9s and f = 1.0.
Downloaded on 2016-05-10 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
1222 J. Electrochem. Soc., Vol. 140, No. 5, M a y 1993 9 The Electrochemical Society, Inc.
z2
i i I 4 I I I
a"
f=l.O_ f=l .-"
2.0
1.8
x.'..
,//,, f=l. 5
Cq
1.6
//I
/
,--I -0.01 /
.... anode o
,.4 -- cathode / /
t/
, ' / V= 0.72 V
V= 0.72 V b
1.2
-0.02
f=l.0
1.0
0.8
, I , ~ , i . i , i , i , i , i , i , i . i , i , i , i ...................
' 2 3 4 5 6 7 8 9 1'0' I'I.....
'2 13 1'4
0 a 2 3 4 5 6 7 8 9 10 11 m 13 14 z, distance from entrance (cm)
z, distance from entrance (cm)
Fig. 8. Net flux of water across membrane. A value of the flux less
Fig. 6. Composition of water in gas channels; f is the air-to-fuel than zero indicates transport from the anode to the cathode; f is the
ratio. Simulation is stopped at the value of z where the cathodegas air-to-fuel ratio. Simulation is stopped at the value of z where the
stream reaches saturation with respect to water. Utilization of hydro- cathode gas stream reaches saturation with respect to water.
gen is then 0.52 and 0.67 far f = 1 and 1.5, respectively.
Downloaded on 2016-05-10 to IP 130.203.136.75 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. 140, No. 5, May 1993 9 The Electrochemical Society, Inc. 1223
not appear practical for transportation applications. Al- o.o o., o; o.4 o; o16 o,
though the model is not able to simulate countercurrent u, fractional utilization of H2
flow of fuel and air streams, our results suggest that this
could be helpful. Finally, we note that while the pressure Fig. 10. Local current density as a function of local utilization of
drop in the z direction would increase, the length of the hydrogen with V, the cell potential, fixed, and the heat-transfer coef-
assembly scales linearly with increasing flow rate. ficient as a parameter; f = 1.
Nonisothermal case.--Temperature has a profound effect
on equilibrium and kinetic properties. In this last section, rent densities, the cell would dehydrate, and the perfor-
we assumed that there are no temperature variations across mance would be poor.
the assembly (the y direction) and that there is no transfer In contrast to the isothermal water profiles of Fig. 4, for
of heat in the z direction. The temperature of the gases at h = 30 W/m s. K, the temperature initially rises sharply with
the inlet is specified, and here we calculate the temperature increasing axial distance, and water evaporates from the
in the z direction based on an energy balance. Heat is re- membrane (Fig. 9). At constant cell potential, the loss of
moved by both an increase in temperature of the flowing water from the membrane (and the consequent decrease in
gases and by heat transfer to an external medium at a tem- conductivity) results in lower current densities and the
perature of 298 K. temperature levels off. Farther down the channel, the tem-
Figure 9 shows water profiles across the membrane with perature decreases slowly, and the water content in the
z as a parameter. The operating conditions are shown in membrane increases (dashed line in Fig. 9). Table IV shows
Table II but also include a heat-transfer coefficient of 30 W/ the average water content of the membrane and the corre-
m s . K. The current density as a function of local utilization sponding membrane resistance, which could be measured
of hydrogen is shown in Fig. 10; the corresponding temper- with a high frequency alternating current.
ature profile is depicted in Fig. 11. The effect of the heat-transfer coefficient on the perfor-
The hydration of the membrane is sensitive to the rate of mance is seen in Fig. 10. For h = 36 W / m 2 - K, the cell
heat removal. Because of the low thermal capacity of the
temperature rises slowly, and the performance is good.
gas streams, the steady-state temperature rises quickly
With a slightly lower value of the heat-transfer coefficient
without adequate removal of heat. At even moderate cur-
(h = 30 W/m s 9K), the temperature increases more sharply,
the membrane dehydrates, and the performance decreases
9 - ' ' " " ' . . . . ' . . . . ' substantially. With little or no removal of heat (h = 10 W/
rp
36o
iJ
: /
h = 30 W/m2-K I /
6
i /
7 --_ ~ V : 0.72 V - - - 36 W / m 2 _ K
d358
14
~ / -- --30
/ .... I0
4J
r~ I /
' /
356 "I
5.0
5
354 I / / ~
, I ; I ' , I , '
40 60 8 lO0 120 140 160
! r ] f
y, distance from cathode (gm) 0 2 3 4 5 6 7 8 9 10
Fig. 9. Concentration profiles of water across the membrane, h is ZI distance from entrance (cm)
the number of water molecules per sulfonic acid group; V= 0.72 V,
and z, the distance from the top of the channel, is a parameter. Fig. ! 1. Temperature in gas channels; f = 1 and V = 0.72 V.
Downloaded on 2016-05-10 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
1224 J. Electrochem. Soc., Vol. 140, No. 5, May 1993 9 The Electrochemical Society, Inc.
Downloaded on 2016-05-10 to IP 130.203.136.75 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.
Alternative Proxies: