International Journal of Hydrogen Energy 32 (2007) 2305 – 2313

www.elsevier.com/locate/ijhydene

Parametric study of solid oxide steam electrolyzer for hydrogen production

Meng Ni, Michael K.H. Leung ∗ , Dennis Y.C. Leung
Department of Mechanical Engineering, The University of Hong Kong, Pokfulam Road, Hong Kong, China

Received 21 February 2007; accepted 1 March 2007

Available online 19 April 2007

Abstract
A theoretical model was developed to study the electrical characteristics of a solid oxide steam electrolyzer (SOSE) for hydrogen production.
The activation and concentration overpotentials at the electrodes as well as the ohmic overpotential at the electrolyte were considered as the
main sources of voltage loss. The Butler–Volmer equation, Fick’s model, and Ohm’s law were applied to characterize the overpotentials.
The theoretical model was validated as the simulation results agreed well with the experimental data from the literature. In the study of the
component thickness effect, anode-support SOSE configuration was identified as the most favorable design. Further parametric analyses were
performed to study the effects of material properties and operating conditions on the anode-supported SOSE cell performance. The results
have shown that increasing electrode porosity and pore size can reduce the voltage loss. In the operation, both temperature and steam molar
fraction can be increased to enhance the SOSE electrical efficiency. The pressure should be regulated depending on the current density. The
electrochemistry model can be used to perform more analyses to gain insightful understanding of the SOSE hydrogen production principles
and to optimize the SOSE cell and system designs.
䉷 2007 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.

Keywords: Solid oxide steam electrolyzer; Hydrogen production; Electrochemistry model; Porous media

1. Introduction a relatively low-energy efficiency. Therefore, expensive solar

electricity generation for sufficiently large output is needed
Hydrogen has been identified as a promising fuel for sustain- to support the renewable hydrogen production. Alternatively,
able energy supply provided that renewable energy is used to high-temperature (873–1273 K) solid oxide steam electrolyzers
derive the hydrogen from its source [1–5]. Efficient and environ- (SOSE) are more advantageous as hydrogen can be produced
mental production of hydrogen becomes the key to the success at a higher chemical reaction rate with a lower electrical en-
of hydrogen economy. Hydrocarbon reforming, the present ma- ergy requirement [9–13]. Therefore, SOSE hydrogen produc-
jor hydrogen production method, is neither renewable nor clean tion has much potential for large-scale hydrogen production
[6]. Using solar energy, hydrogen can be produced in a more [14–16].
environmental-friendly manner from water by thermochemi- Most research works done on SOSE hydrogen production
cal cycles, photocatalysis, or electrolysis. Presently, both ther- were experimental studies with an emphasis on material de-
mochemical and photocatalytic hydrogen production methods velopment to cope with the high operating temperature and
have too low efficiency to be economically competitive [7]. Wa- strongly reducing/oxidizing environments for stable and effi-
ter electrolytic hydrogen production is hence a more practical cient operation [9–16]. The present literature is lacking theo-
technology for large-scale renewable hydrogen production [8]. retical modeling studies of SOSE hydrogen production. In this
Proton exchange membrane (PEM) electrolyzers and alkaline paper, a complete electrochemistry model was developed to
electrolyzers, operating at a low temperature (< 373 K), have characterize the current–voltage (J .V ) characteristics of SOSE
cells. Parametric analyses were performed to determine the ef-
∗ Corresponding author. Tel.: +852 2859 2628; fax: +852 2858 5415. fects of material properties and operating parameters on the cell
E-mail address: [email protected] (M.K.H. Leung). J .V characteristics.

0360-3199/$ - see front matter 䉷 2007 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2007.03.001
2306 M. Ni et al. / International Journal of Hydrogen Energy 32 (2007) 2305 – 2313

2. Model development surfaces, respectively. The standard potential E0 in volt (V) can
be written as [17]:
The fundamental mechanisms of SOSE are illustrated in
E0 = 1.253 − 2.4516 × 10−4 T . (4)
Fig. 1. Steam is fed to the porous cathode. When required elec-
trical potential for water-splitting is applied to the SOSE elec-
trodes, water molecules are dissociated to form hydrogen gas 2.2. Concentration overpotential
and oxygen ions at the cathode–electrolyte interface. The hy-
drogen produced can be collected from the cathode flow chan- The concentration overpotential is caused by the resistance
nel. The oxygen ions are transported through the electrolyte to to the transport of reactant species approaching the reaction
the anode and then oxidized to form oxygen molecules. The site and the transport of product species leaving the reaction
net reaction of SOSE can be expressed as: site. For a SOSE cell, the concentration overpotentials can be
expressed in terms of the gas concentration difference between
H2 O → H2 + 21 O2 . (1) the electrode surface and the electrode–electrolyte interface as
follows:
The required voltage applied to the SOSE can be written as:
I 
RT PH2 · PH02 O
conc,c = ln , (5)
V = E + conc,c + conc,a + act,c + act,a + ohmic , (2) 2F PH02 · PHI 2 O
⎡
1/2 ⎤
where E is the equilibrium voltage; conc,c and conc,a are the RT POI 2
concentration overpotentials at the cathode and anode, respec- conc,a = ln ⎣ ⎦, (6)
tively; ohmic is the ohmic overpotential; and act,a and act,c are
2F PO02
the activation overpotentials at the anode and cathode, respec-
tively. It is noted that the term “overpotential” is often called where P 0 and P I are the partial pressures at the electrode sur-
“polarization” in electrochemistry. face and electrode–electrolyte interface, respectively; and sub-
scripts H2 O, H2 , and O2 represent the steam reactant, hydrogen
and oxygen products, respectively.
2.1. Equilibrium voltage As the chemical reaction takes place at the electrode–electro-
lyte triple phase interface [18,19], the mass of each participating
The equilibrium voltage of a SOSE cell can be expressed by component in the porous electrode is conserved in a steady
the Nernst equation [9], state [20],
⎡  1/2 ⎤
P 0 P0 ∇ · Ni = 0, (7)
RT ⎢ H2 O2 ⎥
E = E0 + ln ⎣ ⎦, (3) where Ni is the rate of mass transfer for species i. In the SOSE
2F PH02 O
porous electrodes the mass transport Ni is mainly by means
of diffusion [17–19]. The diffusion process, which is driven
where R is the universal gas constant (8.3145 Jmol−1 K −1 ); F by concentration gradient or partial pressure gradient, can be
is the Faraday constant (9.6485 × 104 Cmol−1 ); T is the ab- determined by Fick’s model [20]:
solute temperature (K); and PH02 O , PH02 , and PO02 are the par-
tial pressures of steam, hydrogen, and oxygen on the electrode Dieff j(yi P ) D eff j(Pi )
Ni = − =− i , i = 1, 2, . . . , (8)
RT jx RT jx
where P is the operating pressure; Pi , yi , and Dieff are the par-
tial pressure, molar fraction, and effective diffusion coefficient
of species i, respectively. Substituting Eq. (8) into Eq. (7), the
equation governing the mass transport of species i can be ob-
tained:

j Dieff j(Pi )
− =0
jx RT jx
(i = H2 O for cathode, i = O2 for anode), (9)
where x is the depth measured from the electrode surface.
At the cathode–electrolyte interface (x = dc , the thickness
of cathode), the diffusing rate of H2 O to the interface, equal
to H2 O consumption rate in a steady state, is governed by the
electrical current density J. Thus, the mass transfer rate can be
expressed as
J
NH2 O = . (10)
Fig. 1. Hydrogen production by SOSE cell. 2F
 M. Ni et al. / International Journal of Hydrogen Energy 32 (2007) 2305 – 2313 2307

Substituting Eq. (10) into Eq. (8), the boundary condition for At the anode side, oxygen is the only gas transporting in
the governing Eq. (9) can be written as: the porous electrode layer. At the anode–electrolyte interface
 (x = da , the thickness of anode), the rate of O2 diffused away
jPH2 O  J RT
=− . (11) from the interface, equal to O2 generation rate in a steady state,
jx x=dc 2F DH eff
is also governed by the electrical current density,
2O

The Dirichlet boundary condition can be applied for the cathode J

surface: NO2 = − . (17)
4F
PH2 O |x=0 = PH02 O . (12) Similarly, the boundary condition of Eq. (9) for the anode can
be obtained as
Solving Eq. (9) with the above two boundary conditions, the 
jPO2  J RT
partial pressure of H2 O at the cathode–electrolyte interface can
 = . (18)
be obtained: jx x=da 4F D effO2
RTJ The Dirichlet boundary condition can be applied for the anode
PHI 2 O = PH02 O − eff dc . (13)
DH O 2F 2
surface,

As the steam molar consumption rate is equal to the hydrogen PO2 |x=0 = PO02 . (19)
molar generation rate, the cathode total pressure (P = PH2 +
Solving Eq. (9) with boundary conditions expressed in Eqs.
PH2 O ) is constant. Therefore,
(18) and (19), the partial pressure of O2 at the anode–electrolyte


RT J interface can be obtained,
PH2 = P − PH2 O = P − PH2 O − eff
I I 0
dc
DH O 2F RT J
2 POI 2 = PO02 + d .
eff 4F a
(20)
RT J DO
= PH02 + eff dc . (14) 2

DH O 2F The anode concentration overpotential can thus be expressed

2
as:
Substituting Eqs. (13) and (14) into Eq. (5), the cathode con- ⎡
1/2 ⎤
centration overpotential can be obtained: RT RT J
 0 conc,a = ln ⎣ 1 + eff 0 da ⎦. (21)
RT (PH2 + J RT d c /2F D eff 0
H2 O )PH2 O
2F DO PO 4F
conc,c = ln 2 2
2F PH0 (P 0 − J RT d c /2F D eff H2 O )

2 H2 O  eff can be determined in a way similar to D eff
The value of DO H2 O
RT 1 + J RT d c /2F D eff
H2 O PH2
0
[21,22].
2

= ln . (15)
2F 1 − J RT d c /2F D eff 0
H2 O PH2 O
2.3. Activation overpotential
The diffusion in the porous SOSE electrodes is mainly based
on two mechanisms, namely, molecular diffusion and Knud- The activation overpotential is a measure of the activity of the
sen diffusion. Molecular diffusion is the dominant mechanism electrodes. It represents the overpotential required for electro-
if the pore size is much larger than the mean free path of chemical reaction. The relationship between the electrode ac-
the molecular species. In this case, the molecule–molecule in- tivation overpotential and the current density can be expressed
teraction governs the diffusion process. On the other hand, if by the Butler–Volmer equation:
the pore size is much smaller than the mean free path of the     
species, the molecule–pore wall interaction dominates over the zF act,i (1 − )zF act,i
J = J0,i exp − exp − ,
molecule–molecule interaction. Thus, Knudsen diffusion be- RT RT
comes an important mechanism. In most porous structures, both i = a, c, (22)
mechanisms are significant. The effective diffusion coefficient
where J0,i is the exchange current density; subscripts i = a and
of steam can be expressed by combining these two diffusion
i = c represent the anode and cathode, respectively; act is the
mechanisms using the Bosanquet formula [21,22]:
activation overpotential;  is the symmetrical factor or charge


1  1 1 transfer coefficient; and z is the number of electrons produced
eff
= + , (16) per reaction. For water electrolysis,  and z are found to be 0.5
DH O
n DH2 .H2 O DH2 O,k
2 and 2, respectively [18,23]. The activation overpotential of an
where /n is the ratio of cathode tortuosity to porosity; electrode can be expressed explicitly as:
/(nD H2 .H2 O ) is the reciprocal of the effective molecular dif-  
RT J
fusion coefficient for H2 O.H2 binary system; and /(nD H2 O,k ) act,i = sinh−1
F 2J0,i
is the reciprocal of the effective Knudsen diffusion coefficient ⎡  ⎤
 2
for steam. The values of DH2 O,k and DH2 .H2 O can be easily RT J J
obtained by the kinetic theory and Chapman–Enskog theory, = ln ⎣ + + 1⎦ , i = a, c. (23)
F 2J0,i 2J0,i
respectively [21,22].
2308 M. Ni et al. / International Journal of Hydrogen Energy 32 (2007) 2305 – 2313

In electrochemistry, J0,i represents the readiness of an elec- recommendation, a and c are found to be 2.051 × 109 and
trode to proceed with electrochemical reaction. At the cathode, 1.344 × 1010 Am−2 , respectively.
important electrochemical reaction steps include adsorption of
H2 O, charge transfer, desorption of H2 , and incorporation of 2.4. Ohmic overpotential
oxygen ion into the electrolyte. At the anode side, the key
reaction steps are charge transfer and desorption of O2 . The The electrical connecting plates and electrodes generally
above steps depend on not only the operating parameters, i.e. have much higher electrical conductivity than the electrolyte.
temperature and reactant/product partial pressures, but also the Therefore, their effect on the overall ohmic overpotential can
material parameters, i.e. triple phase boundary (TPB) length. be neglected [28]. According to Ohm’s law, the ohmic overpo-
Previous studies have shown that the reactant/product concen- tential of SOSE can be expressed in terms of the properties of
trations have small effect on the exchange current density, while the electrolyte by [28]:
the temperature effect was more significant [24,25]. Consider-  
10300
ing the temperature effect, the exchange current density can be ohmic = 2.99 × 10−5 J L exp , (26)
expressed as: T
  where L is the thickness of the electrolyte in micron (m).
Eact,a
J0,a = a exp − , (24)
RT 3. Model validation and temperature effect
 
Eact,c
J0,c = c exp − , (25) The theoretical electrochemistry model of SOSE presented
RT
in the previous section was used to simulate experimental
where a and c are the pre-exponential factors of anode results for model validation. Among the limited experimental
and cathode, respectively; Eact,a and Eact,c are the activation studies published in the literature, Momma et al.’s work in
energy levels at the anode and cathode, respectively. The SOSE hydrogen production was most suitable for validation
exchange current density is the current density at zero over- of the present model as detailed experimental setup and pro-
potential, where forward and backward reaction rates are the cedures were clearly described [29]. In these experiments, the
same. Therefore, the values of solid oxide fuel cell (SOFC) ex- J .V characteristics of hydrogen production by electrolyte-
change current densities can be used as approximates for steam supported planar SOSE discs were measured. The electrolyte,
electrolysis [26,27]. The values of Eact,a and Eact,c are found cathode, and anode were made of yttria stabilized zirco-
to be 1.2 × 105 and 1.0 × 105 Jmol−1 , respectively [24,25]. nia (YSZ), nickel–YSZ (Ni–YSZ), and mixture of YSZ and
Chan et al. [18] recommended the values of J0,a and J0,c to be strontium-doped lanthanum (LSM–YSZ), respectively. The
2000 and 5300 Am−2 , respectively, at 1073 K. Adopting these electrolyte, anode, and cathode were 1000, 100 and 100 m

Table 1
Parameters used in SOSE electrochemistry modeling analysis

Parameter Value

Operating temperature, T (K) 1273

Operating pressure, P (bar) 1.0
Pre-exponential factor for anode exchange current density, a (Am−2 ) 2.051 × 109
Pre-exponential factor for cathode exchange current density, c (Am−2 ) 1.344 × 1010
Activation energy for anode, Eact,a (Jmol−1 ) 1.2 × 105
Activation energy for cathode, Eact,c (Jmol−1 ) 1.0 × 105
Electrode porosity, n 0.48
Electrode tortuosity,  5.4
Average pore radius, r (m) 1.07

Electrolyte-supported electrolyzer [29]

Electrolyte thickness, L (m) 1000
Cathode thickness, dc (m) 100
Anode thickness, da (m) 100

Cathode-supported electrolyzer
Electrolyte thickness, L (m) 50
Cathode thickness, dc (m) 500
Anode thickness, da (m) 50

Anode-supported electrolyzer
Electrolyte thickness, L (m) 50
Cathode thickness, dc (m) 50
Anode thickness, da (m) 500
 M. Ni et al. / International Journal of Hydrogen Energy 32 (2007) 2305 – 2313 2309

2.0 the temperature increased, due to the significant increase in

oxygen ion conductivity.
Furthermore, the activation overpotential was found to de-
1.5 crease with increasing temperature because a higher tempera-
ture increased the electrochemical reaction rate. The effective
Cell potential,V(V)

Theoretical model diffusion coefficient of gaseous species increased with temper-

T = 1273K ature according to the kinetic theory and Chapman–Enskog the-
1.0
T = 1223K ory [21,22]. However, the gas density decreased as temperature
T = 1173K
increased as indicated by the multiplying term, RT /2F , in Eqs.
Experiment [29] (15) and (21). The opposing effects resulted in insensitivity of
0.5 T = 1273K concentration overpotential to operating temperature.
T = 1223K
T = 1173K
4. Parametric analyses
0.0
0 1000 2000 3000 4000 5000 6000
4.1. Cell component thickness effect and optimal support for
Current density, J (Am-2)
SOSE cell
Fig. 2. Comparison between theoretical simulation results and experimental
data by Momma et al. [22]. A SOSE cell can be either anode-, electrolyte-, or cathode-
supported. The typical dimensions of each support type were
listed in Table 1. As the activation overpotential was con-
sidered independent of the electrode thickness in the present
2.0 study, only the sensitivities of concentration and ohmic over-
T = 1273K
potentials to cell component thickness were analyzed. Fig. 4
conc ,a +conc ,c
act ,a +act ,c shows the concentration and ohmic overpotentials of a cathode-
1.5 ohmic supported SOSE cell. The cathode concentration overpotential
overwhelmed other overpotentials due to the thick cathode used.
Overpotentials, (V)

T = 1173K From modeling results, the limiting current density was found to
conc ,a +conc ,c
act ,a +act ,c
be about 10, 000 Am−2 . The anode concentration overpotential
1.0 ohmic was rather small mainly due to the fact that the reactant species
(oxygen ions) reacted efficiently at the anode–electrolyte in-
terface and caused no transport problem. Moreover, the anode
0.5 was thin in a cathode-supported SOSE cell.
The variations of concentration and ohmic overpotentials
with current density of an anode-supported SOSE are presented
in Fig. 5. The anode concentration overpotential was small de-
0.0 spite the thick anode used. The cathode limiting current density
0 2000 4000 6000 8000 10000
Current density, J (Am-2)
0.30
Fig. 3. Theoretical activation, concentration, and ohmic overpotentials of
electrolyte-supported SOSE cell.
0.25 Cat hode-supported SOSE cell
T = 1273K; Steam molar fraction: 60%
conc,c
Overpotentials, (V)

0.20 conc,a
thick, respectively. The tests were conducted at a constant ohmic
pressure of 1 bar, steam molar fraction of 60%, and various
operating temperature from 1173 to 1273 K. In the theoretical 0.15
simulation, the values of the input parameters are summarized
in Table 1 [17,18,29–31]. As shown in Fig. 2, the good agree- 0.10
ment between the theoretical and experimental results served
as a validation of the theoretical model. 0.05
At a higher temperature, the cell potential decreased result-
ing in a lower electrical loss. Individual overpotentials were
0.00
studied in detail as shown in Fig. 3. The ohmic overpotential 0 2000 4000 6000 8000 10000
dominated the cell overpotential due to the thick (1 mm) elec-
Current density, J (Am-2)
trolyte support used. It explained why the J .V curve shown in
Fig. 2 approximately followed a linear relationship. It was also Fig. 4. Concentration and ohmic overpotentials of cathode-supported SOSE
observed that the ohmic overpotential decreased drastically as cell.
2310 M. Ni et al. / International Journal of Hydrogen Energy 32 (2007) 2305 – 2313

0.5 0.4
Anode-supported SOSE cell

Cathode concentration overpotential, (V)

T = 1273K; Steam molar fraction: 60% Cathode porosity
0.6
0.4 conc,c
0.4
conc,a 0.3 0.2
ohmic
Cell potential,V(V)

0.3

0.2

0.2

0.1
0.1

0.0 0.0
0 20000 40000 60000 80000 100000 120000 0 20000 40000 60000 80000 100000 120000 140000
Current density, J (Am-2) Current density, J (Am-2)

Fig. 5. Concentration and ohmic overpotentials of anode-supported SOSE cell. Fig. 7. Porosity effect on cathode concentration overpotential of anode-
supported SOSE cell.

2.5

0.05
2.0
Anode concentration overpotential, (V)

0.04
Cell potential,V(V)

1.5

0.03
T = 1273K; Steam molar fraction: 60%
1.0
Electrolyte-supported Anode porosity
Cathode-supported 0.02 0.6
Anode-supported 0.4
0.5
0.2
0.01
0.0
0 20000 40000 60000 80000 100000 120000
0.00
Current density, J (Am-2) 0 20000 40000 60000 80000 100000 120000 140000
Current density, J (Am-2)
Fig. 6. Theoretical J .V characteristics of cathode-supported, elec-
trolyte-supported, and anode-supported SOSE cells.
Fig. 8. Porosity effect on anode concentration overpotential of anode-
supported SOSE cell.

was found to be very large (about 100, 000 Am−2 ) as the cath-
ode was thin. Relatively, the ohmic overpotential contributed
most to the overall overpotential at a current density less than tration overpotential increased significantly with decreasing
100, 000 Am−2 . porosity as the gas transport in the porous cathode layer became
The J .V characteristics of electrolyte-, cathode-, and anode- more difficult at a lower porosity. The anode concentration over-
supported SOSE cells were compared as shown in Fig. 6. It was potential showed similar trend to that of the cathode (Fig. 8.
found that the anode-supported SOSE cell had the best elec- More importantly, the SOSE anode did not suffer from limiting
trical performance in terms of the applied voltage required for current density, defined as the current density at which the reac-
water splitting. Therefore, the subsequent parametric analyses tant species at the reaction region approached 0. This was due
as presented in the following sections were conducted for an to the fact that the reactant (oxygen ion) was transported in the
anode-supported SOSE cell. solid electrolyte and reacted efficiently at the anode–electrolyte
interface. Thus, no reactant transport problem was found in the
4.2. Electrode porosity effect porous electrode layer. Combining the porosity effect on anode
and cathode concentration overpotentials, the porosity effect on
The cathode porosity effect on the cathode concentration J .V characteristics was determined as plotted in Fig. 9. A high
overpotential was presented in Fig. 7. The cathode concen- electrode porosity was beneficial, as more space was available
 M. Ni et al. / International Journal of Hydrogen Energy 32 (2007) 2305 – 2313 2311

2.5 3.0

2.5
2.0

Cell potential,V(V)
Cell potential,V(V)

2.0
1.5
1.5
Electrode porosity
Electrode pore radius
1.0 0.6
0.4 2.0x10-6m
1.0
0.2 1.0x10-6m
0.5x10-6m
0.5
0.5

0.0 0.0
0 20000 40000 60000 80000 100000 120000 140000 0 50000 100000 150000 200000
Current density, J (Am-2) Current density, J (Am-2)

Fig. 9. Effect of electrode porosity on J .V characteristics of anode-supported Fig. 10. Electrode pore size effect on J .V characteristics of anode-supported
SOSE cell. SOSE cell.

2.5
for gas transport, resulting in smaller concentration overpoten-
tial. However, recent experimental studies have shown that too
high porosity (0.76) is detrimental as the electrolyte becomes 2.0
non-gas tight, leading to reduction in Nernst potential [32]. As a
Cell potential, V(V)

porosity of 0.57 did not show any reduction in Nernst potential

in Zhao and Virkar’s experiments [32], optimal porosity value 1.5
between 0.5 and 0.6 is recommended. In addition, the present
paper assumes that electrochemical reactions take place only Anode-supported SOSE cell
1.0
at the electrode–electrolyte interface. However, in an advanced T = 1273K; Steam molar fraction:
composite electrode, reaction zone is extended from the inter- 60%
face to the electrode inside [33–36]. Increasing porosity can 0.5
40%
reduce concentration overpotential on the one hand, but can 20%

increase activation overpotential on the other hand due to less

reactive surface area available [33–36]. Therefore, the SOSE 0.0
performance cannot be further enhanced when the porosity 0 20000 40000 60000 80000 100000 120000
exceeds a certain value. Detailed mathematical modeling at Current density, J (Am-2)
micro-scale level is necessary to better simulate the composite
electrodes performance and to more precisely identify the opti- Fig. 11. Cathode steam molar fraction effect on J .V characteristics of
anode-supported SOSE cell.
mal porosity value for SOSE design optimization. The micro-
scale modeling study will be performed in the future.

4.3. Electrode pore size effect 4.4. Steam molar fraction

Besides electrode porosity, pore size is an important param- Fig. 11 shows the effect of steam molar fraction on SOSE
eter for the gas transport characteristics in the porous electrode J .V characteristics. It was found that the cell equilibrium po-
layers. The J .V curves of a SOSE cell with electrode pore tential decreased with increasing steam molar fraction. The
radius ranging from 0.5 to 2.0 m are presented in Fig. 10. finding could be explained by the Nernst equation (Eq. (3)).
The cell potential was found to decrease with increasing pore Furthermore, an increase in steam molar fraction caused an
size. As the pore size increased, the gas transport within pore increase in cathode limiting current density, resulting in a re-
structures became easier, resulting in lower concentration over- duction in cell operating potential. At a higher steam molar
potentials as well as higher cathode limiting current density. fraction, the reactant molar diffusion rate was higher; thus, the
The pore size effect would be more pronounced for a cathode- resistance to gas transport was lower accordingly.
supported SOSE cell because the resistance to gas transport in In order to achieve a higher steam molar fraction, one can
the porous cathode is the dominant factor of overpotential. supply sufficient amount of steam and simultaneously extract
2312 M. Ni et al. / International Journal of Hydrogen Energy 32 (2007) 2305 – 2313

3.0 2.0

2.5
1.5
Cell potential, V(V)

Cell potential,V(V)
2.0

Cathode-supported SOSE cell

1.5 1.0
P = 5.0 bar
Anode-supported SOSE cell P = 1.0 bar
1.0 P = 5.0 bar P = 0.5 bar
P = 1.0 bar
0.5
P = 0.5 bar
0.5

0.0 0.0
0 50000 100000 150000 200000 0 10000 20000 30000 40000 50000
Current density, J (Am-2) Current density, J (Am-2)
Fig. 13. Pressure effect on J .V characteristics of cathode-supported SOSE
Fig. 12. Pressure effect on J .V characteristics of anode-supported SOSE cell.
cell.

5. Conclusion
hydrogen gas from the cathode. Excessive supply of steam can
increase the steam molar fraction but it will absorb heat from An electrochemistry model was developed to characterize
the solid electrode to cause a temperature drop. This may re- the electrical characteristics of SOSE hydrogen production. The
duce the overall cell performance, as the water-splitting reac- model predicts the SOSE operation in reasonably good agree-
tion is endothermic. Apparently, detailed modeling study on ment with experimental data available in the literature. The
heat/mass transfer in the gas channel and electrochemical re- theoretical parametric analyses have shown that the cathode
action in the electrode layer is important to gain understanding concentration overpotential and the ohmic overpotential are
of the complex SOSE mechanisms and its J .V characteristics. sensitive to the cathode thickness and electrolyte thickness, re-
The electrochemistry model presented in this paper can be ex- spectively. As a result, the anode-supported SOSE cell config-
tended to address the above issues. uration is more favorable. In order to further improve the effi-
ciency, thin cathode and electrolyte should be used. Although
4.5. Pressure effect the typical thicknesses of cathode and electrolyte of an anode-
supported SOSE cell are 50 m, thinner layers (10 m thick)
The effect of operating pressure on SOSE J .V characteris- are possible [32]. It is noteworthy that the mechanical and ther-
tics is illustrated in Fig. 12. Increasing the operating pressure mal stresses should be carefully considered when very thin cell
yielded a slight increase in the cell equilibrium potential as a components are used.
result of the Nernst equation (Eq. (3)). According to references The effects of both material properties and operational pa-
[21,22], the molecular diffusion coefficient, DH2 .H2 O , was in- rameters on SOSE electrical characteristics have been analyzed.
versely proportional to the operating pressure, P, but the Knud- It is found that large electrode porosity and large pore size are
sen diffusion coefficient was independent of P. Therefore, the desirable to reduce gas transport resistance. Higher operating
effective diffusion coefficient, Dieff , was less sensitive to P. On temperature is beneficial to reduce the cell potential because of
the other hand, the gas density increased with P. The combined faster electrochemical reaction kinetics and higher oxygen ion
effects resulted in an increase in molar diffusion rate and a de- conductivity of the electrolyte. However, in practical SOSE hy-
crease in concentration overpotential at high P. In practice, the drogen production, electrode sintering and thermal stress may
current density is generally less than 10, 000 Am−2 in SOSE cause problems. Considering all these issues, operating temper-
hydrogen production [9–12]. Therefore, operation at 1 bar is ature at 1273 K is near the optimal condition. Increasing steam
recommended since either increasing the pressure above 1 bar molar fraction can decrease the cell potential due to reduced
or decreasing the pressure below 1 bar requires consumption of Nernst potential and enhanced gas transport. Increasing operat-
energy. ing pressure can reduce the concentration overpotential on one
The J .V characteristics of cathode-supported SOSE cell are hand, but increase the Nernst potential on the other hand. As
presented in Fig. 13 for comparison. The J .V characteris- a result, 1 bar is suggested for anode-supported SOSE hydro-
tics demonstrated similar trend as an anode-supported SOSE gen production at normal operating current density, while high
cell. The limiting current density occurred at a lower value. pressure is preferred for cathode-supported SOSE operating at
As a result, when SOSE hydrogen production is operated at a current density above 5000 Am−2 .
J < 5000 Am−2 , operation at 1 bar is recommended. For oper- In practice, an SOSE hydrogen production system, such as
ation at J > 5000 Am−2 , a pressurized condition is preferred. the Siemens Westinghouse model, is designed in a tubular stack
 M. Ni et al. / International Journal of Hydrogen Energy 32 (2007) 2305 – 2313 2313

configuration. In such configuration, both temperature and gas [16] Hartvigsen J, Swank D, Schade C, Bordia R. Large area cell for hybrid
composition vary along the flow channel of the cathode. In order hydrogen co-generation process. US DOE Hydrogen Program, FY 2005
to model the system operation, the present electrochemistry Progress Report. p. 367–9.
[17] Fuel cell handbook, 7th ed. EG&G Technical Services, Inc., U.S.
model can be modified with additional thermal fluid mechanics Department of Energy, Office of Fossil Energy, National Energy
for a complete modeling of SOSE system performance. Technology Laboratory, P.O. Box 880, Morgantown, West Virginia, USA;
November 2004. p. 57–60.
Acknowledgments [18] Chan SH, Xia ZT. Polarization effects in electrolyte/electrode-supported
solid oxide fuel cells. J Appl Electrochem 2002;32(3):339–47.
[19] Hernandez-Pacheco E, Singh D, Hutton PN, Patel N, Mann MD. A
The work described in this paper was partly supported by macro-level model for determining the performance characteristics of
a Grant from the Research Grants Council of the Hong Kong solid oxide fuel cells. J Power Sources 2004;138(1–2):174–86.
Special Administrative Region, China (HKU 7150/05E). [20] Krishna R, Wesselingh JA. The Maxwell–Stefan approach to mass
transfer. Chem Eng Sci 1997;52(6):861–911.
References [21] Reid RC, Prausnitz JM, Poling BE. The properties of gases and liquids.
4th ed, New York: McGraw-Hill; 1987 ISBN 0-07-051799-1.
[1] Goltsov VA, Veziroglu TN, Goltsova LF. Hydrogen civilization of [22] Veldsink JW, Vandamme RMJ, Versteeg GF, Vanswaaij WPM. The use
the future—a new conception of the IAHE. Int J Hydrogen Energy of dusty-gas model for the description of mass transport with chemical-
2006;31(2):153–9. reaction in porous media. Chem Eng J Biochem Eng J 1995;57(2):
[2] Winter CJ. Into the hydrogen energy economy—milestones. Int 115–25.
J Hydrogen Energy 2005;30(7):681–5. [23] Choi PH, Bessarabov DG, Datta R. A simple model for solid
[3] Sherif SA, Barbir F, Veziroglu TN. Wind energy and the hydrogen polymer electrolyte (SPE) water electrolysis. Solid State Ionics
economy—review of the technology. Sol Energy 2005;78(5):647–60. 2004;175(1–4):535–9.
[4] Sherif SA, Barbir F, Veziroglu TN. Towards a hydrogen economy. Electr [24] Costamagna P, Honegger K. Modeling at solid oxide heat exchanger
J 2005;18(6):62–76. integrated stacks and simulation at high fuel utilization. J Electrochem
[5] Dunn S. Hydrogen futures: toward a sustainable energy system. Int Soc 1998;145(10):3995–4007.
J Hydrogen Energy 2002;27(3):235–64. [25] Zhu HY, Kee RJ. A general mathematical model for analyzing the
[6] Ni M, Leung MKH, Sumathy K, Leung DYC. Potential of renewable performance of fuel cell membrane-electrode assemblies. J Power
hydrogen production for energy supply in Hong Kong. Int J Hydrogen Sources 2003;117(1–2):61–74.
Energy 2006;31(10):1401–12. [26] Momma A. Personal communications; November 2005.
[7] Ni M, Leung MKH, Leung DYC, Sumathy K. A review and recent [27] Momma A, Kaga Y, Takano K, Nozaki K, Negishi A, Kato K. et al.
developments in photocatalytic water-splitting using TiO2 for hydrogen Experimental investigation of anodic gaseous concentration of a practical
production. Renew Sust Energ Rev 2007;11(3):401–25. seal-less solid oxide fuel cell. J Power Sources 2005;145(2):169–77.
[8] Ni M, Leung MKH, Leung DYC. A modeling study on concentration [28] Ferguson JR, Fiard JM, Herbin R. Three dimensional numerical
overpotentials of a reversible solid oxide fuel cell. J Power Sources simulation for various geometries of solid oxide fuel cells. J Power
2006;163(1):460–6. Sources 1996;58(2):109–22.
[9] Salzano FJ, Skaperdas G, Mezzina A. Water vapor electrolysis at high [29] Momma A, Kato T, Kaga Y, Nagata S. Polarization behavior of high
temperature: systems considerations and benefits. Int J Hydrogen Energy temperature solid oxide electrolysis cells (SOEC). J Ceram Soc Jpn
1985;10(11):801–9. 1997;105(5):369–73.
[10] Arashi H, Naito H, Miura H. Hydrogen production from high [30] Lehnert W, Meusinger J, Thom F. Modeling of gas transport phenomena
temperature steam electrolysis using solar energy. Int J Hydrogen Energy in SOFC anodes. J Power Sources 2000;87(1–2):57–63.
1991;16(9):603–8. [31] Divisek J, Froning D, Lehnert W, Meusinger J, Stimming U. Transport
[11] Yildiz B, Kazimi MS. Efficiency of hydrogen production systems processes and methane reforming reactions in the SOFC-cermet anodes.
using alternative nuclear energy technologies. Int J Hydrogen Energy IEA, 10th SOFC Workshop, vol. 1, Les Diableret, Switzerland; 1997.
2006;31(1):77–92. [32] Zhao F, Virkar AV. Dependence of polarization in anode-supported
[12] Hino R, Haga K, Aita H, Sekita K. R&D on hydrogen production by solid oxide fuel cells on various cell parameters. J Power Sources
high-temperature electrolysis of steam. Nucl Eng Des 2004;233(1–3): 2005;141(1):79–95.
363–75. [33] Chan SH, Chen XJ, Khor KA. Cathode micromodel of solid oxide fuel
[13] Licht S. Efficient solar generation of hydrogen fuel—a fundamental cell. J Electrochem Soc 2004;151(1):A164–72.
analysis. Electrochem Commun 2002;4(9):790–5. [34] Chan SH, Xia ZT. Anode micro model of solid oxide fuel cell.
[14] Rund JA. System design and new materials for reversible solid-oxide, J Electrochem Soc 2001;148(4):A388–94.
high-temperature steam electrolysis. US DOE Hydrogen Program, FY [35] Xia ZT, Chan SH, Khor KA. An improved anode micro model of SOFC.
2005 Progress Report. p. 361–2. Electrochem Solid State Lett 2004;7(3):A63–5.
[15] Balachov II, Crouch-Baker S, Hornbostel M, McKubre M, Sanjurjo [36] Chen XJ, Chan SH, Khor KA. Simulation of a composite cathode in
A, Tanzella F. Modular system for hydrogen generation and oxygen solid oxide fuel cells. Electrochim Acta 2004;49(11):1851–61.
recovery. US DOE Hydrogen Program, FY Progress Report. p. 363–6.