Anda di halaman 1dari 8

Journal of Membrane Science 473 (2015) 8–15

Contents lists available at ScienceDirect

Journal of Membrane Science


journal homepage: www.elsevier.com/locate/memsci

Effect of operating conditions on the performance of solid electrolyte


membrane reactor for steam and CO2 electrolysis
Asif Mahmood a,c, Saira Bano a,c, Ji Haeng Yu b,c, Kew-Ho Lee a,c,n
a
Korea Research Institute of Chemical Technology, Daejeon 305-343, Republic of Korea
b
Korea Institute of Energy Research, Daejeon 305-343, Republic of Korea
c
University of Science and Technology, Daejeon 305-350, Republic of Korea

art ic l e i nf o a b s t r a c t

Article history: Solid electrolyte membrane reactor (SEMR) based on the Ni–YSZ (yttria-stabilized zirconia) supported
Received 13 June 2014 YSZ electrolyte membrane was developed and used in the high-temperature electrochemical reduction
Received in revised form of steam, CO2 and a mixture of steam-CO2. Effect of different operating parameters such as temperature,
27 August 2014
feed gas composition, and operating voltage on the performance of SEMR in a high steam and/or CO2
Accepted 3 September 2014
environment was investigated. Experiments were performed in the temperature range of 750–850 1C
Available online 10 September 2014
with 20% H2–80% oxidant as a feed to the Ni–YSZ electrode and air to the LSM electrode. The
Keywords: electrochemical performance was investigated using the current density–voltage (i–V) curves and
Electrolysis electrochemical impedance spectra (EIS). A high area-specific resistance (ASR) was observed in a fuel cell
Steam
mode as compared to the electrolysis mode because of very low fuel content and consequently, a higher
Carbon dioxide
concentration polarization. At 850 1C and 1.5 V, a very high current density of  1.9 A/cm2,  1.5 A/cm2
Membrane reactor
Water-gas shift reaction and  1.2 A/cm2 is observed for steam, steam-CO2 (co-electrolysis) and CO2 electrolysis respectively. The
decrease in electrochemical performance with increasing CO2 content was found to be related to the
mass transport limitation of the fuel electrode. An equivalent circuit model was used to fit the
impedance data and separate the various polarization losses in the cell. EIS results showed that different
performance limiting step is involved depending on the operating conditions. At low operating
temperature and/or cell voltage, the polarization losses at LSM oxygen electrode contributed signifi-
cantly to total ASR of the cell, whereas at high operating temperature and cell voltage, the concentration
polarization at fuel electrode led to a decrease in the cell performance.
& 2014 Elsevier B.V. All rights reserved.

1. Introduction fuels by the well-known Fischer–Tropsch reaction [2–4]. The


production of synthetic fuels by this method is an appropriate
Solid electrolyte membrane reactors (SEMRs) have received way to reduce the consumption of fossil fuels, CO2 emissions and
considerable attention over the last few decades because of the the increasing need for energy storage [5].
depletion of fossil fuels, high oil prices and environmental con- Typically, steam/CO2 electrolysis is carried out at 700–1000 1C
siderations. These are one of the best energy conversion devices and becomes more endothermic at higher temperatures. There-
for future and can be employed to generate chemical energy fore, it requires less electrical energy than low-temperature
carriers like H2, CO, CH4, CH3OH and syn-gas from steam and electrolysis [6–8]. In addition, the availability of waste industrial
CO2 using renewable energy. Consequently, in the absence of heat sources with temperatures over 100 1C and the joule heat
renewable energy sources, the chemical energy carriers can sub- produced due to current flow can further decrease the electricity
sequently be utilized in a SEMR to generate electricity and meet demand and fuel production price [9]. The high operating tem-
the public energy demand [1]. Furthermore, the syn-gas produced perature of SOEC allows the applied voltage to be reduced due to
can be stored and ultimately be converted into liquid hydrocarbon
favorable thermodynamics and fast reaction kinetics [6,10]. Actu-
ally, internal resistance losses decrease at high temperatures and
lead to an improvement in the current densities for a given cell
n
Corresponding author at: Korea Research Institute of Chemical Technology, voltage [9]. However, the high operating temperature limits the
Daejeon 305-343, Republic of Korea. Tel.: þ82 42 860 7240;
choice of materials available for the fabrication of cells or stacks.
fax: þ82 42 861 4151.
E-mail address: khlee@krict.re.kr (K.-H. Lee). The materials used should be chemically inert toward other

http://dx.doi.org/10.1016/j.memsci.2014.09.002
0376-7388/& 2014 Elsevier B.V. All rights reserved.
A. Mahmood et al. / Journal of Membrane Science 473 (2015) 8–15 9

reactor components and stable against any thermal, mechanical or Archimedes method and was around 34% after the reduction of
physical shock [10]. nickel oxide respectively.
The state-of-the art fuel electrode material for hydrogen/hydro-
carbons oxidation is based on a Ni–YSZ cermet due to its high 2.2. Electrochemical performance measurements
activity, electrical conductivity and relatively low cost [3,11–14]. It
has widely been studied for steam electrolysis [15–27] and exhibited The prepared Ni–YSZ supported cell was attached to an alumina
good performance between 800 and 1000 1C [14]. But, a few papers tube with a gold ring as a sealant. A four-probe system was used for
report the use of Ni–YSZ cermet electrode for CO2 electrolysis the electrochemical characterizations of the cell, and the connections
[1,2,5,6,9,28–34] or co-electrolysis [5,9,29,30,33,35–37]. However, were made using Pt wires and meshes. The working electrode (WE)
most of these studies are performed in the presence of either a and working sensor (WS) probes were connected to the LSM oxygen
carrier gas (Ar/N2) or a hydrogen-enriched environment which might electrode whereas the counter electrode (CE) and reference electrode
affect the diffusion properties of the steam/CO2 [29], and hence the (RE) probes were connected to the Ni–YSZ fuel electrode as shown in
electrolysis performance as well. Keeping the goal of SEMR commer- Fig. 1. The cell was heated to 950 1C for proper sealing and then cooled
cialization in mind, the use of argon or high hydrogen contents down to 850 1C. At the start, the nickel oxide was reduced to nickel by
would be impractical, as it would increase the operational cost. Also, flowing 100 mL/min of pure H2 on the NiO–YSZ side. Meanwhile,
the SEMR must operate at high current density to decrease the cost 100 mL/min of air was supplied to the LSM side of the cell. The cell
of synthetic fuel [33]. At present, Ni–YSZ cermet with YSZ electrolyte was initially operated in a fuel cell mode, then in the electrolysis
membrane and LSM oxygen electrode is the only state-of-the art mode. Fuel cell performance was found to increase with time due to
material achieving the desired electrolysis current density [14,32]. the activation of the LSM electrode. After nearly 36 h of operation in a
Hence, the present work is carried out using the conventional fuel cell mode, the cell performance became stable. Hence, its
electrochemical membrane reactor materials. polarization was then switched to the electrolysis mode. Electrolysis
This study primarily focused on the high temperature electrolysis experiments were performed in the temperature range of 850–750 1C.
of steam, CO2 and steam-CO2 in SEMR under a steam/CO2 enriched Steam and/or CO2 with 20% H2 was fed into the Ni–YSZ side, and the
environment. The effect of feed gas compositions, applied voltage and air was fed into the LSM side. The flow rate of the gases was kept the
operating temperature on the electrolysis current density was studied. same at both of the electrodes during the course of experiments.
The electrochemical data were collected in terms of i–V curve and Initially, the steam, steam-CO2 and CO2 electrolysis was performed at
electrochemical impedance spectroscopy (EIS). An equivalent circuit 850 1C followed by performance testing at 800 1C and 750 1C.
model was used to deconvolute the impedance data and to investigate The DC and AC characterizations were performed using the
the processes limiting the electrolysis performances of an oxygen-ion ZIVE SP5 electrochemical analyzer. i–V curves were measured in a
conducting SEMR. potentiostatic mode with a scanning rate of 2 mV/s. The electro-
chemical impedance spectroscopy (EIS) was performed with an AC
amplitude of 10 mV in the frequency range of 1 MHz to 0.1 Hz. The
temperature of the furnace was kept constant, and a thermocouple
2. Experimental was positioned at the center of the oxygen electrode to measure
the actual temperature of the cell.
2.1. Fabrication of cell

A Ni–YSZ supported cell with YSZ-8 (8 mol% Y2O3-stabilized 3. Results and discussion
ZrO2) oxygen-ion transport membrane (electrolyte) and LSM
(La0.7Sr0.3MnO3) positive electrode was fabricated as follows: the 3.1. Microstructure analysis
powders of NiO (99.97% Pure, High Purity Chemicals, Japan) and
YSZ-8 (TZ-8YS, Tosoh, Japan) were taken in 60:40 wt% in a Nalgene Fig. 2 shows the microstructure of the cell after testing in the
bottle containing zirconia balls. Then an appropriate amount of electrolysis mode. The thin layer of YSZ electrolyte membrane
carbon black (L30, Korea Carbon) as a pore former, binder, and shown in Fig. 2a is crack free and dense enough. However, some
ethanol was added and the mixture was ball-milled for 48 h. After closed pores can be seen in the cross-section of the electrolyte
mixing, the solvent was evaporated under reduced pressure in a (Fig. 2b) but that could not allow the fuel crossover. The cross-
rotary evaporator. The material obtained was dried overnight at section image of the cell indicates a good adhesion of electrolyte to
70 1C followed by grinding and screening with a sieve. The Ni–YSZ substrate as well as of the YSZ–LSM layer to electrolyte
prepared NiO–YSZ powder was pressed to form a pellet of 1 in.
diameter and fired at 1175 1C for 3 h to obtain a porous substrate.
The thin YSZ electrolyte membrane was prepared by a dip-coating
method. Initially, the YSZ electrolyte slurry was prepared by
mixing YSZ-8 (TZ-8Y, Tosoh, Japan) with deionized water, isopro-
panol (IPA) and organic binder followed by the ball milling for
about 5 days. The YSZ slurry was coated on the pre-sintered NiO–
YSZ support by dip coating and co-sintered in air at 1400 1C for
3 h. A LSM–YSZ functional layer with 60:40 wt% and an LSM
oxygen electrode layer was applied to the dense YSZ electrolyte
membrane by spray coating, and fired at 1050 1C for 3 h.
The positive electrode area, which is also known as the cell
active area was fixed as 0.5 cm2. At the end, platinum (Pt) paste
was painted on both sides of the cell as a current collector and
cured at 900 1C. The final prepared cell was 19 mm in diameter
with an electrolyte membrane thickness of 10–12 mm, negative
electrode thickness of 1.2 mm, and positive electrode thickness of
25–30 mm. The porosity of substrate was measured by the Fig. 1. Schematic diagram of the experimental set up.
10 A. Mahmood et al. / Journal of Membrane Science 473 (2015) 8–15

Fig. 2. SEM images of the cell after testing in electrolysis mode: (a) surface of YSZ electrolyte dense membrane, and (b) cross-sectional view of the cell.

even after a prolong testing in the electrolysis mode. The connec-


tion between electrodes and electrolyte membrane is critical as
the poor adhesion increases the contact resistance significantly 1.6
and thereby makes the total ohmic resistance quite high. However,

Cell voltage (V)


a critical look at the oxygen electrode/electrolyte interface in 1.2
Fig. 2b indicates that some small cracks are produced which may
further develop into large cracks and ultimately result in mechan-
ical weakening of the electrode/electrolyte interface. This might be 0.8
20H2-80H2O
due to the build of high oxygen partial pressure in or around the
electrode/electrolyte interface under high current density. How- 20H2-40H2O-40CO2
0.4 20H2-80CO2
ever, the extent of delamination of the oxygen electrode was
negligible even after cell testing for about 500 h. Because the cell
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
was not operated under a high current densities for a long time
2
span, but the voltage was periodically swept from SOFC to SOEC Current density (A/cm )
and vice versa, the oxygen accumulated at the LSM/YSZ interface
during electrolysis could be released in the SOFC mode of
operation.
1.6
Cell voltage (V)

3.2. DC characterizations (i–V curve)


1.2
3.2.1. Initial performance testing in the fuel cell mode
The cell was heated to 850 1C at a rate of 1 1C/min in SEMR,
0.8
where nickel oxide was reduced to nickel in the presence of pure 20H2-80H2O
hydrogen. The complete reduction of nickel oxide was confirmed 20H2-40H2O-40CO2
by observing a constant open-circuit voltage (OCV) for different 0.4 20H2-80CO2
flow rates of hydrogen and air but keeping their ratio as constant.
Initially, the cell performance was tested in the fuel-cell mode by -1.6 -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2
providing humidified hydrogen and air to the Ni–YSZ and LSM 2
electrodes respectively, to confirm that the electrochemical cell Current density (A/cm )
worked. The OCV, current density versus voltage (i–V curve in
SOFC mode), potentiostatic operation at 0.7 V, and EIS were
periodically measured during cell testing. At a constant operating 1.6
voltage of 0.7 V, an increase in current density and a decrease in
Cell voltage (V)

area-specific resistance (ASR) were observed over time, due to the


activation of LSM electrode [38]. It took over 36 h to obtain a stable 1.2
performance over time of the SEMR in the fuel-cell mode.
0.8
20H2-80H2O
3.2.2. Performance testing in the electrolysis mode
20H2-40H2O-40CO2
All of the electrolysis experiments were performed with a flow 0.4
of 50 mL/min of the feed gases at the Ni–YSZ electrode to achieve 20H2-80CO2
high reactant utilization. The feed gas always contained 20% H2 to
-1.2 -0.9 -0.6 -0.3 0.0 0.3 0.6 0.9
prohibit the oxidation of nickel into nickel oxide in Ni–YSZ
2
electrode. The remaining 80% was composed of steam and/or Current density (A/cm )
CO2. After successful testing in the fuel cell mode at 850 1C, the
Fig. 3. DC characterization (i–V curves) of cell in H2O, H2O–CO2 and CO2 mixtures
cell was changed to the electrolysis mode. Steam electrolysis was (a) 850 1C, (b) 800 1C, and (c) 750 1C.
performed, followed by the co-electrolysis (steam-CO2) and car-
bon dioxide electrolysis respectively. Next, the electrochemical Fig. 3 shows three sets of i–V characteristic curves for the
performance of the cell was evaluated at 800 1C and 750 1C with 20% H2–80% oxidant (s) in the fuel cell and electrolysis modes
the same feed mixtures and order as well. for different operating temperatures. It is observed that no
A. Mahmood et al. / Journal of Membrane Science 473 (2015) 8–15 11

discontinuity occurs across the OCV in the shift from fuel cell to shown in Table 2.
the electrolysis mode for all the reported mixtures. Hence, it is
confirmed that the SEMR can work reversibly in the presence of H2 þ CO2 ⇋H2 O þ CO ð2Þ
H2-steam, H2-steam-CO2 and H2–CO2 mixtures under the current
operating conditions [6,30]. The theoretical or Nernst potentials Consumption of hydrogen by RWGS leads to the production of
were calculated using Eq. (1) and were found to be similar to all of CO, which could be another fuel for the solid oxide fuel cell (SOFC).
the measured mixture. This is because of the nearly equivalent However, the Ni–YSZ electrode has a lower activity for CO
Gibbs free energies of formation of H2 and CO (from H2O and CO2) oxidation compared to H2 oxidation [30]. Hence, it leads to an
at the current operating temperatures (Table 1) [33]. increase in the polarization resistance (ASR) and consequently, a
decrease in the current density in fuel cell mode.
EN ¼  Δg f =nF ð1Þ In electrolysis mode, i–V curves were almost linear at low
current densities, but a current limitation was observed at higher
EN is the Nernst potential, n is the number of electrons involved in current densities due to the mass transport limitation of the
the reaction, F is the Faraday constant and Δgf is the Gibbs-free electrode [39,40]. The polarization upturn at high current density
energy of formation at the operating temperature. The practically is caused by the diffusion or concentration polarization. This type
achieved open-circuit voltage was 0.868 V, 0.867 V and 0.871 V for of polarization originates from the lack of the reactant at the triple
the 20% H2–80% steam, 20% H2–40% steam-40% CO2 and 20% H2– phase boundary (TPB) due to high reactant utilizations and/or a
80% CO2 feed mixture respectively. The experimentally-measured slow rate of migration of reactants/products. Because CO2 has high
OCV values are almost 10–15 mV more than the calculated Nernst molar mass and a small diffusion coefficient than steam, the
potential as shown in Table 1 and it might be due to the diffusion polarization increases proportionally with the CO2 con-
experimental error. However, the OCV was very stable even in tent of the mixture. Also, a higher resistance at low current density
the presence of the high steam content. Only a fluctuation of occurs when increasing the CO2 concentration (decreasing the
70.5 mV to 71 mV was observed in 40% and/or 80% steam steam content) in the feed gases.
content, which reveals a very fine control of the H2O addition The area-specific resistances (ASRs) were measured at  0.2 A/cm2
during the experiments. for all the curves shown in Fig. 3. The values obtained are summarized
As expected, the i–V curves showed higher polarization resis- in Table 1 and represent the preference of the reduction of steam and/
tance in the fuel-cell mode due to the low proportion of hydrogen or CO2 to form H2 and/or CO respectively. At 850 1C, the lowest ASR of
in the feed mixtures (20% H2–80% oxidant). Intriguingly, different 0.21 Ω cm2 was observed for steam electrolysis (20% H2–80% steam),
current densities were observed in the fuel cell mode for different which signifies the highest activity of the Ni–YSZ electrode for
feed mixtures all containing the same ratio (20%) of feed hydrogen hydrogen production as reported in the literature [1,5,21,28–30,33].
(Fig. 3a–c). In reality, the constant hydrogen feed mixtures contain The highest ASR (0.34 Ω cm2) was observed for the CO2 electrolysis
different hydrogen content at the Ni–YSZ electrode due to the with 20% H2–80% CO2 feed at 850 1C. The Ni–YSZ electrode showed
contribution of the reverse water-gas shift reaction (RWGS). In the the lowest activity for the carbon dioxide reduction and thus requires
presence of CO2, the RWGS reaction consumes hydrogen to a higher reduction potential compared to steam electrolysis. At the
generate CO and steam within the reactor at 850–750 1C (Eq. (2)). same time, the high operating potential may lead to a high CO
It should be noted that the increase in temperature and/or concentration via CO2 electrolysis, which can possibly be further
H2/CO2 concentration can shift this equilibrium in the forward reduced to carbon via the Boudouard reaction [6]. The presence of
direction, resulting in a further decrease in the H2 proportion. The a small amount of H2 may further assist the dissociation of adsor-
feed gas and thermodynamic equilibrium gas compositions are bed CO, thus promoting carbon deposition [41]. However, the carbon

Table 1
Theoretical and measured OCV and ASR in the electrolysis mode for various feed gas compositions at 850–750 1C.

a
Feed gas composition (%) Theoretical OCV (V) Measured OCV (V) ASR in electrolysis mode (Ω cm2)

850 1C 800 1C 750 1C 850 1C 800 1C 750 1C 850 1C 800 1C 750 1C

20H2–80H2O 0.857 0.876 0.895 0.868 0.886 0.905 0.21 0.36 0.68
20H2–40H2O–40CO2 0.856 0.878 0.900 0.867 0.887 0.909 0.27 0.38 0.70
20H2–80CO2 0.854 0.879 0.905 0.871 0.895 0.921 0.34 0.46 0.80

a
ASR was calculated from the DC polarization curve at  0.2 A/cm2 in the electrolysis mode.

Table 2
ASR values obtained from the EIS fitting data in Fig. 4a–c.

Temp. Feed gas composition Equilibrium gas composition Area specific resistance (ASR) Error (%)
(1C) H2:H2O:CO2 (%) H2:H2O:CO:CO2 (%)
Rs R1 R2 R3 Rp Rtotal Err.Rs Err.R1 Err.R2 Err.R3 Err.Rp Err.Rtotal
(Ω cm2) (Ω cm2) (Ω cm2) (Ω cm2) (Ω cm2) (Ω cm2) (%) (%) (%) (%) (%) (%)

850 20: 80: 00 20: 80: 00: 00 0.09 – 0.12 0.07 0.19 0.28 0.16 – 0.25 0.43 0.32 0.27
20: 40: 40 12: 48: 08: 32 0.09 – 0.13 0.10 0.22 0.31 0.25 – 0.39 0.69 0.53 0.45
20: 00: 80 04: 16: 16: 64 0.10 – 0.14 0.13 0.27 0.37 0.36 – 0.42 0.46 0.44 0.42
800 20: 80: 00 20: 80: 00: 00 0.14 – 0.24 0.05 0.30 0.43 0.18 – 0.27 1.36 0.47 0.37
20: 40: 40 12: 48: 08: 32 0.14 – 0.26 0.08 0.34 0.47 0.21 – 0.32 1.54 0.60 0.49
20: 00: 80 04: 16: 16: 64 0.14 – 0.28 0.11 0.39 0.53 0.20 – 0.25 1.41 0.58 0.48
750 20: 80: 00 20: 80: 00: 00 0.19 0.06 0.55 0.03 0.63 0.82 0.18 0.52 2.64 7.65 1.02 0.83
20: 40: 40 13: 47: 07: 33 0.19 0.07 0.57 0.05 0.68 0.87 0.20 0.44 2.32 4.49 0.91 0.76
20: 00: 80 05: 15: 15: 65 0.20 0.07 0.59 0.08 0.73 0.93 0.20 0.36 1.99 2.74 0.83 0.70
12 A. Mahmood et al. / Journal of Membrane Science 473 (2015) 8–15

deposition can be suppressed by increasing the steam concentration in the high thermoneutral voltage (  1.47 V) of the CO2 electrolysis.
the feed mixture. We believe that the complete understanding of this phenomenon
On the other hand, the cell ASR in co-electrolysis (20% H2–40%, is not clear at the moment, and it is a matter of further study to
steam–40% CO2) was 0.27 Ω cm2 that lies in between those of investigate its origin.
steam and CO2 electrolysis as depicted in Fig. 3a. Depending upon
the reduction mechanism of CO2 during co-electrolysis, the inter- 3.3. AC characterizations (EIS)
mediate ASR (0.27 Ω cm2) can be attributed to, either the medium
water contents in 20% H2–40% steam–40% CO2 feed mixture or to An electrochemical impedance spectroscopy is a powerful tool
the lowest activity of Ni–YSZ for carbon dioxide reduction. How- to study the various losses contributions associated with SEMR.
ever, there are some discrepancies in the literature regarding the The impedance measurements were performed at open-circuit
reduction mechanism of CO2 in SEMR. Stoots el. al. stated that CO2 voltage (OCV) as a function of operating temperature and feed gas
is only reduced to CO by the reverse water-gas shift reaction compositions. Thus, the impedance spectra contain information
(RWGS) and not by the electrochemical reduction [42] whereas, about both oxidation and reduction, as the oscillation of voltage
Ebbesen et. al. reported the successful electrochemical reduction occurs across the OCV [5]. Fig. 4a–c shows the impedance spectra
of CO2 during co-electrolysis. Nonetheless, they emphasized that for each of the test mixtures. The corresponding frequencies (Hz)
some CO may be also produced via the reverse water-gas shift are mentioned to have a better understanding of the Nyquist plots.
reaction during co-electrolysis [5].
Based on the results in this study, if we assume that CO2 is not
electrochemically active in 20% H2–80% CO2 mixture and all of the
electrolysis current density is utilized to reduce steam only herein 20H2-80H2O
produced by the RWGS reaction. Then, the hypothetical steam 20H2-40H2O-40CO2
consumption should not exceed about 16%, which is the maximum -0.2 20H2-80CO2

Z (ohm.cm )
2
amount of steam present in 20% H2–80% CO2 feed after establish-
ing the RWGS reaction equilibrium at 850 1C as shown in Table 1.
Considering the 100% Faradic efficiency, a total hypothetical steam
consumption of 8.36% is determined against the  1.2 A/cm2 at -0.1 3 2 1
10 10
//

10
1.5 V and 850 1C for 20% H2–80% CO2 electrolysis, which is about
half of the steam present in the SEMR. Therefore, it manifestly 0.1
indicates that steam electrolysis would occur preferentially during
0.0
20% H2–80% CO2 or 20% H2–40% H2O–40% CO2 electrolysis as it is 0.0 0.1 0.2 0.3 0.4 0.5
relatively easier than electrochemical reduction of CO2 [33] and CO / 2
would mainly produce by the RWGS reaction. However, the Z (ohm.cm )
possibility of the electrochemical reduction of CO2 cannot be
overlooked, especially under the absence or a very low steam
concentration. These results also suggest that the co-electrolysis of -0.3 20H2-80H2O
steam and CO2 is preferred to only CO2 electrolysis irrespective of
20H2-40H2O-40CO2
the involved reaction mechanism.
20H2-80CO2
Z (ohm.cm )
2

-0.2
2
3.2.3. Effect of operating temperature 10
The electrochemical performance of the cell decreased with 3
-0.1 10
//

decreasing temperature from 850 1C to 750 1C as predicted from 10 1


the thermodynamic and kinetic point of view. A 16–23% decrease 0.1
in current density at 1.5 V was observed at 800 1C that was further
decreased to 40–50% at 750 1C, 1.5 V depending on the feed 0.0
mixtures. In fact, the decrease in performance at low temperatures 0.0 0.1 0.2 0.3 0.4 0.5 0.6
is a combined effect of an increase in ohmic resistance (decrease in /
Z (ohm.cm )
2

the O2  ion conductivity) of the electrolyte and the slow reaction


kinetics at the electrodes. Also, a decrease in the slope of the i–V
curve was observed near 1.3 V at a low operating temperature -0.6
(750–800 1C). Although, the possibility of electron leakage through 20H2-80H2O
the YSZ electrolyte under a reducing environment may drop the 20H2-40H2O-40CO2
cell impedance and result in the higher current densities. But as
Z (ohm.cm )

-0.4 20H2-80CO2
2

shown in Fig. 3 that all experiments were performed at less than 2


10
or equal to 1.5 V which corresponds to an oxygen partial pressure
(pO2) of 5.8  10  31 atm, 1.4  10  29 atm and 2.5  10  28 atm at
750 1C, 800 1C and 850 1C, respectively. The electronic transport in -0.2
//

3
8 mol% Y2O3  ZrO2 (YSZ-8) under these conditions has been
10 10
reported to be about 0.2–0.9% [43] which does not seem signifi- 4 1 0.1
10
cant and thus can be neglected. An alternate possible explanation
0.0
may be the Joule heating that is produced within the cell, when
0.0 0.2 0.4 0.6 0.8 1.0
the operating cell voltage exceeded the thermoneutral voltage.
/ 2
Subsequently, this rise in temperature could partially account for Z (ohm.cm )
the decrease in the ASR over 1.3 V. This explanation appears to be Fig. 4. AC characterization (EIS) of cell at OCV in H2–H2O, H2–H2O–CO2 and H2–CO2
closely matched to the result of steam electrolysis, where the mixtures (a) 850 1C, (b) 800 1C, and (c) 750 1C. Experimental data is shown by the
thermoneutral voltage lies around  1.28 V, but it conflicts with symbols and the fitted data by the solid lines.
A. Mahmood et al. / Journal of Membrane Science 473 (2015) 8–15 13

Depending on temperature, the impedance spectra consists of Fig. 4a–c, the symbols shows the measured experimental data,
either two or three distinguishable arcs with slight differences in whereas, the solid lines represent the fitted data. The area specific
their widths suggesting similar reduction phenomena for each of resistances (ASRs) of all the circuit elements and the correspond-
the gas mixtures at a particular temperature. The high-frequency ing errors are shown in Table 2.
arc can be readily recognized at 750 1C, but disappeared at 800–
850 1C. In contrast, the mid-frequency arc and the low-frequency
3.3.1. Ohmic polarization (Rs)
arc can be easily recognized at all the operating temperatures. It
The series resistance (Rs) corresponds to the ohmic resistance
must be noted that out of the three clearly recognized arcs, the
of the electrolyte, electrodes, as well as the resistance of residual
mid-frequency arc seems to be dominant under all operating
lead wires [44]. The electrical conductivities of the Ni–YSZ, LSM
conditions. The total area specific resistances (ASRs) calculated
and the lead wires are orders of magnitude higher than the YSZ
from the low frequency intercept of the Cole–Cole or Nyquist plots
electrolyte. Therefore, the measured ohmic resistance should be
[29] are in close agreement with the rate of change of the i–V
dominated by the resistance of the electrolyte [45]. It is shown in
curves near OCV (Table 1). However, the slight difference in ASRs
Table 2 that at 850 1C, Rs is  0.1 Ω cm2 and it contributes around
shown in Tables 1 and 2 originates from the measurement of the
26–32% to total ASR. As the electrical conductivity of the YSZ
slope of i–V curves near  0.2 A/cm2 instead of OCV.
electrolyte at 850 1C is  0.03 S/cm [46] thus, the ASR of 11.4 mm
An equivalent circuit model was used to deconvolute the
thick electrolyte (YSZ) is estimated to be 0.038 Ω cm2 which is
impedance spectra of Ni–YSZ supported SOC. The proposed model
only 40% of the values shown in Table 2. It means that the ohmic
consists of lumped elements, including inductance (L), resistance
resistance is not only caused by the electrolyte but also by the
(R) etc., and distributed elements such as a constant-phase element
other cell components. Although, the conductivity of Ni–YSZ
(CPE) of complex capacitance that describes the frequency depen-
electrode may be decreased by agglomeration or evaporation of
dent impedance caused by surface roughness or by non-uniformly
the nickel as Ni(OH)2 under high steam SOEC operation. But this
distributed properties of the irregular electrode surface and is
decrease cannot affect the Rs noticeably as the conductivity of
commonly used in the analysis of EIS to account for the non-ideal
Ni–YSZ cermet still remains orders of magnitude higher than that
behavior of capacitive elements. The circuit elements are shown in
of YSZ electrolyte. Therefore, the difference in the calculated and
decreasing order of frequency in Fig. 5a. A graphical representation
measured Rs may be ascribed to the contact resistance and the
of the experimental data, model fitting and the proposed arcs is
inductance effect of the lead wires as reported by Shin et. al. [44].
shown in Fig. 5b for the better illustration of the deconvolution of
Beside it, Rs appears to be independent of the feed gas composi-
impedance spectra.
tion, but varied with temperature for all of the tested gas mixture
The series resistance Rs stands for the ohmic resistance of the
and followed an Arrhenius dependence that is the typical behavior
combined cell, including electrolyte, electrodes, contact resistance
of an electrolyte or a resistive element.
and the wire resistance. The resistive element R1 describes the
charge transfer resistance of Ni–YSZ electrode, R2 describes the
charge transfer resistance of LSM along with the diffusion resis- 3.3.2. Electrode polarization (Rp)
tance of Ni–YSZ, and R3 represents the gas conversion resistance. Three overlapping polarization arcs represented by R1, R2 and
The elements Q1, Q2 and Q3 are the constant phase elements R3 at high, mid and low frequency can be clearly seen in Fig. 4c.
associated with R1, R2, and R3. It should be noted that two circuit However, at high temperatures (800–850 1C), only two arcs can be
elements R1 and Q1 were added to the selected model shown in distinguished at the mid- and low-frequency region, and the high
Fig. 5 to achieve an excellent fitting of impedance data at 750 1C. In frequency arc seems to have disappeared. In fact, the charge
transfer resistance of the two electrodes typically overlaps when
the SOC is operating at high temperatures [47]. The arcs shown in
Fig. 4 represent the charge transfer and concentration polarization
of the electrodes and their sum is called the total electrode
polarization (Rp). It is not easy to separate the anode and cathode
contributions without using a reference electrode or changing the
operating conditions like feed gas composition and/or tempera-
ture [21]. Although four to five different processes are reported to
be involved in SEMRs [45,48–50], we were able to separate only
three to four processes based on the equivalent circuit model
fitting shown in Fig. 5. The parameters calculated from the fitting
data are shown in Table 2.
It is reported that the gas-solid (adsorption, dissociation, and
desorption) and solid–solid reactions (surface diffusion, oxygen
transfer to/from electrolyte) occurs between 100 and 10 kHz,
whereas, the relaxation frequencies of the diffusion or conversion
process lies below 100 Hz [21,48]. Thus, the information obtained by
changing the experimental conditions in combination with the
characteristic frequencies of each arc is used to assign an electrode
process to each arc. The high and low frequency arcs R1 and R3 with
the characteristic frequency of  5 kHz and 1–4 Hz are assigned to
the activation losses and gas conversion losses at the Ni–YSZ
electrode respectively. Meanwhile, the dominating mid-frequency
arc R2 with a peak frequency of 100–501 Hz is assigned to the oxygen
surface exchange kinetics and O2 diffusivity in the LSM electrode as
well as the gas diffusion resistance of the Ni–YSZ substrate.
Fig. 5. Equivalent circuit model to fit the EIS data at OCV (a), EIS measured under Fig. 4a–c reveals that the high frequency arc (R1) with a
OCV with 20% H2–80% CO2 at 750 1C, Fit data and the proposed arcs (b). relaxation frequency 5 kHz that represents the charge transfer
14 A. Mahmood et al. / Journal of Membrane Science 473 (2015) 8–15

reaction at the Ni–YSZ electrode can only be observed at 750 1C o


-0.3 20H2-80CO2, 850 C OCV
but not at higher temperatures. This means that at Z800 1C, the
1.2V
charge transfer reaction of the Ni–YSZ electrode occurs too fast
1.4V
and show no capacitive part in the Nyquest plot. It is reported that

Z (ohm.cm )
2
without properly considering stray impedance, the charge transfer -0.2
impedance of the Ni–YSZ cermet can substantially overlap with 1
3
the ohmic impedance [44]. Furthermore, the comparable values of 10

//
charge transfer resistance (R1) at TPB of Ni–YSZ for steam, CO2 and -0.1
steam-CO2 electrolysis confirms that a similar type of reaction
mechanism is involved in the electrochemical reduction of steam,
CO2 and steam-CO2 at 750 1C [5,29]. Nevertheless, a slightly 0.0
greater resistance for the CO2-containing mixture could be the 0.0 0.1 0.2 0.3 0.4 0.5 0.6
result of different surface and/or gas diffusion properties of the / 2
Z (ohm.cm )
CO2 and a decrease in the reactant concentration at the Ni–YSZ
TPB by the RWGS reaction. Fig. 6. AC characterization (EIS) of cell at different cell voltages in 20% H2–80% CO2
The dominating mid-frequency arc showed a strong dependence at 850 1C.
on temperature and feed gas composition. It is suggested that the
polarization loss associated with the LSM–YSZ oxygen electrode on the electrochemical performance of the cell. The results
and the diffusion resistance of Ni–YSZ electrode dominates the cell obtained with fuel gas composition of 20% H2–80% CO2 at 850 1C
impedance at OCV. Fig. 4a–c indicates how critical the role of the are shown in Fig. 6. It is evident that the total electrode polariza-
diffusion resistance of Ni–YSZ substrate is in this case. An increase tion (Rp), cell area specific resistance (ASR) and the characteristic
in the width of R2 by increasing the CO2 concentration is shown in frequency of each polarization process shows a strong dependence
Fig. 4a–c at all temperatures. It is therefore believed that an increase on cell voltage. An increase in the cell ASR from 0.37 Ω cm2 to
in the width of the R2 by increasing the CO2 concentration at the 0.41 Ω cm2 and 0.56 Ω cm2 with increasing the cell voltage from
Ni–YSZ electrode is related to the relatively lower diffusivity (more OCV to 1.2 V and 1.4 V can clearly be seen in Fig. 6. But the peak
diffusion resistance) of CO2 as compared to steam or hydrogen and/ frequency of high-, mid- and low-frequency arcs is decreased with
or lower steam contents in CO2 containing mixture. Hence, a higher increasing the cell voltage. The ohmic resistance (Rs) of cell shows,
R2 value is obtained for the feed containing 80% CO2, a medium by contrast, a negligible dependence on the operating cell voltage.
value for the co-electrolysis (40% CO2) and small value for the steam However, a slight increase in Rs might possibly be due the
electrolysis. Moreover, a decrease in temperature not only results in endothermic electrolysis reaction. These results are well consistent
an increase in the width of the mid-frequency arc (R2) but also with the dc-characterization (i–V curve) shown in Fig. 3a.
causes a shift in the peak frequency of the mid-frequency arc from The contribution of various polarization losses in Fig. 6 can be
501 Hz (at 850 1C) to 251 Hz and 100 Hz at 800 1C and 750 1C, qualitatively estimated considering the earlier mentioned charac-
respectively. This may be due to the sluggish reaction kinetics at the teristic frequency ranges (Section 3.3.2). Fig. 6 reveals that the
LSM oxygen electrode as the contribution from the diffusion charge transfer and concentration (gas diffusion and gas conver-
polarization is expected to decrease with decreasing temperature. sion) polarization resistances of Ni–YSZ electrode are significantly
A similar shift in the relaxation frequencies of the electrodes with a increased with increasing the cell voltage whereas the LSM
decrease in temperature have been reported [49]. polarization resistance is decreased. This effect may be the con-
The low frequency arc (R3) below 10 Hz is generally attributed to sequence of a decrease in reactant concentration at fuel electrode
gas conversion [5,51]. The gas conversion impedance is an effect of and an increase in the activity of the oxygen electrode under high
the passage of current on an electrode under a finite gas flow rate. current operation [44,52]. It has to be noted that the increase in
Gas conversion impedance is expected to be detected only in a the polarization resistance of Ni–YSZ particularly the gas concen-
measurement setup where the electrode is characterized against a tration (gas diffusion and gas conversion) impedance is more
stable reference potential [51]. Since we were using the four probe pronounced than the decrease in LSM polarization resistance
system for impedance measurement, it was expected to appear in and hence contributed significantly to the total cell ASR. It means
our impedance data. The low-frequency arc (R3) with a relaxation the fuel electrode impedance mainly controls the electrolysis
frequency of about 0.5–3 Hz can easily be identified at 750–850 1C in performance of SOC under high voltage operation. Thus, a new
Fig. 4a–c. It contributes significantly to the total ASR of the SOC, but fuel electrode design with reduce gas concentration impedance
less than the R2 arc. Also, a change in the width of low-frequency arc could be developed to attain a high performance electrolysis
(R3) with changing the temperature or feed gas composition is operation.
evident in Fig. 4a–c. The temperature and feed gas mixture depen- A similar voltage dependence of the EIS for the steam and co-
dence of the low frequency arc (R3) suggests that this arc is electrolysis was observed under the same set of operating condi-
associated with the conversion polarization of the Ni–YSZ fuel tions, but the total cell ASRs were obviously less than that was
electrode as reported in the literature [48,49]. The increase in the observed with 20% H2–80% CO2 fuel composition.
gas conversion resistance with increasing the CO2 concentration is
related to the relatively smaller diffusion coefficient of the CO2 than
steam or hydrogen. This causes a decrease in the partial pressure of 4. Conclusions
the CO2 at TPB and hence a dramatic increase in the conversion
resistance. In contrast, at low temperatures, the relatively high In this work, we have demonstrated the effect of several
effective diffusivities of the gases leads to an increase in the partial operating parameters on the electrolysis performance of SEMR for
pressure of gases at the TPB and hence a low conversion resistance. steam, CO2 and steam-CO2 mixture. The results confirm that
different performance limiting step is involved depending on the
operating temperature. Although, almost similar processes occurred
3.3.3. Effect of operating voltage on EIS during the steam, CO2 and steam-CO2 electrolysis but the electro-
Impedance measurements were also performed as a function of lysis performance was decreased when increasing the CO2 contents
the operating voltage (E) to study the effect of operating voltage in the feed mixtures, with results of H2-steam 4H2–steam–
A. Mahmood et al. / Journal of Membrane Science 473 (2015) 8–15 15

CO2 4H2–CO2 in decreasing order. EIS results showed that this [18] S. Herring, J.E. O’Brien, C.M. Stoots, G.L. Hawkes, J.J. Hartvigsen, M. Shahnam,
decrease in performance is mainly caused by the high concentration J. Int., Hydrog. Energy 32 (2007) 440–450.
[19] A. Hauch, S.D. Ebbesen, S.H. Jensen, M. Mogensen, J. Electrochem. Soc. 155
polarization at the fuel electrode and less by the activation (2008) B1184–B1193.
polarization of the fuel electrode. [20] A. Hauch, S.D. Ebbesen, S.H. Jensen, M. Mogensen, J. Mater. Chem. 18 (2008)
It is concluded that for the high temperature electrolysis of 2331–2340.
[21] A. Brisse, J. Schefold, M. Zahid, J. Int., Hydrog. Energy 33 (2008) 5375–5382.
steam and CO2 in a SEMR, the operating temperature and mass [22] G. Schiller, A. Ansar, M. Lang, O. Patz, J. Appl. Electrochem. 39 (2009) 293–301.
transportation across the substrate (porous membrane) play a key [23] A.V. Virkar, J. Int., Hydrog. Energy 35 (2010) 9527–9543.
role to limit the performance of a reactor. Though, the demand of [24] R. Knibbe, A. Hauch, J. Hjelm, S.D. Ebbesen, M. Mogensen, Green 1 (2011)
141–169.
highly catalytic electrode materials could not be underestimated,
[25] J.J. Schefold, A. Brisse, F. Tietz, J. Electrochem. Soc. 159 (2012) A137–A144.
but in this research the need of an optimization of the fuel [26] V.N. Nguyen, F. Qingping, U. Packbier, L. Blum, J. Int., Hydrog. Energy 38 (2013)
electrode architecture (thickness, porosity) to minimize the con- 4281–4290.
[27] P. Mocoteguy, A. Brisse, Int. J. Hydrog. Energy 38 (2013) 15887–15902.
centration polarization is investigated to be as important as the
[28] S.D. Ebbesen, J. Høgh, M. Mogensen, in: Proceedings of the Risø International
former one. Energy Conference 2009, Risø-R-1712(EN), (2009).
[29] Z. Zhan, W. Kobsiriphat, J.R. Wilson, M. Pillai, I. Kim, S.A. Barnett, Energy Fuels
23 (2009) 3089–3096.
Acknowledgments [30] S.D. Ebbesen, C. Graves, M. Mogensen, Int. J. Green Energy 6 (2009) 646–660.
[31] S.D. Ebbesen, C. Graves, A. Hauch, S.H. Jensen, M. Mogensen, J. Electrochem.
Soc. 157 (2010) B1419–B1429.
The authors would like to acknowledge the Korea Research [32] S.D. Ebbesen, M. Mogensen, Electrochem. Solid-State Lett. 13 (2010)
Institute of Chemical Technology (KRICT) (grant no. KK-1301-D0) B106–B108.
[33] C. Graves, S.D. Ebbesen, M. Mogensen, Solid State Ion. 192 (2011) 398–403.
for the financial support. [34] P. K-Lohsoontorn, J. Bae, J. Power Sources 196 (2011) 7161–7168.
[35] W. Li, H. Wang, Y. Shi, N. Cai, J. Int., Hydrog. Energy 38 (2013) 11104–11109.
References [36] M. Chen, J.V.T. Høgh, J.U. Nielsen, J.J. Bentzen, S.D. Ebbesen, P.V. Hendriksen,
Fuel Cells 13 (2013) 638–645.
[37] X. Sun, M. Chen, Y.-L. Liu, P. Hjalmarsson, S.D. Ebbesen, S.H. Jensen, M.B. Mogensen,
[1] M.B. Mogensen, in: Proceedings of the19th World Energy Congress 2004, P.V. Hendriksen, J. Electrochem. Soc. 160 (2013) F1074–F1080.
Sydney-Australia 5–9 Sep. (2004). [38] V.A.C. Haanappel, A. Mai, J. Mertens, Solid State Ion. 177 (2006) 2033–2037.
[2] Z. Zhan, L. Zhao, Short communication, J. Power Sources 195 (2010) [39] Y. Jiang, A.V. Virkar, J. Electrochem. Soc. 150 (2003) A942–A951.
7250–7254. [40] F. Bidrawn, G. Kim, G. Corre, J.T.S. Irvine, J.M. Vohs, R.J. Gorte, Electrochem.
[3] X. Yue, J.T.S. Irvine, J. Electrochem. Soc. 159 (2012) F442–F448. Solid-State Lett. 11 (2008) B167–B170.
[4] M.E. Dry, Catal. Today 71 (2002) 227–241. [41] V. A-Restrepo, J.M. Hill, J. Power Sources 195 (2009) 1344–1351.
[5] S.D. Ebbesen, R. Knibbe, M. Mogensen, J. Electrochem. Soc. 159 (2012) [42] C. Stoots, J. O’Brien, J. Hartvigsen, Int. J. Hydrog. Energy 34 (2009) 4208–4215.
F482–F489. [43] J.-H. Park, R.N. Blumenthal, J. Electrochem. Soc. 136 (1989) 2867–2876.
[6] S.D. Ebbesen, M. Mogensen, J. Power Sources 193 (2009) 349–358. [44] E.C. Shin, P.A. Ahn, H. Ho Seo, J.M. Jo, S.D. Kim, S.K. Woo, J.H. Yu, J. Mizusaki,
[7] M. Mogensen, S.H. Jensen, A. Hauch, I. Chorkendorff, T. Jacobsen, in: Proceed- J.S. Lee, Solid State Ion. 232 (2013) 80–96.
ings of the 7th European SOFC Forum (2006). [45] J.-C. Njodzefon, D. Klotz, A. Kromp, A. Weber, E.I. Tiffee, J. Electrochem. Soc.
[8] S.H. Jensen, (Ph.D. thesis), Risø National Laboratory, Roskilde, Denmark, 2006. 160 (2013) F313–F323.
[9] S.H. Jensen, P.H. Larsen, M. Mogensen, Int. J. Hydrog. Energy 32 (2007) [46] T. Suzuki, M. Awano, P. Jasinski, V. Petrovsky, H.U. Anderson, Solid State Ion.
3253–3257. 177 (2006) 2071–2074.
[10] H. Uchida, N. Osada, M. Watanabe, Electrochem. Solid-State Lett. 7 (2004) [47] S.H. Jensen, A. Hauch, P.V. Hendriksen, M. Mogensen, N. Bonanos, T. Jacobsenb,
A500–A502. J. Electrochem. Soc. 154 (2007) B1325–B1330.
[11] H.S. Spacil, U.S. Patent 3503809; (1970). [48] R. Barfod, M. Mogensen, T. Klemensø, A. Hagen, Y.-L. Liu, P.V. Hendriksen,
[12] Y. Matsuzaki, I. Yasuda, J. Electrochem. Soc. 147 (2000) 1630–1635. J. Electrochem. Soc. 154 (2007) B371–B378.
[13] M. Homel, T.M. G̈ ur, J. Koh, A. Virkar, J. Power Sources 195 (2010) 6367–6372. [49] A. Leonide, V. Sonn, A. Weber, E.I. Tiffee, J. Electrochem. Soc. 155 (2008)
[14] X. Yue, J.T.S. Irvine, Solid State Ion. 225 (2012) 131–135. B36–B41.
[15] K. Eguchi, T. Hatagishi, H. Arai, Solid State Ion. 86–88 (1996) 1245–1249. [50] H. Schichlein, A.C. Muller, M. Voigts, A. Krugel, E.I. Tiffee, J. Appl. Electrochem.
[16] J.E. O’Brien, C.M. Stoots, J.S. Herring, J. Hartvigsen, J. Fuel Cell Sci. Technol. 3 32 (2002) 875–882.
(2006) 213–219. [51] S. Primdahl, M. Mogensen, J. Electrochem. Soc. 145 (1998) 2431–2438.
[17] A. Hauch, S.H. Jensen, S. Ramousse, M. Mogensen, J. Electrochem. Soc. 153 [52] P.-A. Ahn, E.-C. Shin, J.-M. Jo, J.-H. Yu, S.-K. Woo, J.-S. Lee, Fuel Cells 12 (2012)
(2006) A1741–A1747. 1070–1084.

Anda mungkin juga menyukai