Professional Documents
Culture Documents
Article history: An experimental study has been conducted at Idaho National Laboratory to demonstrate
Received 22 August 2012 recent improvements in long-term durability of solid oxide electrolysis cells (SOEC) and
Received in revised form stacks. Results of five stack tests are presented. Electrolyte-supported SOEC stacks were
25 September 2012 provided by Ceramatec Inc. and electrode-supported SOEC stacks were provided by
Accepted 25 September 2012 Materials and Systems Research Inc. (MSRI), for these tests. Long-term durability tests were
Available online 16 November 2012 generally operated for durations of 1000 h or more. Stack tests based on technologies
developed at Ceramatec and MSRI have shown significant improvement in durability in the
Keywords: electrolysis mode. Long-term degradation rates of 3.2%/khr and 4.6%/khr were observed for
Solid oxide electrolysis stack MSRI and Ceramatec stacks, respectively. One recent Ceramatec stack even showed
Durability negative degradation (performance improvement) over 1900 h of operation. Optimization
High temperature electrolysis of electrode and electrolyte materials, interconnect coatings, and electrolyteeelectrode
Hydrogen generation interface microstructures contribute to improve the durability of SOEC stacks.
Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
reserved.
* Corresponding author. Tel.: þ1 208 525 5409; fax: þ1 208 525 5754.
E-mail address: James.Obrien@inl.gov (J.E. O’Brien).
0360-3199/$ e see front matter Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijhydene.2012.09.176
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 0 e2 8 21
including sulfur-iodine (SI), high temperature electrolysis evaluation of initial performances and long-term durability in
(HTE), and hybrid sulfur process (HyS), were initially selected the electrolysis mode. The objective of the test program is to
by US DOE for further investigation. The goal was to down identify degradation mechanisms and to find ways to improve
select the most suitable technology for integration with the SOEC stack durability.
next generation nuclear plant (NGNP) for large-scale hydrogen
production [6]. Based on the recommendation of an indepen-
dent review team [6], HTE was selected by the US DOE as the 2. Materials, experimental apparatus and
most appropriate advanced nuclear hydrogen production methods
technology to be supported for further development and early
deployment. The SOEC stacks used in this study were provided by Ceram-
The US Department of Energy, Office of Nuclear Energy has atec Inc. and Materials and Systems Research Inc. (MSRI).
supported development of HTE at the Idaho National Labo- Ceramatec used electrolyte-supported cells for the stacks,
ratory (INL) since 2003. INL has demonstrated HTE at the while MSRI used electrode-supported cells. Three 10-cell
15 kW scale with a hydrogen production rate in excess of Ceramatec stacks and two 5-cell MSRI stacks were tested at
5000 NL/hr [7]. However, long-term performance degradation INL. The configurations of the stacks and testing fixtures are
is the biggest barrier to scale up of HTE technology. Contin- described below in detail.
uous improvements on materials, seals, and microstructures
have been done and various SOEC stacks have been tested to 2.1. Configuration of stacks
investigate ways to mitigate degradation at INL [8e11].
The major problems found in the previous stack tests were The Ceramatec stack consists of 10 planar electrolyte-
air electrode delamination, Cr vapor poisoning, microstruc- supported SOECs, interconnects, flow channels, seal gaskets,
ture degradation, and seal leakage. Air electrode delamination and end plates. An exploded view of the stack is shown in
is often observed during high temperature electrolysis. Fig. 1. The stack is externally manifolded with a cross-flow
Oxygen partial pressure build-up in the electrodeeelectrolyte configuration. The air outlet face is open, allowing for the
interface is a proposed mechanism leading to air electrode placement of supplemental instrumentation such as inter-
delamination [12e15]. Cr vapor poisoning mainly comes from mediate voltage taps and internal stack thermocouples, as
the stack interconnects which are usually made of a chro- needed. The standard materials used in Ceramatec SOECs are
mium-rich alloy, typically a ferritic stainless steel or scandia-stabilized zirconia (ScSZ) electrolyte, Ni-Ceria
specialty alloy such as Crofer. Investigations of the interaction steam/hydrogen electrode, and LaeCoeFe oxide based
between metal interconnects and cell components and perovskite air electrode. Each SOEC has an active area of
methods to mitigate poisoning are well documented [16e21]. 45 cm2. Corrugated metallic flow fields fabricated from nickel
Microstructure degradation caused by interactions among the and ferritic stainless steel (alloy 441) were used for gas
material constituents results in a decrease in conductivity and distribution as well as current collection on the steam/
electrochemical activity [22,23]. Seal leakage is another hydrogen side and air side, respectively. The separator plate is
important factor that can cause identical cells to behave also ferritic stainless steel. The ferritic stainless components
differently in a stack. Seal failure often happens after thermal are treated with a rare-earth coating/firing to support devel-
cycles during the tests. Mismatch of coefficient of thermal opment of a stable electronically conducting protective oxide
expansion (CTE) between the seal and other components can layer. Several different gasket materials were tested to mini-
cause cracks of the seal [24,25]. mize leakage issue.
Much of the research on SOECs studying performance and An exploded view of a 5-cell MSRI stack is provided in Fig. 2.
durability stays at single cell level. Besides our efforts, only The stack is internally manifolded with process gases entering
limited investigation addressed the performance and dura- and exiting at the bottom of the stack, flowing in an inverted-
bility issues of SOEC stacks. Ebbesen et al. recently reported U pattern in cross-flow. The oval flow passages around the
SOEC stack tests for steam electrolysis and CO2/steam co- periphery of the cells, interconnects and gaskets provide gas
electrolysis [26]. The stack and glass seals used in their tests distribution to the micro-channel flow patterns in the inter-
were designed by Topsøe Fuel Cells. Trace amounts of impu- connects. The MSRI stack uses electrode-supported cells. Each
rity in the inlet stream was claimed to have significant impact MSRI stack has five identical Ni/YSZ-supported SOECs
on the performance and durability of the stacks. (Ni/YSZ-YSZ-LSCF) with an active area of 100 cm2. MSRI SOECs
This paper provides a summary of results for the improved were fabricated using their state-of-the-art fabrication
durability of SOEC stacks tested at INL. Stack tests included processes and the materials and the microstructure of the
22 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 0 e2 8
cells was optimized specifically for operation in the electrol- a maximum steam content of about 55%. One advantage of
ysis mode. Micro-channel flow patterns (12 cm 12 cm) were using the CEM system to deliver steam is that it can achieve
directly machined into both surfaces of each interconnect, higher steam content, as high as 90%. The other advantage of
providing gas separation, current collection, and flow distri- using the CEM system is that the steam content can be
bution. A contact aid was used to enhance electric conduction. adjusted promptly compared to the humidifier, which usually
Sealing was provided by a means of compliant seal developed takes more than 1 h before stabilizing at a specified bath
by MSRI. temperature. The carrier gases used in the CEM are either pure
hydrogen or a hydrogen-nitrogen mixture. Two CEM systems
with different flow ranges, 100 g/hr and 1000 g/hr, were used
2.2. Experimental apparatus and methods
in the tests to evaporate liquid demineralized/deionized (DI)
water. The carrier gas contains at least 10% hydrogen in order
A piping and instrument diagram (P&ID) for stack testing is
to prevent oxidation of the steam/hydrogen electrode.
shown in Fig. 3. Steam is generated either by bubbling a carrier
Humidity transducers are placed at the inlet and outlet of the
gas through the humidifier, or by a controlled evaporation and
stack in the steam/hydrogen line to provide a direct
mixing (CEM) system (Bronkhorst USA Inc.). Bubbling gas
measurement of steam consumption rate, which is equal to
through the humidifier is simple, but it is limited to
the hydrogen production rate on a molar basis. The humidity
measurements therefore provide an independent verification
of the hydrogen production rate for comparison to the rate
determined by application of Faraday’s law, based on electric
current.
The test fixture for Ceramatec stacks is shown in Fig. 4. It
shows a Ceramatec stack, mounted on its inconel test fixture,
which rests on the furnace base. The power leads are inconel
rods insulated with alumina tubing. The hydrogen/steam and
air inlet tubes are coiled to provide additional length for heat
transfer upstream of the stack. The cross bar on top of the
stack is used to apply a mechanical compression load. The
load can be adjusted by tightening the springs underneath the
test stand through the linkage rods. More detailed description
of this experimental set-up can be found in [9].
A totally different test fixture was developed for testing
MSRI stacks. Fig. 5 shows a 3D overview drawing of the test
fixture. Two inconel bars were pre-welded vertically onto two
Fig. 2 e Exploded view of an MSRI 5-cell stack. inconel plates, working as the current collectors. The bottom
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 0 e2 8 23
Fig. 3 e Piping and instrument diagram for the INL high temperature steam electrolysis.
alumina tube sitting on a stand (not shown) supports the conditions for each stack test are listed in Table 1. In the
whole test fixture. The top alumina tube is used to transfer the Ceramatec stack tests, steam was generated by bubbling gas
mechanical compression load from the springs and to insulate through a humidifier. Typically 56% inlet steam content was
the top inconel plate from the other parts. The spring loading obtained by setting the humidifier bath temperature to 80 C. In
assembly stays outside the furnace, so that the compression the MSRI and St. Gobain stack tests, steam was produced by
load can be adjusted during the curing process. The stack was a CEM system. 70% steam content was recommended for oper-
sandwiched between two inconel plates, serving as the ating MSRI stacks as well as maintaining the reducing condition
current collectors and the end plates for compression. Coiled at the electrode.
tubes were welded onto the bottom inconel plate for pre- For all the tests, Agilent units (34972 A LXI data acquisi-
heating of the inlet gases. Internal gas flow channels were tion/data logger switch unit with two 34901 A 20 channel
also machined into the bottom inconel plate for flow multiplexer modules) were used for data acquisition. A DC
distribution. power supply (Lambda GEN20-38) was used to provide elec-
Prior to each test, the furnace temperature was slowly ram- trical power to the stacks during electrolysis. An electronic
ped up to 800 C or 850 C with dry gases flowing through the load (AMREL FCL400-20-100E) was used for operating the
stack. After reaching the operating temperature, each stack was stacks in the fuel cell mode. All the instruments were
conditioned in a reducing atmosphere for curing of the seals and remotely controlled by computer using a Labview based
reduction of the steam/hydrogen electrode. During the electrode virtual instrument (VI). All of the ultra high purity (UHP)
reduction process, the downstream humidity sensor indicates hydrogen cylinders, the liquid nitrogen dewars, and a pre-
an elevated dew point temperature associated with reduction of mixed safety gas cylinder were purchased from Norco Inc.
nickel oxide to nickel metal in a hydrogen atmosphere, which Air was supplied by the air compressor in the building.
produces steam. By the end of the curing and reduction
processes, both the inlet and the outlet humidity sensors will
indicate 0 C dew point. After the conditioning procedure, the 3. Results and discussion
furnace temperature was set at 800 C. Then the initial perfor-
mance of each stack was evaluated by DC polarization scans in 3.1. Ceramatec stacks
the electrolysis mode and the fuel cell mode. Both electrolysis
and fuel cell sweeps were conducted at ambient pressure with A series of DC polarization sweeps were performed with
different steam concentrations. The long-term durability test in different steam contents before the long-term tests. The
the electrolysis mode was performed immediately after the initial VI sweeps were conducted in both the fuel cell and the
initial performance characterization. The long-term operating electrolysis modes, in order to evaluate the initial
24 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 0 e2 8
H2 N2 Air
a N2 was used instead of air as the air-electrode sweep gas during MSRI stack tests.
Fig. 8 represents results of a 1000 h electrolysis test of Fig. 9 illustrates the performance of Ceramatec stack #3
Ceramatec stack #2. The stack was operated galvanostatically during long-term electrolysis. This stack was operated galva-
at 0.25 A/cm2 during the long-term test. The average per-cell nostatically at 0.25 A/cm2 for about 1900 h and it was by far the
ASR increased from 1.69 U cm2 to 1.77 U cm2. Stack perfor- best performing Ceramatec stack that has ever been tested at
mance increased during the first 70 h, which may be due to INL. Unfortunately the stack test was shut down due to an off-
initial conditioning of the stack. Thereafter the stack perfor- normal event that happened at 1900 h and the performance
mance started degrading. Based on the time period from 70 h did not recover after that. The stack degraded over the first
to the end of the test, the degradation rate is 6.87%/khr based 200 h, but thereafter the performance started increasing. After
on voltage. The degradation rate is slightly higher than that of 1600 h of operation, the stack voltage became stabilized. Based
the stack #1, mainly due to nonuniform behaviors of the cells on the time period from 200 h to the end of the test, the
in the stack. The intermediate voltages shown in Fig. 8 indi- performance increased 7.4% over 1700 h. Inlet and outlet dew
cate that some cells degraded faster than others. As point temperatures were constant over the period, indicating
a comparison, intermediate voltage #5 degraded 11.3% over a constant hydrogen production rate consistent with the
1000 h, while intermediate #1 voltage degraded only 1.5% over constant current conditions of the test. The overall area
1000 h. It is difficult to identify the exact reason why those specific resistance (ASR) was used to quantify cell perfor-
cells behaved differently. It might be attributed to the defect mance. ASR is calculated based on the following equation:
formed during cell fabrication and stack assembling, the flow
ASR ¼ ðV Vo Þ=J (1)
distribution in the stack, the leakage, etc.
Compared to the stack #1, a modified composite functional where J is the current density; V and Vo are the operating
layer that incorporates YSZ was added in each cell of the stack voltage and the open circuit voltage, respectively. This defi-
#2 to improve bonding between the electrolyte and the air nition of ASR is actually an apparent ASR since it is based on
electrode. In addition, a novel coating was applied on each the observed open-cell voltage at zero current density. The
interconnect to further mitigate oxidation and Cr vapor initial and final average ASR values for this test were
formation. Either one of the improvement did not further 1.59 U cm2 and 1.76 U cm2, respectively. The ASR reached
mitigate the degradation, at least for some cells. a peak of 1.92 U cm2 after a brief power outrage that occurred
around 500 h. Intermediate voltages in stack #3 behaved
differently over the first 500 h. After that time, the
Fig. 9 e Ceramatec stack #3 1900 h test at 0.25 A/cm2 in Fig. 11 e MSRI stack #1 1000 h test at 0.2 A/cm2 in
electrolysis mode. electrolysis mode.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 0 e2 8 27
values, corresponding to higher energy efficient during stacks may be more robust during long-term operations. Our
electrolysis. experience shows that the electrolyte-supported stacks have
Fig. 11 represents the results of a long-term electrolysis test better chance to survive through thermal cycles due to power
on MSRI stack #1. The stack was operated galvanostatically at outages. The electrode-supported stacks may also be more
0.2 A/cm2 for 1000 h. The overall degradation rate was 8.9%/khr sensitive to unintentional redox events which can cause irre-
based on voltage, which is calculated based on linear curve versible damage to the thick steam-hydrogen electrodes.
fitting. The average ASR increased from 0.41 U cm2 to 0.90 U cm2.
This degradation rate was higher than expected, mainly due to
the rapid degradation of the top cell, as shown in Fig. 11. The
degradation rate of the top cell was 26.8%/khr, while other cells 4. Conclusion
were close to 3.7%/khr. The rapid degradation on the top tell
might be caused by Cr poisoning since it has direct contact with Significant improvements on the durability of the SOEC stacks
inconel current collector on the air side. A special surface have been achieved. The state-of-art stack design, using
treatment was suggested to mitigate the top cell degradation. advanced materials, and the novel treatment of interconnects
Compared to the test of MSRI stack #1, the operating all help to mitigate the degradation of the SOEC stacks in elec-
conditions were adjusted in the test of the stack #2. The fuel- trolysis operations. Three electrolyte-supported SOEC stacks
side carrier gas flow rate was decreased from 5 slpm to provided by Ceramatec were tested. The degradation rates of
0.7 slpm by switching from the large CEM system to the small the first two stacks were between 4.62%/khr and 6.87%/khr. The
CEM system. Also, pure hydrogen was used instead of third stack demonstrated exceptional performance, which
hydrogen/nitrogen mixture. The adjustment resulted in showed negative degradation over 1900 h test. To the best of the
a higher steam utilization value of 40%. Fig. 12 shows the authors’ knowledge, this is the first SOEC stack ever docu-
results of a long-term electrolysis test on MSRI stack #2. The mented in the literature for which the performance increased
stack was operated galvanostatically at 0.2 A/cm2 for 1100 h steadily during a long-term electrolysis tests. However, the
with an overall degradation rate 3.2%/khr. The improvement factors that caused this exceptional result remain for further
of the overall stack durability was mainly attributed to the investigations. Two electrode-supported SOEC stacks provided
modification of the top cell. Examination of Fig. 12 reveals that by MSRI showed promising long-term durability. The most
all the cells behaved similarly during the long-term test, recent stack demonstrated an overall degradation rate of only
which indicates that the treatment on the top cell significantly 3.2%/khr, which was the best result obtained to date at INL. A
mitigated its degradation. An alternative way to mitigate the degradation rate of w1%/khr or lower is needed for commer-
top cell degradation is by adding a dummy cell to separate the cialization of SOEC stacks for high temperature electrolysis.
top cell and the inconel plate. The dummy cell works as More stacks are under testing at INL to verify various strategies
a barrier as well as a current collector. for further mitigating the degradation.
Generally, both the electrode-supported and the electrolyte-
supported SOEC stacks tested showed improved durability in
the long-term HTE operations. No general conclusion can be
made on which cell configuration is better than the other, since Acknowledgments
each configuration has its own advantages. The electrode-
supported SOEC stacks have much lower ASR values due to This work was supported by the U.S. Department of Energy,
the thin electrolyte layers, which results in a lower power input Office of Nuclear Energy, Next Generation Nuclear Plant
requirement for a specified current density or hydrogen Program under DOE Operations Office Contract DE-AC07-
production rate. However, the electrolyte-supported SOEC 05ID14517.
28 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 2 0 e2 8