international journal of hydrogen energy 34 (2009) 6603–6608

Available at www.sciencedirect.com

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

Proton exchange membrane (PEM) electrolyzer operation

under anode liquid and cathode vapor feed configurations

Scott D. Greenway, Elise B. Fox*, Amy A. Ekechukwu

Savannah River National Laboratory, 999-2W, Aiken, SC 29808, United States

article info abstract

Article history: Proton exchange membrane (PEM) electrolysis is a potential alternative technology to crack
Received 6 April 2009 water in specialty applications where a dry gas stream is needed, such as isotope
Received in revised form production. One design proposal is to feed the cathode of the electrolyzer with vapor phase
10 June 2009 water. This feed configuration would allow isotopic water to be isolated on the cathode side
Accepted 13 June 2009 of the electrolyzer and the isotope recovery system could be operated in a closed loop.
Available online 17 July 2009 Tests were performed to characterize the difference in the current–voltage behavior
between a PEM electrolyzer operated with a cathode water vapor feed and with an anode
Keywords: liquid water feed. The cathode water vapor feed cell had a maximum limiting current
Electrolysis density of 400 mA/cm2 at 70  C compared to a current density of 800 mA/cm2 for the anode
Proton exchange membrane liquid feed cell at 70  C. The limiting current densities for the cathode water vapor feed cell
Cathode vapor were similar to those predicted by a water mass transfer model. It is estimated that
Isotope separation a cathode water vapor feed electrolyzer system will need to be between 5 and 8 times larger
in active area or number of cells than an anode liquid feed system.
ª 2009 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights
reserved.

1. Introduction One of the technological issues with the use of a PEM

electrolyzer in isotope separation is the development of an
Commercially available PEM electrolyzers can generate H2 overall system design that will maintain the oxygen and
pressures of up to 200 psi and have production rates from 400 hydrogen streams separate, maintain a closed loop hydrogen
to 7900 kg/yr [1]. These units have the advantages of system, and allows easy integration of the electrolyzer. One
producing high purity hydrogen (99.999%) at high conversion design option for the isotope recovery system that would meet
efficiency (95%). Fast response time as well as the ability to these criteria is to feed water vapor to the PEM electrolyzer on
accept variations in load makes the technology desirable for the cathode side of the membrane. In order to keep the
on-demand delivery and combining with photovoltaic power hydrogen contained in a closed loop, hydrogen gas would be
sources [2]. Disadvantages of operation include high capital used as the carrier for the isotopic water vapor. Oxygen gas
cost and the hydrogen production costs. The production costs would be produced at the anode and be exhausted by natural
are dependent on the cost of electricity. A large energy supply convection or be removed by a sweep gas, such as nitrogen. A
is often needed due to high anode overpotentials [3]. In certain proposed operating configuration, using tritiated water, is
specialty applications, such as isotope separation, a dry gas shown in Fig. 1, where T represents the hydrogen isotope
supply is needed and the capital and energy costs have less tritium. The water vapor would diffuse across the membrane
effect on the relative efficiency of the system. and be oxidized at the anode to produce oxygen gas. The

* Corresponding author. Tel.: þ1 803 507 8560.

E-mail address: [email protected] (E.B. Fox).
0360-3199/$ – see front matter ª 2009 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijhydene.2009.06.048
6604 international journal of hydrogen energy 34 (2009) 6603–6608

+ -
2. Experimental

2.1. Cathode water vapor feed operation

N2 HTO+HT
Gas Membrane Gas
Diffusion Diffusion A bench scale PEM electrolyzer in cathode water vapor feed
Layer Layer configuration was tested to determine the electrochemical
performance of the cell. Nitrogen was used as the carrier gas
for water vapor to the cathode. The dry flow rate of nitrogen
for all cathode water vapor feed experiments was 1000 sccm.
When using a 1000 sccm carrier gas flow rate, water vapor was
carried into a cell at a rate well above 1.3 stoich at all
temperatures. Nitrogen was humidified using a Nafion tube
N2+O2 HTO+HT humidifier from Fuel Cell Technologies. The humidifier
temperature and water level were controlled throughout the
experiments. The humidifier temperature was set to the
Anode Cathode desired dewpoint for each experiment. All cathode water
2HTO O2 2H 2T 4e 2H 2T 4e 2HT vapor feed experiments were performed at 100% relative
humidity (RH). To prevent condensation, a heated transfer
Fig. 1 – Illustration of a cathode water vapor feed PEM line carried the gas from the humidifier to the electrochemical
electrolyzer, where T represents tritium, an isotope of cell. Before entering the cell, the dewpoint of the inlet gas was
hydrogen. measured using a humidity sensor from Vaisala HMT 334
humidity sensor. The humidity sensor was mounted in
a heated and insulated T fitting that was thermostated at
protons and cations produced by the reaction would be con- 75  C. The outlets of the cell were at atmospheric pressure.
ducted back to the cathode and reduced to form isotopic The cell exhaust was vented into a fume hood with adequate
hydrogen gas. When an electrolyzer in this configuration is ventilation to maintain the atmospheric concentration of
operated such that it electrolyzes all the isotopic water that hydrogen well below 1%. The cathode water vapor feed
crosses the membrane, the isotopic water and hydrogen gas equipment described above was integrated and assembled by
should be confined to the cathode side of the electrolysis cell. Giner Electrochemical Systems, Inc. and operation of the
Operating the cathode water vapor feed in this mass transport system was controlled using a LabView program. All tests
limited condition will limit the reaction rate in the cell to the were run with DI water as a surrogate for tritiated water.
rate of water diffusion from the cathode to the anode minus
the rate of water electroosmotic drag from the anode to the 2.2. Anode liquid feed operation
cathode. The cathode water vapor feed configuration is
significantly different than traditional PEM electrolyzer oper- In the anode liquid feed configuration, water was fed to the
ation with a feed of liquid water to the anode [4]. With a liquid anode of the electrolyzer from a temperature controlled water
feed of water to the anode, the reaction rates in the cell at reservoir. A Thermo Scientific NESLAB 7 constant temperature
a given cell voltage are limited more by the kinetics of the bath was used as the water reservoir. Water was pumped from
reactions and the ohmic losses in the membrane and are not the reservoir to the cell using a Watson–Marlow 403U/R1
significantly limited by mass transfer. peristaltic pump. Flexible tubing with a 1/1600 bore was used
Previous research on vapor electrolysis has been limited to for delivery of the water to the cell. With this size tubing, the
high temperature solid oxide electrolyzers operating at the pump accurately delivered water flow rates between 2 ccm
anode [5–7]. While this work focuses on the experimental and 11 ccm as per the manufacturer’s directions. During
comparison of a cathode water vapor feed electrolyzer with anode liquid water feed experiments, nitrogen with a 25  C
a traditional anode liquid feed system at 80  C or below using dewpoint was flowed through the cathode. The dry nitrogen
PEM based electrolysis. These experimental results are flow rate during these experiments was 500 sccm. The heated
compared with a model for mass transfer in a cathode water transfer line was heated to 75  C to keep water from
vapor feed electrolyzer that was developed in previous work condensing.
[8]. Models of the cathode water vapor feed electrolyzer
performance were developed using physics that are well 2.3. Electrochemical cell
accepted in PEMFC systems [9]. Polarization (current–voltage)
testing of a single electrolysis cell is performed to compare the The cells tested had a 50 cm2 circular active area cell and were
steady-state performance of a cathode water vapor feed cell prepared by Giner Electrochemical Systems, Inc. The cell
with an anode liquid feed cell for design purposes. The housing was made out of aluminum and was heated by sili-
polarization of a cathode water vapor feed electrolyzer was cone heating pads placed on the outside of the cell. The
characterized for different cell temperatures at 100% relative heating pads were circular and of similar size as the active
humidity for the cathode feed gas. For comparison, an iden- area. The temperature of the cell was monitored by a ther-
tical cell was operated with an anode liquid water feed at mocouple within the cell housing. The Membrane Electrode
different cell temperatures and water flow rates. Assembly (MEA) in the cell had 2 mg/cm2 of Pt catalyst at the
 international journal of hydrogen energy 34 (2009) 6603–6608 6605

cathode and a 2 mg/cm2 of PtIrO2 catalyst at the anode. The 80

membrane in the cell was Pt-treated Nafion 117 (1100 equiv

Dewpoint Temperature (°C)

weight, 7 mils in thickness). Porous titanium gas diffusion 70
media were used for mass transport and electrical contact.
The cell used for cathode water vapor feed experiments only 60
had an anode oxygen outlet port and thus did not allow for
testing with an anode sweep gas. The cell used for anode 50
liquid water experiments had water inlet and outlet ports on
the cathode. Both the cathode water vapor feed cell and the 40
anode liquid feed cell were assembled by Giner Electro-
chemical Systems, Inc. 30

2.4. Polarization testing 20

0:00 0:30 1:00 1:30 2:00
Time (hr:min)
The cell polarization was tested using a Bio-Logic HCP-803
high-current potentiostat capable of applying currents up to Fig. 2 – Cathode humidifier dewpoint temperature profiles
80 A and voltages of 3 V. The anode of the cell was connected with time. Humidifier set point (>) 30 8C (,) 40 8C (B) 50 8C
to the potentiostat as the working electrode and the cathode (6) 60 8C (3) 70 8C.
of the cell were connected as the counter and reference elec-
trodes. After setting the flows and temperatures for an
experiment, the cell was allowed to come to thermal equilib-
rium for 15–20 minutes with no electrochemical load. At the corresponding spikes in the cell current. The spikes in the cell
start of the electrochemical testing, the cell voltage was held current are caused because the membrane water content
at 1.8 V for 30 minutes to allow concentration gradients in the increases with relative humidity and as the membrane water
cell to fully develop. The cell voltage was changed in a step- content increases more water diffuses across the membrane
wise fashion between 1.80 V and 1.20 V with between 2 and to react at the anode. These results show that maintaining
12 minutes given for the cell to achieve steady-state. In good control of the inlet gas humidification in a cathode water
general, longer step times were used in the cathode water vapor feed electrolyzer is important to achieve a constant
vapor feed configuration because humidifier transients occa- reaction rate.
sionally caused data fluctuation. The variation of the inlet dewpoint also led to difficulties in
interpreting the polarization data collected during cathode
water vapor feed experiments. The polarization curves were
3. Results and discussion performed in using fixed voltage control. For experiments at
30  C and 40  C, the points on the polarization curve were
An important consideration in cathode water vapor feed taken where the current density was stable. For the 50–70  C
experiments is the relative humidity of the cathode feed gas. data, the points on the polarization curve were determined
This is due to potential degradation of the Nafion membrane from average current density values at each voltage.
under low humidification conditions [10,11]. The humidity Fig. 3 shows the current/voltage behavior (i.e., polarization)
from the Fuel Cell Technologies humidifier is measured by of the electrolysis cell at temperatures between 30  C and
a dewpoint meter to monitor its variability. Representative
samples of the transient gas dewpoint data collected at
humidifier temperatures between 30  C and 70  C are shown 450
in Fig. 2. The flow rate of nitrogen through the humidifier was 400
1000 sccm for all cases. The profiles show that the humidifier
Current Density (mA/cm2)

350
does not produce a steady dewpoint and that the transient
spikes seen in the dewpoint of the cathode feed gas depend on 300
the humidifier temperature. It is believed that the variability
250
in the dewpoint temperature is related to the construction of
the humidifier. In the humidifier, the dry gas flows through 200
a long coil of Nafion tubing submerged in heated water. As the
150
gas travels through the tube, water diffuses through the
Nafion and saturates the gas. The saturated gas then flows out 100
of the humidifier and to the cell. The thermocouple for 100% RH
50
monitoring the humidifier temperature is placed in the gas Nafion 117
0
space of the humidifier near the gas exit. The location of the
1.2 1.4 1.6 1.8 2.0
thermocouple in the humidifier may be one possible cause of
Cell Voltage (V)
the dewpoint instability. If there are temperature differences
between the liquid temperature and the temperature of the Fig. 3 – Cathode water vapor feed electrolyzer cell
gas in the humidifier then the humidity transients may result. polarization at 100% RH for temperatures between 30 8C
The spikes in the cathode inlet dewpoint temperature cause and 70 8C. (>) 30 8C (,) 40 8C (6) 50 8C (B) 60 8C (3) 70 8C.
6606 international journal of hydrogen energy 34 (2009) 6603–6608

70  C. The polarization for the cell was tested in the range 800
10 ccm - anode liquid
between 1.20 V and 1.80 V. In the voltage range below
water flow rate

Current Density (mA/cm2)

approximately 1.4 V the cell is kinetically limited and the
current increases exponentially with the cell potential. 600
Between 1.4 V and 1.7 V the cell is transitioning to a mass
transfer limited mode of operation. Above 1.7 V, the cell
current is completely limited by the diffusion rate of water 400

across the membrane and further increases in the cell voltage

do not result in higher cell current. The steady-state current
200
that is reached above 1.7 V is known as the mass transfer
limited current density. At the mass transfer limiting current 25°C Cathode Dewpoint
Nafion 117
density, the rate of water diffusion across the membrane
0
minus the rate of electroosmotic drag is equal to the reaction 1.2 1.4 1.6 1.8 2.0
rate of water at the anode. Cell Voltage (V)
As the cell temperature is increased in cathode water vapor
feed experiments, the current density of the reaction increases Fig. 5 – Anode liquid water feed electrolyzer polarization at
in both the kinetically limited region and the mass transfer temperatures between 30 8C and 70 8C with a liquid water
limited region. The increase in current density in the kineti- flow rate of 10 ccm. (>) 30 8C (,) 40 8C (6) 50 8C (B) 60 8C (3)
cally limited region is due to the dependence of the exchange 70 8C.
current density on temperature. The increase in the limiting
current with temperature is due to the dependence of the
membrane water diffusion coefficient on temperature. The for the cathode vapor feed cell. Also, in the anode liquid feed
temperature dependence of the limiting current density is system, a mass transfer limiting current was never reached. In
shown in Fig. 4. Also shown in Fig. 4 are limiting current the kinetically limited region of the curve, the current density
density predictions from a 1D mass transfer model [8]. The produced by the cell does not vary with temperature as much
experimental limiting current densities follow the same as for the water vapor feed cell. Also, a limiting current density
general trend as the mass transfer model, but show a higher was not reached in the voltage range tested. At cell voltages
dependence on the cell temperature. The differences between between 1.4 and 1.8 V, the polarization was mostly linear. The
the mass transfer model and the experimental data may be linear polarization is due to the ohmic losses in the Nafion
due to condensation in the pre-heater line caused at cold spots membrane. These ohmic losses decrease as the temperature
at high temperatures. increases. This is consistent with the fact that the conduc-
For comparison with the cathode water vapor feed polari- tivity of Nafion increases with temperature [12]. The anode
zation data, anode liquid feed testing of a similar PEM elec- liquid water feed during the experiments in Fig. 5 was 10 ccm.
trolyzer was conducted at temperatures between 30  C and The 10 ccm water flow rate was large enough that at all
70  C. Fig. 5 shows the results of polarization testing of the experimental conditions there was more than 30 times the
liquid feed electrolyzer for temperatures in this range. The amount of water reacted flowing to the cell. The relationship
current densities generated in the anode liquid feed cell were of the stoichiometric excess ratio to the cell current density
significantly larger than in the cathode water vapor feed cell. for water flow rates of 10 ccm and 2 ccm is shown in Fig. 6.
At 1.8 V, the current densities for the anode liquid feed cell The cell polarization for a liquid water feed of 2 ccm is
were all above 500 mA/cm2 compared to less than 110 mA/cm2 shown in Fig. 7. The cell polarization with the 2 ccm water
flow rate was nearly identical to the 10 ccm case. A
600

1000
Current Density (mA/cm2)

500
Stoichiometric Excess Ratio

400

100
300

200
10
100

0
280 300 320 340 360 380
1
Temperature (K) 0 200 400 600 800 1000
Current Density (mA/cm2)
Fig. 4 – Temperature dependence of the mass transfer
limiting current density for a cathode water vapor feed PEM Fig. 6 – Relationship of the stoichiometric excess ratio and
electrolyzer and comparison to model predictions. (>) Exp. current density for a 50 cm2 cell for liquid water feeds of
limiting current densities (–) Model predictions. 2 ccm and 10 ccm. (,) 2 ccm (6) 10 ccm.
 international journal of hydrogen energy 34 (2009) 6603–6608 6607

800 600
2 ccm - anode liquid Cell Temperature
water flow rate 30°C
Current Density (mA/cm2)

Current Density (mA/cm2)

600 Nafion 117
400

400

200
200
25°C Cathode Dewpoint
Nafion 117
0 0
1.2 1.4 1.6 1.8 2.0 1.2 1.4 1.6 1.8 2.0
Cell Voltage (V) Cell Voltage (V)

Fig. 7 – Anode liquid water feed electrolyzer polarization at Fig. 9 – Cell polarization comparison for anode liquid feed
temperatures between 30 8C and 70 8C with a liquid water and cathode vapor feed PEM electrolyzer cells at 30 8C. (,)
flow rate of 2 ccm. (>) 30 8C (,) 40 8C (6) 50 8C (B) 60 8C (3) Anode liquid feed; (6) Cathode vapor feed.
70 8C.

density is desired and if it was determined that the higher

comparison case illustrating the similarity in polarization voltage did not significantly affect the lifetime of the MEAs.
with varying water flow rate at 70  C is shown in Fig. 8. Therefore, the cell current density produced by the anode
The lack of variation in cell polarization with liquid water liquid water feed system and thus the water reaction rate
flow rate demonstrates that the cell is capable of operating at could be between 5 and 8 times larger than a similarly sized
low anode flow rates without mass transfer limitations. cathode water vapor feed system. To get an equivalent water
Maintaining a low water feed rate during the operation of processing rate between the two systems, either the reaction
a PEM electrolyzer allows the amount of water recirculation in area of the cathode water vapor feed system or the number of
the system to be minimized. Future testing could investigate cells used for processing the water would need to be
using anode water flow rates to near stoichiometric flow rates. increased proportionally to the difference in current density.
The mass transfer model presented in previous work [8] If it is assumed that the tritium system needs to process
predicted that the current for an anode liquid water feed 190 kg of tritiated water per year, this means that a single
electrolyzer would be around 8 times larger than the limiting electrolysis cell would need to operate at an average current
current in a cathode water vapor feed electrolyzer. The of 65 A to meet this demand. Theoretically, all of this water
difference in the cell polarization between these two feed could be processed by a single 50 cm2 cell at 1300 mA/cm2 if
configurations is shown in Fig. 9. the cell operated continuously with an anode liquid water
The mass transfer limiting current density for the cathode feed. More likely, is that several cells would be needed in
water vapor feed system is around 92 mA/cm2 while the order to handle a more transient water processing rate. In
current for the cathode water vapor feed system is near contrast, a cathode water vapor feed system would require 5–
475 mA/cm2 at 1.8 V at 30  C. This current density could most 8 cells operating continuously to process the 190 kg/yr of
likely be increased to 1000–1400 mA/cm2 if a higher current tritiated water and this would be increased if the processing
rate was transient.

1000
Cell Temperature
70°C 4. Conclusions
Current Density (mA/cm2)

800
Experiments have been conducted to characterize the
600 current–voltage (i.e. polarization) operating characteristics for
a Proton Exchange Membrane (PEM) electrolyzer. As predicted
400 by a mass transfer model, a PEM electrolyzer fed by a cathode
water vapor feed reaches a limiting current at potentials
above 1.6 V. Tests performed on a single Nafion 117 MEA
200
25°C Cathode Dewpoint showed that the limiting current could be as high as 400 mA/
Nafion 117 cm2 at 70  C. The experimental values for the limiting current
0
density were similar to those predicted by a mass transfer
1.2 1.4 1.6 1.8 2.0
model for a PEM electrolyzer that was previously published.
Cell Voltage (V)
The differences between predictions and the experimental
Fig. 8 – Anode liquid water feed electrolyzer polarization at data are hypothesized to be due to lack of consideration for
70 8C with flow rates of 2 ccm and 10 ccm. (>) 2 ccm (,) other significant mass transfer limitation sources or inappli-
10 ccm. cability of fuel cell operating parameters such as the
6608 international journal of hydrogen energy 34 (2009) 6603–6608

electroosmotic drag coefficient. Experiments with an anode [3] Choi P, Bessarabov DG, Datta R. A simple model for solid
liquid water feed with a similar cell showed that the current polymer electrolyte (SPE) water electrolysis. Solid State
density was not limited by mass transfer up to 1.8 V. At 1.8 V, Ionics 2004;175:535–9.
[4] Barbir F. PEM electrolysis for production of hydrogen from
the current density was 800 mA/cm2 and was still in the ohmic
renewable energy sources. Sol Energy 2005;78:661–9.
region where current increases linearly with voltage. The [5] Ni M, Leung MKH, Leung DYC. Parametric study of solid
50 cm2 cell could possibly reach current densities of 1200– oxide steam electrolyzer for hydrogen production. Int
1400 mA/cm2 before mass transfer limitations occur. J Hydrogen Energy 2007;32:2305–13.
[6] Sakai T, Matsushita S, Matsumoto H, Okada S, Hashimoto S,
Ishihara T. Intermediate temperature steam electrolysis
using strontium zirconate-based protonic conductors. Int
Acknowledgments J Hydrogen Energy 2009;34:56–63.
[7] Hashimoto S, Liu Y, Mori M, Funahashi Y, Fujishiro Y. Study
The authors wish to acknowledge Bob Rabun, Program of steam electrolysis using a microtubular ceramic reactor.
Manager for NNSA funded WSRC PDRD programs and Leah Int J Hydrogen Energy 2009;34:1159–65.
Arrigo for data verification. This document was prepared in [8] Fox EB, Greenway SD, Ekechukwu AA. Hydrogen isotope
recovery using a cathode water vapor PEM electrolyzer.
conjunction with work accomplished under Contract No. DE-
Fusion Sci Technol 2008;54:483–6.
AC09-08SR22470 with the U.S. Department of Energy. [9] Jeon DH, Greenway SD, Shimpalee S, Van Zee JW. The effect
of serpentine flow-field designs on PEM fuel cell
performance. Int J Hydrogen Energy 2008;33:1052–66.
references [10] Collier A, Wang H, Yuan XZ, Zhang J, Wilkinson DP.
Degradation of polymer electrolyte membranes. Int J
Hydrogen Energy 2006;31:1838–54.
[1] Ivy J, Summary of electrolytic hydrogen production, DOE [11] Zhang S, Yuan X, Wang H, Merida W, Zhu H, Shen J, et al. A
report NREL/MP-560-36734. National Renewable Energy review of accelerate stress tests of MEA durability in PEM fuel
Laboratory: Golden, CO. cells. Int J Hydrogen Energy 2009;34:388–404.
[2] Badwal SPS, Giddey S, Ciacchi FT. Hydrogen and oxygen [12] Springer TE, Zawodzinski TA, Gottesfeld S. Polymer
generation with polymer electrolyte membrane (PEM)-based electrolyte fuel-cell model. J Electrochem Soc 1991;138:
electrolytic technology. Ionics 2006;12:7–14. 2334–42.