Ionics (2006) 12: 7–14

DOI 10.1007/s11581-006-0002-x

ORIGINA L PA PER

S. P. S. Badwal . S. Giddey . F. T. Ciacchi

Hydrogen and oxygen generation with polymer electrolyte

membrane (PEM)-based electrolytic technology

Received: 25 January 2006 / Accepted: 9 February 2006 / Published online: 26 April 2006
# Springer-Verlag 2006

Abstract With dwindling liquid fuel resources, hydrogen inherent embedded energy or from water by supplying
offers a credible alternative. The use of hydrogen in a fuel energy. Linking of hydrogen generation with renewable
cell offers the highest fuel conversion efficiency compared energy would provide a totally sustainable energy cycle.
with all other technologies and it also has the potential to However, most renewable sources are intermittent sources
substantially reduce greenhouse gas and particulate emis- of power and require energy storage for load levelling and
sions at least at the end-user sites. One of the major barriers to reduce overall plant cost. Hydrogen is arguably the best
to the introduction of the hydrogen economy and its wider energy storage medium and is an excellent alternative to
acceptance is the lack of the rather costly hydrogen carbon-based transport fuels.
generation, transportation and distribution infrastructure to The current world H2 generation capacity is about 500
meet the local transport fuel demands. On-site or billion m3/year=42 million tons/year (=1.6% of the global
distributed hydrogen generation would remove the need energy consumption) (90 billion m3/year in the USA) [1].
for this up-front infrastructure requirements and assist with Most H2 (∼95%) is produced from fossil fuels, mainly
the early large-scale trials of the fuel cell technology for natural gas. The major uses of hydrogen are:
both transport and stationary applications and also intro-
– Over half of the total hydrogen produced is utilised for
duction of the hydrogen economy. In this paper, the
ammonia (fertilizer and explosive) production
development of polymer electrolyte membrane electrolysis
– About one third in oil refineries—for impurity removal
technology for on-site, on-demand hydrogen generation
and upgrading of heavy oil fractions into lighter and
has been discussed. The major emphasis is given on
more valuable products
reducing catalyst cost; interface design and modifications;
– Remaining in methanol production, chemical and
interconnect materials, design and fabrication; and
metallurgical industries and space programs
investigation of the sources of degradation. Stacks to
2 kWH2 capacity have been constructed and tested and The use of hydrogen as a power generation source is
show initial efficiencies of >87% at 1 A cm−2. minimal at this stage due to its high cost of production, lack
of hydrogen transport, storage and distribution infrastruc-
Keywords Hydrogen production . Electrolysis . ture and technological challenges related to production,
Renewable energy . Polymer electrolyte membrane storage and end-use of hydrogen. The current total global
hydrogen production is barely enough to feed the US
transport fleet. Global oil production is likely to peak in the
Introduction next 5–10 years (it has already peaked in Australia, USA
and Europe) and soon the world will run short of oil [2–4].
As fossil fuel and especially liquid fuel reserves start The demand for transport fuels and energy is increasing at
declining and the gap between demand and supply widens, an alarming rate especially in developing and heavily
the cost of transport fuels would increase at an alarming populated countries such as India and China. As the gap
rate and there will be an increasing need to find alternative between supply and demand for liquid fuels increases, oil
transport fuels. Hydrogen is the cleanest fuel available to prices will increase and there will be increasing require-
mankind but it has to be generated from fossil fuels with ments to use alternative clean and low- or zero-emission
fuels such as hydrogen.
S. P. S. Badwal (*) . S. Giddey . F. T. Ciacchi Hydrogen is considered to be the best energy storage
CSIRO Manufacturing and Infrastructure Technology,
Private Bag 33, Clayton South 3169, media due to its flexibility for end-use and regenerative and
Victoria, Australia zero-emission properties. Hydrogen combusted in a fuel
e-mail: [email protected] cell can give combined heat and power efficiencies of more
8

than 80%, with zero greenhouse gas and pollutant emis- compared to other technologies such as direct water split-
sions at end-use sites with water as the only by-product. ting (photo-electrochemical), biological routes, etc., and
Hydrogen can be alternatively used as a fuel in an internal the electrical/water infrastructure already exists. On-site
combustion engine, for example, in a car to power the hydrogen generation can reduce the need for up-front ex-
mechanical drive train. pensive hydrogen transmission/distribution infrastructure.
Several technologies are under development for hydro- The major issues associated with large-scale adoption of the
gen generation including: electrolysis technology are the high capital cost of the
electrolyser (>A$5,000/kW), low system efficiencies, life-
– Reforming fossil fuels (natural gas, liquified petroleum
time performance and integration with renewable
gas, gasoline, diesel, coal, methanol, etc.) with appro-
sources [5, 6].
priate fuel processing/cleaning
In this paper, developmental work on solid state water
– Electrolysis of water (energy source: nuclear, fossil,
electrolysis technology based on polymer electrolyte
renewable)
membrane for small-scale distributed hydrogen generation
– Low-temperature (alkaline and polymer electrolyte has been described. Small-scale distributed production of
membrane, PEM) and hydrogen will enhance its availability at consumer level
– High-temperature ceramic membranes (transport, distributed generation, fuelling of micro de-
vices, etc.) without the cost of large hydrogen distribution
– Hybrid fossil fuel/renewable technologies (e.g. solar
and storage facilities. It will, therefore, accelerate the
thermal reforming)
hydrogen economy by providing local hydrogen at more
– Biomass gasification
attractive investment costs. PEM-based electrolysis sys-
– Photo-electrochemical/photo-catalytic splitting of water
tems offer a number of attributes such as modular aspect
– Thermolysis of water
(there is minimal penalty on efficiency due to unit size), all
– Thermo-chemical processes
solid state system (no alkaline liquid electrolytes or its
Photo-electrolysis is considered to be an ideal prospect recycling involved and water and electricity are the only
for hydrogen generation using only sunlight. However, the inputs required), pure hydrogen and oxygen generation
technology is in the early stages of research on materials (due to physical separation by solid electrolyte membrane),
development. There are substantial technical challenges ability to produce hydrogen at a pressure (electrochemical
related to efficiency, cost and lifetime. It is a long way compression), small footprint due to high current densities
away from meeting the collective technical and commercial achievable (>1 A cm−2 as compared to ∼0.4 A cm−2 by
targets for hydrogen production [5, 6]. There are other alkaline system) [7–9]. PEM electrolysis technology due to
issues such as the logistics of hydrogen collection from its fast response time and start-up/shut-down character-
large surface areas (small current densities for hydrogen istics (hydrogen generation starts immediately at ambient
production). Other proposed technologies such as thermol- conditions) and ability to accept large variations in load is
ysis and thermochemical processes suffer from severe ideal for integration with intermittently available sources of
material-related problems [5, 6]. electricity (renewable and off-peak grid). In addition, due
Although in the short to medium term, increased hydro- to the similar aspects of the PEM electrolysis and fuel cell
gen demand is likely to be met by fossil fuels (e.g. natural technologies, the impact on development and system cost
gas reforming, coal gasification with CO2 separation/ reduction can be enormous.
sequestration), in the medium to long term, renewable
energy offers the best possible solution. With time, the use
of renewable energy (solar, wind, tidal, wave, etc.) as a Experimental procedures
clean alternative power is likely to increase as the cost of
production declines. These renewable sources are inter- Electrolysis cell/stack construction
mittent sources of power. In the absence of energy storage,
these renewable systems have to be grossly over-designed Membrane electrode assemblies (MEAs) for water elec-
in terms of capacity. Hydrogen offers the flexibility of trolysis varying in dimension from 9 up to 150 cm2 active
energy storage for long duration. electrode area and stacks up to multi-kilowatts equivalent
Hydrogen can be easily produced from water by elec- hydrogen generation capacity were constructed for evalua-
trolysis using intermittent renewable energy and off-peak tion as described below. The hydrogen electrode of the
electricity from nuclear, hydro or thermal power plants. cells was prepared using Pt/C catalyst with platinum
Water electrolysis is considered to be one of the key loadings of up to 0.4 mg.cm−2 as detailed elsewhere [10].
technologies for hydrogen generation as it is compatible The oxygen electrode was prepared using various noble
with existing and future power generation technologies and metal catalysts with loading in the 0.2–0.4 mg.cm−2 range
a large number of renewable technologies (solar, biomass, on a metallic support. The polymer electrolyte membranes
hydro, wind, tidal, wave, geothermal, etc.). In particular, used were Nafion N112 and N115. Membrane electrode
hydrogen enables the long-term storage of energy for load assemblies for electrolysis cells were prepared by hot
levelling. pressing the membrane between the hydrogen and oxygen
The major advantages of producing hydrogen via this electrodes. Metallic plates with parallel cross channel flow
route are that the technology is relatively more mature fields were used as interconnects. The electrolysis cells
 9

with interconnect plates were stacked and assembled with Impedance spectroscopy was performed with the Zahner
the end interconnects on both sides of the stack, serving model IM6e over the frequency range of 0.1 Hz to 1 kHz at
dual functions—as interconnects as well as current col- 75°C on selected 9 cm2 active area cells at the start and
lectors for the stack. after operation of the cell for different times at a current
Several single cells or stacks with 50, 100 and 150 cm2 density of 1 A cm−2. The software package Thales pro-
active area per cell were fabricated and assembled to vided by Zahner was used for data acquisition and fitting of
investigate the issues related to cell scale-up, stacking, the data to a simple equivalent circuit consisting of a series
performance, degradation and lifetime. Stack modules resistor representing all ohmic losses in the cell in series
ranging in hydrogen generation capacity from 100 WH2 to with constant phase angle element in parallel with a resistor
2 kWH2 were constructed and evaluated. For the 9 cm2 cells representing the electrode/electrolyte interface, after cor-
and 50 cm2 stacks, Nafion membrane N112 (50 μm thick) recting the data for lead resistance and measuring circuit
was used whereas for the 100 cm2 stacks, thicker Nafion parameters.
membrane N115 (125 μm thick) was used.

Results and discussion

Cell and stack evaluation
Single cells
Test stations with built-in diagnostic tools were constructed
in-house with multiple levels of operator and stack safety Several single cells of 9 cm2 active electrode area, with
redundancy and capability to test electrolysis cells/stacks different catalysts, loadings, membrane type and thickness,
up to 5 kWH2 capacity. One test station was constructed to interface modification, oxygen electrode support material,
operate stacks to 20 barA pressure. The system design etc., were investigated. Figure 1 shows typical voltage–
allows for a continuous unattended operation of the current characteristics and efficiencies as a function of
electrolysis cells/stacks for extended periods of time with current density of a 9 cm2 active area cell operating at
continuous monitoring of voltages of individual cells and 75°C. The cell produced efficiencies of 82% at 1 A cm−2
stacks, stack current and water temperatures in different and 69% at 2 A cm−2, respectively. It should be noted that
parts of the flow circuit. All key operator safety and stack there is no oxygen or hydrogen generation below the
safety system components (vent pressure, gas and temper- thermo-neutral voltage of 1.48 V. There is almost a linear
ature sensors, low water level and flow switches) are relationship between voltage and current once the current
interlocked with the power source to stop the hydrogen/ starts flowing through the cell but more specifically above
oxygen generation in the event of hydrogen leak or any 0.5 A cm−2. It is worth noting that this linear relationship
other safety-related issues. continues even up to 2 A cm−2 current density with no sign
For screening and evaluation of electrolysis cell of limiting currents (due to diffusion limiting processes).
materials, design parameters and operating conditions, These results indicate that most of the voltage losses across
cells of 9 cm2 active area, were used. Cells of this size were the MEA are ohmic in nature (the membrane, electrode
also used for electrochemical diagnostic analysis as even a supports, electrode materials (anode and cathode) and
small current signal can produce reasonably high current contact resistances). The polarisation losses appear to be
densities. The small cells (9 cm2) were operated at different relatively small. The surface-specific resistance value
temperatures between room temperature and 80°C to determined from this linear relationship at current densities
determine the optimum operating temperature. The cell above 0.5 A cm−2 was 0.335 Ω cm2.
temperature was maintained with external heating. The
effect of catalyst loading, membrane thickness and inter- 2.4 100
face modification on the cell performance was studied at a
typical cell operating temperature in the 75–85°C range.
2.2
Electrochemical investigations included current–voltage 90
Cell voltage, V

Efficiency, %

characteristics, electrochemical impedance spectroscopy

on selected cells, efficiency calculations and performance 2.0

degradation as a function of time at current density of 1 A 80

cm−2. The hydrogen generation efficiency was calculated 1.8
based upon thermo-neutral voltage, Vthn, of 1.48 V [11].
70
Current efficiency was determined by measuring the 1.6
hydrogen gas flow exiting and that expected for a given
charge flow. 1.4 60
Stack testing involved determining current–voltage 0.0 0.4 0.8 1.2 1.6 2.0
curves at different operating times of its operation, current Current density, A.cm-2
efficiency, overall stack hydrogen generation efficiency,
Fig. 1 Voltage–current characteristics (○) and efficiencies (△) of a
lifetime testing, operation in a thermally self-sustaining 9 cm2 active area electrolysis cell (Nafion membrane N112, 50 μm
mode (no supplementary heating of water) and evaluation thick) operating up to 2 A cm−2 current density. Data were recorded
of the degradation mechanisms. at 75°C
10
1.9
In general, higher performance was achieved with 80% efficiency
thinner membranes (50 μm). However, such membranes
are found to be more fragile, have short lifetime and need to 1.8
85% efficiency
be handled very carefully. There is also a complex

Cell voltage, V
relationship between catalyst loading, the method of cata- 1.7
lyst deposition, performance and lifetime. The three phase 90% efficiency

boundary area between the reactant species, electrode and 1.6

electrolyte need to be optimised to reduce catalyst loading,
attain high hydrogen generation efficiency and increase
1.5
lifetime.

1.4
Scale-up and stack evaluation 0.0 0.2 0.4 0.6 0.8 1.0
Current density, A.cm-2
The electrolysis cells of sizes 50 and 100 cm2 active areas Fig. 2 Voltage–current characteristics of electrolysis cells showing
were fabricated and stacks of varying capacity assembled the technology scale-up from 9 to 100 cm2 active area cells and
to investigate the issues related to scale-up and stacking. from single cells to a 14-cell stack. In the case of stacks, the data
Figure 2 compares the V–I characteristics of 9 cm2 active have been normalised to a single cell voltage. In all cases, data were
recorded at 80°C except the 9 cm2 cells for which data were
area single cell, 50 cm2 active area three-cell and six-cell recorded at 75°C. ▲, 9 cm2 active area single cell; △, three-cell stack
stacks, and 100 cm2 active area two-cell and 14-cell stacks with 50 cm2/cell active area; □, six-cell stack with 50 cm2/cell active
ð2kWH2 Þ . In the case of stacks, Y-axis represents average area; ◇, two-cell stack with 100 cm2/cell active area; ○, 14-cell stack
voltage per cell in the stack. The horizontal lines in the with 100 cm2/cell active area
figure have been drawn at cell voltages corresponding to
80, 85 and 90% efficiencies. The data in this figure shows a the water exiting the stack of 85–86°C. This may have been
significant improvement in efficiencies of 50 and 100 cm2 even higher had the stack been insulated. At 80 A (0.8 A
stacks as compared to 9 cm2 active area single cells. cm−2) current flow, the temperature of the water exiting the
Furthermore, the figure shows that, at 1 A cm−2 current stack was 80°C, at 60 A (0.6 A cm−2) it was 73°C and at
density, the efficiency of the 50 cm2, six-cell stack is 87% 50 A (0.5 A cm−2) it was 67°C. It is clear that a significant
as compared to 84% for the three-cell stack, and the amount of heat is generated within the stack due to internal
efficiency of the 100 cm2, 14-cell stack is 86% as com- resistive losses, which is sufficient to keep the stack in a
pared to 83% for the two-cell stack. These results indicate thermally self-sustaining mode. The final stack temperature
that the increase in number of cells in the stack improves would be determined by the internal cell losses within the
the efficiency of the stack. It appears that increasing the stack, current density, level of stack insulation or losses to
number of cells in the stack results in a more uniform the surroundings and the water flow rate.
temperature distribution and also lower average contact
resistance per cell (higher ohmic losses are expected for
end cells and at current collection plates). This is also Current or Faradaic efficiency
reflected in the area-specific resistances calculated from the
V–I curves (0.21 Ω cm2 for the 50 cm2 six-cell stack and The current efficiency was measured by relating the flow
0.27 Ω cm2 for the 100 cm2, active area 14-cell stack). It rate measurements (the amount of hydrogen generated per
should be noted that for the 9 cm2 cells and 50 cm2 stacks, unit time) with the current flow per unit time. Most of the
Nafion membrane N112 (50 μm) was used, whereas for the cells showed near 100% Faradaic efficiency during oper-
100 cm2 stacks, thicker Nafion membrane N115 (125 μm)
was used and, thus, somewhat higher specific resistance. 30 110
80oC
The relationship between current and voltage is again 25
reasonably linear for all cells and stacks above the current 100
density of 0.5 A cm−2 indicating the major contribution
Stack voltage, V

Efficiency, %

20
from ohmic resistance within the stack. 90
The 14-cell stack generated about 11 l/min of hydrogen 15
and 5.5 l/min of oxygen at an initial efficiency of over 85% 80
at 1 A cm−2 (Fig. 3). This stack was first operated without 10

any insulation but with an auxiliary heater supplying hot 5

70
water at 80°C for electrolysis. The stack was subsequently
operated at different current densities without any sup- 0 60
plementary heating in a thermally self-sustaining mode 0.0 0.2 0.4 0.6 0.8 1.0
with input water at room temperature (22°C). The results Current density, A.cm-2
are shown in Fig. 4. As the current was slowly increased to Fig. 3 Voltage–current characteristics (○) and efficiencies (●) ob-
100 A (1 A cm−2), within a few minutes, the stack reached tained from a 2 kW hydrogen equivalent electrolysis stack (Nafion
its maximum operating temperature with the temperature of membrane N115, 125 μm thick)
 11
35 90
(Rel, the left intercept of the respective arc on the real
impedance axis) with time with very little effect on the
33 80
electrode resistance, Rp, the difference of the intercepts of
Stack voltage, V

the respective arc on real impedance axis (Fig. 6). The

Stack T, oC
31 100A 70
impedance data were recorded at the start of the operation
80A and after 170 h of hydrogen and oxygen generation at a
29
60A
60
current density of 1 A cm−2. Further electrochemical tests,
interconnect surface examination and membrane electrode
27 50
50A assembly analysis are being carried out, in addition to
monitoring pH and conductivity of water at various
25 40 locations in the water supply sub-circuit, to understand
0 50 100 150 200 250 300 350
the exact mechanisms of this cell/stack degradation.
Time, min Another catastrophic degradation has also been observed
Fig. 4 The operation of the 14-cell (100 cm2 per cell active area), for cells that have been operated for long periods of time
2 kW hydrogen equivalent stack in a thermally self-sustaining mode. (typically >1500 h). This, in fact, is related to internal
The stack voltage (●) and stack temperature (▲) are plotted as a electronic short circuiting through the electrolyte mem-
function of time for different current values brane and loss of Faradaic efficiency as discussed below.
For a normal cell with near 100% current efficiency (that
is, all the current that flows in the cell goes towards
ation. However, over a period of operation, some cells from hydrogen and oxygen generation), no current flows in the
time to time developed internal electronic short circuits cell up to the thermo-neutral cell voltage of about 1.48 V
which led to a drop in the current efficiency, as discussed in (Fig. 7) [11]. Above this voltage, the behaviour is that
the next section. shown by the near-linear portion of the voltage–current plot
(○, Fig. 7), recorded after 1,728 h of accumulated operation
at a current density of 1 A cm−2 and at 80°C. However, after
Lifetime performance and degradation mechanism 1,782 h of accumulated operation at 1 A cm−2, it was
observed that the current flowed through this cell even at
A 50 cm2 active area four-cell stack with Nafion membrane voltages below 1.48 V (△, steeper section in Fig. 7). Such
N115 ð0:25kWH2 Þ has been operated at 1 A cm−2 current cells also showed lower current efficiency (the amount of
density for lifetime performance evaluation for a period of hydrogen generated for the electric charge flow was lower
more than 1,500 h at 80°C. The hydrogen and oxygen than that expected theoretically). It is believed that this
gases were generated at atmospheric pressures. The stack behaviour is caused by physical damage to the membrane
was usually operated continuously during weekdays for over a period of operation at high current densities and
approximately 100 h and shut down over the weekends. internal short circuiting of the membrane. However, the
Figure 5 shows the lifetime performance data of the stack in electronic short circuit paths initially still have quite high
terms of stack voltages at the start (Es) and after 6 h (E6) of resistance and the voltage across the cell can reach
continuous operation into the 100 h cycle to simulate the 1.48 V at relatively low current densities. Once this
operation of the electrolyser with an intermittent source of
energy such as solar photovoltaic. Up to 400 h of operation,
the stack efficiency was between 80 and 85%. The data also 12.5
4-cell stack
shows that there is about 3% drop in the stack efficiency
during the 6 h operation, and about 2.5% of this is 10.0
recovered at the start of the next cycle of operation. The
Stack voltage, V

average rate of drop in the stack efficiency during 1,500 h 6h operation

7.5
of operation was 1.30% and 1.04% per 100 h of operation Start of cycle
for Es and E6, respectively. There are several possible
reasons for this degradation in performance, such as an 5.0
increase in the contact resistance between the oxygen
electrode support and the interconnect due to surface 2.5
oxidation, changes in interfacial conditions (de-bonding of
electrode/membrane, changes in catalyst surface/structural
0.0
properties), membrane conductivity loss due to contami- 0 400 800 1200 1600
nation from other ionic species entering the water sub- Accumulated operation, hour
circuit, etc. The temporary (recoverable) drop in efficiency
of the stack appears to be related to charging/discharging of Fig. 5 Lifetime performance data for a 0.25 kW hydrogen
the interface, saturation of the water with oxygen and/or equivalent four-cell (50 cm2 per cell active area) electrolysis stack
(Nafion membrane N115, 125 μm thick). ○, voltage data at the start
adsorption of oxygen molecules on the reaction sites. of the hydrogen generation cycle at 1 A cm−2 current density and □,
Preliminary electrochemical impedance studies on 9 cm2 after 6 h of continuous operation to simulate the solar PV cycle.
active area single cells indicate an increase in ohmic losses Data were recorded at 80°C
12
0.3
Rel Rp
observed behaviour (△, Fig. 7). When the cell voltage is
below 1.48 V, the current initially flows through electronic
2

2
- Zimag, ohm.cm

(ohm.cm )
0.2 At start 0.291 0.216 short circuit paths in the membrane and there is no ionic
After 170h 0.393 0.228 current flow and, thus no hydrogen generation occurs. With
increasing applied voltage, more and more current flow
0.1
through these short circuit paths represented collectively by
Rshort [consisting of a number of resistors (Rs1, Rs2–Rsn) in
0.0 parallel]. Once the applied voltage exceeds 1.48 V, part of
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 the total current starts flowing as ionic current and it goes
Zreal, ohm.cm 2 towards hydrogen and oxygen generation. We have fitted
the data in Fig. 7 to the equivalent circuit models and
Fig. 6 Impedance diagram (real impedance, Zreal) vs imaginary equations given in Fig. 8 (○, circuit 1 and △, circuit 2) and
impedance (Zimag) for a 9 cm2 electrolysis cell (Nafion membrane derived the values of 0.0125 Ω for Rp (the polarisation
N112, 50 μm thick) operating at 75°C, showing increase in ohmic resistance at the electrode/electrolyte interface) plus Rel
losses with time. ○ at the start and □ after 170 h of operation at 1 A
cm−2 current density (representing the total ohmic losses in the cell) for both sets
of data, and the value of 0.165 Ω for Rshort for the data
represented by △ in Fig. 7. The solid curves in Fig. 7 are the
threshold voltage is reached, part of the current goes fitted data and show an excellent fit to the experimental data.
towards hydrogen generation and the cell behaviour is that For fitting the data to these circuit models, we have made
shown in the less steep section of the 1,782 h plot (△, Fig. 7). the following assumptions: The major voltages and effi-
This electronic current flow is responsible for the loss of ciency losses across the cell are ohmic in nature (electrolyte
current or Faradaic efficiency. An optical examination of membrane, support materials, current collection plates and
some damaged cells indeed confirmed membrane rupture. contact resistances) with relatively low contribution from
To further understand this behaviour, we have tried to polarisation losses at interfaces between electrodes and the
qualitatively model the overall cell behaviour in terms of electrolyte as obvious from the near-linear voltage–current
electrical equivalent circuits shown in Fig. 8. Circuit 1 in relationship above 1.48 V. It should be noted that, in
the figure can be used to describe the behaviour observed general, the major efficiency losses in electrolysis cells
when there is no internal short circuiting of the membrane occur as a result of activation polarisation and ohmic
and the current efficiency is near 100%. No current flows resistance, with insignificant contribution from concentra-
through the cell when the applied voltage is below 1.48 V tion polarisation effects. However, it has been reported and
represented by Vthn in circuit 1. Once the applied voltage as indeed also observed in the present work that ohmic
exceeds the thermo-neutral voltage, the current starts losses are much higher compared to activation polarisation
flowing through the cell and almost a linear voltage current losses especially for higher current densities [12, 13].
relationship is observed as shown in Fig. 7 (○). We have also assumed that, at least during the collection
For the case when there is an electronic short through the of V–I data, Rshort remains reasonably constant. This
membrane, circuit 2 in Fig. 8 can be used to describe the assumption may, however, be somewhat flawed as it will
be shown later that, once there is an electronic short
through the membrane, the cell degradation at higher
2.5
After 1728h operation
current densities increases more rapidly with time and it is
After 1782h operation possible that during collection of the current voltage curve,
2.0
Rshort may have changed slightly.
Cell voltage, V

In Fig. 9, values of total (It), short circuit (Ishort) and

1.5
Faradaic or ionic (IFaradaic) currents have been plotted as a
function of the applied voltage. These results show that
1.0
although the cell had developed electronic conductivity, it
would still generate hydrogen once the voltage across the
0.5 cell exceeds 1.48 V. For example, at a current density of
1.0 A cm−2, about 76% of the current flowed through the
0.0 cell as ionic current or part of the current that contributed to
0 10 20 30 40 50
the hydrogen and oxygen generation. However, it was
Current, A observed that, once electronic short circuiting occurred
Fig. 7 Voltage–current characteristics of a cell in the four-cell through the electrolyte membrane, the slope of the initial
(50 cm2 per cell active area) electrolysis stack (Nafion membrane part of the curve decreased with time of the current
N115, 125 μm thick) after 1,728 and 1,782 h (○ and △, respectively) passage, indicating lower Rshort values and, thus, increased
operation at 1 A cm−2 current density and 80°C. The solid lines are electronic short circuit paths. Typical behaviour is shown in
fitted data as per models described in Fig. 8 (Rp+Rel=0.0125 Ω for
the 1,728 h fitted curve and Rp+Rel=0.0125 Ω and Rshort=0.165 Ω for Fig. 10 for a cell which had developed electronic short
the 1,782 h fitted curve). The dotted line has been drawn at the circuiting. The plots after 1,644 and 1,650 h (○ and ●,
thermo-neutral voltage of 1.48 V [11] respectively) of current passage at 1.0 A.cm−2 current
 13

Circuit 1 Perfect cell – no damage to the membrane

Cdl
Case 1: Vt < Vthn
Vthn It = 0
Rel
RP Case 2: Vt > Vthn
It = IFaradaic = (Vt – Vthn)/(Rp + Rel)

Circuit 2 Cdl
Degraded cell – damaged membrane
Vthn
Case 1: Vt < Vthn
Ifaradaic

Rel
IFaradaic = 0
RP It = Ishort = Vt/Rshort
1/Rshortt=1/Rs1+1/Rs2+…1/Rsn
Ishort

Case 2: Vt > Vthn

Rs1
It = IFaradaic + Ishort
Rs2 IFaradaic = (Vt – Vthn)/(Rp + Rel)
Ishort = Vt/Rshort
Rsn

Fig. 8 Equivalent circuit representation of a normal cell (no internal layer capacitance at the electrode/electrolyte interface; Rel=total ohmic
short circuiting), circuit 1, and a damaged cell with internal electronic losses consisting of resistance of electrolyte membrane, support
short through the membrane, circuit 2. It=total current flowing through materials, current collection plates and contact resistances; Rshort=total
the circuit; IFaradaic=Faradaic current which goes towards hydrogen and short circuit resistance consisting of a number of parallel short circuit
oxygen generation; Ishort=short circuiting current responsible for loss in resistors (Rs1, Rs2—Rsn), each representing a short circuit path through
hydrogen generation efficiency; Rp= polarisation resistance (activation the membrane; Vt=total voltage drop across the cell; and Vthnthermo-
and concentration) at the electrode/electrolyte interface; Cdl=double neutral voltage [11]

density clearly show a decrease in the value of Rshort with potential for coupling with intermittent sources of renew-
increased time of current passage. able electricity with minimal power electronics. Thus, it is
important to investigate the response of the PEM elec-
trolyser to an intermittent and variable power source and its
Technology demonstration effect on the electrolyser performance and efficiency over a
period of time. As hydrogen offers the flexibility of energy
There is substantial market for small-scale distributed storage for long duration, the PEM electrolyser technology
hydrogen generation. PEM electrolysis technology, due to will be demonstrated as a remote area power supply as
its fast response time and start-up/shut down characteristics shown in a conceptual diagram (Fig. 11) in a totally
and ability to accept large variations in load, has the sustainable energy cycle.
2.5

50
2.0
Cell voltage, V

40 It
1.5
Current, A

30
1.0
IFaradaic
20 0.5

10 Ishort 0.0
0 10 20 30 40 50
0
Current, A
0.0 0.4 0.8 1.2 1.6 2.0 Fig. 10 Voltage–current characteristics of a cell in the four-cell
Cell voltage, V (50 cm2 per cell active area) electrolysis stack (Nafion membrane
N115, 125 μm thick) after 1,644 and 1,650 h (○ and ●, respectively)
Fig. 9 Plots of total (□), Faradaic (◆) and short (●) circuit currents operation at 1 A cm−2 current density and 80°C, showing rapid
vs cell voltage calculated from parameters derived for the fitted increase in cell degradation and increased electronic short circuit
curve to the 1,782-hour operation data in Fig. 7 and as per the model paths after initial onset of electronic short circuiting of the
equivalent circuit 2 in Fig. 8 electrolyte membrane
14
Fig. 11 Concept diagram for Load
demonstration of the electrolysis Leveling
technology in a sustainable en-
ergy cycle as a remote area Hydrogen
power supply Renewable Energy: Electricity Water
H2 / O2 storage
Solar, tidal, wind Electrolysis Oxygen
hydro, biomass
H2 O2
Power / Heat
Water (H2O)
Power to Grid Remote Area
Fuel
Power
Cell
Residential
Transport

Conclusions References

Electrolysis cells have been scaled from 9 to 150 cm2 1. International Atomic Energy Agency (1999) Hydrogen as an
active area with very little loss in hydrogen generation energy carrier and its production by nuclear power. Report
IAEA TECDOC-1085. Available at http://www-pub.iaea.org/
efficiency. Test stations to test stacks to 5 kWH2 capacity MTCD/publications/PDF/te_1085_prn.pdf
and pressures to 20 barA have been constructed with 2. Campbell CJ (2002) World: oil and gas industry-peak oil: an
multiple levels of safety redundancy. The system design outlook on crude oil depletion. Revised, February 2002, http://
allows for an unattended safe operation of stacks for www.mbendi.com/indy/oilg/p0070.htm
3. Campbell CJ, Leherrère JH (1998) The End of Cheap Oil.
extended periods. Stacks of varying dimensions and Scientific American, March 1998, 6 pp
hydrogen generation capacity have been built and tested, 4. Duncan RC, Youngquist W (1998) The world petroleum life
some for periods exceeding 1,500 h at a current density of cycle. PTTC workshop on OPEC oil pricing and independent
1 A cm−2. Efficiency of up to 87% initially at 1 A cm−2 has oil producers. Petroleum Technology Transfer Council, October
been achieved through optimisation of interface, catalyst, 1998, http://www.dieoff.com/page133.pdf
5. US DOE. Energy efficiency and renewable energy, hydrogen fuel
MEA fabrication, fluid/current flow and stack design. The cell and infrastructure technologies program, multi-year research,
largest size stack tested to date has hydrogen generation development and demonstration plan: planned program activities
capacity of 11 l/min and oxygen generation capacity of for 2003–2010, technical plan—hydrogen production, http://www.
5.5 l/min. The operation of this stack has been demonstra- eere.energy.gov/hydrogenandfuelcells/mypp/pdfs/production.
pdf. US DOE hydrogen, fuel cells and infrastructure technologies
ted in a thermally self-sustaining mode from cold start with program review proceedings 2004 and 2005. Available at http://
water and electricity as the only inputs. Temperatures www.eere.energy.gov/hydrogenandfuelcells/2004_annual_review.
exceeding 80°C can be easily achieved at a current density html and http://www.hydrogen.energy.gov/annual_review05_pro
of 0.8 A cm−2 or higher. Causes of degradation have been ceedings.html
6. McHugh K (2005) Hydrogen production methods. February
discussed and the major cause of hydrogen generation 2005, MPR-WP-0001, MPR Associates, Alexandria, Virginia,
efficiency loss has been established to be the electronic USA
short circuiting through the polymer membrane. The 7. Millet P, Andolfatto F, Durand R (1996) Design and per-
results have been explained in terms of electrical equivalent formance of a solid polymer electrolyte water electrolyzer. Int J
circuit models. Future work, apart from optimisation of Hydrogen Energy 21:87–93
8. Friedland RJ, Speranza AJ (1999) Integrated renewable
stack components and system design, is to reduce cost and hydrogen utility system. Proceedings of the 1999 US DOE
performance degradation. It would also involve demon- Hydrogen Program Review, NREL/CP-570-26938
stration of the technology initially as a remote area power 9. Prince-Richard S, Whale M, Djilali N (2005) A techno-econ-
supply by mapping the total available solar or wind power omic analysis of decentralized electrolytic hydrogen production
for fuel cell vehicles. Int J Hydrogen Energy 30:1159–1179
for a site on a yearly basis and matching this with daily, 10. Giddey S, Ciacchi FT, Badwal SPS (2004) Design, assembly
weekly and yearly power demand and hydrogen generation and operation of polymer electrolyte membrane fuel cell stacks
and storage requirements. to 1 kWe capacity. J Power Sources 125:155–165
11. Onda K, Kyakuno T, Hattori K, Ito K (2004) Prediction of
production power for high-pressure hydrogen by high-pressure
Acknowledgements The authors would like to thank Kristine water electrolysis. J Power Sources 132:64–70
Giampietro and Pon Kao for their assistance with the experimental 12. JO’M Bockris (1980) Energy options. Wiley, New York, p. 333
work. The manuscript was reviewed by Robin Clarke and Pon Kao. 13. Ströbel R, Oszcipok M, Fasil M, Rohland B, Jörissen L, Garche
The work described in this paper forms part of CSIRO’s Energy J (2002) J Power Sources 105:208–215
Transformed Flagship research program into positioning Australia
for a future hydrogen economy.