Water Electrolysis

T Smolinka, Fraunhofer Institute for Solar Energy Systems ISE, Freiburg, Germany
& 2009 Elsevier B.V. All rights reserved.

General Concept An overview of the different equations for the

hydrogen evolution reaction (HER) and the oxygen
Water electrolysis is an electrochemical process in which evolution reaction (OER), the typical temperature win-
electricity is used to split water into two gases, hydrogen dow and the ions acting as the charge carrier through the
and oxygen. The overall reaction of water electrolysis is diaphragm/membrane is given in Table 1.
expressed as
Short Historical Overview
H2 Oðl=gÞ-12 O2 þ H2 ½I
The basic equations for water electrolysis indicate that
Three main technologies can be distinguished according this technique can produce clean hydrogen by utilizing
to the electrolyte used in the electrolysis cell: electricity from renewable energy sources without the
emission of carbon dioxide, in contrast to the industrial
• alkaline electrolysis, with a liquid electrolyte; steam reforming of fossil fuels. In recent years, water
• with
polymer electrolyte membrane (PEM) electrolysis,
an acidic ionomer electrolyte – often called solid
electrolysis has attracted great interest due to the pro-
gress that fuel cells (FCs) have made in different appli-
polymer electrolysis (SPE); cations and, consequently, the anticipated higher demand
• solid
high-temperature (HT)
oxide electrolyte.
steam electrolysis, with a for clean hydrogen. Water electrolysis is, however, quite
an old technology; the principle has been known for
more than 200 years. Shortly after Alessandro Volta in-
The underlying general process is the same in all three vented the voltaic pile in 1800, Anthony Carlisle and
technologies: water is fed to an electrochemical cell in William Nicholson in London, England, used such a
which hydrogen evolves at the negative electrode (cath- device to decompose water into hydrogen and oxygen. In
ode) and oxygen at the positive electrode (anode) when a the same year, Johann Wilhelm Ritter performed com-
sufficiently high voltage is applied to the cell. The basic parable experiments in Jena, Germany. Nevertheless, it
principle for an acidic electrolysis cell is depicted in took decades before the process found its first
Figure 1. The anode and cathode are not separated by a application.
diaphragm/membrane. Oxygen and hydrogen gases are In 1890, Charles Renard constructed a water elec-
found in the space above the electrolysis cell. trolysis unit to generate hydrogen for French military
airships. As hydrogen could be produced at lower cost by
steam reforming, water electrolysis technology advanced
only slowly. It has been estimated that more than 400
industrial electrolyzers were operating worldwide around
1900. Different types of alkaline electrolyzer were de-
veloped in the 1920s and 1930s for the commercial
production of hydrogen based on low-cost hydro-
electricity, primarily for the manufacture of ammonia
fertilizers. After World War II, the development of al-
H2 O2 kaline electrolyzers proceeded and by the 1970s, well-
engineered systems for industrial applications were
offered by several companies that included Bamag,
H 3O +
Demag, Electrolyzer Corp., Norsk Hydro, Stuart Energy,
Oerlikon/Brown, Boveri & Cie, and Lurgi. For economic
reasons, the share of the hydrogen market remained
small.
The development of the so-called advanced alkaline
– Cathode Anode +
electrolyzers started in the years after the Organization of
the Petroleum Exporting Countries (OPEC) oil embargo.
Figure 1 Basic principle of an electrolysis cell in an acidic The resulting shock over the future security of supply
medium. demonstrated that the era of cheap petroleum would end

394
 Fuels – Hydrogen Production | Water Electrolysis 395

Table 1 Half-cell reactions, typical temperature ranges, and ionic charge carriers for different types of water electrolysis
Technology Temperature (K) Cathode (HER) Charge carrier Anode (OER)
Alkaline electrolysis 333–363 H2 O þ 2e -H2 þ 2OH OH 1
2OH - O2 þ H2 O þ 2e
2
PEM electrolysis 293–373 2Hþ þ 2e -H2 Hþ 1
H2 O- O2 þ 2Hþ þ 2e
2
HT steam electrolysis 973–1273 H2 O þ 2e -H2 þ O2 O2 1
O2 - O2 þ 2e
2
OER, oxygen evolution reaction; HER, hydrogen evolution reaction; PEM, polymer electrolyte membrane; HT, high temperature.

and that the world needed alternative fuels. The research 300
resulted in better cell design, improved electrode ma-

Free energy of reaction (kJ mol–1)

275

Enthalpy of reaction (kJ mol–1)

terials, and new system concepts for electrolyzers. ΔH R
Meanwhile, in the 1970s, the development of the PEM 250
electrolyzer commenced at General Electric (GE). The 225
emerging technology offered several advantages over
200 ΔG R
alkaline electrolyzers. Nevertheless, due to expensive
materials and a limited lifetime, PEM electrolyzers be- 175
came established only in niche applications and labora- 150
tory equipment with small hydrogen and oxygen
125
production capacities. H2O(I) H2O(g)
According to the US Department of Energy, elec- 100
trolysis (EL) only accounts for about 1.6 million tons of 200 400 600 800 1000 1200
the present global hydrogen production of 50 million tons Temperature (K)
per annum. Most of the hydrogen is used on site for the Figure 2 Thermodynamic properties DHR and DGR as a
production of ammonia for fertilizers through the Haber function of temperature for the decomposition of liquid and
process, and for the conversion of heavy petroleum gaseous water at 101.325 kPa (1 atm).
sources into lighter fractions through hydrocracking.
96 485 C mol1. For standard conditions of temperature
Fundamentals of Water Electrolysis (298.15 K) and pressure (101.325 kPa), eqn [2] can be
written as
The energy required to decompose one mole of water  
DGR0 1 1
into hydrogen and oxygen corresponds to the enthalpy of Vr0 ¼ ¼ Gf0;H2 þ Gf0;O2  Gf0;H2 OðI Þ ½2a
nF nF 2
formation of one mole of water. According to the second
principle of thermodynamics, a part of this enthalpy of where G0f,i is standard free energy of formation for com-
reaction, DHR, can be applied as thermal energy with a ponent i. The standard reversible cell voltage V 0r is
maximum equal to the product of the thermodynamic 1.229 V with respect to liquid water. Values under other
temperature T and the entropy of reaction DSR. The conditions are given by the Nernst equation (see Cor-
difference between the entropy term TDSR and the en- rosion, Thermodynamic). Equation [1] reveals that the
thalpy of reaction is defined as the free energy of reaction total amount of energy required for the decomposition of
DGR (often called the change in Gibbs free energy) water can be supplied by a combination of electricity and
heat. If, however, heat cannot be integrated into the pro-
DGR ¼ DHR  T DSR ½1 cess at low temperatures, the minimum cell voltage re-
quired for water electrolysis is related to the enthalpy of
and can be regarded as the minimum amount of the en- reaction and is called the thermo-neutral cell voltage, Vth:
thalpy of reaction that has to be applied as electrical en-
ergy. From a thermodynamic point of view, the minimum DHR
Vth ¼ ½3
cell voltage required for the decomposition of water is the nF
reversible cell voltage Vr which can be expressed as
For standard conditions, V 0th ¼ 1.481 V with respect to li-
quid water. The enthalpy of reaction DHR and free energy
DGR
Vr ¼ ½2 of reaction DGR as function of temperature at 101.325 kPa
nF
(1 atm) is shown in Figure 2. The relationship demon-
where n is the number of moles of electrons per mole of strates why higher temperatures are desirable for water
products, that is, n ¼ 2; and F is the Faraday constant, electrolysis  at least from a thermodynamic point of
396 Fuels – Hydrogen Production | Water Electrolysis

view. DHR depends only weakly on temperature. Owing to (cathode) electrodes, ZAn and ZCath, respectively, that is,
the phase change from liquid water to water vapor (at
373 K and 101.325 kPa), the enthalpy of reaction of the Vcell ¼ Vr þ jZAn j þ jZCath j þ iRA ½4
water decomposition is reduced by the value of the en-
thalpy of evaporation, and thus the minimum of energy for The components of the real cell voltage are illustrated in
water electrolysis diminishes. Subsequently, the values for Figure 3 as a function of the current density for a PEM
DHR rise again slightly with increasing temperature. With electrolysis cell working near room temperature. Ac-
respect to the required total amount of energy, water cording to Ohm’s law, the voltage drop iRA depends
electrolysis is not advantageous at these HTs when com- linearly on the current density. The cell resistance is the
pared with temperatures slightly higher than 373K. sum of single ohmic resistances from cell components
On the contrary, the free energy of reaction DGR (e.g., membrane, electrodes, and current collector) and
displays different behavior. At higher temperatures, the contact resistances at the interfaces between different
value of DGR becomes significantly smaller, which means subcomponents of the cell.
that the required minimum amount of electrical power The overpotentials ZAn and ZCath in case of the water
for water electrolysis can be reduced and replaced by electrolysis are also called oxygen and hydrogen over-
heat, for example, by that obtained as a by-product from potentials, respectively, and are caused by irreversible
another process. The reversible voltage Vr of the water processes at the electrode|electrolyte interface during
electrolysis is directly linked to DGR, see eqn [1], and operation, for example, electrokinetic (‘activation’) limi-
thus the required minimum cell voltage decreases with tations. As a consequence, the potential of the anode
increasing temperature. becomes more positive and that of the cathode becomes
The important values of DHR, DGR, Vth, and Vr for more negative. Catalysts are used in both electrodes to
water electrolysis at different temperatures are sum- accelerate the respective reactions. Oxygen evolution at
marized in Table 2. the anode involves the overall transfer of four electrons
The cell voltage of an operating electrolysis cell is and is therefore a more complex reaction compared with
significantly higher than the theoretical reversible cell hydrogen evolution at the cathode. For this reason, the
voltage derived from thermodynamics. The real cell overpotential at the anode becomes dominant at higher
voltage Vcell can be regarded as the sum of the reversible current densities, as seen in Figure 3. Apart from the
cell voltage Vr, the voltage drop iRA caused by the area- activation overpotential, additional losses can be caused
specific ohmic resistance (cell resistance normalized by by a concentration overpotential due to limitations in the
its area, RA) of the cell at a certain current density i, and mass transport of reactions species, and by a so-called
the overpotentials at the positive (anode) and negative bubble overpotential (see below) at the electrode (not

Table 2 Thermodynamic data for water electrolysis at different temperatures and 101.325 kPa
DHR (kJ mol1) Vth (V) DGR (kJ mol1) Vr(V)
H2O(l) at 298.15 K 285.9 1.48 237.2 1.23
H2O(g) at 373.15 K 242.6 1.26 225.1 1.17
H2O(g) at 1273.15 K 249.4 1.29 177.1 0.92

2.0

1.8
 Cath
Cell voltage (V)

1.6
 An
1.4
iR A
1.2
Vr
1.0
0 200 400 600 800 1000
Current density (mA cm–2)

Figure 3 Typical cell voltage versus current density characteristic of a polymer electrolyte membrane (PEM) electrolysis cell with
illustration of the contribution of different potential losses during operation.
 Fuels – Hydrogen Production | Water Electrolysis 397

included in Figure 3). All losses in the electrolysis cell Note that in fuel cell applications, the reversible cell
result in the need to supply extra electrical energy to voltage Vr is always the reference point for calculation of
effect water decomposition. the voltage efficiency.
An alkaline electrolysis cell shows behavior compar- The application of only eqn [6] assumes that the water
able with that depicted in Figure 3 for a PEM elec- decomposition takes place with complete conversion
trolysis cell. The ohmic voltage drop iRA in an alkaline according to Faraday’s law. The amount of produced
electrolysis cell is, however, more dominant. Among hydrogen nH2,real, however, will be less than the theoretical
other factors, the electrolyte has a higher resistance due value nH2,ideal due to parasitic stray currents within the cell,
to the larger cell dimensions, for example, distance from diffusion of gases through the diaphragm/membrane,
electrode-to-electrode and thickness of the diaphragm. leakage to the surroundings, and so on. The current ef-
At high current densities, the bubble overpotential plays ficiency eI, or the so-called Faraday efficiency, accounts
a role as the evolving gas bubbles displace the electrolyte for these losses, that is,
in an alkaline cell and thereby reduce the effective area
 
for current flow that, in turn, increases the local current nH 2;real
nH 2;real
eI ¼ ¼ ½7
density. nH2;real I ðnF Þ1


The energy efficiency eE of the water electrolysis

Thus the overall cell (or stack) efficiency ecell (or estack) is
process is often defined as the ratio of the energy content
the product of the voltage efficiency and the current ef-
of hydrogen produced per unit time (with V̇ as flow rate)
ficiency, see eqn [8]. The difference to the previously
to the electrical power Pel needed for its production, as
introduced overall efficiency etotal of the system is that
expressed by eqn [5]. Whereas the lower heating value
losses due to the power consumption of peripheral de-
(LHV) of hydrogen (LHV ¼ 3.00 kWh N m3) is most
vices are not considered.
often used as the reference, efficiency values based on the
higher heating value (HHV ¼ 3.54 kWh N m3) are also
ecell ¼ eV eI ½8
reported in the literature. Thus, the choice of heating
value used should be indicated clearly to avoid mis- Now that the fundamentals of water electrolysis have
understanding. Another common measure of the elec- been discussed, the following sections will provide an
trolysis efficiency is the power consumption, which overview of the technical implementation of different
denotes how much electricity is consumed by the process electrolysis technologies.
to produce one normal cubic meter of hydrogen (kWh N
m3 H2):
Alkaline Electrolysis
DHR V̇ H2 General Principles
eE ¼ ½5
Pel
As their name implies, alkaline electrolyzers use an
aqueous alkaline solution. This is potassium hydroxide
To calculate the overall efficiency etotal at a system level, with a typical concentration of 20–40 wt% as the basic
the power consumption of peripheral devices (e.g., for electrolyte that usually circulates through the cells, as
water pretreatment, hydrogen drying, and purification) has shown schematically in Figure 4. The general equations
to be taken into account. Hence, it is essential that system of the half-cell reactions are listed in Table 1. Typical
boundaries and operating conditions (e.g., hydrogen outlet operation temperatures are between 343 and 363 K. The
pressure) are similar if overall efficiencies of different operating pressure is mostly atmospheric but commercial
electrolysis systems are to be compared with each other. systems with pressures up to 3 MPa are also on the
At a cell level, the voltage efficiency eV is used most market. As stated earlier, alkaline electrolysis is a mature
commonly as it can be measured easily. It is defined as the technology and today nearly all of hydrogen that is
ratio of the minimum theoretical cell voltage to the real produced through electrolysis originates from the em-
voltage during operation of the cell. The thermo-neutral ployment of alkaline electrolyzers.
cell voltage Vth based on the HHV – and not the reversible
cell voltage Vr  is mostly taken as the reference point, Cell Components
see eqn [6]. This approach might be justified as long as The two gas sides of an alkaline electrolysis cell are
only an electrolysis process at near room temperature is separated by a diaphragm. In the past, the standard dia-
considered and that no additional ambient heat is included phragm was made of asbestos, but this material is now
in the entropy term TDSR, refer to eqn [1]: prohibited by health regulations. Alternative materials
have to be impermeable to oxygen and hydrogen, ion
Vth ðT Þ 1:481V conductive, and stable in a caustic environment at high
eV ¼ E ½6
Vcell ðT Þ Vcell ði Þ temperatures of at least 373 K and elevated pressures.
398 Fuels – Hydrogen Production | Water Electrolysis

H2 O2

8 3 5 8

7
4
KOH(aq) KOH(aq)

OH−

1 2
Figure 5 Cross section of a diaphragm made of glass-
reinforced polyphenylene sulfide compound. Material courtesy of
6
Wasserelektrolyse Hydrotechnik.

Figure 4 Simplified schematic diagram of a conventional

alkaline electrolysis cell: anode compartment (1), cathode
compartment (2), diaphragm (3), electrodes (4), cell frame (5),
direct current (dc) power supply (6), end-plates (7), and gas
separators (8).

Most of the developed substitutes for asbestos are com-

posite materials based on microporous polymers or cer-
amic materials, for example,

• onreinforced, microporous polymer membranes based

polysulfones (PSf), for example, polyethersulfone
(PES);
• glass-reinforced polyphenylene sulfide (PPS) com-
pounds, see Figure 5;
• a porous matrix consisting of (sintered) nickel oxide
layers on a mesh or net made of nickel as a permeable
support structure, which may exhibit improved
chemical stability due to added titanium oxide;
• fine-pored, predominantly ceramic, layers made of
potassium titanate supported by a mesh-like support
Figure 6 Detail of a perforated electrode made of nickel-plated
iron for alkaline electrolysis cells. Material courtesy of
Wasserelektrolyse Hydrotechnik.
structure or framework.

Electrocatalysts are usually employed in electrochemical materials for the cathode and have been used for
reactions to lower the activation energy and thus accel- decades in alkaline electrolyzers, as well as in a large
erate the reaction rate. In alkaline electrolysis cells, the number of other industrial processes and inorganic syn-
kinetics of both the HER and the OER depend strongly on theses. A Raney nickel electrode is produced by rolling or
the activity of the electrocatalysts. Electrodes made of hot-pressing a Ni–Al alloy to an electrode support and
iron, nickel, nickel-plated iron, or electrodes activated by then sintering this layer. In a second step, the layer is
nickel sulfide are state-of-the-art in conventional elec- treated with concentrated sodium or potassium hydroxide
trolyzers. It has been demonstrated that activated nickel at an elevated temperature of 343–363 K. This treatment is
sulfide electrodes can lower the overpotential from 250 to called activation and dissolves most of the aluminum out
125 mV compared with iron at a typical current density of of the alloy. For alkaline water electrolyzers, the
250 mA cm2. The perforation of a conventional cathode process can be carried out in situ within the electrolysis
made of nickel-plated iron (System Demag marketed by cells. The remaining porous nickel structure has a large
Wasserelektrolyse Hydrotechnik) is shown in Figure 6. surface area, which gives a high catalytic activity. The
Raney nickel electrodes serve as high-performance durability of such an activated electrode can be enhanced
 Fuels – Hydrogen Production | Water Electrolysis 399

by adding cobalt or molybdenum to the alloy. Other ap- 2

313 K
proaches are the galvanic deposition of Ni–Zn, Ni–Co– 323 K
1.9
Zn, or Fe–Zn alloys on the electrode support (perforated 333 K
plates), or the vacuum plasma spraying (VPS) technique as 343 K

Cell voltage (V)

1.8
353 K
developed at the German Aerospace Center (DLR). Re-
363 K
cently, several combinations of transition metals, such as 1.7
Pt2Mo, Hf2Fe, and TiPt, have been used as cathode ma-
1.6
terials and have shown significantly higher electrocatalytic
activity than state-of-the-art electrodes. 1.5
A large number of mixed oxides have been investi-
gated with the goal of minimizing the anode over- 1.4
potential. Unfortunately, however, few electrocatalysts 0 200 400 600 800
–2
with sufficient long-term stability and acceptably Current density (mA cm )
low cost have been identified. The oxide ruthenium oxide Figure 7 Performance data at different temperatures for an
(RuO2) displays the highest activity toward OER in al- alkaline electrolysis cell with an activated molybdenum-containing
kaline and acidic media but is quite unstable under the Raney nickel cathode and an anode produced by the vacuum
harsh conditions of alkaline electrolysis. Mixed oxides plasma spraying (VPS) technique of a compound of Raney nickel
and cobalt spinel. Data courtesy of DLR.
containing nickel and cobalt, with a perovskite structure
(e.g., lanthanum nickel oxide (LaNiO3) or a spinel
structure (e.g., nickel cobaltite (NiCo2O4), have been
3.9
employed as electrocatalysts in alkaline water elec- 313 K
trolysis. In particular, the cobalt spinel cobalt(III) oxide 333 K
3.4
(Co3O4) displays high activity and good stability for the 353 K
Cell voltage (V)

OER in an alkaline medium. Nevertheless, Raney nickel

2.9
remains the preferred option for industrial applications.
In conventional electrolyzers, the electrodes do not
2.4
have any direct contact with the diaphragm, as indicated
in Figure 4 for a single cell. The electrolyte circulates
1.9
separately in the anode and cathode compartments be-
tween the diaphragm and the end plate of the respective
1.4
cell. The electrodes are made as perforated sheets and are 0 100 200 300 400 500
fixed on the end plates in the middle of the compartments –2
Current density (mA cm )
at a distance of a few millimeters from the diaphragm. The
Figure 8 Performance data of a conventional alkaline
main disadvantage of this configuration is the large dis-
electrolyzer with electrodes made of untreated iron (cathode) and
tance between the anode and the cathode, which results in nickel-plated iron (anode). Data courtesy of Wasserelektrolyse
high ohmic losses. For this reason, the so-called ‘zero-gap’ Hydrotechnik.
electrode design was developed and first implemented in
the Lurgi electrolyzer during the mid-1970s. In this de-
sign, perforated plates with porous and activated surfaces from more than 5.0 kWh N m3 H2 to o4.0 kWh N m3
are fixed on the diaphragm from both sides in a sandwich- H2 at temperatures around 353 K.
like manner. In this way, the distance between the elec- The core component of an electrolysis system is the
trodes is minimized and thus the voltage drop due to the cell stack, in which several cells are assembled to form
electrolyte resistance is reduced. The perforated holes are one unit. The simplest of past designs is based on a
of different sizes for the anode and the cathode to ac- unipolar concept in which cells were connected elec-
commodate the different diameters of the evolving bub- trically in parallel so that high currents and low voltages
bles. The zero-gap cell has become state-of-the-art in had to be handled. This concept is no longer produced
modern alkaline electrolyzers. Plots of cell voltage versus commercially. Instead, all industrial alkaline electro-
current density at different temperatures for an optimized lyzers are now based on a bipolar series connection of
electrolysis cell developed at DLR are presented in the cells – the so-called ‘filter-press assembly’. The
Figure 7. Both electrodes consist of highly active Raney metal separating plate between two adjacent cells serves
nickel made by the VPS technique. Similar data are pre- as the anode for one cell and the cathode for the other.
sented in Figure 8 for a conventional alkaline electrolysis Currents are low and higher voltages can be applied to
cell with electrodes made of untreated iron (cathode) the stack, which generally simplifies the power con-
and nickel-plated iron (anode) for comparison. Using ditioning. The active area of a cell in an alkaline elec-
optimized electrodes, the power consumption is reduced trolysis stack can be several square meters (up to 4 m2)
400 Fuels – Hydrogen Production | Water Electrolysis

and is mainly limited by the gas discharge, which would in order to retain fine droplets of liquid potassium hy-
cause unduly high losses due to bubble overpotential at droxide in the electrolyte circulation. Control valves
the electrodes if the active area, and thus the hydrogen after the demister regulate the pressure inside the system.
production capacity, were too high. Stacks operating at The remaining potassium hydroxide is washed out in
higher pressures (e.g., the Lurgi system) take advantage a gas scrubber. Without a deoxidizer as an additional
of the smaller bubbles evolving at the electrodes. For a purification unit, the typical purity of hydrogen and
given current density, such a stack can be designed to be oxygen is in the range of 99.8–99.9% and 99.3–99.8%,
more compact than a stack operating at atmospheric respectively. High purity can be guaranteed only if cir-
pressure. For larger hydrogen production capacities, one culation in the cells is fast enough. For this reason, the
stack can be assembled easily from several hundreds of lower range of operation is limited to >20% of the
cells. nominal hydrogen production rate in most of the systems.
Often a (low-pressure) gas reservoir for hydrogen is in-
Peripheral Components of Industrial stalled to guarantee a constant hydrogen flow for the
Electrolyzers downstream application.
The electrolyte from both sides is remixed after the
Apart from the stack, an electrolysis system consists separators and impurities are filtered out. A circulation
of a number of peripheral components, as illustrated in pump can be used to assist the natural convection of the
Figure 9. The general principles of the main components electrolyte. Forced convection of the electrolyte is ad-
are described in this section: vantageous with regard to heat dissipation and uniform
• and
power conditioning, which consists of a transformer
rectifier if energy is available from an ac power
electrolyte concentration in the cells. Nevertheless, most
of the modern alkaline electrolyzers operate without a
supply; circulation pump.
• gas separation and conditioning; A feed water system is necessary to maintain the
• feed water system;
electrolyte (lye) cycle; concentration of the alkaline solution at a constant level.
• gas purification andandcompression. With the help of a pump, a process control system adds as
• much water to the solution as is removed by the water
decomposition. The quality of the feed water is guaran-
The ascending gases drag the electrolyte along and force teed by deionization in order to prevent fouling in the
circulation on each side of the stack. Separators, in which system. The compact design of an atmospheric alkaline
the gas is separated from the electrolyte, are positioned electrolyzer is presented in Figure 10.
on top of, or above, the stack. Inside these gas separators, Additional components for hydrogen conditioning are
heat exchangers cool the gas and the electrolyte. More- gas driers to reduce the water content in the product
over, the separators can be equipped with special internal stream, deoxidizers (catalytic gas purifiers) to lower the
components to reduce the aerosol level of hydrogen and oxygen concentration in hydrogen to a few ppm
oxygen. Subsequently, the gases flow through a demister (510 ppm), gas compressors to deliver the storage

H2
Control
valve Demister Gas
scrubber Gas
reservoir

Gas
separator
Water
purification

Alkaline
Filter H2O
Rectifier electrolysis
stack Feed water
pump
Transformer
Lye
pump Lye
tank

Figure 9 Simplified illustration of an alkaline electrolysis system. Balance-of-plant does not include gas purification and compression.
 Fuels – Hydrogen Production | Water Electrolysis 401

pressure (e.g., 20 MPa), a high-pressure gas reservoir and, atmospheric electrolyzers. In general, however, the first
if necessary, a refilling station for gas bottles or tanks. stage of a mechanical compressor can be economized if a
Major manufacturer’s of industrial alkaline electrolyzers pressurized electrolyzer is used and hydrogen has to be
are listed in Table 3. stored under high pressures (>20 MPa). For electrolyzers
Although considerable research and development operating at atmospheric pressure, the power con-
progress was made in the 1980s and 1990s, present-day sumption decreases weakly at high production rates,
alkaline electrolyzers are typically operated at a cell whereas pressurized systems show a stronger dependence
voltage of 1.8–2.1 V and a current density of approxi- on the production rate.
mately 200 mA cm2. The power consumption of alka-
line electrolyzers with a nominal capacity higher than
10 N m3 h1 is around 4.1–5.0 kWh N m3 of produced Proton-Exchange Membrane Water
hydrogen, which includes power conditioning, electrolyte Electrolysis
cycling, and gas treatment, but not cooling water, gas fine
General Principles
purification, and mechanical compression. This power
demand corresponds to an energy efficiency of 60–73% If an acidic solid polymer is used as the electrolyte, the
based on the LHV. Power consumption as a function technology is called polymer electrolyte membrane
of the hydrogen production rate is summarized in (PEM) electrolysis. It was first described in a patent by
Figure 11 and reveals that smaller systems are charac- GE in 1970. The membrane functions both as the gas
terized by higher power consumption. In addition, pres- separator and the solid electrolyte. Therefore, only de-
surized electrolyzers have a higher power demand than ionized water without any electrolytic additive is fed to

Figure 10 Front and side views of a compact alkaline, atmospheric electrolysis system (EV150 system of Demag, Wasserelektrolyse
Hydrotechnik).

Table 3 Manufacturers of alkaline electrolyzer systems (a selection), database 2007/2008

Company System name (operation pressure) H2 production
capacity (per module)
(N m3 h1)
Accagen (Switzerland) AGE (1.0/3.0 MPa) 1–100
ELT (Germany) Bamag (atmospheric) Lurgi (3.0 MPa) 3–330 100–760
Hydrogen Technologies (Statoil Hydro) 50xx (atmospheric) HPE (1.2 MPa) 10–485 10–65
(Norway)
Hydrogenics (Canada) HySTAT-A (1.0/2.5 MPa) 1–60
Idroenergy (Italy)  (0.18/0.39 MPa) 0.4–64
SAGIM (France) BP/MP (0.4/0.8/1.0 MPa) 0.1–10
Teledyne Energy Systems (USA) TITAN (0.79/0.89 MPa) 2.8–56
Wasserelektrolyse Hydrotechnik (Germany) Demag (atmospheric) 0.12–250
402 Fuels – Hydrogen Production | Water Electrolysis

6.5 the anode. Carbon-based materials, which are used in

Power consumption (kWh N m3 H2) Pressurized electrolyzer
PEM fuel cells, cannot be employed since the carbon
6.0 Atmospheric electrolyzer
undergoes electrochemical oxidation at voltages higher
5.5 than 0.9 V through the following reaction:

5.0 C þ 2H2 O-CO2 þ 4Hþ þ 4e ½II

4.5 Consequently, materials such as titanium and coated

stainless steel have to be used for constructing the bipolar
4.0 plate, current collector, and, if necessary, the support for
catalysts.
3.5
1 10 100 1000
Membrane–Electrode Assembly
Hydrogen production rate (N m3 h–1)
As already explained, the heart of the cell consists of an
Figure 11 Power consumption of different industrial alkaline
electrolyzers operating under pressure or under atmospheric
MEA with precious metals and their oxides as electro-
conditions. Revised data from Wittstadt. catalysts (platinum, iridium, iridium dioxide (IrO2),
rhodium, rhodium oxide (RhO2). State-of-the-art solid
polymers are nonreinforced ionomers comprised sulfonic
H2 0.5O2 acid-functionalized, perfluorinated side chains on a
polytetrafluoroethylene backbone (perfluorosulfonic acid
Current
Membrane
distributor
polymers – PFSA membranes) such as Nafions, Fle-
(solid polymer)
H+ mions, and fumapems. These PFSA membranes are
− +
known for high oxidative stability as well as high proton
Cathode Anode conductivity and good durability: lifetimes approaching
Bipolar several 10 000 h and proton conductivities as high as 0.1 S
Electrode H2O plate (BiP) cm1 have been reported. The typical membrane thick-
(electrocatalysts)
ness varies from approximately 100 to 200 mm (in a
nonswollen state), which is a good compromise between
Figure 12 Main principle of a polymer electrolyte membrane ohmic drop, mechanical stability, and gas permeability.
(PEM) electrolysis cell. According to the literature, Nafion 115 and 117 from
DuPont are the most commonly employed membranes in
PEM electrolysis cells to date. The gas purity of the
the cell – one of the great advantages of this technology. produced hydrogen reaches values of up to 99.99% on a
Protons are transported in the acidic membrane and the dry basis and the number, n, of water molecules per
half-cell reactions (OER and HER) can be expressed proton transported across the membrane has been ob-
according to the equations in Table 1. The fundamental served to be in the range n ¼ 2–5 due to electro-osmotic
design of a PEM electrolysis cell, as shown in Figure 12, water drag. The production of the ionomers is in-
is similar to that of a PEM fuel cell. The two half-cells trinsically very expensive and the membranes suffer from
are separated by the membrane. If the electrodes are relatively low mechanical strength. Given that PFSA
directly coated on the membrane, then this core com- membranes have low glass-transition temperatures and
ponent is called the membrane–electrode assembly decreasing ionic conductivities at temperatures 4373 K,
(MEA). Porous current collectors (distributors) are used cell operation is usually limited to o373 K. The use of
to enable an electric current to flow between the bipolar inorganic fillers such as silica has been found to improve
plates and the electrodes and, simultaneously, the the water retention inside a perfluorosulfonic membrane
evacuation of the generated gas bubbles from the elec- at temperatures 4373 K and results in enhanced
trodes. As in most PEM fuel cells, the electrically con- performance.
ductive bipolar plates have flow-field structures (e.g., To reduce material costs, alternative nonfluorinated
parallel channels) to transport liquid water to the elec- membranes were investigated recently. Sulfonated aro-
trode (at the anode) and oxygen and hydrogen out of the matic polymers such as polyether ether ketone (PEEK),
electrolysis cell. PSf, and polybenzimidazole (PBI) show adequate chem-
The main difference between PEM fuel cells and ical and electrochemical properties at low production
electrolysis cells is the choice of material for the cell costs. On the contrary, the high degree of sulfonation
components. Owing to the high cell voltage of Vcell 4 leads to excessive water swelling at elevated temperatures
1.5 V in an acidic environment, the components of the and, in turn, a significant reduction in the mechanical
electrolysis cell have to be corrosion resistant at least at strength of the membrane. To overcome this drawback,
 Fuels – Hydrogen Production | Water Electrolysis 403

2.1 10
Cell resistance

At 1 KHZ [mΩ]
8
2.0 6

4
1.9

Current voltage (V)

2
20 40 60 80
1.8 Temperature (°C)

1.7

1.6

1.5 25 °C 45 °C 65 °C 85 °C

1.4
0 100 200 300 400 500 600 700
−2
Current density (mA cm )

Figure 14 Performance data of a standard electrolysis

Figure 13 Cross section of a membrane–electrode assembly membrane  electrode assembly at different temperatures for
consisting of a Nafion 117 membrane and electrodes made of atmospheric operation. Inset shows corresponding cell
pure platinum (cathode) and pure iridium (anode), respectively resistance measured at 1 kHz.
(Fraunhofer ISE).

covalently cross-linked aromatic ionomer blends have 2.0

Optimized MEAs Metal
recently been proposed and investigated. (M)
1.9 Ir1Ru2OxM0.1 Fe
The acidity in Nafion membranes is similar to that of Ir1Ru2OxM0.1 Ni
a 20 wt% sulfuric acid solution. Therefore, acid-resistant, 1.8 Ir1Ru2OxM0.1 Co
Cell voltage (V)

noble metal catalysts must be used as electrocatalysts. Ir1Ru2OxM0.1 Sn

(Ir5Ru1)Ox
Typically, the membrane is coated directly with noble 1.7
metal electrodes at a total loading of about 3–8 mg cm2.
Ir
Another approach is to apply the electrocatalysts on the 1.6
current collector, for example, by spraying. In a second
1.5
step, the electrode and current collector are fixed to the
membrane. The loading of the catalysts is biased such
1.4
that the anode has up to 6 mg cm2 and the cathode 0.0 0.2 0.4 0.6 0.8
about 1–2 mg cm2, because of the slower kinetics of the Current density (mA cm ) −2

OER. Unlike PEM fuel cell catalysts, which are sup-

Figure 15 Performance data of an optimized electrolysis
ported on conductive substrates, catalysts used in PEM
membrane–electrode assembly operated at 353 K and
electrolysis are mostly unsupported, at least at the anode. atmospheric pressure. Cathode consists of platinum black.
The cross section of a standard MEA for PEM water Binary and ternary catalyst systems used for the anode.
electrolysis manufactured by the hot-pressing method is
shown in Figure 13. The membrane is Nafion 117 coated
with platinum black on the cathode and iridium black on of metals from the platinum group, for example, mixed
the anode. The loading on both sides is approximately oxides of iridium and ruthenium. The improved per-
2 mg cm2. Performance data for this MEA in a titanium- formance of MEAs with binary and ternary unsupported
based, laboratory, single cell (active area: 50 cm2) at dif- catalyst systems at the anode is demonstrated in
ferent temperatures and atmospheric conditions are Figure 15. The use of a mixed-oxide anode catalyst for
presented in Figure 14. The cell efficiency increases with the MEA has been found to increase the voltage effi-
temperature due to the faster kinetics (mainly of the ciency to 93% at 0.6 A cm2, 353 K, and 101.325 kPa.
OER) and the lower resistance of the membrane; cell (Note, unless otherwise stated, all cell efficiency values
impedance measured at 1 kHz is given in the inset. reported here are based on the HHV.) Similar, or even
Starting from the above standard configuration, con- better, results at 363 K with voltage efficiencies of about
siderable progress has been made in improving the per- 90–96% have also been reported. At 353 K, cell voltages
formance of PEM electrolysis MEAs and reducing the of o1.6 V at 1 A cm2 have recently been observed with
noble metal loading of the electrodes. In particular, the anode catalysts based on IrxRuyTazO2. The best result
kinetics of the OER can be improved by using other was obtained with an Ir0.6Ru0.4O2 anode and a 20 wt%
catalyst systems. The electrocatalytic activity for the platinum–carbon cathode (Vcell ¼ 1.567 V at 1 A cm2,
OER in acidic media is enhanced by using mixed oxides Nafion 115).
404 Fuels – Hydrogen Production | Water Electrolysis

Figure 16 Typical current collectors as used in polymer electrolyte membrane (PEM) electrolysis cells/stacks. (a) foam made of
titanium offers best performance as an anode current collector. (b) Expanded titanium sheet also used in electrolysis cells.

A promising option to reduce the noble metal loading expensive titanium foam as a current distributor/col-
of the electrodes is the application of a supported cata- lector, single- or multilayered expanded titanium sheets
lyst. At 363 K, carbon-supported palladium catalysts on offer an alternative for cell construction, see right-hand
the cathode (40 wt% palladium–carbon, 2.4 mg cm2; side of Figure 16. The metallic components in the
1.65 V at 1 A cm2) yielded higher efficiency compared electrical path can be lightly coated with noble metals
with carbon-supported platinum catalysts (30 wt% (e.g., sintered, porous, Pt-plated titanium plates) to
platinum–carbon, 2.0 mg cm2; 1.72 V at 1 A cm2). In minimize losses with time due to contact resistances.
both cases, the anode of the electrolysis cell consisted of Similar to the bipolar plates of alkaline electrolyzers, a
iridium (2.4 mg cm2). Recent results have even shown bipolar plate of a PEM electrolyzer separates two ad-
that a comparable performance in a small laboratory test jacent cells of the stack and is at the same time the anode
cell could be achieved with carbon-supported platinum of one cell and the cathode of the adjacent cell. A typical
and palladium nanoparticles at noble metal loadings of construction of a bipolar plate made of titanium is pre-
o1.0 mg cm2 at the cathode. sented in Figure 17. Flow-field structures with parallel
channels are machined on both sides of the plate.
Moreover, the plate includes gasket grooves, and inlets
Current Collectors, Bipolar Plates, and Stack
and outlets for water circulation and gas evacuation.
Design
Obviously, the design is suitable only for laboratory
The main task of the current distributors/collectors electrolyzers, as the costs for the material and manu-
(sometimes also called distributor screens) in an elec- facturing process for such a bipolar plate are very high.
trolysis cell is to supply electric power uniformly to the Relatively few data have been published concerning
MEA from the bipolar plate. In addition, water must be the construction and materials used for bipolar plates,
fed to the electrode surface (anode) and the generated sealings, and stacks. This is because as most companies
gases have to be released smoothly to the channels of the have developed proprietary cells and stack designs. In
bipolar plates. The first choice for the anode of labora- principle, the stack of a PEM electrolyzer is based on the
tory electrolysis cells/stacks are porous foams or sintered filter-press configuration. The number of cells typically
plates made of titanium. This is because the oxide layer ranges from 10 to 120. To facilitate flow distribution and
on the surface provides very high corrosion resistance gas evacuation, cells are sometimes grouped into mod-
and acceptable conductivity. Hydrogen embrittlement is ules, and several modules are electrically connected in
an issue at the cathode, particularly at HTs and pressures. series to form stacks. The active area of a PEM elec-
Only valve metals, a few stainless-steel alloys, and trolysis cell is appreciably smaller than that of an alkaline
graphite are known to be sufficiently resistant to cor- electrolysis cell: Although prototypes of PEM electro-
rosion. Graphite shows the greatest resistance to em- lyzers with active areas of up to 2500 cm2 have been
brittlement for hydrated PEM conditions whereas reported, most present developments aim for an active
titanium is suitable to only a limited extent. For this and area that is clearly lower than 1000 cm2.
economic reasons, (sintered) carbon-based paper and To avoid high manufacturing costs caused by ma-
felts, which are employed in PEM fuel cells, are used chining a titanium bipolar plate, frame constructions
commonly as long as the overpotential of the cathode are used to build up the cell in most developments of
remains acceptably small. To avoid the application of electrolyzers. A thin metallic plate or foil serves as the
 Fuels – Hydrogen Production | Water Electrolysis 405

bipolar separator plate, which may be made of titanium thermoplastic, nonconductive separator plates with
or another corrosion-resistant and conductive material, structures for water feeding and gas evacuation. The
for example, coated stainless steel. Frames on the electric current flows through conductive bridges or pins
hydrogen and oxygen sides fix the MEA and comprise embedded in a plate that connects the current collectors
the current-distributors/collectors, manifolds, and seal- of two adjacent cells.
ings. Different concepts for the frames are known: die-cut Two different stack concepts are shown in Figure 18. A
rubber, metal laminate, or plastic-molded frames. Bipolar 15-cell stack from Hydrogenics with a metallic plate de-
plates can also be produced by super-plastic mold sign and an active cell area of 90 cm2 is presented on the
forming of a titanium alloy, which is a method suitable left. The nominal hydrogen production rate is 940 L h1
for the economic production of large quantities. Another for an electric power input of 4.2 kW. At a current density
approach, particularly suited for smaller stack design and of 1.7 A cm2, the stack has a power consumption of 4.5
thus relatively small thermal loads, is the use of molded kWh N m3 hydrogen. A 15-cell stack from the Fraun-
hofer ISE that employed injection-molded separator plates
with an active area of 80 cm2 is shown on the right of
Figure 18. The current density is relatively low, namely,
0.2 A cm2, and results in a hydrogen production rate of
approximately 100 L h1 for an electric power input of
0.44 kW. At this operating point, the stack has a power
consumption of 4.4 kWh N m3 hydrogen. The operating
pressure for the two stacks is 0.8 and 1.0 MPa, respectively.
Compared with alkaline electrolyzers, PEM water
electrolyzers offer a significant advantage in terms of the
high-power density of the stacks. At present, current
densities from 1.0 to 2.0 A cm2 at a cell voltage of 1.7–
1.85 V are usual. The current efficiency of the cells is
quite high, with values ranging from 95 to 98%. Thus,
the power consumption at a stack level varies from 3.9 to
4.5 kWh N m3 hydrogen. In most cases, the PEM elec-
trolysis stacks operate at temperatures from 333 to 363 K
and the very compact design of the stacks allows high-
pressure operation. For example, the companies Tread-
well and Giner have jointly developed an oxygen-
generation plant for submarines that produces oxygen
Figure 17 Conventional construction of a bipolar plate made of
(and hydrogen) at approximately 20.7 MPa. The stack is
titanium for pressurized PEM electrolysis at 3.0 MPa (Fraunhofer
ISE). Bipolar plate has an active area of 80 cm2 and includes enclosed in a pressure container that is filled with ni-
machined flow-field structures, gasket grooves, and inlets and trogen at the same pressure. When the membrane is
outlets for water circulation and gas evacuation. mechanically supported, a differential pressure of several

Figure 18 Two different stack designs for polymer electrolyte membrane (PEM) electrolyzers. Left: 91 E series stack made of 15 cells
from Hydrogenics with hydrogen production rate of 0.94 N m3 h1, metallic plate design. Photo courtesy of Hydrogenics. Right: 15-cell
stack from Fraunhofer ISE with hydrogen production rate of 0.11 N m3 h1, injection-molded plate design.
406 Fuels – Hydrogen Production | Water Electrolysis

hundred kilo pascals is possible over the membrane. With part of a PEM electrolysis stack due to ionomer thinning,
the help of a pressure compensation system for the MEA, with subsequent loss in gas purity, and finally failure
pressure differences between the hydrogen and oxygen (leakage). The lifetime of PEM electrolysis stacks varies
side of up to 17.2 MPa have been reported. Such a stack from several thousands of hours of operation up to more
design has recently been developed at Giner and its than 33 000 h; the latter lifetime, however, was achieved in
application allows replacement of the high-pressure a military application where economic issues were of
water feed pump. secondary importance. The high durability of an 11-cell
The lifetime of a PEM electrolysis stack is a less stack from Hydrogenics running at 2.0 A cm2 under ac-
critical issue than for a PEM fuel cell stack. In spite of celerated test conditions is demonstrated in Figure 19.
very high current densities, the stack efficiency is similar,
or even superior, to that of PEM fuel cells, and the water
PEM Water Electrolyzer System
decomposition is an endothermic reaction. Moreover, the
liquid flow regime of the feed water enables effective A PEM electrolyzer has a similar layout to an alkaline
temperature control of the stack. Consequently, thermal electrolysis system, see Figure 20. The system consists
management of the stack is easier to control over a broad of the electrolysis stack, gas/liquid separators for
range of operation, provided that an appropriate cell de- hydrogen and oxygen, a circulation loop for the feed
sign provides a homogeneous current flow over the elec- water at the anode with a heat exchanger for the thermal
trodes. The membrane is considered to be the weakest management, pressure control valves, a transformer, and
rectifier for power-conditioning and a control system
including safety devices. An ion-exchanger can be inte-
3.0 grated into the circulation loop to maintain high purity
2.7 of the feed water in the system. Additional downstream
components have already been explained in a previous
Cell voltage (V)

2.4 On-going testing

section. If necessary, a second circulation loop can be
2.1 added at the hydrogen side for better stack cooling. In
most cases, however, the hydrogen side operates pas-
1.8
sively. Owing to electro-osmotic drag, water accumulates
1.5 in the gas–liquid separator on the hydrogen side and is
transported back to the oxygen side from time to time.
1.2
0 5000 10 000 15 000 20 000 The hydrogen side is pressurized in most PEM elec-
Operating time (h) trolyzers. The oxygen side can be either pressurized or
operated at near-atmospheric conditions. In the former
Figure 19 Long-term performance data of a polymer electrolyte
membrane (PEM) electrolysis stack for more than 17 000 h of case, the mechanical stress for the membrane can be kept
operation. Data courtesy of Hydrogenics. Stack is operated at a small but the peripheral components of the oxygen side
current density of 2.0 A cm2. have to be compression-proof and the feed water pump

O2 H2
Control
valve Demister Condenstate
trap
Feed water Gas
pump reservoir
Gas
H2O separator

PEM
Circulation Ion Electrolysis
pump exchanger stack

+ −

Rectifier Transformer

Figure 20 Typical layout of a polymer electrolyte membrane (PEM) electrolysis unit. In this case, the oxygen side operates under
pressure.
 Fuels – Hydrogen Production | Water Electrolysis 407

has to work against the system pressure, and thereby Hamilton Sundstrand are advancing the development of
invokes higher investment costs. For oxygen at near-at- this technology.
mospheric conditions, the membrane has to withstand Available information on the power consumption of
the differential pressure between the hydrogen and the PEM electrolysis stacks and systems as a function of the
oxygen sides. Whereas a sophisticated mechanical sup- hydrogen production rate is summarized in Figure 21. It
port for the membrane is necessary, the peripherals of is clear that stacks of different size exhibit the same ef-
the oxygen side do not need to be compression-proof. ficiency. The power consumption of the stacks decreases
Compared with alkaline electrolysis systems, the need only slightly with higher production rates, with an
for gas-purification steps beyond the PEM stack output is average value of 4.2 7 0.3 kWh N m3 hydrogen. The
less due to the absence of residual lye in the gas flow (no stack efficiency of 67–77% (based on LHV) corresponds
gas scrubber). In addition, the PEM electrolysis stack to the efficiency of advanced alkaline electrolysis stacks.
possesses a higher power density, which means that the Nevertheless, it must be remembered that PEM elec-
PEM electrolyzer footprint per unit of gas produced is trolysis stacks are operated at significantly higher current
smaller. Different system designs are available in the densities. At the system level, Figure 21 reveals a strong
literature. correlation between the power consumption and the
Up to now, PEM electrolyzers have been com- hydrogen production rate. Smaller systems require the
mercially available only for smaller hydrogen pro- same balance-of-plant as larger systems and this results
duction rates such as those from alkaline electrolyzers. in a higher power consumption of 7–8 kWh N m3
Apart from gas generators for laboratory equipment hydrogen.
(hydrogen production rates 51 N m3 h1), today only a As a result of the emerging FC industry, it is expected
few manufacturers offer PEM electrolyzers with that electrolyzers will attract increasing attention in the
a hydrogen production rate of up to approximately near future. In particular, PEM electrolyzers offer the
10 N m3 h1 or have systems available for demonstration possibility of on-site hydrogen and oxygen generation in
projects and early-adopter partnering, for example, small, highly efficient units that are suitable for distrib-
Proton Energy Systems (subsidiary of Distributed Power uted operation. Hence, PEM water electrolyzers are
Systems, USA), Hydrogen Technologies (subsidiary of considered to be an important component in a future
StatoilHydro, Norway), Hydrogenics (Canada), and h- energy industry where renewable energy makes a major
tec (Germany). The main fields of application are contribution to the power supply.
similar to those for small alkaline electrolyzers, namely,
production of semiconductor electronics, generator Unitized Regenerative Fuel Cells Based on PEM
cooling in power plants, meteorological applications Technology
(registering balloons), and materials processing (coatings,
The idea of an energy supply based on renewable energy
metallurgy, etc.). In the United States, PEM electro-
sources such as wind and solar power offers a new field
lyzers are well established in niche markets for military
for hydrogen production from electrolysis. In this con-
and space applications. Here, companies such as
text, the following two fields of application are under
investigation.
10
• overcome
The use of hydrogen for temporary energy storage to
Power consumption (kWh N m–3 H2)

the mismatch between the intermittent

Systems
8
power supply from renewable energy sources and the
power demand. This could be implemented either in
large, grid-connected, energy storage systems, for
6 example, near off-shore wind farms or in smaller off-
grid installations for remote settlements, huts, tele-
communication stations and so on.
4
Stacks • The production of hydrogen as a fuel for automotive
applications in large centralized electrolysis instal-
2
lations using the surplus electricity from wind farms
0.1 1.0 10.0 100.0 or other renewable energy sources. Subsequently, the
Hydrogen production rate (N m3 h–1) hydrogen has to be distributed to refilling stations.
Figure 21 Power consumption of different polymer electrolyte For smaller applications (100 W to a few 10 kW), such as
membrane (PEM) electrolysis stacks and systems as a function
of the hydrogen production rate (’ commercial systems; m
autonomous power supplies (APS), the PEM technology
systems under development; & commercial stack; D stacks is highly attractive. In general, the combination of a water
under development). electrolyzer, a storage tank for the produced gases, and an
408 Fuels – Hydrogen Production | Water Electrolysis

FC to generate electricity is called a regenerative fuel In the recent Revcell Project of the European Com-
cell (RFC) and is the basis for an electricity storage mission (Project ENK6-CT-2002-00661), a laboratory
device using hydrogen. In APS applications, an RFC prototype of a URFC was constructed and subjected to a
brings several advantages compared with conventional field test. The URFC system consisted of a URFC stack,
systems with large batteries and a diesel generator as the conventional pressure tanks for hydrogen storage, a
backup: pressure tank filled with a zeolite adsorbent with a pore
diameter of 5 Å (capacity: 0.13 N m3 oxygen) for oxygen
• acity
the required power can be decoupled from the cap-
of the energy storage unit;
storage, and a bi-directional dc–dc converter.
The URFC stack was constructed by the Energy Re-
• high specificself-discharging
minimized allows seasonal storage; search Centre of the Netherlands (ECN) and consisted of
• low noise and no local emission of pollutants.
energy; and 20 cells based on PEM technology, as shown in Figure 23.
• Each cell had an active area of 156 cm2. In FC mode, the
stack had a nominal electric output of 250 W at 0.2 MPa,
The main drawback is the rather low overall efficiency of whereas in EL mode it operated at 0.9 MPa with a nominal
the RFC. If the electrolyzer and the FC are combined in electric input of 530 W. The hydrogen side employed
one cell stack, the device is called a unitized regenerative graphite-based materials of the type used in PEM fuel
fuel cell (URFC) and offers potential for further re- cells. On the oxygen side, titanium-based components
duction in costs, volume, and weight. The principle of a were used to resist the high potentials. A mixture of
URFC for a grid independent, autonomous power supply platinum and iridium gave the highest electrocataly-
(APS) using renewable energy sources is depicted in tic activity for oxygen reduction and water splitting.
Figure 22. Performance data of the URFC stack are presented in
During periods of solar radiation, the photovoltaic Figure 24. The nominal operation point was set at 15.6 A
generator primarily supplies the electric load with energy and corresponded to a current density of 100 mA cm2.
through the charge controller. Excess energy is used in the The performance of the URFC stack was characterized by
EL mode to generate hydrogen and oxygen in the URFC its round-trip efficiency, which was defined as
and to load the battery. The produced gases are retained
in storage tanks. When the electricity generated by the VFC ði Þ
eURFC ¼ ½9
photovoltaic panels cannot meet the energy demand of the VEL ði Þ
electric load (e.g., during the night, cloudy weather, winter),
the URFC runs in the FC mode. The produced water can where VFC is the voltage of the URFC stack in FC mode
be stored for the next EL cycle. The battery is used to and VEL is the stack voltage in EL mode. A round-trip
bridge the start-up time and the switching time from the EL efficiency of 42% was achieved at the nominal operation
to the FC modes. Moreover, the battery can be used for point. Even higher efficiencies were reported in literature.
short-term storage and/or for peak shaving. A dc–dc con- The bi-directional dc–dc converter has to level the
verter equalizes the stack voltage and the voltage of the dc operation voltage between the URFC in both modes and
busbar. For a URFC stack, a bi-directional dc–dc converter the dc busbar voltage defined by the battery. In the FC
has to be used as the electric power has to change direction mode (battery charger), a step-up converter is required
between the EL and FC modes. The complete system has to level the different voltages, whereas in the EL
to be controlled by an energy-management system to op- mode (power supply), a step-down converter is used. A
timize the interaction between the different devices. bi-directional working device was developed by the

EL mode
Photovoltaic Charger
H2 O2 generator controller
FC mode

dc busbar

Unitized
regenerative
H2O dc–dc Electrical
fuel cell Battery
converter load

Figure 22 Simplified schematic diagram of a unitized regenerative fuel cell system for autonomous power supplies. The autonomous
power supply is controlled by an energy-management system (not shown here) dc, direct current.
 Fuels – Hydrogen Production | Water Electrolysis 409

Fraunhofer ISE for the Revcell Project. All components sequence of different steps to remove all the water out of
were selected in order to achieve the highest efficiencies. the small pores of the electrodes. In the Revcell Project, a
In particular, the selection of the MOSFETS and the switching time of approximately 5 min was achieved.
construction of the inductors were most important. The
bi-directional dc–dc converter reached efficiencies of up
to 97% and 98% in the FC and the electrolyzer modes, High-Temperature Steam Electrolysis
respectively.
The most critical aspect in operating a URFC is From the thermodynamic point of view, higher tem-
switching from the EL mode to the FC mode. In the peratures are favorable for the electrolysis process, as
electrolysis mode, the porous structures of the current discussed above. The total energy demand (DHR) for
collectors and electrodes are flooded completely with water electrolysis increases slightly with temperature,
water. At the start, the FC mode requires a certain while the electricity demand (DGR) decreases signifi-
cantly, as shown earlier in Figure 2. The heat promotes
the decomposition of the water molecules in this endo-
thermic process. Thus, the higher the temperature, the
less the electricity needed for the decomposition. At
1073–1273 K, for example, the amount of electrical en-
ergy required to split water is reduced by about 25%.
Consequently, electrolysis technology can be favorable
when HT heat is available as a waste product from other
processes, for example, of fossil, geothermal, solar, or
nuclear origin. The greatest disadvantages of HT oper-
ation are long start-up times and mechanical/chemical
compatibility issues of the type experienced with solid
oxide fuel cells (SOFCs).
One of the first investigations on hydrogen production
by HT steam electrolysis based on solid oxide electrolysis
cells (SOECs) was carried out in the HOT ELLY Project
described by Dönitz and Schmidberger in 1982, but the
project terminated around 1990. Over the past few dec-
ades, research and development has focused on the ad-
Figure 23 20-cell stack of unitized regenerative fuel cell vancement of SOFCs and has resulted in considerable
system designed and constructed by ECN. Each cell has an progress made within materials science. The development
active area of 156 cm2. Photo courtesy of ECN. of SOECs has benefitted from such work, particularly as

18

16
EL mode

14
Stack voltage (V)

Nominal
operation
12 point

10
FC mode
8

6
0 20 40 60 80 100 120
Current density (mA cm–2)

Figure 24 Performance data for 10-cell unitized regenerative fuel cell (URFC) stack in the electrolysis (EL) mode (at 0.2 MPa and
297 K) and in the fuel cell (FC) mode (at 0.1 MPa and 338 K).
410 Fuels – Hydrogen Production | Water Electrolysis

H2
Membrane
separator
90% H2 /
10% H2O
H2O
Feed water
pump O2

High-temperature
Steam electrolysis cell
Helium loop hydrogen
mixture

HT heat
HT source exchanger − +
10% H2 / Rectifier Transformer
90% H2O

Figure 25 Simplified schematic diagram of an high temperature (HT) electrolysis system.

an SOFC can work reversibly as both an FC and an Materials: Electrolyte and Electrodes
SOEC. Recently, there has been a revival of interest in
SOEC technology and different research groups in Eur- Similar to other electrolysis systems, the core com-
ope, America, and Asia are working in this field as dem- ponents of an SOEC are the solid ionic-conducting
onstrated by the Hi2H2 Project (FP6-503765) funded by electrolyte and two porous electrodes. The HT operation
the European Union and the US DOE Nuclear Hydrogen demands materials for the electrolyte and electrodes with
Initiative. Moreover, the electrolysis of water (steam) similar thermal expansion coefficients to minimize the
and carbon dioxide according to the following overall thermal expansion mismatch and resulting mechanical
reaction stress or even failure in the cell. Yttria-stabilized zirconia
(YSZ) is the electrolyte material most commonly used in
an SOEC. Typical dopants for zirconia dioxide are yt-
H2 O þ CO2 -H2 þ CO þ O2 ½III trium oxide (Y2O3) and ytterbium oxide (Yb2O3). These
materials exhibit good mechanical strength and high
has been proposed and is under investigation. This offers oxygen ion conductivity at temperatures above 1073 K.
an attractive option to produce synthesis gas that can then Sc2O3-doped ZrO2 (ScSZ) displays higher ionic con-
be used for the production of synthetic fuels through the ductivity but the material is more expensive and rarely
Fischer–Tropsch process. used as an electrolyte. The hydrogen electrode is mainly
A simplified schematic diagram of an HT electrolysis based on nickel because it exhibits high electrochemical
system is shown in Figure 25. Thermal energy to reactivity. A ceramic metal compound (the so-called
heat steam for the electrolysis process is supplied by an cermet) is widely used for the cathode to enlarge the
HT source, for example, by a gas-cooled nuclear reactor. active electrocatalytic surface area (triple-phase bound-
The same nuclear reactor also provides heat to the ary) and to provide a very porous structure for the
gas turbine of a power cycle (not shown in Figure 25). transport of steam and gas. The cermet consists of a
The generated electricity is used for the electro- mixture of nickel and yttria-stabilized zirconia (Ni-YSZ).
lysis process. The HT heat-exchanger supplies super- Mixed oxides with a perovskite structure, such as
heated steam to the SOEC at a temperature of about strontium-doped lanthanum manganite (LSM), for ex-
1123 K and a pressure of up to 5.0 MPa. The input ample, La0,6Sr0,4MnO3, represent the state-of-the-art for
gas contains both steam and hydrogen to maintain re- anode materials. Just as for the cathode, a cermet is
ducing conditions at the electrolytic cathode. The employed for the oxygen electrode, but is made of
steam–hydrogen mixture at the outlet of the SOEC is LSM-YSZ.
fed to a separator, which is based on membrane tech- As in the case of SOFCs, the cell design of an SOEC
nology, and thereby provides pure hydrogen for further can be either tubular or planar. The tubular geometry
applications. facilitates sealing but the power densities are rather low.
 Fuels – Hydrogen Production | Water Electrolysis 411

Conversely, the planar geometry yields higher power chromium species from the bipolar plate to the bonded
densities but sealing is more demanding. Nevertheless, layer of the electrode; (2) delamination of the oxygen
most of the recent research has been performed on SOECs electrode.
with a planar cell design, due to their easier manu- In summary, whereas promising progress has been
facturability and superior performance. Typical planar made in SOEC technology substantial research and de-
configurations for SOECs are either electrolyte-supported velopment is required to obtain inexpensive, high-per-
or cathode-supported. In the former case, the YSZ or ScSZ forming, and durable electrolysis cells. Optimization of
electrolyte has a layer thickness of B100–200 mm and the electrode materials and a better understanding of deg-
electrodes are relatively thin at 10–30 mm. Electrolyte- radation mechanisms are essential.
supported SOECs have to operate at close to 1273 K to
achieve good ionic conductivity. Cathode-supported cells
are supported by a Ni-YSZ layer of several 100 mm in Concluding Remarks
thickness and the solid electrolyte has a layer thickness of
only 10–15 mm. The operation temperature for such cells At present, hydrogen is used mainly as a chemical raw
is typically 1023–1123 K. The inter-connectors (bipolar material and only very rarely as a fuel. Hydrogen can be
plates) are fabricated from ferritic stainless steel and produced in several different ways, but commercial bulk
protective coatings can be applied to prevent chromium hydrogen is usually produced by steam reforming of
release and migration. natural gas in petrochemistry. Some hydrogen is also
delivered as a by-product from chlor-alkaline elec-
trolysis. Production of hydrogen through water elec-
Solid Oxide Electrolysis Cell Performance and
trolysis is negligible at the present but has been an
Durability
industrial activity for more than 100 years. During the
To date, all reported SOEC performance data are based on past few decades, water electrolysis has attracted new
investigations only with laboratory cells and stacks. Typical interest since hydrogen could be an essential secondary
current densities in an HT electrolysis cell/stack range fuel in a future energy economy.
from 0.25 to 0.5 A cm2. The corresponding cell voltages Today, large-scale production of hydrogen by elec-
are distinctly lower than the thermoneutral cell voltage, trolysis is dominated by the alkaline electrolysis tech-
thereby demonstrating the main advantage of HT elec- nology. Different manufacturers offer systems with a
trolysis. A cell voltage of 1.06 V has been achieved at a hydrogen production capacity of up to several 100 N m3
current density of 0.5 A cm2 and an operating temperature h1. The systems can be operated under either increased
of 1123 K. If the temperature drops to 1023 K, the cell pressure or atmospheric conditions. The power con-
voltage increases to more than 1.26 V. The highest current sumption lies between 4.1 and 6.3 kWh N m3 hydrogen
density ever reported has been 3.6 A cm2 at a cell voltage for systems with a production rate higher than 1 N m3
of 1.48 V and a cell temperature of 1223 K. The HT steam h1. Some larger hydrogen-generation plants have been
electrolysis program at the Idaho National Laboratory has constructed near hydropower stations at various locations
set target values for an integrated, laboratory scale, SOEC around the world.
system based on 12 stacks. Each stack consisted of 60 cells Compared with alkaline electrolyzers, PEM electro-
with an active cell area of 64 cm2. The average cell lyzers offer an option for hydrogen generation in small
voltage was calculated to be 1.3 V at a current density of and efficient units that are particularly suitable for
0.27 A cm2 and a temperature of 1103 K. With a hydrogen distributed, as well as autonomous, operation. The main
production rate of 5.22 N m3 h1 and a nominal power of advantages of this technology are the high efficiency at
16.2 kW, the power consumption was estimated to 3.1 kWh high current densities, a simpler system design, and
N m3 H2. thus very high power densities. Moreover, this tech-
Long-term performance tests of laboratory SOEC nology allows high-pressure operation and the excellent
cells/stacks range from several 100 h up to about 3500 h, partial-load behavior enables direct coupling with re-
but durability is rather poor at the present. In particular, newable energy sources. Thus, PEM electrolyzers
a strong degradation/passivation of the cell/stack per- can be regarded as a bridging technology in an emerging
formance occurs in the early stages of long-term en- FC market that will allow local, carbon-free hydrogen
durance tests. During a period of 1000 h, between 2 and generation with renewable energy sources before the
23% of the initial nominal power is lost. It has been introduction of more centralized, large-scale hydrogen
found that a passivated SOEC can be partly reactivated generation and distribution. Almost all developed PEM
by operation in the FC mode. Post-test analysis of de- systems operate under pressure with hydrogen pro-
graded oxygen electrodes has revealed two major con- duction rates up to 20 N m3 h1. At the stack level, the
tributions to the rapid deterioration in performance: (1) power consumption has an average value of 4.2 7 0.3
an increase in electrical resistance due to the migration of kWh N m3 h1 hydrogen, even at current densities
412 Fuels – Hydrogen Production | Water Electrolysis

several times higher than for alkaline systems. Today, eI current efficiency
PEM electrolyzers are close to broader commerciali- estack stack efficiency
zation and are already found in laboratory and niche etotal overall efficiency – on a system level
applications. Nevertheless, challenges remain in terms eURFC round-trip efficiency of a unitized
of lowering the high cost of materials and achieving regenerative fuel cell
durability comparable with that of an alkaline eV voltage efficiency
electrolyzer. gAn overpotential at positive electrode
Owing to the high efficiency of water decomposition (anode)
at elevated temperatures, HT steam electrolysis could be gCath overpotential at negative electrode
an option in the future, but only in the long term. Since (cathode)
HT heat (e.g., from a nuclear or solar power plant) and Abbreviations and Acronyms
base-load operation are required, this technology would APS autonomous power supply
be favorable for centralized and large-scale hydrogen dc direct current
production plants. At present, research and development DLR German Aerospace Center
work is focused mainly on the realization of long-lasting (Deutsches Zentrum für Luft- und
materials to extend both the lifetime and the perform- Raumfahrt e.V.)
ance of electrolysis stacks. Reduction in system com- ECN Energy Research Centre of the
plexity also remains a major challenge. Netherlands (Energieonderzoek
Centrum Nederland)
EL electrolysis
Nomenclature FC fuel cell
GE General Electric
Symbols and Units
HER hydrogen evolution reaction
F Faraday constant
HHV higher heating value
GOf ;i standard free energy of formation of
HT high temperature
component i (J mol  1)
LHV lower heating value
i current density (A cm  2)
LSM strontium-doped lanthanum
n number of moles of electrons per
manganite
mole of products
MEA membrane–electrode assembly
ndH2ideal theoretical molar flow of hydrogen
OER oxygen evolution reaction
(mol s  1)
OPEC Organization of the Petroleum
ndH2real real molar flow of hydrogen
Exporting Countries
(mol s  1)
PBI polybenzimidazole
Pel electrical power (W)
PEEK polyether ether ketone
RA area-specific ohmic resistance
PEM polymer electrolyte membrane;
(O cm2)
proton-exchange membrane
T temperature (K)
PES polyethersulfone
V̇ flow rate (N m3 h  1)
PFSA perfluorosulfonic acid
Vcell cell voltage (V)
PPS polyphenylene sulfide
VEL cell voltage of a unitized regenerative
PSf polysulfone
fuel cell in electrolysis mode (V)
RFC regenerative fuel cell
VFC cell voltage of a unitized regenerative
ScSZ Sc2O3-doped ZrO2
fuel cell in fuel cell mode (V)
SOEC solid oxide electrolysis cell
Vth thermo-neutral cell voltage (V)
SOFC solid oxide fuel cell
V 0th thermo-neutral cell voltage at STP
SPE solid polymer electrolysis
(V)
URFC unitized regenerative fuel cell
Vr reversible cell voltage (V)
VPS vacuum plasma spraying
V 0r reversible cell voltage at STP (V)
YSZ yttria-stabilized zirconia
DGR free energy of reaction (J mol  1)
DG 0R standard free energy of reaction at
STP (J mol  1) See also: Applications – Transportation: Satellites:
DHR enthalpy of reaction (J mol  1) Batteries; Electrochemical Theory: Hydrogen Evolution;
DSR entropy of reaction (J K  1 mol  1) Oxygen Evolution; Energy: Energy Storage; Hydrogen
ecell cell efficiency Economy; Fuel Cells – Exploratory Fuel Cells:
eE energy efficiency – in general Regenerative Fuel Cells.
 Fuels – Hydrogen Production | Water Electrolysis 413

Further Reading Marčeta Kaninski MP, Stojić DL, Šaponjić DP, Potkonjak NI, and Miljanić
ŠS (2006) Comparison of different electrode materials – Energy
requirements in the electrolytic hydrogen evolution process. Journal
Antonucci V, Di Blasi A, Baglio V, et al. (2008) High temperature
of Power Sources 157: 758--764.
operation of a composite membrane-based solid polymer electrolyte
water electrolyser. Electrochimica Acta 53: 7350--7356. Marshall A, Sunde S, Tsypkin M, and Tunold R (2007) Performance of a
PEM water electrolysis cell using IrxRuyTazO2 electrocatalysts for the
Divisek J and Emonts B (2003) Energy storage via electrolysis/fuel cells.
oxygen evolution electrode. International Journal of Hydrogen
In: Vielstich W, Gasteiger HA, and Lamm A (eds.) Handbook of Fuel
Energy 32: 2320--2324.
Cells  Fundamentals, Technology and Applications, vol. 3, ch. 24,
Mawdsley JR, Carter JD, Kropf AJ, Yildiz B, and Maroni VA (2009) Post-
pp. 416--431. Chichester: John Wiley & Sons, Ltd. ISBN 0-471-
test evaluation of oxygen electrodes from solid oxide electrolysis
49926-9.
stacks. International Journal of Hydrogen Energy 34(9): 4198–4207.
Dönitz W and Schmidberger R (1982) Concepts and design for scaling
up high temperature water vapour electrolysis. International Journal Millet P, Andolfatto F, and Durand R (1996) Design and performance of
a solid polymer electrolyte water electrolyzer. International Journal of
of Hydrogen Energy 7: 321--330.
Hydrogen Energy 21: 87--93.
Grigoriev SA, Millet P, and Fateev VN (2008) Evaluation of carbon-
Mogensen M, Jensen SH, Hauch A, Jacobsen T, and Chorkendorff I
supported Pt and Pd nanoparticles for the hydrogen evolution
(2006) Performance of reversible solid oxide cells: A review. In:
reaction in PEM water electrolysers. Journal of Power Sources 177:
Proceedings of the 7th European SOFC Forum. Lucerne,
281--285.
Hauch A, Jensen SH, Ramousse S, and Mogensen M (2006) Switzerland, 3–7 July.
Ni M, Leung MKH, and Leung DYC (2008) Technological development
Performance and durability of solid oxide electrolysis cells. Journal of
of hydrogen production by solid oxide electrolyzer cell (SOEC).
the Electrochemical Society 153: A1741--A1747.
International Journal of Hydrogen Energy 33: 2337--2354.
Herring JS, O’Brien JE, Stoots CM, Hawkes GL, Hartvigsen JJ, and
Rasten E, Hagen G, and Tunold R (2003) Electrocatalysis in water
Shahnam M (2007) Progress in high-temperature electrolysis for
electrolysis with solid polymer electrolyte. Electrochimica Acta 48:
hydrogen production using planar SOFC technology. International
3945--3952.
Journal of Hydrogen Energy 32: 440--450.
Sandstede G (1989) Moderne Elektrolyseverfahren für die Wasserstoff-
Isenberg AO (1981) Energy conversion via solid oxide electrolyte
Technologie. Chemie Ingenieur Technik 61: 349--361.
electrochemical cells at high temperatures. Solid State Ionics 3–4:
Schiller G, Henne R, Mohr P, and Peinecke V (1998) High performance
431--437.
electrodes for an advanced intermittently operated 10-kW alkaline
Jensen SH, Larsen PH, and Mogensen M (2007) Hydrogen and
water electrolyzer. International Journal of Hydrogen Energy 23:
synthetic fuel production from renewable energy sources.
761--765.
International Journal of Hydrogen Energy 32: 3253--3257.
Kerres J, Eigenberger G, Reichle S, et al. (1996) Advanced alkaline Smolinka T, Grootjes S, Mahlendorf F, Hesselmann J, and Makkus R
(2006) Prototype of a reversible fuel cell system for autonomous
electrolysis with porous polymeric diaphragms. Desalination 104:
power supplies. In: Proceedings of the 3rd European PV-Hybrid and
47--57.
Mini-Grid Conference. Aix en Provence, France, 11–12 May.
Kondoh M, Yokoyama N, Inazumi C, et al. (2000) Development of solid
Stucki S, Scherer GG, Schlagowski S, and Fischer E (1998) PEM water
polymer-electrolyte water electrolyser. Journal of New Materials for
electrolysers: Evidence for membrane failure in 100 kW
Electrochemical Systems 3: 61--65.
demonstration plants. Journal of Applied Electrochemistry 28:
LaConti AB and Swette L (2003) Special application using PEM-
1041--1049.
technology. In: Vielstich W, Gasteiger HA, and Lamm A (eds.)
Wendt H and Plzak V (1983) Electrocatalytic and thermal activation of
Handbook of Fuel Cells – Fundamentals, Technology and
anodic oxygen- and cathodic hydrogen-evolution in alkaline water
Applications, vol. 4 ch. 55, pp. 745--761. Chichester: John Wiley &
electrolysis. Electrochimica Acta 28: 27--34.
Sons, Ltd. ISBN 0-471-49926-9.
Winter CJ and Nitsch J (1986) Wasserstoff als Energieträger – Technik,
Lax E and Synowietz C (1967) D’Ans-Lax Taschenbuch für Chemiker
Systeme, Wirtschaft. Berlin: Springer-Verlag. ISBN 3-540-15865-0.
und Physiker, 3rd edn., vol. 1. Berlin: Springer-Verlag.
LeRoy RL (1983) Industrial water electrolysis: Present and future. Wittstadt U (2008) Electrolysis: Hydrogen production using electricity.
In: Züttel A, Borgschulte A, and Schlapbach L (eds.) Hydrogen as a
International Journal of Hydrogen Energy 8: 401--417.
Future Energy Carrier, pp. 155--161. Weinheim: Wiley-VCH Verlag.
Maget HJR (1970) Process for gas purification (13 January). US Patent
ISBN 978-527-30817-0.
3,489,670.
Wittstadt U, Wagner E, and Jungmann T (2005) Membrane electrode
Makkus R, Grootjes S, Brey J, and Rodriguez C (2005) Development of
assemblies for unitised regenerative polymer electrolyte fuel cells.
a new reversible fuel cell stack: Revcell. In: Proceedings of the 2nd
Journal of Power Sources 145: 555--562.
EHEC. Zaragoza, Spain, 22–25 November.