Robert Phillips PHD Submission

Z ERO GAP CELL DESIGN FOR
ALKALINE ELECTROLYSIS
A P H D T HESIS
submitted in fulfillment of the requirements
for the award of the degree of
D OCTOR OF P HILOSOPHY
S UBMITTED BY
Robert Phillips
S UPERVISED BY
Charles W. Dunnill
E NERGY S AFETY R ESEARCH I NSTITUTE

S WANSEA U NIVERSITY
F EBRUARY, 2019
Abstract
As the world moves towards an energy landscape based around renewable electricity
sources, the requirement to store this electricity becomes key. Hydrogen, produced by
the electrolysis of water, has a pivotal role to play in this new era, with the potential
to provide a cheap, efficient and scalable energy storage solution. This will allow
intermittent renewable energy sources to provide dispatchable and reliable low carbon
energy.
Electrolysis in an alkaline environment permits the use of cheap and abundant
materials, but currently suffers from low power densities and poor efficiencies. This
research focuses on increasing the current density of the technology using advanced
cell design. Though the use of cutting edge membrane technology, high performing
electrocatalyst coatings, and zero gap cell design, an alkaline electrolysis cell working
at 1 A·cm−2 below 2.2 V has been developed.
Using a zero gap cell arrangement is shown to significantly reduce the Ohmic
resistance of the setup, as well as decreasing the effect of gas evolution on the per-
formance of the cell. This allows the cell to operate at higher current densities when
compared to the conventional finite gap setup. This research quantifies the benefit of
employing zero gap cell design over the traditional finite gap approach, with a 30 %
reduction in Ohmic resistance observed when compared to the previous setup.
Electrocatalysts for both the hydrogen evolution and oxygen evolution reactions
have been developed and shown to display high electrocatalytic performance, good
adhesion to the substrate and good electrochemical stability at current densities ap-
propriate for alkaline electrolysis. Simple, one step deposition methods are used, to
allow simple scaling up of the electrode size. The use of a nickel-iron oxyhydroxide
electrocatalyst at the anode is demonstrated with an overpotential of 280 mV at 10
mA·cm−2 and a Tafel slope of 37 mV·dec−1 . Raney nickel has been shown to be an
effective hydrogen evolution electrocatalyst, with a current density of 93 mV at 10
mA·cm−2 and Tafel slope of 62 mV·dec−1 .
The evolution of two-phase flow inside the zero gap electrolyser is investigated,
with two models of the behaviour being validated against visual data. Using high
speed photography, the flow regimes have been visualised using bespoke flowfield
designs. This information gives an insight into the two phase flow regime inside the
cell at current densities up to 2 A·cm−2 . A Matlab program is developed to quantify
the flow regime transition between the bubbly and slug flow regimes.
Finally, the employment of an anion exchange membrane permits a significant
reduction in the Ohmic resistance of the test cell. A zero gap cell running at 2 A·cm−2
below 2.5 V is developed using a Sustainion membrane. This membrane was charac-
terised in terms of it gas separation and electrochemical performance and compared
to the current commercial membrane choice for alkaline electrolysis.
Declaration
I hereby declare that the work contained in the project is original. The work has not
previously been accepted in substance for any degree and is not being concurrently
submitted in candidature for any degree. This thesis is the result of my own investiga-
tions, except where otherwise stated and other sources are acknowledged by footnotes
giving explicit references and a bibliography is appended. I hereby give consent for
the thesis, if accepted, to be made available online in the Universitys Open Access
Repository and for inter-library loan, and for the title and summary to be made avail-
able to outside organisations.
Robert Phillips
Date: February, 2019

Certificate
This to certify that the project entitled, “Zero gap cell design for alkaline electroly-
sis” has been carried out by Robert Phillips under my supervision. He has fullfilled
the mandatory requirements for the submission of this project work. In my opinion
this work is fully adequate, in scope and quality, as a dissertation for the degree of
Doctor of Philosophy. To best of my knowledge, the results contained in this project
has not been submitted in part or full to any other University for the award of any
degree.
(Charles W. Dunnill)
Signature Supervisor
Date: February, 2019

Acknowledgements
I would like to thank everyone who made the PhD experience such a good one and
in particular Dr Charlie Dunnill for his enthusiasm, guidance and support across the
three years.
Thanks to the full ESRI crew for making it such a fun lab to be involved with and
creating a positive research environment. In particular the Dunnill group for the good
times and also for the research advice - Bertie, Mike, Dan, Bill, Nassoss and Adam for
some good times on the water, in the water and on wind street. Jon, Ryan, Kat, Max,
Lisa, Suzanne and Gregg for keeping it fresh in the office. Francesco, Lorn, Amir,
Marcin and the rest of the battery guys for good electrochemical chat.
Thanks to the high speed camera expert, Deren, for his help getting some great
images and giving up his time to help a brother out with Matlab coding, and Bjørnar
for letting me use his lab.
I am very grateful to those who helped pass along the mysteries of electrochem-
istry - particularly the guru Sunihyk for passing on his extensive knowledge and en-
thusiasm for the subject. Thanks also to Enrico, Jennifer and Bill who all came up
with key knowledge when it was needed.
I appreciate the solid advice from Ian Evans and Mark Watkins in the workshop,
and for always making the highest quality components for my setup, seemingly with
ease.
To any outsiders this could look like a very dull topic, so thanks to my dad for
reading the thesis and appearing to be enthused by it. Thanks to my mum for the
support outside of the PhD life, which was perhaps needed more than anticipated and
very much appreciated. Thanks to Jane and Lucy for the encouragement, grounding
and distraction. And to Sarah for always having adventures planned to keep me sane.
Contents
Certificate iv
1 Introduction 3
1.1 Energy source transition . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Fossil fuel usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Demand for renewable energy . . . . . . . . . . . . . . . . . . . . . 5
1.4 Decarbonisation of the energy sector . . . . . . . . . . . . . . . . . . 6
1.5 Energy storage and distribution . . . . . . . . . . . . . . . . . . . . . 7
1.6 Hydrogen production . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.7 Project aims . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2 Theory 12
2.1 Cell potential of water electrolysis cell . . . . . . . . . . . . . . . . . 12
2.2 Alkaline electrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 PEM electrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4 Thermodynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5 Actual cell voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.6 Cell efficiencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.7 Electrolyte . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.8 Kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.9 Hydrogen evolution reaction . . . . . . . . . . . . . . . . . . . . . . 25
2.10 Oxygen evolution reaction . . . . . . . . . . . . . . . . . . . . . . . 26
2.11 Cell configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3 Literature review 30
3.1 Electrocatalyst materials . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2 Gas separators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3 Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.4 Cell design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.5 Full cell performance . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.6 Transport losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.7 Current density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.8 Technology gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4 Experimental techniques 52
4.1 Electrochemical impedance spectroscopy . . . . . . . . . . . . . . . 52
4.2 Linear sweep voltammetry . . . . . . . . . . . . . . . . . . . . . . . 58
4.3 Tafel slope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
C ONTENTS vii
4.4 Four-terminal sensing . . . . . . . . . . . . . . . . . . . . . . . . . . 61

4.5 High speed imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.6 SEM imaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.7 Compositional analysis . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.8 Gas chromotography . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5 Minimising the Ohmic resistance of an alkaline electrolysis cell 63

5.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.3 Methods and materials . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . 72
5.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6 Development of catalysts 85
6.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.3 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.5 Overall cell model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
6.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
7 Characterisation of two phase flow 106

7.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
7.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
7.3 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
7.4 Mathematical models . . . . . . . . . . . . . . . . . . . . . . . . . . 116
7.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
8 Zero gap electrolysis with an anion exchange membrane 135

8.1 Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
8.2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
8.3 Theory of gas mixing . . . . . . . . . . . . . . . . . . . . . . . . . . 140
8.4 Experimental methods . . . . . . . . . . . . . . . . . . . . . . . . . 145
8.6 Cell Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
8.7 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
8.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
9 Conclusions and outlook 163

9.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
9.2 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
A List of Publications 167

List of Figures
1.1 2015 energy consumption of Scotland . . . . . . . . . . . . . . . . . 7

1.2 Energy storage technologies . . . . . . . . . . . . . . . . . . . . . . 8
1.3 Electrolysis technology comparison . . . . . . . . . . . . . . . . . . 10
2.1 Alkaline electrolysis cell . . . . . . . . . . . . . . . . . . . . . . . . 14

2.2 PEM cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Actual cell voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Cell configurations: monopolar and bipolar . . . . . . . . . . . . . . 27
3.1 Electrical circuit analogy . . . . . . . . . . . . . . . . . . . . . . . . 30

3.2 Zero gap cell design . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3 Catalyst coated substrate cell layout . . . . . . . . . . . . . . . . . . 44
3.4 Catalyst coated membrane cell layout . . . . . . . . . . . . . . . . . 47
3.5 Effect of cell compression . . . . . . . . . . . . . . . . . . . . . . . 49
4.1 EIS signal and response . . . . . . . . . . . . . . . . . . . . . . . . . 52

4.2 Nyquist plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.3 EIS signals, resistor and capacitor . . . . . . . . . . . . . . . . . . . 55
4.4 Impedance,resistor and capacitor in series . . . . . . . . . . . . . . . 56
4.5 Impedance,resistor and capacitor in parallel . . . . . . . . . . . . . . 56
4.6 Impedance, equivalent circuit . . . . . . . . . . . . . . . . . . . . . . 57
4.7 Impedance, Randles circuit . . . . . . . . . . . . . . . . . . . . . . . 58
4.8 Linear Sweep Voltammetry . . . . . . . . . . . . . . . . . . . . . . . 59
4.9 Tafel slope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.10 Four electrode sensing . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.1 Finite gap vs zero gap cell design . . . . . . . . . . . . . . . . . . . . 64

5.2 Zero gap cell components . . . . . . . . . . . . . . . . . . . . . . . . 70
5.3 Finite gap cell components . . . . . . . . . . . . . . . . . . . . . . . 70
5.4 Test rig . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.5 Effect of electrode gap on voltage . . . . . . . . . . . . . . . . . . . 72
5.6 Effect of current density on resistance . . . . . . . . . . . . . . . . . 73
5.7 Effect of flow rate on resistance . . . . . . . . . . . . . . . . . . . . . 75
5.8 Ohmic resistance, finite gap vs zero gap . . . . . . . . . . . . . . . . 76
5.9 Effect of electrolyte and temerature on performance . . . . . . . . . . 78
5.10 SEM, porous electrodes . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.11 Interfacial contact resistance measurement method . . . . . . . . . . 81
5.12 Interfacial contact resistance . . . . . . . . . . . . . . . . . . . . . . 82
L IST OF F IGURES ix
6.1 SEM, Ni-Fe electodes (x300) . . . . . . . . . . . . . . . . . . . . . . 92

6.2 SEM, Ni-Fe Electodes (x1800) . . . . . . . . . . . . . . . . . . . . . 93
6.3 Polarisation data, Ni-Fe electrodes . . . . . . . . . . . . . . . . . . . 94
6.4 Tafel plot, Ni-Fe electodes . . . . . . . . . . . . . . . . . . . . . . . 95
6.5 Chronopotentiometry, Ni-Fe electrodes . . . . . . . . . . . . . . . . . 96
6.6 SEM, Raney nickel electrodes . . . . . . . . . . . . . . . . . . . . . 97
6.7 Polarisation data, Raney nickel electrodes . . . . . . . . . . . . . . . 98
6.8 Tafel plot, Raney nickel electrode . . . . . . . . . . . . . . . . . . . 99
6.9 Chronopotentiometry, Raney nickel electrodes . . . . . . . . . . . . . 100
6.10 Overall cell model (I-V) . . . . . . . . . . . . . . . . . . . . . . . . 103
6.11 Cell contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
7.1 Flow regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

7.2 Viusalisation cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
7.3 Flowfield design, serpentine and parallel . . . . . . . . . . . . . . . . 113
7.4 Viusalisation cell setup . . . . . . . . . . . . . . . . . . . . . . . . . 114
7.5 Image processing protocol . . . . . . . . . . . . . . . . . . . . . . . 115
7.6 Serpentine visual data . . . . . . . . . . . . . . . . . . . . . . . . . . 120
7.7 Void fration model, three dimensional plot . . . . . . . . . . . . . . . 121
7.8 Flow regime transitions . . . . . . . . . . . . . . . . . . . . . . . . . 123
7.9 Superficial velocities model, theoretical . . . . . . . . . . . . . . . . 124
7.10 Superficial velocities model, observed . . . . . . . . . . . . . . . . . 125
7.11 Superficial velocities model, individual channel observations . . . . . 126
7.12 Image processing data, slug flow coverage in channels . . . . . . . . 127
7.13 Image processing data, time dependent slug flow coverage . . . . . . 128
7.14 Serpentine visual data . . . . . . . . . . . . . . . . . . . . . . . . . . 130
7.15 Parallel visual data . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
7.16 Effect of flow rate on resistance, zero gap cell . . . . . . . . . . . . . 132
8.1 Gas analysis setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

8.2 Gas crossover data, Zirfon and Sustanion . . . . . . . . . . . . . . . 148
8.3 Rate of gas crossover rate, Zirfon and Sustanion . . . . . . . . . . . . 149
8.4 Gas crossover data, Sustanion, oxygen in hydrogen stream . . . . . . 151
8.5 Rate of gas crossover data, Sustanion, oxygen in hydrogen stream . . 152
8.6 Gas crossover comparison with theory . . . . . . . . . . . . . . . . . 153
8.7 Turndown ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
8.8 I-V performance comparison . . . . . . . . . . . . . . . . . . . . . . 155
8.9 Resistance performance comparison . . . . . . . . . . . . . . . . . . 157
8.10 Membrane layers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
8.11 Model, Zirfon gas crossover under diffusion conditions . . . . . . . . 160
List of Tables
2.1 Cell configurations: advantages and disadvantages . . . . . . . . . . . 29
3.1 Electrocatalyst performance for HER . . . . . . . . . . . . . . . . . . 34

3.2 Electrocatalyst performance for OER . . . . . . . . . . . . . . . . . . 36
3.3 Technology gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.1 Composition and performance of Ni-Fe electrode . . . . . . . . . . . 93

6.2 EDX results for Raney nickel coatings, the incorporation of chromium
into the surface layer is clear for the coating using a stainless steel
counter electrode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6.3 Performance of Raney nickel electrode . . . . . . . . . . . . . . . . . 99
6.4 Model variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
7.1 Flow regime model properties . . . . . . . . . . . . . . . . . . . . . 116
8.1 Diffusivity of Zirfon . . . . . . . . . . . . . . . . . . . . . . . . . . 143

8.2 Key operating conditions for the gas chromatograph . . . . . . . . . . 145
8.3 Zirfon resistance and gas crossover . . . . . . . . . . . . . . . . . . . 158
Acronyms
AEM Anion Exchange Membrane
CCM Catalyst Coated Membrane
CCS Catalyst Coated Substrate
EDX Energy-dispersive X-ray Spectroscopy
EIS Electrochemical Impedance Spectroscopy
HER Hydrogen Evolution Reaction
LSV Linear Sweep Voltammetry
MEA Membrane Electrode Assembly
OER Oxygen Evolution Reaction
PEM Proton Exchange Membrane
SEM Scanning Electron Microscopy

Chapter 1
Introduction
As the world moves towards an energy landscape that is dominated by renewable

sources, time synchronisation between generation and demand is not always possible,
and so the need to store energy becomes clear. This research investigates a potentially
cheap and efficient method to store large quantities of such energy; the electrolysis of
water in alkaline conditions.
Decreasing availability of fossil fuels, the competitive price of renewable energy

versus traditional sources and the legislative commitments from world governments
to reduce carbon dioxide conditions all combine to drive the increasing demand for
renewable energy and its storage, giving a strong motivation for this research from
both a commercial and societal point of view.
1.1 Energy source transition

During the last 200 years, global energy consumption has undergone a transition from
primarily biofuels (wood) to an energy supply dominated by fossil fuels (coal, oil and
natural gas).1 Key changes in the historical consumption of energy occurred during the
industrial revolution of the late 1800’s, where coal was widely used to power steam
engines, heat homes and generate electricity.2, 3 The mid 1900’s saw the rise of oil
used particularly in transport, and global energy use in 1970 was 38 Gbboe per year
(barrels of oil equivalent).4
Global energy demand has more than doubled between 1970 and 2010 to 86
1.2. F OSSIL FUEL USAGE 4
Gbboe per year. This has been driven by an increasing population, as well as the
number of people living in developed nations, with access to domestic power and
personal transport. Worldwide energy consumption is predicted to rise by another
48% by 2040, driven by the rapid development of highly populated countries currently
undergoing fast economic growth such as China and India.5, 6
Since the mid 2000’s, a new shift in the energy sources used by the world has
begun, and over the next 50 years, a remarkable and rapid shift to low carbon tech-
nologies will occur. The exact details of this shift yet to be finalised, allowing current
innovation and new technologies to take a place the forefront of this energy transition.
1.2 Fossil fuel usage
The widespread use of fossil fuels has been crucial in the rapid development of the
world, providing a cheap and plentiful energy resource that has allowed widespread
use of personal transport, heating and electricity. As the use of fossil fuels increased
globally, the issues caused by their combustion became more prominent, and it is
accepted that their use causes dangerous risks to both the health of humans, and the
planet. Notwithstanding the finite nature of the resources, the pollution of air with
toxic particulate, and the emissions of greenhouse gases, in particular carbon dioxide,
leave fossil fuels in an untenable position as the most widely used energy resource.
The particulate emissions from vehicles, power stations and gas and wood burn-
ing stoves causes harmful air pollution, with particular issues associated with nitrogen
dioxide. This pollution is found to be particularly bad within densely populated areas,
and has been linked to over 40,000 deaths per year in the .7, 8
The combustion of fossil fuels releases carbon previously stored as geological de-
posits, and this has caused an increase in the level of carbon dioxide in the atmosphere
from 280 ppm in 1850 to 404 ppm in 2017.9 Carbon dioxide acts as a greenhouse
gas, trapping solar radiation that would have otherwise been reflected back out of the
1.3. D EMAND FOR RENEWABLE ENERGY 5
atmosphere, causing an increase in the temperature of the earth and its lower atmo-
sphere10 . Even moderate rises in the temperature has caused increasing frequency and
magnitude of extreme weather events around the world. The melting of polar sea ice
has been well documented over the last 40 years, in turn causing sea levels to rise,
leaving low lying areas vulnerable to flooding. The temperature increase has been
0.85◦ C from 1880 to 2012, and the years 2014, 2015, 2016 and 2017 all feature in the
top 4 warmest years on record.11
Action to phase out the use of fossil fuels has been both technologically and po-
litically led. The recent Paris agreement aims to limit the overall temperature increase
to well below 2◦ C, with a focus on minimising the effects of the temperature rise on
the earth.
1.3 Demand for renewable energy

The past 15 years has seen a paradigm shift in attitudes towards renewable electric-
ity generation, most notably wind and solar PV. In the UK, during 2018, 33.4% of
electricity was generated from renewable sources, with 17.4% coming from wind,
3.9% from solar, and 10.7% from bioenergy. The unprecedented roll out of renewable
generation has driven down costs, such that Lazards 2017 levelised cost of electricity
comparison showed that onshore wind was the cheapest form of electricity generation
at $ 40 per MWh, and solar PV was comparable to natural gas at $51 and $48 per
MWh respectively.12 Offshore wind technology shows promise to realise further cost
savings in the near future; recent price auctions in the UK have shown that the cost of
the technology has dropped over 50% in the past 5 years.13
Renewable energy production allows for potential energy independence for a

large number of countries around the world, rather than the current dependence on
states with large oil and natural gas resources. The geo-political tensions associated
with this energy resources make it an attractive prospect, as well as the chance to keep
money spent on energy inside the local economies.
1.4. D ECARBONISATION OF THE ENERGY SECTOR 6
Despite 2018 being the UK’s greenest year for electricity generation, natural gas
still dominated as the principal source of electricity generation, due to its cheap and
flexible supply.14 This ability to provide flexible load has helped the electricity grid
to accommodate the high penetration of renewables, but there have been points when
high rates of production from renewable sources, coupled with low demand, have led
to an oversupply of electricity to the grid, leading to wholesale electricity prices be-
coming low and even becoming negative for short periods of time. As this behaviour
will become more frequent as the capacity of renewable generation on the grid in-
creases, the role of energy storage technologies to provide a useful and financially
profitable service becomes more important.
1.4 Decarbonisation of the energy sector

Global energy consumption can be subdivided into 3 main categories: electricity gen-
eration, transport and heating. Figure 1.1 shows the energy consumption for Scotland
during 2015, with both electricity and transport each contributing around a quarter of
the total energy demand, with heat responsible for the rest, making up just over 50%
of total energy demand.15
Great strides have been made in identifying and implementing a pathway towards
the decarbonisation of the electricity sector based around renewable electricity gener-
ation, and it appears that sales of electric vehicles are beginning to gain momentum
towards the initial stages of the decarbonisation of the transport sector. The ’electrifi-
cation of everything’ is a strategy often discussed, with the thought that each of these
three energy categories will slowly be electrified. There are notable applications which
appear difficult to electrify, such as long distance travel on ships and planes. Energy
dense solutions such as hydrogen can offer unique benefits in these situations. The
requirements of heat supply, where a large amounts of energy need to be supplied for
short timescales, suggest that it could be another area where hydrogen could provide
a meaningful solution. Heat is a decarbonisation problem where a concrete solution is
1.5. E NERGY STORAGE AND DISTRIBUTION 7
Figure 1.1: The breakdown of final energy consumption in Scotland for 2015
shows that electricity makes up only a quarter of total energy consumption.15
yet to emerge, with almost all heat currently coming from the combustion of natural
gas. This is a sector where hydrogen has the potential to have a great effect, with its
possible use as a direct replacement for natural gas, including use as a cooking fuel.16
Focus on this sector of energy consumption will be key in the near future, with inter-
esting technologies including heat pumps, district heating and hydrogen technologies
all looking like promising options.
1.5 Energy storage and distribution

The mismatch of real time renewable electricity generation and its consumption has
created an industry for energy storage technologies to bridge this gap. The time shift-
ing of this energy from periods of high supply to high demand can be broken down
in different time frames, short term load shifting is needed to meet the daily morning
and evening peaks of demand. Solar PV provides a good example, where the peak
electricity generation is in the middle of the day, storage can time shift this energy
to be used during the time of peak demand in the early evening. Long term energy
storage will be key to take advantages of days and weeks when there are abundant
wind and solar supplies, and store it for periods of time where there is low supply.17
A large number of energy storage technologies exist covering a wide range of both
1.5. E NERGY STORAGE AND DISTRIBUTION 8
energy capacities and storage duration’s.

Pumped hydro storage is the most mature energy storage technology and already
provides a reliable and cheap method, but the geographical limitations make it unsuit-
able to be considered to provide the magnitude of energy storage that will be required
over the next decades .18, 19
Lithium-ion battery prices are being driven down due to the economies of scale
associated with the rapid increase in sales of electric cars.202122 This has enabled
the technology to provide affordable rapid-response, short duration energy storage,
as demonstrated at the Hornsdale windfarm in South Australia, opened in 2017, it
provides 100 MW/129 MWh storage and provides a grid stabilisation service to reduce
power outages.
Figure 1.2: An outline of current energy storage technologies at there rate power
rating and estimated storage duration.23, 24
Hydrogen provides a unique option among the energy storage technologies, with
the potential to store GWh worth of energy for weeks and even months. Current
investigations into the use of underground salt caverns for large scale hydrogen storage
opens up the possibility for seasonal energy storage.25, 26 Huge quantities of hydrogen
could be produced, with its potential applications spread across the electricity, heat and
1.6. H YDROGEN PRODUCTION 9
transport applications.27
The ability to soak up large quantities of renewable generations, and be able to

supply gigawatts of energy for a period of days or weeks provides an opportunity for
which hydrogen appears well suited.28 Figure 1.2 visualises current technologies and
their storage duration and power rating. It can be seen that while a number of technolo-
gies can fulfill the low power, small duration requirements, high power storage that
can last days or weeks is unfulfilled. Pure hydrogen or synthetic natural gas (based on
hydrogen from electrolysis) have the potential to fulfill these requirements.29
1.6 Hydrogen production

Suggestions of a hydrogen economy have been around since the oil crises of the 1970s,
and the recent implementation of hydrogen refuelling stations in Europe and the USA,
as well as a power-to-gas energy storage plant in Germany, show initial steps in that
direction30 . The idea of hydrogen used a universal energy vector is an attractive idea,
with the potential to provide electricity storage, transport fuel and heating fuel with
one material.
This holistic approach appears limited, however, and it is likely that hydrogen
will provide more specific roles in the decarbonised society alongside a large number
of other energy storage technologies. Hydrogen’s use as a heating fuel is particularly
attractive, with trials already underway for the Leeds hydrogen heating system. Other
applications such as long distance travel, including trains, trucks and shipping are all
developing promising fuel cell based propulsion systems.31
The electrolytic splitting of water has been known for centuries, yet hydrogen
produced in this way only currently contributes ∼4% of worldwide hydrogen produc-
tion due to more economical methods such as steam reformation of fossil fuels. The
origin of the rest of the hydrogen by percentage is natural gas at 48%, oil with 30%
and coal 18%.32
There are three principal technologies associated with the electrolytic splitting of
1.6. H YDROGEN PRODUCTION 10
water: alkaline, acidic and solid oxide. Alkaline electrolysis uses an aqueous alkaline
electrolyte, Proton Exchange Membrane (PEM) electrolysis uses a acidic solid poly-
mer electrolyte, and solid oxide uses high temperatures and a ceramic electrolyte.33–36
Each technology has its associated benefits and drawbacks, with alkaline elec-
trolysis being a mature technology, with cheap materials and excellent durability, al-
though it is associated with the lowest efficiencies. PEM electrolysis operates at high
efficiencies and power densities, but the acidic environment currently requires the use
of expensive noble metal catalysts. Solid oxide electrolysis has the potential for the
highest efficiencies, but the high temperatures needed accelerate the degradation of
the materials, and durability is a major issue.33 Figure 1.3 shows a comparison of
the three technologies in terms of the cell voltage and current density, related to the
efficiencies and costs associated with the systems. The higher the voltage of the sys-
tem, the lower the efficiency, and the higher the current density, the lower the capital
cost of the electrolyser. Research is focused on moving the electrolyser performances
towards high current densities at lower voltages.
Figure 1.3: Alkaline, PEM and solid oxide electrolysis technologies comparison.37
1.7. P ROJECT AIMS 11
1.7 Project aims

The objective of this project is to improve the performance of the alkaline electrolysis
system towards the level currently associated with the PEM technology. Creating a
zero gap cell that runs at current densities up to 2 A·cm−2 is the focus of the research,
and in terms of cell efficiency, a cell operating with 1 A·cm−2 at less than 2 V is
suitable target, as current commercial alkaline systems are rarely operated above 0.5
A·cm−2 at the same voltage. Investigating the effects of running at current densities
up to 2 A·cm−2 has attracted little research in the alkaline field, and this thesis aims
to fill in gaps in the research associated to this. More specifically, the project has five
main targets:
1. Develop and build an alkaline electrolysis cell with zero gap cell design, iden-
tifying and optimising materials associated with its construction. Establish a
reliable cell test station, with electrolyte and gas management systems to allow
comprehensive electrochemicial characterisation of the cell.
2. Characterise each contribution to the cell voltage using electrochemical tech-

niques, and develop a mathematical model of the electrochemical performance
of the cell which extends to high current densities.
3. Identify and establish methods for the deposition of high performing materials
for both the cathodic and anodic catalysts, utilising simple and scalable deposi-
tion techniques. Characterise the materials using electrochemical and imaging
techniques.
4. Investigate the progression and effects of high rates of gas evolution associated
with the cell running at up to 2 A·cm−2 . Utilise visual, electrochemical and
mathematical methods to characterise the two phase flow in the flow channels.
5. Evaluate the potential use of anion exchange membranes in a zero gap alkaline
cell. Perform electrochemical and gas composition analysis to compare state of
the art membranes with currently available gas separators.
Chapter 2
Theory
2.1 Cell potential of water electrolysis cell

The basic process of electrolysis is the splitting of water into its two elemental com-
ponents of hydrogen and oxygen according to the following:38
2H2 O(l)− > 2H2 (g) + O2 (g) (2.1)
This is achieved by connecting a metallic anode and cathode to a DC power

source and immersing them in a conductive electrolyte. In a neutral or acidic elec-
trolyte, the processes occurring at the electrodes, can be described as follows:
At the cathode:
2H + + 2e− − > H2 (E 0 = 0.00V vsSHE) (2.2)
At the anode:
1
H2 O− > H2 O + 2H + + 2e− (E 0 = 1.23V vsSHE) (2.3)
2
For alkaline electrolysis, a strong base is used as the electrolyte to avoid the corro-
sion problems associated with the electrodes in an acidic media; potassium hydroxide
(KOH) or sodium hydroxide (NaOH) are usually the electrolytes of choice.39 The use
of an alkaline medium allows the use of cheap materials for the electrodes, compared
2.2. A LKALINE ELECTROLYSIS 13
to those necessary to withstand electrolysis in an acidic system. The electrochemical

potentials shift by -59 mV·pH−1 , following the Nernst equation, such that in strongly
alkaline media (pH = 14), the half-cell reactions are as follows:40
At the cathode:
2H2 O + +2e− − > H2 + 2OH − (E 0 = −0.83V vsSHE) (2.4)
At the anode:
1
2OH − − > H2 O + O2 + 2e− (E 0 = 0.4V vsSHE) (2.5)
2
The total reversible cell voltage is calculated using the cell potentials of the two half-
reactions. Following convention such the half reactions on the left and right are con-
sidered to be an oxidation and reduction respectively:
ERev = ERight − ELe f t = ECathode − EAnode (2.6)
It can therefore be seen that for both conditions, ERev = -1.23 V.
2.2 Alkaline electrolysis

Commercial alkaline electrolysers have been operational since the early 1900’s, with
the first large industrial system built in 1927 by a Norwegian company, Norsk Hydro
Electrolyzers, where the hydrogen produced was used as a precursor for ammonia
production. Early alkaline systems were built with the focus on a robust and reliable
system that could operate with minimal maintenance, with lifetimes of more than 10
years widely reported. The early systems were implemented in locations with cheap
electricity sources close by (principally hydro-electric plants) to keep operational costs
low. Due to this, the historical focus of the developemnt of alkaline electrolysers has
never been on system efficiency. With modern energy systems placing value on energy
storage, the technology development so far acts as a firm foundation for the innovative
2.2. A LKALINE ELECTROLYSIS 14
adjustments needed to meet the new requirements for the technology. This provides
the main incentive for this research project as, although the technology is mature, an
efficiency focus has never been a priority, whereas in this case, efficiency is placed as
the key priority.
The widespread take up of alkaline electrolysis has been limited due to the low
cost of hydrogen obtained by steam reformation of natural gas, making cost com-
petitive hydrogen from electrolysis difficult. With the new requirements for energy
storage increasing daily, the widespread commercialisation of this technology now
has a promising future.
Figure 2.1: An alkaline electrolyser requires two metallic plates separated by a

liquid electrolyte.41
As shown in Figure 2.1, alkaline electrolysers use two metallic plates separated
by liquid electrolyte and a diaphragm to minimise gas mixing. The gap between the
plate and the diaphragm is up to 42mm in conventional systems.39 A strong liquid
base is used as the electrolyte and generally operated at 50 - 80◦ C. The performance
of alkaline cells has been limited to <250 mA·cm−2 at 1.8 V due to the focus on
operational robustness.42 However, work in the past 5 years at the laboratory scale
2.3. PEM ELECTROLYSIS 15
has seen performances of 1 A·cm−2 at 2.01 V reported, with novel membranes and
high performing catalyst coatings.23, 43
2.3 PEM electrolysis

The development of proton exchange membrane (PEM) electrolysers was accel-
erated in the 1960’s, when the synthesis and commercialisation of a sulphonated
tetrafluoroethylene-based fluoropolymer, under the brand name of Nafion, was made
to replace the liquid alkaline electrolyte of the traditional alkaline electrolysers. The
proton exchange membrane showed excellent proton conductivity combined with a
low gas crossover rate, all achieved with a thin (50 -250 µm thick) yet chemically and
mechanically stable membrane.44, 45 Figure 2.2 shows the makeup of a PEM elec-
trolyer, where a catalyst layer is deposited onto the thin acidic polymer membrane,
with a metallic gas diffusion layer providing an electrical connection between the cat-
alyst layer and bipolar plate.
PEM Electrolysers have the advantage of high current densities, 4 A· cm−2 has
been demonstrated at laboratory scale with low Ohmic losses, high gas purity, and
a compact system design. PEM electrolysers exhibit high voltage efficiency, even
at high current densities, Marshall et al. reported a cell voltage of 1.57 V at 1 A·
cm−2 at 80◦ C and atmospheric pressure, and Yanaguch et al. showed 1.53 V for 1 A·
cm−2 , and 1.67 V at 3 A· cm−2 respectively for and the same operating conditions as
Marshall.46, 47 This high performance is due to low kinetic overpotentials in the acidic
environment, and low Ohmic resistance of the Nafion membranes.
The high performance of the system comes with a cost, however. Due to the cor-
rosive acidic environment caused by the PEM membrane, platinum group metals are
necessary due to their chemical stability. Iridium and platinum are used as catalysts
for the electrodes and titanium is used for the separator plates necessary to achieve
the zero gap arrangement. Iridium is 40 times rarer than gold, and it is thought that
large scale commercialisation of PEM technology would have a significant effect on
2.3. PEM ELECTROLYSIS 16
Figure 2.2: A PEM electrolyser has a catalyst layer deposited directly onto the
membrane, and a porous metallic gas/liquid diffusion layer provides an electrical
connection between the catalyst layer and the bipolar plate.
its demand, and hence drive up the price, as has been seen with the recent price in-
creases of lithium and cobalt associated with the acceleration of lithium-ion battery
production.48
Rapid progress has been made in recent years, with the main focus on minimising
precious metal loading and optimising the efficiency of the systems to reduce the
operational cost. Comprehensive data regarding the durability and stability of the
PEM stacks has revealed good durability, with commercial systems reporting 60,000
hours of operation with only minimal voltage decay.48–50
The high performance and compact stack design make PEM electrolysers an at-
tractive alternative to their alkaline counterpart, and PEM technology currently dom-
inates sales of new electrolyser systems. Cost reductions by reduced loading of pre-
cious metals (or appropriate use of cheaper alternative materials) and stack lifetime
extension are needed to push PEM electrolysers further into the commercial domain.
2.4. T HERMODYNAMICS 17
2.4 Thermodynamics
The process of water electrolysis involves the conversion of electrical and heat energy
to chemical energy, which is stored in the form of hydrogen. The total energy required
for the reaction described in equation 2.7 is called the enthalpy ∆H. A certain amount
of the energy, called the Gibbs free energy change ∆Gd must be supplied in the form
of electrical energy. The remainder can be provided in form of heat, which is the
product of the temperature T and entropy change ∆S . The change in enthalpy can be
expressed as:
∆H = ∆G + T ∆S (2.7)
The enthalpy change necessary to dissociate one mole of water is ∆H = 285.84 kj

and so using equation 2.7 in standard conditions, and knowing that ∆S = 163 J·K−1 , the
energy that must be provided as electrical energy is ∆Gd = +237.2 kj·mol−1 , whereas
the rest of the energy can be provided as heat from the environment.51
The electrical work done by the cell (at a constant pressure and temperature) is
known as the Gibbs free energy change:
∆Gd = −nFECell (2.8)
where F is the Faraday constant (96,485 C) and n is the number of electrons

exchanged during the electrochemical splitting of one water molecule (2 electrons).
At standard conditions, it has been calculated that ∆Gd = +237.2 kj·mol−1 . For a
spontaneous cell reaction, a negative value of ∆Gd is required, and in this case, the
positive value relates to the minimum amount of electrical energy required to split one
mole of water into its molecular components.
In this way, performing electrolysis at the minimum Gibbs free energy, the re-
action is an endothermic process, and ERev can be calculated from equation 2.8 to be
1.23 V. The process becomes thermoneutral when the change in enthalpy is supplied
2.5. ACTUAL CELL VOLTAGE 18
entirely as electrical energy. It follows that since ∆H = +285.840 kJ·mol1 , the ther-
∆H ∆H
moneutral voltage can be calculated: ET hermoneutral = nF = 2F = 1.48V . It should be
noted that the change in Gibbs free energy is a function of temperature and pressure,
and changes to these two parameters changes the free energy electrolysis voltage. For
example at 100◦ C the reversible cell voltage decreases to 1.18 V, and down to ≈ 0.9
V at 1000◦ C. This is not overly useful except in the rare case where there is a constant
supply of heat, and so it can be assumed that low temperature electrolysis will be run
at voltages above ET hermoneutral .
The value of energy required to drive reactions in the cell does not take into ac-
count the rate at which the reactions are taking place, which can be considered to be
directly proportional to the current. To achieve desirable reaction rates an overpoten-
tial is necessary to overcome the electrical resistances present in the system (eg energy
to overcome the activation energies of the reactions) and achieve these current densi-
ties,. The overpotentials due to the electrode reactions is expressed as η = ECell – E0
– IRCell , and increase with current density.
2.5 Actual cell voltage

The actual cell voltage is distinct from the reversible cell voltage due to inefficiencies
in the system. The actual cell voltage can be broken down into its contributing factors:
ECell = ERev + ηAnode + ηCathode + I · RCell + ηMT (2.9)
Where ηAnode is the overpotential at the anode, ηCathode the overpotential at the
cathode, I is the current, RCell is the resistance of the cell and ηMT is the voltage
inefficiency caused by mass transport limitations at the electrodes. Figure 2.3 shows
an approximate breakdown of the voltage contributions for a conventional alkaline
electrolysis cell. It can be seen the the voltage contribution from Ohmic losses begins
to dominate at a current density of 500 mA·cm−2 .
2.6. C ELL EFFICIENCIES 19
Figure 2.3: A rough breakdown of the contributions to the cell voltage. The mass
transport overpotential is poorly defined.
To optimise the cell voltage, each of these contributing factors must be charac-
terised and optimised individually. Strategies to provide information on each of these
factors are outlined in the rest of this thesis.
2.6 Cell efficiencies

Energy efficiency is an important figure used for the comparison of systems, and is
generally defined as the ratio of useful energy output to the total energy input. There
are, however, a number of different ways of expressing the efficiency of the electroly-
sis process.
For the overall cell, when operating in an exothermic way, the useful energy
output is the energy content of one mole of hydrogen, and the total energy input is the
electrical energy supplied by the power source to produce that one mole of hydrogen:
WH2 285.8(kJ)
εH2 = = (2.10)
Winput UIt
Where WH2 is the higher heating value of one mole of hydrogen, U is the cell
voltage, I is the current and t is the time taken to produce one mole.
2.6. C ELL EFFICIENCIES 20
Another method compares the theoretical amount of energy with the actual mea-
sured amount of energy required to split one mole of water:
Wt Et · I · t Et
η= = = (2.11)
Wa ECell · I · t ECell
Where Wt is the theoretical energy required to split one mole of water, and Wa is
the actual energy required using the electrolysis cell in question. This leads to a neat
comparison of voltage efficiency. The theoretical energy Wt can be chosen at either
the free energy electrolysis voltage EFree or the thermoneutral voltage ET hermoneutral ,
giving two separate efficiencies, the free energy efficiency and the thermoneutral effi-
ciency.
EFree 1.23V
ηFree = , atT = 25◦C, = (2.12)
ECell ECell
ET hermoneutral 1.48V
ηT hermoneutral = , atT = 25◦C, = (2.13)
ECell ECell
As practical cells will be working in the exothermic region, the most appropriate
voltage efficiency to use is ηT hermoneutral .
The free energy efficiency is a measure of cell efficiency purely from a voltage
perspective, and electrolysis between the two theoretical voltages requires that the
electrolysis is an endothermic reaction, the Gibbs free energy will be supplied by
electrical energy, but the additional entropy will be taken in as heat energy by the cell
from the surroundings to maintain its temperature. The thermal efficiency assumes
that the full enthalpy is supplied as electrical energy, and so the cell does not consume
any heat energy from its surroundings.
The efficiency can also be measured by comparing the hydrogen production rate
to the total electrical energy applied to the system over a defined time period:
VH2 VH2
ηH2 = = (2.14)
WElectrical ECell it
Where VH2 is the rate of hydrogen production per unit volume of the cell, and
2.7. E LECTROLYTE 21
WElectrical is the electrical energy applied to the system over time period t.
The natural aim of research and development is to increase the efficiency of the
electrolysis process such that less energy is needed to split one mole of water into its
component parts. Defining the efficiency is key to providing a benchmark from which
to compare improvements to the technology.
Comparisons between different electrolysis cells can be made by comparing cur-

rent densities or cell voltages at fixed temperatures and pressures. For example, take
two cells that both have current densities of 0.5 A· cm−2 with a cell voltage for
Cell 1 of 1.68 V and of 1.8 V for Cell 2. Cell 1 has a Free Energy efficiency of
1.23x100 1.48x100
1.68 = 73.2% and thermal efficiency of 1.68 = 88.1%, whereas the correspond-
1.23x100 1.48x100
ing efficiencies for Cell 2 would be 1.8 = 68.3% and 1.8 = 82.2% . This
allows comparisons between both different iterations of a cell, as well as other devel-
opments reported in the literature.
2.7 Electrolyte
Water is a notoriously poor conductor of electricity, and so an alkaline electrolyte is
added to increase the specific conductivity of the solution. The electrolytic current in
alkaline electrolysis is transported by the hydroxyl anion.The ionic conductivity of the
electrolyte depends on two main factors; temperature and concentration.
A higher ionic conductivity clearly relates directly to a lower resistance of the

electrolyte, which affects the overall performance of the electrolytic cell.
The voltage drop due to the electrolytic solution is expressed using Ohms law:
Il il
IR = = (2.15)
Aκ κ
Where I is the current in amperes, i is the current density in A· cm−2 , l is elec-

trode spacing in cm, A is the cross-sectional area in cm2 , and κ is the conductivity in
S · cm−1 . It is clear from Ohms law that to reduce voltage drop at the same current
density it is key to increase conductivity.
2.8. K INETICS 22
Historically, aqueous electrolytes have been used due to their wide availability
and ease of use inside an electrolysis cell. Sodium hydroxide (NaOH) and potassium
hydroxide (KOH) are both used in literature, and a comparison of the two candidates
was made by measuring conductivities at different concentrations and temperatures;
the peak conductivity of the KOH solution is 1.50 S · cm−1 compared to 1.08 S cm−1
for NaOH.52
Gilliam et al. reviewed and compared research into the relationship between
the concentration of a KOH electrolyte and its specific conductivity. At 25◦C the
conductivity peaks at a concentration of 6 molar KOH. A comparison of operational
alkaline electrolysers suggest that 25 -30 wt. % KOH is the preferred solution and
concentration.52
The effect of temperature of specific conductivity was also reported, showing a
near linear increase of specific conductivity with temperature. It is interesting to note
that the optimum concentration of KOH also varies with temperature; the optimum
concentration of 6 molar at 25◦C is compared to 8 molar at 80◦C.
2.8 Kinetics
To target a reduction in overpotentials due to the anodic and cathodic charge exchange,
it is important that the processes at the electrode-electrolyte junction are understood.
The basic electrode reaction can be considered as Red− > Ox + ne, and using Fara-
days 1st law of electrolysis, the number of moles altered at an electrode is directly
proportional to the charge transferred at the electrode:
Q It
N= = (2.16)
nF nF
Where Q is the charge transferred, n is the number of electrons exchanged in each

half reaction and F is the Faraday constant. The number of moles of each reactant
electrolysed per unit time is the rate of reaction, v. As reactions occurring at the
interface of the electrode and electrolyte are considered heterogeneous, the reaction
2.8. K INETICS 23
rate is dependent on the surface area of the electrode, A, for a current density, j:
1 dN I j
v= · = = (2.17)
A dt nFA nF
From a chemical kinetics approach, the rate of oxidation is expressed as va =

ka cRed and the rate of reduction as vc = kc cOx where ka and kc are the rate constants of
the reactions, and cOx and cRed the concentrations of the reacting substances.
Rate constants of electrode reactions can be expressed using the Arrhenius equa-
tion:
∆Ha
ka = Pa exp (− ) (2.18)
RT
∆Hc
kc = Pc exp (− ) (2.19)
RT
Where Pa and Pc are pre-exponential factors independent of electrode potential,

and ∆Ha and ∆Hc are the activation enthalpies of the oxidation and reduction reactions.
Bringing together the two derivations of the rate of reaction, and acknowledging
that the overall current is a summation of the anodic and cathodic currents j = ja + jc
(where the currents are vectors across the interface), we see that:
ja = nFka cRed (2.20)
jc = nFkc cOx (2.21)
and hence
j = nF(ka cRed − kc cOx ) (2.22)
When there is no net current, the rate of oxidation and reduction must be equal,
2.8. K INETICS 24
and the electrode will be at the equilibrium electrode potential E e , which can be ex-
pressed using the Nernst equation.
RT co
Ee = E0 + ln( ) (2.23)
nF cR
The relationship between the current on an electrode (I) and the applied poten-
tial (E) is best described by the Butler-Volmer equation. This particular form of the
equation is correct when the rate of reaction is determined by the transfer of electri-
cal charge at the surface of the electrode, and not by the mass transfer from the bulk
electrolyte to the electrode surface.
(1 − α)nFη αnFη
j = j0 e( − e( ) (2.24)
RT RT
Where the overpotential η = E − E0 .

RT
When η F either the first, or second, exponential term of the above equation
will become small and can be neglected. The equation then takes the form of the Tafel
equation which is written below.
RT RT
η= ln j − ln j0 (2.25)
αc F αc F
which can be re-written in the form:
j
η = B · ln (2.26)
j0
Where B is the Tafel slope.
Another common form of the equation is written as η = a + b log j, which can be

plotted to show the relationship of overpotential to current density. The parameter a is
linked with the exchange current density j0 , which refects the intrinsic rate of electron
transfer, and b, the gradient of the asymptote at high current, is known as the Tafel
slope, which is the rate of change of current density with overpotential. For cathodic
2.9. H YDROGEN EVOLUTION REACTION 25
processes:
2.3RT
a= αnF log j0 and b = − 2.3RT
αnF
−2.3RT 2.3RT
a= (1−α)nF log j0 and b = − (1−α)nF
Tafel diagrams are commonly used in electrochemistry to provide meaningful

comparisons of electrocatalysts, with the Tafel slope offering an insight into the ability
of the catalyst to scale to high current densities with a reasonable increase in overpo-
tential. The Tafel slope can be used to infer the rate determining process within the
reaction, which can then be used to guide catalyst and electrode substrate design.
2.9 Hydrogen evolution reaction

The hydrogen evolution electrode reaction has historically been the subject of much
research due to its technological and fundamental importance. The overall reaction at
the cathode was stated as:
2H2 O + 2e− − > H2 + 2OH − (2.27)
The process of hydrogen evolution is accepted to be via the Volmer-Tafel and

Volmer Heyrovsky mechanisms.53 The hydrogen is adsorbed via the Volmer reaction:
H2 O + e− − > Hads + OH − (2.28)
Which is followed by either the chemical Tafel reaction or the electrochemical

Heyrovsky desorption step:
2Hads − > H2 (2.29)
Hads + H2 O + e− − > H2 + OH − (2.30)
Both mechanisms involve the adsorption and subsequent desorption of hydrogen

2.10. OXYGEN EVOLUTION REACTION 26
atoms, and it is therefore important that the electrocatalyst surface creates M-H bonds
of an intermediate strength. Suitable electrocatalyst materials based on this criteria
include platinum and iridium rank, which rank the highest, with nickel featuring as
the highest ranking earth abundant metal.54
Both pathways are two step processes, and either the first or second step can be
rate defining. If the Volmer reaction is the rate defining step, the Tafel slope is in the
order of 120 mV·dec−1 . In this case, electrodes with a higher effective surface area
(through edges and cavities in the material surface) will create more sites for hydrogen
adsorption.
If the Tafel step is rate defining, a Tafel slope of 30 mV · dec−1 will be ob-
served, and if the Heyrovsky step is rate defining the expected Tafel slope will be 40
mV·dec−1 .55
At low overpotentials, the electrode performance tends to be limited by the hy-
drogen desorption step, with initial Tafel slopes of 30-40 mV·dec−1 . At higher over-
potentials, the Tafel slope increases to 120 mV·dec−1 due to the rate limiting hydrogen
adsorption step. Electrocatalyst development is focused on increasing the overpoten-
tial at which the transition to the higher Tafel slope occurs.
2.10 Oxygen evolution reaction

The process of oxygen evolution in alkaline media is a multi-step reaction. Although
there is no confirmed scheme, the proposed reaction pathways are notably similar.
One such reaction pathway follows the below:
OH − − > OHads + e− (2.31)
OHads + OH − − > Oads + O2 (g) + e− (2.32)
Oads + OH − − − > OOHads + e− (2.33)

2.11. C ELL CONFIGURATION 27
OOHads + OH − − > OO−

ads + H2 O (2.34)
OO−
ads − > O2 (g) + e
−
(2.35)
There are four different types of adsorbed species (O, OH, OOH and OO-) and
four of the steps involve electron transfer, where the potential of the electrode is used
to drive the reaction. In contrast to the hydrogen reaction, the many possibilities for
rate determining steps make identifying the reason for slow kinetics is more difficult,
and this is a key reason that the overpotentials associated with this reaction are higher
than with its counterpart.56
2.11 Cell configuration
Figure 2.4: The monopolar cell configuration is characterised by each electrode

being connected to the power supply, whereas the bipolar configuration just requires
the two end electrodes to be connected
Alkaline electrolysers have two principal cell configurations, monopolar and

bipolar, as visualised in Figure 2.4. The monopolar assembly operates with each elec-
trode directly connected to the DC power supply, so that the configuration gives a
number of individual cells in parallel. The voltage of the whole monopolar electrol-
yser is the same as that applied to the individual pairs. Either a reduction or oxidation
reaction will occur at each electrode.
In the bipolar arrangement, only the two end electrodes are connected to the DC
power supply. This creates a line of unit cells, with the current flowing through the
electrolyte, and polarising each individual electrode such that there is an oxidation on
one side, and a reduction on the other.
Individual cell voltage was defined in Equation 2.9 as ECell voltage. The overall
voltages for the two systems will be ECell for the monopolar design, and ECell ·(n-1)
for the bipolar assembly (n is the number of electrodes). To supply power to a system
with multiple cells, using P = I ·V where P is power, I current and V the cell voltage
ECell , the monopolar configuration will experience high electrical currents compared
to the bipolar equivalent. This provides a key disadvantage: the monopolar design
will experience high Ohmic losses which have a significant effect on the efficiency of
the systems.
Advantages and disadvantages on each system are discussed in Table 2.1, with
the main disadvantages of the bipolar system being attributed to more complex cell
design. With the maturing of the technology, and applying lessons learned from the
chloro-alkali industry that have developed similar systems, the ’filter press’ bipolar
design is an effective and well established system that can be relied upon for long
term operation with low maintenance requirements.46
Almost all current commercial electrolysers use the bipolar design, and further
developments to reduce the cell voltage whilst increasing current densities are being
investigated.
Monopolar Bipolar
Advantages
Simple design Low electrical current and high voltages
Easy to maintain Compact physical design
No parasitic currents Possible operation at high temperatures
and pressures
Disadvantages
High current densities at low voltages Complex design - possible electrolyte leak-
causing high Ohmic losses age dangerous due to caustic electrolyte
Difficult to achieve small interelectrode Difficult to locate malfunctioning cells
gaps
Table 2.1: The advantages and disadvantages of monopolar and bipolar cell de-
signs.46
Chapter 3
Literature review
To target the research effectively, the state of the art performance and current status of
research must first be reviewed. Gaps in research will be identified, and the research
targeted to ensure that it adds value to the field of alkaline electrolysis; providing
meaningful information to researchers and others involved in the development of the
technology.
Figure 3.1: The electrochemical cell can be viewed as an electrical circuit, where
the inefficiencies in the cell are broken down into a series of components resisting
the flow of current.
The electrochemical cell can be modelled as a electrical circuit as seen in Figure

3.1, to provide a breakdown of the inefficiencies of the system. Each contribution
has a negative impact and needs to be minimised, The main considerations for cell
performance are broken down here, and will be addressed individually:
1. Electrocatalytic materials
2. Gas seperator materials

3.1. E LECTROCATALYST MATERIALS 31
3. Cell design and full cell performance
4. Operating conditions
At the end of the chapter, an overview of the current gaps in the technology will
be made, and goals for this research will be set out.
3.1 Electrocatalyst materials

The choice of electrocatalytic material is key to achieving highly efficient electrolytic
cell performance. Research has been focused on developing bespoke electrocatalysts
for both the hydrogen and oxygen evolution reactions individually. The criteria for ef-
fective materials goes beyond high electrocatalytic activity; the material must exhibit
high corrosion resistance in alkaline environments, low electrical resistance, high se-
lectivity towards the desired reaction and good stability under intermittent power con-
ditions, with these requirements all fulfilled at a reasonable cost.
Electrocatalytic performance is a poorly defined characteristic, and as such, there
are a number of ways that the electrocatalytic activity of a material is reported and
compared with other materials. Polarisation curves show the overpotentials required
to achieve certain current densities, however a change in current density by a factor
of 10 can lead to the overpotential of each electrode varying by <40 mV to >2000
mV, and so polarisation curves must cover a wide range of current densities (0-100
mA·cm−2 ). The rate of change in overpotential with current density is displayed on a
Tafel plot, with the ovepotential change across a tenfold increase in current is known
as the Tafel slope, with units mV·dec−1 . Another key metric for evaluating the perfor-
mance of a catalyst is by observing its onset potentials, but as this doesn’t constitute
an exact value, the overpotential (η) at a specified current density, often 10 mA·cm−2
is used. This is the most widely reported characteristic, and allows the most effective
comparison between studies.
One of the principle advantages of alkaline electrolysis over its acidic counterpart
is the stability of non-precious metals in this environment. As a result, while platinum
group metals are suitable materials due to their associated low overpotentials. The
incorporation of these materials into commercial electrolysers is unattractive due to
the associated cost. Extensive research into earth abundant electrocatalysts has led to
the development of promising candidates for application in large scale electrolysers.
One of those materials, nickel, is a key electrode material for both reactions in alkaline
electrolysis due to its corrosion resistance in an alkaline environment and relatively
low cost. It features as both a substrate material as well as part of a co-deposit for
both gas evolving reactions.
3.1.1 Materials for the hydrogen evolution reaction

Nickels credentials make it an excellent candidate as an electrocatalyst for the hydro-
gen evolution reaction due to its stability in an strongly basic environment, although
its overpotential of ∼300-400 mV at appropriate current densities mean that adjust-
ments to the material are necessary to bring this overpotential down.57, 58 During the
1980’s electrocatalytic studies identified Raney nickel, produced by the codeposition
and subsequent leaching of nickel with aluminium or zinc as the leachable compo-
nents, as a high performing electrocatalytic material.59 It was recognised that the high
surface area of this material provides more sites for hydrogen adsorption than a flat
substrate, increasing the reaction rate and reducing the overpotential of the material.
Key research was done by Suffredini et al., who used a simple and cheap electrodepo-
sition methodology to produce Raney nickel with roughness factors of 2200 for pure
nickel, and 4400 for Raney Ni-Co alloy. Overpotentials of 100 and 90 mV respectively
were reported at 70◦C and a current density of 135 mA·cm−2 .60
Arul raj et al., compared a number of nickel alloys, obtained through electrode-
position methods onto steel strips, and ranked the alloys on performance as follows:
Ni-Mo > Ni-Zn > Ni-Co > Ni-W > Ni-Fe > Ni-CR > Ni plated steel.61 An over-
potential reduction of 300 mV was reported when compared to conventional steel
cathodes with the cathode based on the Ni-Mo alloy, which showed an overpotential
of 180 mV for over 1500 h of continuous electrolysis in industrial conditions of 6 M
KOH, at 300 mA·cm−2 and 80◦C.
The high catalytic performance of the nickel-molybdenum alloy made it a pop-

ular choice as electrode coating, and there has been much research into fabrication
methods and performance characterisation.62 Traditional preparation methods for Ni-
Mo electrodes often result in brittle electrodes, although Tang et al. report a novel
way to fabricate Ni-Mo cathodes using NiMoO4 powder as a precursor.63 This ma-
terial demonstrated a current density of 700 mA·cm−2 at an overpotential of just 150
mV, which is very high performance in comparison to other literature reports. Hu et
al. stated that composite coated Raney nickel and thermally deposited Ni-Mo coated
electrodes have proved to be very effective hydrogen electrodes, with the Ni60 Mo40
electrocatalyst found to exhibit overpotential between 60 and 80 mV at 0.5 mA · cm−2
for 10000 h in 30 wt. KOH at 70◦C , although the electrodes were deactivated during
intermittent electrolysis.64 Recent work by Schalenbach et al. offers an excellent in-
sight into the role of molybdenum in the high activity of the Ni-Mo deposit - finding
that it was the high surface area morphology of the deposit, rather than any intrinsic
activity of molybdenum which caused the high activity of the coating.65 This leads to
the conclusion that maximising the surface area of a nickel deposit is vital to achieving
a highly performing Raney nickel electrocatalyst.
Pletcher et al. compared Ni-Mo, RuO2 and Pt coated nickel and stainless steel
mesh cathodes in a zero-gap electrolyser, with both Ni-Mo and RuO2 showing good
stability during electrolysis continued over 10 days, at an operating current of 1
A·cm−2 .42 The cell containing the RuO2 cathode shows a slightly lower voltage at
1 A·cm−2 with 2.06 V in comparison to 2.14 V for Ni-Mo. As reference, an un-
coated Ni cathode had a cell voltage of 2.48 V. The long term performance of the
cells showed that platinum degraded quickly and along with its price, platinum can
quickly be dismissed. At high current densities, the RuO2 based cell showed com-
parable performances to Ni-Mo. In line with Schalenbach’s statement regarding the
effect of molybdenum, this infers that Raney nickel could show long term stability at
η at 10 mA·cm−2 Tafel slope

Material Electrolyte Reference
(mV) (mV·dec−1 )
Pt 128 124 1M NaOH 66
Ni 311 113 1M KOH 67
Ni-Mo 28 36 1M KOH 68
Ni-Al 62 107 1M KOH 69
Ni-Fe(OH)2 214 n/a 1M NaOH 66
Table 3.1: Performance of selected materials for hydrogen evolution reaction
high current densities.
Studies investigating the long term stability of Raney nickel in a controlled three
electrode environment are lacking, although recent work by Gannon et al. have sought
to address this by applying accelerated degradation testing protocols to stainless steel,
nickel and Raney nickel electrodes, finding only minor decreases in electrode perfor-
mance of the Raney nickel, even under long term intermittant power conditions. The
stainless steel electrode saw visual degradation after the 2000 cycles.
The lack of a standard characterisation criteria makes catalyst comparisons diffi-

cult between different labs. Table 3.1 summarises some key investigations, with nickel
based catalysts demonstrated to outperform platinum. The highest performing cata-
lyst, Ni-Mo, has been shown to just represent a very high surface area nickel electrode,
and as such investigations focused on high surface area nickel electrodes appear to be
the most interesting prospect for the hydrogen evolution reaction in alkaline condi-
tions, due to their low cost and high performance.66
3.1.2 Materials for the oxygen evolution reaction
The oxygen evolution reaction has a larger contribution to the cell voltage than that
of the hydrogen evolution reaction. The added complexity of the 4 electron transfer
has left this reaction less well understood. Nevertheless, this reaction has been the
subject of a wide body of research, and progress has been made in identifying suitable
materials for use in working alkaline electrolysis cells.
Miles et al. compared the effectiveness of a number of metal oxide elec-

trodes for oxygen evolution, they were rated as follows Ru∼Ir∼ Pt∼Rh∼Pd∼Ni∼
Os>Co>Fe.70 Due to nickels price and abundance, it became the obvious choice for
anode material.
Early research by Hall found that the commonly used unactivated nickel anode
has an oxygen overpotential of about 400 mV, approximately 20% of a average cell
voltage. An overpotential reduction was reported through the use of high surface
area Raney alloys, with Raney nickel about 60 mV more efficient than low surface
area nickel anodes, with a comparable performance to Raney nickel-cobalt and Raney
cobalt. Hall reported that the nickel based anodes had a high corrosion resistance,
which made them an attractive option for commercial use.57, 58
Whilst researching nickel hydroxide electrodes for use in alkaline batteries, it

was discovered that nickel was always found with iron impurities, and that these im-
purities reduced the overpotential of oxygen evolution, to the detriment of the batteries
performance71 . This inspired research into the positive performance of Ni-Fe towards
the oxygen evolution reaction, and its possible use for alkaline electrolysis. Much
research has been focused recently in this area, with studies looking at the optimal
balance of Ni and Fe in the deposits, and facile deposition techniques to allow for
larger scale depositions.72 The combination of nickel, iron and chromium in stainless
steel give it acceptable catalytic properties at a very low price, with good long term
stability.73
The use of Ni-Fe as a co-deposit became a key area for research due to its initial
promising performance, and in 2012, Li et al. compared a selection of anode materials
in more rigorous conditions, using a current of 0.5 A·cm−2 in 1M NaOH at 80 ◦C in
a zero-gap cell configuration, reporting reference overpotentials for stainless steel and
smooth Ni of 400 mV and 389 mV respectively. An overpotential of 265 mV at stated
conditions for NiFe(OH)2 concluding that this can be easily deposited onto nickel and
steel substrates, even if the substrate has a complex shape, which makes it particularly
η at 10 mA·cm−2 Tafel slope

Material Electrolyte Reference
(mV) (mV·dec−1 )
Ni 395 n/a 30% wt KOH 58
Ni-Co 420 113 0.1 M KOH 77
Ni-Co2 O4 360 n/a 1M KOH 78
Ni(OH)2 312 n/a 1M NaOH 74
Ni-Fe(OH)2 220 28 1M KOH 75
Table 3.2: Selected electrocatalysts performance for the oxygen evolution reaction
attractive from a zero gap point of view. The catalyst was shown to be completely
stable over 10 days of electrolysis at 1 A·cm−2 .74
More recent work by Lu et al. reported excellent results for electrodeposited

amorphous mesoporous Ni-Fe composite nanosheets on macroporous Ni foam, with
overpotentials of 240 mV and 270 mV at current densities of 0.5 and 1 A·cm−2 .75 It
is claimed that this is the most efficient oxygen evolution electrode reported to date.
Long term stability tests of Ni-Fe(OH)2 are still yet to appear in the literature, although
initial indications suggest good corrosion resistance.
Other notable materials include spinel oxides, with Singh et al. reporting a thin
film of Co3 O4 prepared on a Ni substrate by spray pyrolysis that had an oxygen over-
potential of 360 mV at a current of 1 A·cm−2 in 30 wt. % KOH at 70◦C.76 These
electrodes showed a Tafel slope of 51-68 mV·dec−1 at low overpotentials, and 120-
140 mV·dec−1 at high overpotentials.
Figure 3.2 shows selected perfomance of electrocatalysts for the oxygen evolu-
tion reaction, with nickel and iron based hydroxides showing excellent promise, with
low initial overpotentials well below those of pure nickel. Consistant good perfor-
mance is seen with the Ni-Fe(OH)2 co-deposit, showing a Tafel slope of less than 30
mV·dec−1 . This, coupled with the potential for facile deposition techniques on porous
surfaces, including thermal and electrodeposition techniques, make it an attractive op-
tion for further research.
3.2. G AS SEPARATORS 37
3.2 Gas separators
There are two types of separator used in alkaline electrolysers, porous gas separators
and polymer based anion exchange membranes (AEM). Both have the same primary
functions; to be ion-permeable whilst keeping the product gases (H2 and O2 ) separate.
Porous gas separators can utilise both finite gap and zero gap design and have been
used commercially for many decades, whilst the AEM membranes require a zero gap
arrangement, and have only begun to reach commercialisation within the past decade.
An appropriate gas separator must demonstrate good levels of gas separation,

including at working current densities (more than 4% oxygen in hydrogen can be
potentially explosive but gas purities above 99.5% are preferable for use with a fuel
cell), it must have low ionic resistance (certainly less than 0.5 Ω·cm2 ), be durable
in caustic alkali conditions, be relatively cheap and with a long lifetime. The main
performance indicator of concern is the ionic resistance of the material, and where
this is not reported, the performance of the full cell gives an indication of how the
membrane is impacting performance.
Rosa et al. investigated the effect on ionic transport in solution of prospec-

tive separator materials including polyphenylene sulfide (PPS), polytetra fluorethylene
(PTFE or better known as Teflon) in both woven and felter forms, polysulfone (PSF),
asbestos (as well as asbestos coated with PSF) and ion exchange membranes (in this
case Nafion). Rosa concluded that based on conductivity that Asbestos performed
best, and suggesting that woven PTFE and PPS could be considered as separators
for alkaline electrolysis, based on their performance.79 Asbestos became the separa-
tor material of choice during the late 1970’s, but due to the well documented negative
health effects associated with asbestos, its commercial use has now been phased out.80
Gas separators based on polysulfones were investigated due to their high chem-
ical and mechanical stability. Kerres et al. found that polysulfone membranes pro-
duced with a polymeric additive (PVP) show high ionic conductivity due their high
porosity and hydrophilicity.81 Kerres concluded that commercially available polysul-

fone UDEL was a good choice as a gas separator due to its long term stability in caustic
conditions, as well as good gas separation performance. The optimum specific resis-
tance of the membrane was approx. 1 Ω·cm2 , which is too high to be used in present
day electrolyers due to the need to work at current densities above 500 mA·cm−2 .
The addition of zirconium oxide to the polysulfone matrix by Vermeiren et al.

led to the development and subsequent commercialisation of the Zirfon gas separator.
Zirfon showed excellent gas separation characteristics, long term stability, and initial
prototypes had an ionic resistance of 0.2 Ω·cm2 (30◦C, 30 wt. % KOH).82 Gas purities
for both H2 and O2 of greater than 99.9% at 70◦ C 5.3 M KOH were reported at current
densities between 100 and 400 mA·cm−2 . Alongside this report, Vermeiren et al.
also reported a high performance cell obtaining a current density of 800 mA·cm−2
at 90◦C with a cell voltage of just 1.68 V using a 0.5 mm thick Zirfon diaphragm.
After further testing, the commerically available version of Zirfon had an increased
thickness, leading to a slightly higher resistance of 0.3 Ω· cm2 (80◦C, 30 wt. %
KOH).80, 82, 83
Cermet diaphrams were characterised by Wendt et al., showing good chemical

stability, and adequate gas separation performance. The ionic resistance for the di-
aphragms was adequate at 0.2 Ω· cm2 , but the mechanical stability of the membranes
was deemed too low.84
Of the microporous gas separators investigated, Zirfon has demonstrated good

electrochemical and gas separation performance as well as long term stability in the
strongly basic environment. The robust nature, and durable quality of the separator
has made it the principal choice or use in commercial systems. The limitations of
microporous gas separators are their inability to work at differential pressure due to
gas mixing, as well as the safety concerns around the high concentration alkaline
solution used. The development of the AEM works around these limitations, and has
the potential to enable compact alkaline cells running at differential pressures, with
lower concentration electrolyte.85
The first publication of a solid-state alkaline electrolyser was be Leng et al., us-
ing the commercially available Tokuyama A201 alkaline membrane. The cell was
operated on a feed of 1 M KOH. At 50◦C, the cell using iridium oxide as the an-
odic catalyst, and Pt black as the cathodic catalyst exhibited a current density of 399
mA·cm−2 at 1.80 V. The membrane resistance was 0.27 Ω· cm2 , useful in the con-
text of higher current densities, but the membrane showed a significant decrease in
performance after 27h.86
Wu and Scott employed an AEM with quaternary ammonium function groups,

the operational thickness was 50 µm when fully wet, and the resistivity was 0.16 Ω·
cm2 at 25◦C. With high performance catalyst coatings, the electrolysis cell demon-
strated 1 A·cm−2 at 1.8 V in 1M KOH at 25 ◦C.87
The recent development of a imidazole-based anion exchange membrane Sustan-

ion by Dioxide Materials, USA has showed promising results, with a resistance of
0.045 Ω· cm2 reported by Liu et al. being one of the lowest reported in the literature.
Further development of the membrane with a PTFE support for mechanical stability,
and zirconium oxide layer for greater compression strength increase the usibility of
the membrane, but have increased the membrane resistance to 0.090 Ω· cm2 in its
current iteration.88, 89
A novel membrane-less electrolyser was reported by Gillespie et al..90 The cell

achieves gas separation using the flow of electrolyte to push the hydrogen and oxygen
product gases through the mesh electrodes.91 Published results lack information about
the Ohmic resistance of the system, and although high current densities have been
achieved, the high cell voltage suggests that an unacceptably high Ohmic resistance is
seen due to the necessary gap between the electrodes to maintain gas separation.
Overall, AEM-based water electrolysers are commercially attractive, although

they are held back by their poor stability in the alkaline environemnts.92 Break-
throughs in this area could allow AEM electrolysis to combine the benefits of the
3.3. T EMPERATURE 40
earth abundant materials used for traditional alkaline and the high performance of
PEM technologies.
3.3 Temperature
Running electrolysis cells at higher temperatures has the benefit of increasing the per-
formance of the cell through increased ionic conductivity of the membrane, as well as
faster kinetics minimising the activation losses. Temperatures above 100 ◦C comes as-
sociated with higher corrosion levels of the materials, and so a balance must be struck
between the two. Liquid alkaline systems are run at 60−80 ◦C, whereas newer AEM
cells are run at below 60◦C due to stability concerns associated with the membranes.
3.4 Cell design

Work on zero gap alkaline electrolysis within the last 10 years has suggested that a
significant increase in cell performance can be achieved. Through the employment
of the more compact cell design, this could offer the potential for a step change in
the performance of alkaline electrolysis.35 Figure 3.2 shows that the main difference
between the traditional setup and the zero gap design is the employment of porous
electrodes rather than solid metal plates. This allows cells with a very small inter-
electrode gap, compact design and high efficiency. It forces gas bubbles to be released
from the backside of the electrodes, reducing their contribution to the cell voltage.41
Figure 3.2: Schematic showing reduction of inter-electrode gap from employing a

zero gap cell design, significantly reducing the cell resistance.
3.4. C ELL DESIGN 41
3.4.1 Zero gap alkaline electrolysis
Zero gap alkaline electrolysis works by compressing two porous electrodes either side
of a hydroxide ion conducting membrane or gas separator. This achieves a gap be-
tween the two electrodes equal to the thickness of the membrane (<0.5 mm) rather
than (>2 mm) for the traditional setup, thus significantly reducing the Ohmic resis-
tance contribution from the electrolyte between the two electrodes. A gas diffusion
layer provides an electrical connection from the porous electrode to the bipolar plate,
whilst simultaneously allowing a feed of electrolytic solution, and the removal of the
gas products.
Zero gap alkaline electrolysis was first proposed in 1967 by Costa and Grimes,
using mesh electrodes either side of a microporous gas separator. Significant research
was undertaken during the 1980’s showing a large increase in current density. Re-
cent research has been principally centred on the development of the anion exchange
membrane, which offer advantages of lower resistance and potential for operation at
differential pressures.84, 86, 88 Novel cell designs have also been developed including
the use of high surface area foam electrodes, and adopting fuel cell type electrodes
deposited directly onto the membrane.93
The zero gap design is more complex than the traditional approach, and as a result
the structural components are somewhat different. The zero gap cell employs porous
electrodes compressed against a engraved flow plate instead of the single flat plate that
is used used for finite gap electrolysis. The catalyst material is either deposited directly
onto the porous electrode, or onto the membrane, whereas the traditional design has
the catalyst deposited directly onto the flat electrode plate. The electrocatalytic mate-
rials listed earlier are all directly transferable to the new design, although deposition
methods may need to be adapted to deposit the materials onto porous substrates. Both
cell designs can use the same gas separator, however the zero gap design opens up to
possibility of working with an anion exchange membrane.
Porous electrodes based on nickel mesh/foam are used in many systems reported
3.4. C ELL DESIGN 42
in the literature. Coated stainless steel is also a possibility on the cathodic side, al-
though stainless steel experiences corrosion when subjected to high potentials in the
presence of oxygen. Carbon cloth is employed in alkaline fuel cells although similarly
due to oxidation at high potentials, it is not suitable for use on the anodic side.
Bipolar plates must provide good electrical conductivity, low contact resistance
and corrosion resistance. Titanium is commonly used in PEM cells, however plates
based around stainless steel and nickel are attractive options for alkaline electrolysers
due to lower cost. Graphite is used for bipolar plates in alkaline fuel cells however at
the high potentials experienced on the anodic side make graphite unsuitable for this
side of the alkaline electrolysis. Karimi et al. made a comparison of materials with
regard to interfacial resistance for solid polymer fuel cells, stating that the increase in
resistance over time is due to the formation of an insoluble oxide layer on the surface
of the plate.94
Whilst individual characterisation of the electrocatalytic and gas separator mate-

rials is important to guide appropriate development in the individual areas, their appli-
cation into full working cell is important to demonstrate their viability at larger scale.
The development and testing of full cells, both on the laboratory and demonstration
scale have been reported, and these results are outlined below.
3.4.2 Membrane Electrode Assembly

The central part of the zero gap design is called the membrane electrode assembly
(MEA) and includes the two porous electrodes in contact with either side of the mem-
brane. There are different methods of producing the MEA which can be divided into
two principal categories: catalyst-coated substrate (CCS) and catalyst-coated mem-
brane (CCM).
Both methods employ porous electrodes, which offer the added benefit of in-
creased active surface area, such that for the same material and cell design, a higher
geometric current density can be achieved. The performance of the cell is dependent
on the choice of catalysts and membranes as well as cell design, making comparisons
3.5. F ULL CELL PERFORMANCE 43
between different cell designs difficult to quantify. The area resistances (Ω· cm2 ) of
the zero gap cells includes the membrane resistance, as well as any interfacial contact
resistances between the cell layers, this value allows a degree of comparison between
similar zero gap cells, and can help to guide the importance of cell design in overall
cell performance. Where published, this value is reported.
3.5 Full cell performance
In this section, the benefits and drawbacks of each type of assembly will be outlined,
published examples introduced and the cell design bespoke to each system discussed.
3.5.1 Catalyst-coated substrates (CCS)
Figure 3.3 shows a three-dimensional schematic of a catalyst coated substrate zero gap
cell, the two porous electrodes are individually coated with catalysts, and compressed
onto either side of the gas separator. The flow channels in the bipolar plates allow a
path for electrolyte, and also permit efficient removal of product gases from the cell.
The setup is compressed together to ensure good connections and gaskets are used to
prevent leaking, although care must be taken in this setup to avoid deformation of the
membrane when applying compression.
Schiller et al. developed a high performing cell based around catalysed perforated
nickel sheets, with circular electrodes of 600 cm2 showing good performance and
stability (300 mA·cm−2 at 1.65 V and 80◦C), even in intermittent conditions.95 This
high performance is based around Raney nickel electrodes, and an early iteration of
Zirfon with low resistance.
Figure 3.3: Schematic showing the CCS setup, with porous electrodes either side
of a gas separator
Another cell based on perforated sheet electrodes was reported by Li et al., who
developed a 9 cm2 test cell based on coated mesh, SS flow plates and an experimen-
tal alkaline anion exchange membrane from ITM Power, with current densities of 1
A·cm−2 at an initial voltage of 2.12 V, the cell showed stability during long term
testing. The high current at such low voltage is attributed to the thin experimental
membrane being used in this system, although no data is available on the resistance of
the membrane.
Ahn et al. electrodeposited small amounts of nickel directly onto carbon paper
substrates and characterised a CCS test cell using graphite bipolar plates. An optimal
performance of 2.04 V at 250 mA·cm−2 was reported with the influence of cell com-
pression on the cell performance recognised as significant. The use of the Tokuyama
A201 AEM offers an acceptable resistance of 0.27 Ω·cm2 , but the use of pure nickel
as the electrode materials lead to large overpotentials being the limiting contribution
to the cell performance.96
3.5.2 Nickel foam
A variation of the CCS setup is achieved by using of high surface area electrodes such
as nickel foam, which has the advantage of a much higher active surface area than
mesh substrates. The cell design is slightly altered as flow-field etched bipolar plates
are no longer necessary, due to electrolyte flow through the porous material, although
they are often still employed.
Gas management becomes an important factor due to the small pore size char-
acteristic of the metal foam. When high current densities are applied, gas removal
must be effective to stop the gas bubbles covering parts of the material, and reducing
its available surface area. The large surface area provides a high number of sites for
catalyst deposition, and one of the highest performing anodic electrodes reported is
Ni-Fe(OH)2 deposited onto a nickel foam substrate.
Xiao et al. used Ni-Fe catalysed nickel foam and Ni-Mo catalysed stainless steel
fibre felt hot pressed either side of an alkaline polymer electrolyte. The cell has a cell
voltage of 1.85 V at a current density of 0.4 A·cm−2 when operating at 70◦C. The
Ohmic resistance of 0.375 Ω· cm2 is high in comparison to other systems, and is due
to the experimental membrane used in the cell. Nevertheless, the performance of the
cell is promising, and the choice of electrocatalysts, and electrode substrates facilitate
the high performance of the cell.97
Kim et al. altered the nickel foam to fabricate an asymmetric porous nickel elec-
trode, which had small pores (∼5 µm) in contact with the membrane to provide the
maximum active surface are, and a more open structure (pore size 100 µm) on the
backside, to facilitate gas bubble removal from the bulk. Performance of 0.5 A·cm−2
was reported at a cell voltage of 1.8 V and 80◦C.98 The gaskets were used as multi-
functional sealants and electrolyte flow channels, such that the foam was compressed
directly onto un-etched bipolar plate. This innovative approach to the alteration of
nickel foam could minimise the effect of gas build up inside the cell when compared
to similar setups.
A research group at the Technical University of Denmark (DTU) have devel-

oped high temperature and pressure metal foam based alkaline electrolysis cells.99
Charzichristodoulou et al. report a high temperature (250◦C) and pressure (40 bar)
alkaline electrolysis cell with catalysed nickel foam based/metal alloy gas diffusion
electrodes. The metal foams were compressed to a thickness of 0.5 mm, and were
assembled either side of a novel electrolyte matrix tape. The high performance cell
(3.75 A·cm−2 at 1.75 V) showed stability for 400 h, with a specific cell resistance of
just 0.15 Ω·cm2 .100 Despite the high performance, the high temperature and pressure
open up concerns about the stability of the materials in such harsh conditions.
3.5.3 Catalyst-coated membranes (CCM)

With the development of the anion exchange membranes, the use of a catalyst coated
membrane offers a novel arrangement to allow for efficient use of the electrocatalysts.
The CCM setup has been used in PEM electrolysis since its introduction, and is cur-
rently widely used in both PEM and alkaline fuel cells. This setup involves a catalytic
layer consisting of catalyst nano-particles mixed with an ionomer/binder and disper-
sion solvent being deposited directly onto each side of the membrane. A gas diffusion
layer is employed to provide an electrical connection from the catalyst layer to the
bipolar plate, whilst also allowing the produced gas bubbles to escape; electrolyte is
flowed through or behind the gas diffusion layer to facilitate gas removal. Figure 3.4
shows how the cell differs from the CCS setup, with the electrocatalyst deposited di-
rectly onto the membrane. Recent work, particularly since the improvement of AEMs,
has applied this setup to alkaline electrolysers with the aim of combining the benefits
of PEM electrolysers with the less harsh alkaline environment.
Leng et al. prepared a CCM cell by hand spraying prepared catalyst ink onto
either side of the membrane, the ink consisted of a catalyst (IrO2 for anode, Pt for
cathode), de-ionized water, n-propanol and AS-4 ionomer suspension. Titanium foam
was used for the anodic gas diffusion layer and plain carbon paper for the cathodic gas
diffusion layer, and they were mechanically pressed against the CCM when preparing
the cell hardware.86 Initially the cell was fed with a pure water feed, 399 mA·cm−2
achieved at 1.8 V but better performance was observed with 1 M KOH. The cell resis-
tance was 0.27 Ω·cm2 , and, whilst the cell resistance is acceptable, the use of precious
metals as the catalysts takes away from one of the key advantages of the alkaline
environment.
Figure 3.4: Schematic showing the CCM setup, with the catalyst layer deposited
directly onto the membrane, the other components of the cell stay similar to the
CCS setup
Wu et al. airbrushed catalyst inks onto either side of the membrane, and used
stainless steel mesh as both gas diffusion layers and current collectors on each side.
Using an anion exchange membrane, the cell demonstrated 100 mA·cm−2 at 1.9 V.
Using EIS the electrolytic resistance was found to be 0.85 Ω· cm2 , and had a consid-
erable effect on performance, especially towards 0.5 A·cm−2 .87 Wu proposed that a
large proportion of the cell resistance was due to the contact resistance between the
catalyst layer and the gas diffusion layer, further optimisation of this aspect is key in
the future development of CCM cells.
Pandiarajan and Ravichandran brush coated a spinel ferrite and nickel powder
onto the anodic and cathodic sides of the commercial AEM Fumasep FAA-3-PK-
130, and used a pair of platinum coated titanium mesh as the current collectors. The
cell exhibited a current density of 300 mA·cm−2 at 1.8 V in deionised water, with a
lifetime of 100 hrs, the cell resistance was seen to be approx. 0.5 Ω· cm2 .101 The
durability of the cell is of note, although the resistance is towards the upper limit of
what is applicable to high performance alkaline cells.
In 2017, Masel et al. demonstrated the use of the Sustainion AEM in a cell with
NiFeCo cathode and a Ni-Fe anode, with 1 A·cm−2 achieved at 1.9 V, with the voltage
stable for 2000 hours. This state of the art performance exceeds all others reviewed in
this section, and brings together the developments of the other systems, combining a
low resistance membrane, with highly performing base metal catalysts, and the long
term stability is an exciting finding. Further research using the membrane necessary
to investigate it’s full potential.88, 89
Overall, the potential for CCM alkaline electrolysers has been demonstrated by
a number of groups. The performance associated with these CCM electrolysers is
has historically been limited by the high Ohmic resistance associated with the AEM
membranes, however recent advances in this area suggest that the CCM setup has
great potential.
3.5.4 Cell design considerations

The move from finite gap to zero gap electrolysis introduces new components which
must be understood and optimised to ensure optimal performance of the setup. Com-
pression of the cell has an impact of cell performance due to the interfacial contact
resistances inside the cell.
Figure 3.5 shows the effect of cell compression on overall cell performance. Ahn
et al. found that compression of the components affected cell voltage, which was the
result of interfacial contact resistances between the gas diffusion layer and the bipolar
plate, as well as between the catalyst layer and the gas diffusion layer.96 This area
has attracted plenty of research for PEM electrolysers, although no research currently
exists on this topic for zero gap alkaline electrolysers.
The criteria for bipolar plates materials include corrosion resistance, and have a
low contact resistance, current choices include nickel, titanium and graphite.102 In
3.6. T RANSPORT LOSSES 49
Figure 3.5: The effect of cell compression on the polarisation perfomance of the
cell was demonstrated by Ahn et al..96
PEM cells, the cost of the bipolar plates make up 50% of the whole stack cost ,103 and
much research has been focused on developing cheaper alternatives to the currently
used titanium. Nickel coated stainless steel is an attractive option for zero gap alkaline
electrolysis due to its potential low cost.
3.6 Transport losses

Limited data has been published regarding the voltage contribution from the mass
transfer losses inside electrolysis cells, and no studies are currently available regarding
mass transport in a zero gap alkaline cell. Indeed there is yet to be a universally applied
technique that is reliable at isolating this aspect.
Dedigama et al. used an optically transparent flow channel to observe the two
phase flow properties inside a PEM electrolysis cell, and proposed using EIS to iso-
late the mass transport resistance of the cell, stating that enhanced mass transport was
observed in the presence of slug flow.104, 105 Van de Merve et al. also suggested
that using the low frequency tail on EIS spectra demonstrated the presence of mass
3.7. C URRENT DENSITY 50
transport limitation at high current densities, however no quantified value was asso-
ciated with this observation.106 Suermann et al. extrapolated full cell Tafel plots to
reveal the mass transport contributions, although the contributions for the Tafel slope
changes associated with the catalysts was left unaccounted for.107
This is an area that requires further research to understand and isolate the inter-
action of two phase flow and the losses associated with mass transport.
3.7 Current density

As the current density of a cell increases, the efficiency of the cell decreases. An
optimum balance must be sought that allows a sufficient production rate of hydrogen,
but also a high enough efficiency that doesn’t waste too much of the input energy.
At a cell voltage of 1.8 V, PEM cells can operate at above 1 A·cm−2 , whereas
alkaline cells tend to only operate below 0.5 A·cm−2 . In this situation PEM cells
require half of the surface area to achieve the same production rate, which allows for
reduced capital expenditure. Masel et al. reported a cell running at 2 A·cm−2 without
any major mass transport limitations, but operation at or above this current density
will require investigations into the effects of gas management.88, 108
Furthermore, the use of high current densities dilutes any foreign gas that per-
meates through the membrane due to the higher rates of gas evolution. As a result,
operation at higher current densities may permit the use of thinner membranes, in turn
allowing for an increase in the efficiency of the cell.109
3.8 Technology gaps

Figure 3.3 outlines the key gaps in knowledge related to alkaline electrolysis. As dis-
cussed in this chapter, the adoption of zero gap design looks to be a key factor in
increasing the efficiency and current density of the technology. Understanding each
contribution to the cell voltage from the zero gap design, including the interfacial con-
tact resistance and the resistance associated with the bubble formation and removal,
3.8. T ECHNOLOGY GAPS 51
Technology gap Understanding Materials Durability Cost

OER catalyst x x x
HER catalyst x x
Gas seperators/AEMs x x x x
Zero gap cell design x x
Two phase flow effects and
x x
trasnport losses
Degradation mechanisms x x x
Higher current densities x x x x
Gas crossover mechanisms x x
Effects of intermittant
x x x x
operation
Table 3.3: An overview of the gaps in knowledge relating to alkaline electrolysis
is crucial to implementing strategies to mitigate these effects. Related to this, increas-

ing current density to the 1-2 A·cm−2 region is key to reducing the capital cost, and
the effect of this on both the electrocatalytic materials as well as the gas management
properties inside the cell needs to be better understood to optimise the running of the
cell.
Much work has been focused on the development of electrocatalytic materials.
The development of HER catalysts shows promising results, with only small contri-
butions to the cell voltage, although durability of these materials in dynamic operating
conditions is another factor that should be investigated further. The relatively high
contribution from the OER catalyst suggests that further work is needed in this aspect
to reduce contributions below 200 mV at current densities above 100 mA·cm−2 .
The resistance of the gas separator currently makes a large contribution to the
overall cell voltage at current densities above 0.5 A·cm−2 , and novel materials and
operating strategies should be investigated to reduce these effects, and allow efficient
running of the technology above 1 A·cm−2 .
Overall, whilst alkaline electrolysis is considered a mature technology, many
gaps in the research still exist. With the new application of electrolysers in a en-
ergy storage setting, the focus on efficiency and cost of the electrolysers bring new
challenges for the technology.
Chapter 4
Experimental techniques
4.1 Electrochemical impedance spectroscopy

Electrochemical impedance spectroscopy (EIS) is an AC technique widely used as a
diagnostic tool to assess performance parameters of electrochemical systems. EIS in-
volves the application of a small pertubation of either the applied potential or current
superimposed on a DC offset. The measurement of the phase difference and ampli-
tiude of the response to the applied signal allows the analysis of electrode processes.
Input Amplitude
φ
Output Amplitude
Input
Output
Figure 4.1: Viusualisation shows the signal and response signals demonstrating the
change in amplitude, and the phase shift (φ )
EIS typically uses perturbations in the range of 10 mV or 10 mA amplitude, with

4.1. E LECTROCHEMICAL IMPEDANCE SPECTROSCOPY 53
the small amplitude necessary to invoke a pseudo-linear response from the system.
The frequency of the perturbation is varied across a preset frequency window, and
the response gives insight into the resistive, capacitative and inductive features of the
system. In this thesis, EIS will be used primarily to gain information regarding the
resistive features of the electrochemical cell.
Figure 4.1 shows the sinusoidal signal and response, and the associated phase
shift (φ ).
Using an equation analogous to Ohms law, the impedance of the system can be
calculated:
E E0 sin(ωt) sin(ωt)
Z= = = Z0 (4.1)
I I0 sin(ωt + φ ) sin(ωt + φ )
The impedance can therefore be expressed in terms of its magnitude, Z0 , the

radial frequency ω which is related to the frequency such that ω = 2π f and the phase
angle, φ , between the signal and response.
-Z"
|Z|
φ
Z'
Figure 4.2: The Nyquist plot is a convenient way to display the real and imaginary
components of the impedance
For convenience, these equations can be re-written in complex form using Eulers
formula giving the signals as follows:
E = E0 exp(iωt) (4.2)
I = I0 exp(iωt)exp(− jφ ) (4.3)
The impedance can then be represented as a complex number:
E
Z= = Z0 exp( jφ ) = Z0 (cos(φ ) + j sin(φ )) (4.4)
I
A Nyquist plot is widely used to display the impedance, with the real and imag-
inary components of the impedance displayed on the X and Y axes respectively. In
vector form, the magnitude of the impedance |Z| is the distance from the origin, and
the phase angle as the angle between the vector and X-axis.
As shown in Figure 4.2 the Nyquist plot below,
Z 0 = |Z| cos(φ )
−Z” = |Z| sin(φ )
Electrochemical systems can be modelled as electrical circuits using simple elec-

trical components including resistors, capacitors and inductors.
The AC responses of these components are well known, and electrochemical

systems can be modelled as combinations of these components.
The impedance of a resistor has no imaginary component as the AC voltage and

current response are in phase (φ = 0), and so the impedance is given by Ohms law
such that
E0
ZR = =R (4.5)
I0
This is shown as a single point on the Nyquist plot in Figure 4.3.
The impedance of a capacitor has no real component as the phase angle is be-
π
tween the signal and response is 2 . The current and voltage relationship of a capacitor
is:
δV (t)
I(t) = C · (4.6)
δT
-Z"
1
0 1 2 3 4
Z'
Figure 4.3: Impedance of a resistor (black) and a capacitor (red) and their fre-
quency dependance
For the initial voltage of V = V0 e jωt , the current is:
I(t) = C ·V0 jωe jωt (4.7)
the magnitude of the impedance can therefore be calculated as:
V 1 −j
ZC = = = (4.8)
I jωC ωC
It can be seen that the impedance of a capacitor has no real component, and is
represented in Figure 4.3 as a vertical line along the line where Z 0 = 0.
To model realistic systems, the impedance from multiple components are calcu-
lated using combinations analogous to simple resistor circuits such that:
In series:
Z = Z1 + Z2 + Z3
and in parallel:
1
Z= Z1 + Z12 + Z13
Using this information, combinations of components can be displayed on the

Nyquist plot. For a series combination of a resistor and capacitor:
1
Z = ZR + ZC = R + (4.9)
jωC
2
-Z"
0 1 2 3 4
Z'
Figure 4.4: Impedance of a resistor and capacitor in series
Figure 4.4 shows how the combination of a resistor and capacitor in series appears
on a Nyquist plot, witha vertical line offset from the Y axis by the magnitude of the
resistance, R,
Figure 4.5: Impedance of a resistor and capacitor in parallel
For a parallel combination of resistor and capacitor, the impedance is therefore:
1 1 (1 − jωRC)
Z= 1
= =R (4.10)
ZR + Z1C 1
R + jωC (1 + ω 2 R2C2 )
when a high frequency perturbation is applied (ω = ∞), the impedance tends to

zero, the capacitor cannot charge and discharge at high frequencies, and acts a short
circuit. At low frequencies (ω = 0), the impedance tends to R, as the impedance of
the capacitor becomes very large, and so the current flows through the resistor. Figure
4.5 demonstrates how this appears on a Nyquist plot.
Figure 4.6: The equivalent circuit of a electrode/electrolyte interface
A common real system experienced in electrochemistry is the charge transfer at

an electrode/electrolyte interface, the equivalent circuit is shown in Figure 4.6 and
the corresponding Nyquist plot is shown in Figure 4.7. The double layer capacitance
that occurs at the electrode/electrolyte interface is modelled as a capacitor (CDL ), the
Faradic charge transfer between the two phases is modelled in parallel with the ca-
pacitor as a resistor (RCT ), and they are both in series with the series resistance of the
system (RS ).
The observed response on the Nyquist plot is represented by the parallel resistor
capacitor combination, offset by the series resistance. The magnitude of the charge
transfer resistance is represented as the diameter of the semicircle. The frequency
that corresponds to the maximum point on the -Z” axis is know as the RC time con-
stant, and can be divided by the charge transfer resistance to give the double layer
capacitance.
The impedance at each frequency can be canculated by using the information

regarding resistor/capacitor combinations as:
4.2. L INEAR SWEEP VOLTAMMETRY 58
ω max = (R CT CDL) -1
-Z" (Ω)
RS RS + RCT
0
0 1 2 3 4
Z' (Ω)
Figure 4.7: Represents the impedance response for the above Randles circuit with
Rs =1.5 Ω, RCT =2 Ω and CDL =1 µ F.
1
Z = R1 + 1
(4.11)
ZR2 + Z1C
Such that for the equivalent circuit:
1 (1 − jωRCT CDL )
Z = RS + 1
= RS + RCT 2 C2 )
(4.12)
RCT − ωCDL (1 + ω 2 RCT DL
At high frequencies it can be seen that Z = RS , allowing the series resistance of

the system to be easily extracted.
4.2 Linear sweep voltammetry

Linear sweep voltammetry is a sensitive technique that involves the measurement of
the current flowing through a system as a function of the applied voltage.
As seen in Figure 4.8a, the voltage at the working electrode is swept across a
preset range at a chosen sweep rate, while Figure 4.8b shows the simultaneous mea-
surement of the current flowing through the system. The oxidation or reduction of
4.2. L INEAR SWEEP VOLTAMMETRY 59
V2
Voltage
Time
V1
(a) Linear sweep voltammetry is a sweep between two preset voltage

Current
V1 V2
Voltage
(b) The current response to the applied voltage is recorded and gives insight in the electrochemical
performance of the working electrode.
Figure 4.8
species is observed as a peak or trough in the current signal, and the observations
can be attributed to appropriate oxidation and reduction reactions based on a good
knowledge of materials within the system, as well as comparisons to historic data.
The characteristics of the output results depend on three main factors:
• The voltage scan rate
• The rate of electron transfer
• The reactivity of the chemical species at the working electrode

4.3. TAFEL SLOPE 60
As the scan rate is a user choice rather than a property of the materials, it is im-
portant that its effect on the results is minimiesed. This is usually achieved by utilising
a low scan rate in relation to the rate of electron transfer, and prevents distortion of the
data by diffusion limiting phenomena. For the reactions investigated in this thesis, a
scan rate below 5 mV· s−1 is appropriate.
4.3 Tafel slope

The Tafel slope is the gradient of the plot of the log of the current against the over-
potential, and provides an insight into the rate of change of overpotential with current
density. The acquisition of data displayed in a Tafel plots allows an insight into the
limiting step in the electron transfer, and allows comparisons between catalysts.
The relation between the overpotential and the current can be expressed using the
Tafel equation derived in Chapter 2, and is shown in Figure 4.9 where:
I
η = B log (4.13)
I0
where B is the Tafel slope, and logI0 is the extrapolated intersection with the log
axis.
The Tafel data is obtained using galvanostatic analysis (0.1 mA·cm−2 to 100
mA·cm−2 ). Each current step is left for 20 seconds to ensure that it reaches steady
state. Data is recorded for both ascending currents and descending currents, and the
data is compared to ensure stability of the material, even at the highest current densi-
ties.
4.4. F OUR - TERMINAL SENSING 61
Figure 4.9: The Tafel slope is extracted from a plot of the overpotential vs log
current
4.4 Four-terminal sensing

Four terminal sensing is a measurement technique widely used to determine the elec-
trical resistance of a sample. It permits the measurement of small resistances to a
higher precision when compared with the two terminal method, as the technique elim-
inates contact and lead resistances from the measurement.
As shown in Figure 4.10, four terminal sensing is achieved by supplying a current
via a pair of source leads, and measuring the voltage drop across these leads and their
connections, which obeys Ohm’s law V = IR. The voltage is measured using two
sense leads, connected on the target sample, and due to the fact that almost no current
flows through the leads, there is no voltage drop across the leads or connections.
Four-terminal sensing is key in the electrolyser setting as a diagnostic tool, as
high currents are supplied to the cell, and the method allows accurate measurement of
resistances between component interfaces.
4.5. H IGH SPEED IMAGING 62
Figure 4.10: The four-terminal sensing method applies a current across the full
circuit, but measures the voltage just across the target sample
4.5 High speed imaging

A Photron Fastcam 1024-PCI camera with a maximum resolution of 1024x1024 pixels
was used to image the two-phase behaviour of the system. Images were taken at a
rate of 1000 frames per second. The flowfields were illuminated using two PhotoSel
LES600 lights. Post processing of the images was undertaken using Matlab.
4.6 SEM imaging

A Hitachi TM3030 Plus scanning electron microscope was used to image materials -
with a standard setting of accelerating voltage of 15 kV.
4.7 Compositional analysis

Energy Dispersive X-ray Spectroscopy (EDX) was used for elemental analysis of
samples using an Aztec EDX extension to the Hitachi SEM described above, with
AztecOne software allowing for automatic peak detection.
4.8 Gas chromotography

Elemental analysis of the product gases were characterised using an Agilent Tech-
nologies 7820A GC System.
Chapter 5
Minimising the Ohmic resistance of an

alkaline electrolysis cell
5.1 Abstract
This chapter describes the motivation for zero gap cell design, and outlines the devel-
opment and optimisation of a zero gap alkaline electrolysis test cell. The benefits of
employing a zero gap design over the finite gap equivalent are introduced, and experi-
ments confirm and quantify the benefits. Investigations are focused around minimising
the Ohmic resistance of the cell, and investigations show how operating parameters
such as increased temperature as well as the choice of electrolyte have an effect on
the Ohmic resistance. Furthermore, the morphology of the electrode is shown have an
effect on the Ohmic resistance of the cell.
5.2 Introduction
Traditional ’finite gap’ alkaline electrolysis based on two electrode plates separated by
a liquid alkaline electrolyte suffers from low current densities (<250 mA·cm−2 ) at low
efficiencies (<60%), principally due to high internal resistance losses.39, 43, 110 Other
water splitting technologies, most notably PEM electrolysis, have been developed,
and demonstrate higher current densities (>2000 mA·cm−2 ) at higher efficiencies
(>72%).34, 103, 111, 112 Recent installations on the commercial scale are dominated
by the PEM technology; however their widespread adoption is limited by the high
5.2. I NTRODUCTION 64
Figure 5.1: Schematic showing reduction of inter-electrode gap from employing a

zero gap cell design, significantly reducing the cell resistance.
cost of both the Nafion membrane, and the noble metal catalysts such as platinum or
iridium needed because of the acidic environment. Intense development of new cata-
lysts and the availability of a cheap and stable gas separator, together with modern cell
design, has led to a resurgence of interest into alkaline electrolysis, with the prospect
of both low cost and highly efficient electrolysis being a strong attraction.74, 75, 113–115
Previous work regarding the optimisation of cell design has been principally
based around investigations by Nagai et al., who used DC voltage polarisation data
to show that a decrease in electrode gap causes cell voltage reductions at low current
densities, and demonstrated the experimental exception to Ohms law at high current
densities due to the increase of the electrolyte void fraction from the evolved gas bub-
bles .116 Nagai et al. proposed the existence of an optimal gap between electrodes of
2 mm at high current densities (>500 mA·cm−2 ).117
LeRoy et al. showed that when gas bubbles are being evolved, the increase of the
volume fraction of bubbles between the electrodes will increase the electrical resis-
tance of the electrolyte, whilst the effect of electrolyte flow on finite gap cell perfor-
mance was investigated by Zhang and Zeng as well as Bongenaar-Schlenter, showing
electrolyte flow to be key to reducing the effect of gas bubbles, especially at high
current densities.118, 119 Zhang investigated cell performance based on the cell po-
larisation data, which is a limited approach as the Ohmic resistance is not isolated
from electrode performance, wheras Bongenaar-Schlenter used a segmented nickel
electrode, with multiple reference electrodes to build up a holistic view of the perfor-
mance inside the cell. Dedigama et al. used EIS to isolate the Ohmic resistance of
a PEM electrolyser, and studied the effects of different flow rates and flow regimes,
indicating that the localised cell heating can be an important factor at low flow rates.
No studies on the cell performance have been done for alkaline electrolysis using
EIS.104–106, 120
The employment of the zero gap cell design, as discussed recently by Pletcher
and Li, looks to push the performance of alkaline cells towards that of PEM whilst
maintaining the benefits of cheaper cell materials, and works by compressing two
porous electrodes either side of a hydroxide ion conducting membrane or gas sepa-
rator43 . This achieves a gap between the two electrodes equal to the thickness of the
membrane (≤0.5 mm) rather than (≥2 mm) for the traditional setup (Figure 5.1). Zero
gap cell design has been discussed by a number of authors,93, 100, 102, 121 although its
benefits over conventional designs are yet to be directly quantified.
Following the discussion of zero gap alkaline electrolyers by Pletcher and Li in

2011, further research into the zero gap technology has been varied and inexhaustive.
Much research has been focused on the development of catalyst materials, and full
cell characterisation has been more a display of catalyst performance rather than in-
vestigating and optimising cell design. Less common but no less important has been
the research focused on membrane development, with either porous gas seperators
(Zirfon) or AEMs the focus of the research. Both catalyst choices and cell design are
similar for both types of membrane, hence, can be continued almost independent of
these developments.
Many different porous electrode morphologies have been reported in the litera-
ture regarding alkaline electrolysers, including perforated plates, expanded mesh, wo-
ven mesh and metal foam, although no direct comparisons of the structures has been
made.42, 93, 102 The effect of the structural properties of porous gas diffusion layers
have been investigated for PEM systems, with the pore size shown to effect the Ohmic
resistance of the cell, as well as its performance at high current densities.122, 123
Information regarding the Ohmic resistance of alkaline electrolysis systems is

lacking in the literature, with adjustments of design and operating parameters gener-
ally investigated only with regards to its effect of cell voltage.This chapter quantifies
the Ohmic resistance savings achieved by employing a zero gap design, as well as
demonstrating the effects of current density and electrolyte flow rate on the Ohmic
resistance of both finite gap and zero gap cell designs. Furthermore, the effect of the
electrode morphology on the cell resistance is investigated.
The use of EIS in this study allows the cell resistance to be decoupled from
the overall cell voltage, allowing the effects of cell design to be investigated without
the influence of electrode performance. This is particularly important when making
comparisons between setups using different materials, as well as with other setups
reported in literature. Current density i, (mA·cm−2 ) and area specific resistance R,
(Ω· cm2 ) are used at all times to allow straightforward comparison of data to other
systems.
The resistance of an electrolysis cell is expressed as
ρcell l
R= (5.1)
A
where R is resistance (Ω), A is cross sectional area (cm2 ), l thickness of cell (cm),
and ρcell is the specific electrical resistance of the cell (Ω·cm). To allow for easy
comparison between setups, area specific electrical resistance ROhmic is used such that
ROhmic = RA = ρcell l (5.2)
with units of (Ω·cm2 ).

5.2.1 Ohmic resistance

The voltage drop between the electrodes can be expressed using Ohms law
V = iROhmic (5.3)
Where i is the current density in A·cm−2 , ROhmic is the area specific resistance of
the cell in (Ω·cm2 ). When no gas bubbles are being evolved,
ROhmic = RMembrane + RElectrolyte (5.4)
The area specific resistance of the membrane is constant (due to constant thick-
ness and resistivity), and the area specific resistance of the electrolyte is a function of
electrolyte resistivity and distance, such that
ROhmic = RMembrane + ρElectrolyte l (5.5)
Consequently, when the concentration of the electrolyte is fixed during alkaline

electrolysis, ROhmic varies with l, the distance between the two electrodes.
5.2.2 Electrolyte resistivity

Treating the electrolyte resistivity (ρelectrolyte ) as a fixed value does not hold true in
working conditions. LeRoy et al. showed that when current is flowing and gas bub-
bles are being evolved, the increase of the volume fraction of bubbles between the
electrode will increase the electrical resistance of the electrolyte.124 Nagai et al.
demonstrated that when the electrode gap is small and the volume fraction is large,
the electrical resistance of the electrolyte becomes particularly large, and an optimum
distance between electrodes was found to exist at high current densities.117
The employment of the zero gap design aims to minimise the Ohmic resistance
by minimising l, the distance between the electrodes, as well as allowing gas bubbles
to be released from the backside of the electrodes, reducing the increase in ρcell due to
rise of the void fraction. The ultimate aim of advanced cell design is to eliminate the
resistance contribution from the electrolyte, even at high current densities, such that
ROhmic = RMembrane .
5.2.3 Effect of gas bubbles
The effect of electrolytic gas bubbles on the cell performance was investigated fur-
ther by Zeng and Zhang, who modelled the bubble formation on a metal electrode
in a stagnant electrolyte, showing the critical diameter of bubble detachment to be
a function of buoyancy, expansion force and interfacial tension force, and showing
electrolyte concentration to have a key effect.118 With electrolyte circulation, the ad-
ditional forces on the gas bubble cause early detachment of the bubbles, and Zeng
and Zhang demonstrate a small reduction in the cell voltage when flowing electrolyte
is used; attributed to early detachment of the bubbles. The bulk of the resistance
was still present and this was attributed to the bubble curtain between the two elec-
trodes. Bongenaar-Schlenter et al. showed that the electrolyte flow rate reduces the
cell resistance in a finite gap cell, with large reductions in cell voltage was reported
at high current densities (> 500 mA·cm−2 ), suggesting that the void fraction in the
inter-electrode gap being the key factor, and that the effects from the bubble curtain
suggested by Zhang could be mitigated.119
5.2.4 Electrode surface morphology
The electrode morphology can have a synergistic effect on both the overpotentials (η)
and the ion transport resistance (RElectrolyte ) when using porous electrodes. As such,
the electrode morphology must be subdivided into the macroscopic structural mor-
phology and the microscopic surface morphology. The overpotential at an electrode
is affected by both the intrinsic activity of the catalyst material, and the effective sur-
face area of the electrode. Modifications of electrode surface morphology are made
to increase the real electrode surface area; producing active sites such as cracks and
crevices which promote the release and capture of electrons.125 Raney nickel is a key
5.3. M ETHODS AND MATERIALS 69
example, with its excellent catalytic performance when compared to smooth nickel
demonstrated by a number of research groups.42, 95, 126–128 This high performance
is attributed to its large effective surface area, resulting from its high porosity and
nanocrystalline structure. The use of high surface area electrodes with favourable
structural morphologies, such as meshes and foams, provide higher surface area sub-
strates than flat plate electrodes; this contributes to a further increase in real surface
area, as well as a possible reduction in the Ohmic resistance due to smaller ion con-
duction distances.
5.3 Methods and materials

The first aim of the cell development was to quantify the benefit of employing zero gap
design over finite gap design, and use electrochemical methods to isolate the factors
which contribute to the increase in performance. The first iteration of the zero gap cell
was constructed to allow a direct comparison to the finite gap cells often discussed in
literature.
5.3.1 Zero gap cell

Due to its cheap price, corrosion resistance in alkaline conditions, and its use in the
finite gap electrolysers being used in our lab at the time, stainless steel 316 flow plates
were fabricated by machining a 10 cm2 serpentine flow field, with channels (1.5 mm
deep, 1 mm wide) into plates (3 mm thick). Electrodes were chosen to be stainless
steel expanded mesh, and the gas separator to be the commercially available Zirfon
Perl UTP 500 gas separator (Agfa). Silicone sealing gaskets were used with the thick-
ness chosen to be 10% thicker than the expanded mesh itself, such that under com-
pression, a connection between the flow plate and the mesh was made. Figure 5.2
shows the components for the zero gap setup.
5.3.2 Finite gap cell

A square cell was constructed with a geometric surface area of 10 cm2 ; Grade 316
stainless steel plates were used as electrodes, and separated by spacers sandwiching
Figure 5.2: Components for the zero gap cell, including the machined flow field
plates, silicone gaskets, mesh electrodes, and Zirfon gas separator.
Figure 5.3: 3d drawing of the finite gap cell; cell spacers of varying thicknesses
allow the interelectrode gap to be varied from 20 - 2 mm.
a gas separator as seen in Figure 5.3. The spacers allowed the adjustment of the
inter-electrode gap from 2-20 mm and silicone gaskets facilitated cell sealing. The
electrodes were polished using 150, 300 and 600 grit sandpaper, then sonicated for 20
mins in acetone and ethanol before being used as electrodes. A Zirfon Perl UTP 500
gas separator (Agfa) was hydrated for 48 hours in 1 M NaOH before use.
5.3.3 Test station
The test station is shown in Figure 5.4), and consists of two separate electrolyte reser-
voirs for anodic and cathodic feeds, with diaphragm pumps (Anself 12 V) suppling
electrolyte feed rates up 200 ml·min−1 . The rate of electrolyte flow was controlled
via variable area rotameters (Dwyer 20 - 200 ml·min−1 ) and was set to 80 ml·min−1
unless stated otherwise.
Figure 5.4: The test rig features separate electrolyte reservoirs for the anodic and
cathodic feeds, diaphragm pumps to provide up to 200 ml·min−1 , and a variable
area rotameters to control the electrolyte flow speed.
5.3.4 Electrochemical characterisation
All electrochemical characterisation was performed using an Ivium-n-Stat multi-

channel potentiostat. The Ohmic resistance of the cell (Rs ) was determined using
galvanostatic electrochemical impedance spectroscopy. A signal amplitude of 5% of
the applied current and a frequency range (100 kHz to 1 Hz). Rs is equated to the
impedance at high frequencies, when the phase angle is closest to zero.129 All mea-
surements were made at 20 ◦C. The cell was allowed to reach equilibrium by applying
the desired current density for 300 s before any readings were made.
Four point probe measurements were made using the potentiostat as the current
supply, and a BK Precision 5491B Multimeter to monitor voltage drop.
5.4. R ESULTS AND DISCUSSION 72
5.4 Results and discussion

5.4.1 Effect of the distance between electrodes
The effect of l, the distance between the electrodes, on the cell voltage is demonstrated
in Figure 5.5. When operating at 250 mA·cm−2 in 1 M NaOH, it is seen to be the dom-
inant contribution to cell voltage. The clear benefit of reducing l is demonstrated by
the reduction in cell voltage associated with reduced l. The onset potential is the same
throughout the experiments at around 1.75 V, well above the reversible potential of
water (1.23 V); this is a function of the intrinsic activity of the electrode materials,
which is stainless steel in this setup, and remain consistent for all designs. I-V char-
acterisation shows an overview of the performance of the different cell designs, but
to isolate the exact reason behind the performance increases, further characterisation
must be done. EIS is used to offer an insight into the Ohmic losses associated with
this system.
Cell Voltage at Varying Interelectrode Gaps

250
Current Density (mA·cm -2 )
200
150
100
20mm
9mm
50 4mm
2mm
Zero Gap
0
1.5 2 2.5 3 3.5 4 4.5 5

Cell Voltage (V)
Figure 5.5: Plot comparing the cell voltage at chosen current densities for different
cell setups; the increase in cell performance with reduced electrode gap is demon-
strated, with the best performance achieved with a zero gap cell configuration.
5.4.2 Isolating the Ohmic resistance of different cell designs

The Ohmic resistance was initially measured at 1 mA·cm−2 , such that the influence of
evolved bubbles could be neglected. The experimental results in Figure 5.6 show that
Ohmic Resistance vs Electrode Gap

10
Ohmic Resistance (Ω·cm 2 )

8
0
0 5 10 15 20
Electrode Gap (mm)
(a) The Ohmic resistance follows a linear trend with electrode gap up to 10 mm, with the
lowest resistance achieved using zero gap design, although it is still 42% larger than the
membrane resistance.
Ohmic Resistance vs Electrode Gap
10
4 10 mA·cm -2
50 mA·cm -2
2 125 mA·cm -2
250 mA·cm -2
0
0 5 10 15 20
Electrode Gap (mm)
(b) The comparison of Ohmic resistance at different current densities shows a differing
increase in resistance for all cells, notably, the rise in Ohmic resistance for the zero gap
cell is much lower than for the finite gap designs, indicating that a bubble curtain does not
form on the frontside of the porous electrodes
Figure 5.6: The effect of electrode distance and gas evolution rates on the Ohmic
resistance of the cell was demonstrated.
the internal resistance of the cell varied linearly with the distance between electrodes
from 2-10 mm. Figure 5.6a shows the deviation from this linearity at 20 mm agrees
with similar results from Nagai et al..117 The proportionality of the resistance and
electrode distance matches well with the theory in section 2, and the figure shows that
the lowest internal cell resistance may be achieved when l is equal to the thickness
of the membrane. The line of best fit is extrapolated backwards to an electrode gap
of 0.5 mm, estimating the membrane resistance to be 0.85 Ω·cm2 . The zero gap
arrangement achieves the lowest cell resistance, although it is still 42% above the
membrane resistance. This difference is principally attributed to the morphology of
the porous mesh, such that the transport of ions through solution is not always by the
shortest path.
Figure 5.6b shows that for all designs, the Ohmic resistance of the cell increased
with rising current density due to effects of gas bubbles causing an increase in the
void fraction of the electrolyte. The Ohmic resistances rose by around 0.4 Ω·cm2
for all electrode gaps with the exception of the zero gap cell, which had a Ohmic
resistance change of just 0.1 Ω·cm2 . This result shows the clear benefit of employing
the zero gap cell setup, and can be attributed to the bubbles being transported away
from the mesh electrodes via the backside, minimising the contribution of the bubbles
to the inter-electrode void fraction. The cell assembly, including the choice of gasket
thickness and cell compression were seen to be important factors in the performance
of the zero gap cell; the cell tested here was in an optimised state.
These results show a different approach to cell design can overcome the effects
that both Zhang and Bongenaar-Schlenter reported regarding the effects of bubbles.
The zero gap arrangement substantially reduces the effect of the bubble curtain, and
will require a lower electrolyte flow rate to maintain low Ohmic resistance at high
current densities when compared to the finite gap cell.
5.4.3 Effect of flow rate of Ohmic resistance of a finite gap cell

Figure 5.7 shows the reduction of the Ohmic resistance by flowing electrolyte for a
cell with 2 mm spacing; a 0.4 Ω·cm2 resistance saving was made by increasing the rate
of electrolyte flow from 20 ml·min−1 to 80 ml·min−1 at 250 mA·cm−2 . The Ohmic
resistance savings become more pronounced at higher current densities. Further re-
sistance savings may have been made by increasing the flow rate further, although the
energy required to maintain high flow rates acts to negate the energy saving from the
cell resistance. These results are consistent with previous studies, and validate the hy-
pothesis that the Ohmic resistance is governed by the void fraction of the electrolyte
Effect of Electrolyte Flow Rate on Ohmic Resistance (2mm)

0.8

0.6
0.4
0.2 20 ml·min-1
40 ml·min-1
80 ml·min-1
0
0 50 100 150 200 250

-2
Current Density (mA·cm )
Figure 5.7: Plot comparing the Ohmic resistance increase of the 2 mm gap cell at
chosen flow rates; increasing flow rate is shown to reduce the Ohmic resistance by
removing the bubbles from the electrolyte, reducing the void fraction in the inter-
electrode gap.
between the planar electrodes, rather that the more general suggestion of bubble cur-
tain by Zhang et al.. This is evidenced by the sharp reduction in Ohmic resistance
associated with flow rate, attributed to the reduced time that each bubble spends in
the inter-electrode gap. As the rate of bubble generation is constant, the faster flow
rate clears each unit volume of gas in a shorter time, causing a smaller void fraction
between the electrodes.
5.4.4 Effect of current density
Both the 2 mm and zero gap cells were tested to higher current densities to investigate
the effect on the cells’ performance. Figure 5.8a show that the zero gap cell had a
significantly lower voltage at all current densities, even above 500 mA·cm−2 ; this im-
provement is likely attributable to the excellent gas management properties of the zero
gap cell. This suggestion is supported by 5.8b, which shows that only small increases
in Ohmic resistance occurred over the measured range of cell current, even when large
amounts of gas evolution was observed. This is consistent with other research showing
the large effect of the bubble curtain formed in front of the planar electrodes at high
Cell Performance at High Current Densities
600
Zero Gap Cell
Current Density (mA·cm-2 )

2mm Gap Cell
500
400
300
200
100
0
1.5 2 2.5 3 3.5 4
Cell Voltage (V)
(a) This plot compares the cell polarisation voltage at selected current densities for the 2 mm and zero
gap cell’s. The data shows that the improved performance from utilising the zero gap arrangement
continues, even above 500 mA·cm−2 .
Ohmic Resistance vs Current Density
3
Zero Gap Cell (Mesh)
2.5 2mm Gap Cell

Membrane Resistance
2
1.5
0.5
0
0 100 200 300 400 500 600
-2
(b) The figure plots the Ohmic resistance at chosen current densities for both cells; and shows the
Ohmic resistance of the 2 mm cell increases faster than the zero gap cell, which has a 18% increase in
Ohmic resistance from 0-625 mA·cm−2 .
Figure 5.8: The cell polarisation and Ohmic resistance results show the benefit of
employing a zero gap cell design, even at high current densities.
current densities; the zero gap cell avoids this issue by forcing the gas bubbles to col-
lect in the flow channels away from the inter-electrode gap. These results demonstrate
that the zero gap arrangement overcomes the ’optimal’ electrode gap demonstrated by
Nagai et al., and sets the modern standard for alkaline electrolyser cell design.117
The previous experiment demonstrates the performance increase possible when

zero gap cell design is used, and as such zero gap cell design will be considered the
only cell design used in the rest of this thesis.
5.4.5 Effect of temperature and electrolyte

Cell design is certainly not the only factor affecting the Ohmic resistance of the cell -
key effects are seen from temperature, choice of electrolyte and choice of membrane.
In this section, the effects of temperature and electrolyte are investigated, such that the
information can be used to guide operating strategies.
Effect of temperature
The effect of temperature on the performance of the zero gap cell was investigated, and
its effect on the Ohmic resistance of the cell was quantified to extract exact benefits
of operating at elevated temperatures. Figure 5.9a shows that the Ohmic resistance
is reduced by 29% when the temperature is increased from 30 ◦ C to 80 ◦ C from 1.19
Ω·cm2 to 0.85 Ω·cm2 from 30 ◦ C to 80 ◦ C respectively. This is attributed to an increase
in the conductivity of the electrolyte with increased temperature. The increase of the
Ohmic resistance with current density remains constant, reflecting the fact the bubble
effect is not greatly influenced by temperature.
The increase in performance is also reflected with improved I-V performance due
to both the reduction of the Ohmic resistance, but also the improved electrode kinetics
at increased temperature reducing the gas evolution overpotentials.
These results offer an insight into the cell performance during the running of a
working electrolyser. Optimal operating conditions of 80 ◦ C are widely used, however
when the system is starting up/shutting down the stack runs at lower temperatures.
Our results demonstrate the key effect that temperature has on the cell performance.
Effect of electrolyte
The choice of electrolyte has a key role in the performance of the electrolysis cell -
Both sodium hydroxide and potassium hydroxide have both been used as electrolyte
in previous studies. Figure 5.9b shows that our results confirm data from previous re-
searchers that potassium hydroxide is more conductive than sodium hydroxide, and
that more concentrate solutions of KOH provide best performance up to 30% wt.
Effect of Cell Temperature on Ohmic Resistance
Ohmic Resistance (Ω·cm 2) 1.2
0.8
0.6
30°C
0.4 40°C
50°C
60°C
0.2 70°C
80°C
0
0 50 100 150 200 250
-2
(a) The figure compares the Ohmic resistance of the zero gap cell at different temperatures, higher temperatures are clearly
favourable due to increase in conductivity of the electrolyte
Effect of Electrolyte Concentration on Ohmic Resistance
1.4
1.2
Resistance (Ω·cm 2)
0.8
0.6
1 M NaOH
0.4 1 M KOH
2 M KOH
0.2 4 M KOH
30 %wt KOH
0
0 50 100 150 200 250
Current Density (mA·cm -2)
(b) Plot demonstrating the reduction in Ohmic resistance possible through choice of electrolyte
Figure 5.9: The cell polarisation and Ohmic resistance results demonstrate the
effect of cell temperature on the performance of the zero gap cell.
KOH. The reduction in Ohmic resistance of the cell can be seen, with the resistance
of a Zirfon membrane tested at 20 ◦C and 30% wt. KOH as low as 0.45 Ω·cm2 .
The safety of using high concentration caustic solution must be considered, as it has
a risk to operators, and ideally lower concentrations of KOH would be used - The
widespread used of alkaline electrolysis in industry suggests that with appropriate
precautions, the safe use of caustic solution can be managed appropriately.
5.4.6 Performance of porous electrodes
A variety of structures, including foams, woven meshes and expanded meshes have
been reported as electrode substrates, although the effect of the structure on the Ohmic
resistance of a full cell has not been explored. Nickel foam is widely used as a sub-
strate in studies of OER and HER electrocatalysts due to its exceptionally high surface
area vs other substrates. As shown in Figure 5.10a, the electrode morphology is seen
to affect the Ohmic resistance of the cell, with the nickel foam exhibiting the lowest
initial Ohmic resistance, 2% lower than the expanded mesh. This can be attributed
to the lower ion conduction resistance due to the small pores and high surface area
of the nickel foam in the immediate vicinity of the gas separator. The coarse woven
mesh demonstrates the largest Ohmic resistance as expected, due to the large average
distance for ion conduction through the electrolyte. These are important result, as it
highlights the effect of porous electrode morphology on the Ohmic resistance of the
cell, a key parameter if alkaline electrolysis is going to be able to compete with the
high current densities currently displayed by PEM electrolysers. Figure 5.10b shows
how the surface area of the electrodes differs between substrate, the foam clearly has
the largest surface area of the 4 substrates.
When operating at high current densities, the chosen foam became clogged with
gas, and showed a steep rise in the cell resistance. This is attributed to both its high
porosity and thickness, providing many locations for bubble entrapment, as well as
unoptimised operating conditions; further increasing the electrolyte flow rate may
Ohmic Resistance vs Current Density

1.5
1
Expanded Mesh
Nickel Foam
0.5 Coarse Woven Mesh
Fine Woven Mesh
Membrane Resistance
0
0 50 100 150 200 250 300
-2
Current Density mA·cm
(a) This plot shows the effect of electrode morphology on the Ohmic resistance. The initial Ohmic
resistance of the cell with nickel foam as the electrodes is 2% lower than the next best of the expanded
mesh, the coarse woven mesh has the lowest initial value. This is attributed to the reduced distance for
ion conduction between the electrodes. The poor performance of the foam cell at high current densities
is ascribed to thickness of the nickel foam, causing gas entrapment in the large quantity of pores.
(b) SEM images of the used substrates including A) Expanded mesh, B) Coarse woven mesh (mesh
#12), C) Fine woven mesh (mesh #40) and D) Nickel foam (40 pores per inch)
Figure 5.10: The effect of the morphology of various electrodes on the Ohmic
resistance of the cell, SEM images provide a visual comparison of electrodes used.
mitigate this increase in Ohmic resistance. Other research has demonstrated effec-
tive integration of nickel foam into laboratory cells,98 indicating that the particular
foam tested in this study has some unwanted characteristics. Woven mesh was also
found show an increase in the Ohmic resistance with current density. The findings
show that overall the expanded mesh is most suited for utilisation as an electrode sub-
strate in the zero gap cell, due to its low Ohmic resistance, increased surface area and
gas management properties.
5.4.7 Interfacial contact resistances

The employment of the zero gap design introduces an added component in terms of
the porous electrode. Characterisation of the interfacial contact resistance between the
bipolar flow plate and the mesh electrode has not been made in the literature for alka-
line cells, yet knowledge of the interfacial contact resistance is useful when targeting
efficiency improvements in the cell. Using the four point probe method, an in-situ
method for measuring the contact resistance was developed and was investigated for
a range of cell compressions by utilising the bolt torque.
Figure 5.11: Experimental setup for in-situ measurement of the interfacial contact
resistance between the porous electrode and the flow plate - The voltage drop was
measured across the junction, with a preset current applied to the cell.
Bolt torque is not an ideal unit to compare against, however investigating the true
force at the mesh/bipolar plate interface is particularly tough, especially in situ, due to
its dependence on factors including the compressibility of the silicone gasket and the
membrane.
The contact resistance for nickel mesh is below 6 mΩ·cm2 for bolt torque above
Contact resistance of mesh - bipolar plate interface

20
Nickel
Contact Resistance (mΩ·cm2 )

Stainless steel 304
15
10
0
0 5 10 15 20
Torque (Nm)
Figure 5.12: Interfacial contact resistance between the mesh and bipolar plate,
plotted against the compression bolt torque
4 Nm, which is lower than the 16 mΩ·cm2 for the stainless steel mesh. Both values
are significantly lower than figures reported for titanium foam/titanium bipolar plate
in PEM electrolysers.130131 The high performance of nickel in this regard appears to
be a feature of the very thin nickel oxide layer that forms on the surface of the metal,
which provides a conductive pathway when compared to the oxides of other materials.
Stainless steel is known to have a fairly thick chromium oxide protective layer, which
may explain its worse relative performance.
Larger bolt torque led to a reduction in the contact resistance, explained by the
increase in contact surface area between the two surfaces under higher pressure, but
the improved performance quickly reaches a plateau.
Bolt torque of above 11 Nm led to deformation of the silicone gasket into the
flow channels - so for future experiments, a torque of 9 Nm was chosen as a balance
between interfacial contact resistance and interference in other aspects of the cells
performance.
5.5. C ONCLUSIONS 83
5.5 Conclusions
Using EIS to decouple the cell resistance and cell voltage, this chapter demonstrates
the clear benefit of employing zero gap cell design over the conventional finite gap ap-
proach, with both a 30% lower initial Ohmic resistance as well as smaller resistance
increases at high current densities. These benefits are reinforced by the cell polari-
sation data with higher efficiencies achieved at higher current densities for the zero
gap cell. The use of porous mesh electrodes allow easy design, fabrication and as-
sembly of the cell, and the results show that the existence of the optimal electrode gap
exhibited by Nagai et al. can be overcome by employing a zero gap cell design.117
The influence of the flow rate on the cell resistance is shown, with increased flow
rate causing a decrease in Ohmic resistance, reinforcing results from previous studies.
Furthermore, design parameters for the zero gap cell area investigated, and the
effect of morphology of porous electrodes on the Ohmic resistance is demonstrated,
with foam electrodes reducing the Ohmic resistance of the cell by 2% when compared
to the mesh electrodes at low current densities. The excellent gas management proper-
ties, as well as low Ohmic resistance contributions of expanded mesh electrodes make
them currently the most attractive choice for electrode substrate going forward. This
shows that a necessary area of future research should be focused on the optimisation
of electrode morphology with regards to the Ohmic resistance of the cell, with the aim
of low cell Ohmic resistance and good gas management properties.
Investigations into the interfacial contact resistance of the porous electrode inter-
face with the bipolar place are made, finding nickel to have a lower contact resistance
than stainless steel. The overall contributions to the cell resistance can be considered
as low when compared to that associated with that of the Zirfon separator, with a total
cell contact resistance of less than 8 mΩ·cm2 expected at appropriate cell compres-
sion when taking into account the contact resistance contributions from the anodic and
cathodic sides of the cell.
These results show the potential of zero gap alkaline electrolysis to achieve high
current densities currently only associated with the PEM technology; allowing this
technology to move towards the key target of low cost and highly efficient electrolysis.
Chapter 6
Development of catalysts
6.1 Abstract
In this chapter, the development of electrocatalysts for both the hydrogen evolution
and oxygen evolution reactions is made with a focus on facile one step deposition
methods. The deposits display high electrocatalytic performance, good adhesion to
the substrate and good electrochemical stability at current densities appropriate for
alkaline electrolysis.
The electrocatalyts are characterised by electrochemical methods, with the Tafel
slopes, current-voltage profile and short term stability of the materials investigated, as
well as compositional analysis by EDX and SEM imaging to reveal the mophology
and composition of the catalytic coatings.
Furthermore, a electrochemical model of a full cell is made based on these results,
and the breakdown of the cell contributions reveal that the membrane contributes a
significant amount to the cell voltage, becoming the largest cell contribution above a
current density of 0.8 A·cm−2 .
6.2 Introduction
A key advantage of using alkaline electrolysis is the possibility of using electrocata-
lysts made from low cost materials for both the hydrogen and oxygen evolution reac-
tions. The main criteria for the selection of catalysts include stability to corrosion in
the working conditions of the cell, low cost, and widely available materials - all to be
fulfilled whilst demonstrating a low overpotential. A literature review of suitable ma-

terials has been made in chapter 3, with electrodes based around nickel identified as
being attractive options due to its low cost and stability in the strongly basic environ-
ment. The use of high performing and stable catalysts is key to further investigations
into the zero gap system design, and as such, applying simple deposition procedures to
deposit high performing catalysts onto porous substrates is crucial in the development
of the test cell.
In this chapter, the fabrication of cathodic and anodic electrocatalysts using one
step deposition processes will be undertaken using work from the literature, and they
will be compared to the baseline catalysts of nickel and stainless steel, which are often
still used in commercial electrolysers. Each catalyst will be deposited on a suitable
porous substrate for use in a zero gap cell, with expanded mesh identified as an attrac-
tive option in the previous chapter. The catalysts will be characterised by SEM and
EDX, as well as using linear sweep voltammetry, Tafel plots and chronopotentiometry
to reveal the overpotentials, Tafel slopes and electrochemical stability of the materials.
Finally an electrochemical model will be developed to predict the performance of the
full cell, and reveal the key sources of inefficiencies.
6.2.1 OER Catalyst

The oxygen evolution reaction overpotential has a large contribution to the overall
cell voltage, with pure nickel demonstrating an overpotential of 400 mV. Preliminary
tests on nickel mesh showed that the overpotential increased a further 50 mV within
5 minutes of electrolysis, seemingly due to a passivation on the surface of the nickel,
and this phenomenon affected accurate investigations of the cell. Based on the litera-
ture review, catalysts based around nickel and iron are attractive options due to their
reported synergistic relationship.
A stable and efficient option was identified through using the incorporation of
iron into nickel hydroxide. Previous work using the one step electrodeposition of
the nickel-iron oxyhydroxide from a nitrate solution was adapted and utilised. The
optimal ratio of each material has not been agreed in the literature, with Li finding
that an optimal ratio was around 20 % iron in the electrolyte solution, whereas Lu
found the optimal molar ratio to be 1:1. To address this discrepancy, investigations
into the optimal composition of the Ni-Fe(OH)2 will be undertaken by comparing
electrochemical performance to the elemental composition of the deposits.74, 75
The co-deposition of a mixed metal oxyhydroxide occurs in the presence of

metal-nitrate salts where the nitrate is reduced at the surface of the metal substrate
such that:
NO− − −
3 + H2 O + 2e − − > NO2 + 2OH
−
(6.1)
M 2+ + 2OH − − > M(OH)2 (6.2)
The standard redox potentials of Ni2+ and Fe2+ area -0.25 V and -0.44 V vs RHE
respectively, and experiments were performed to optimise the deposition potential of
the catalyst. It was found that depositions a potentials too close to the redox potential
of Fe resulted in poor adhesion of the deposit to the electrode substrate. On the other
hand, depositions at potentials below 1.4V resulted in high rates of hydrogen evolu-
tion which again disrupted the quality of the deposit. Therefore, an optimal deposition
potential of -1 V vs RHE provided the optimal conditions. Nonetheless, using a poten-
tial for deposition causes difficulty in making the quantity of the deposition constant
between samples, and so a current density became the key parameter to achieve an
optimal deposit. A deposition current density of -250 mAcm2 was found to provide
a strong balance between adhesion of the deposit and the equal deposition of Nickel
and Iron. This current density was achieved at a potential of between -1 and -1.1 V,
the variation attributed to the changing Ohmic resistance of the solution due to bubble
effects, and the change in mass transfer resistance due to the depletion of salts in the
vicinity of the electrode surface.
6.3. M ATERIALS AND METHODS 88
6.2.2 HER Catalyst

During preliminary testing of the zero gap cell, it became clear that stainless steel
electrodes became discoloured when being used as a cathode, suggesting corrosion in
the cell. The literature suggests that Raney nickel is an effective and cheap catalyst.
The incorporation of molybdenum into the Raney nickel deposit has previously been
suggested to increase performance, but recent work by Schalenbach attribute this ef-
fect to the high surface area structures obtained when depositing Ni-Mo, rather than
any intrinic activity of the Mo itself132 .
The depostion of Raney nickel is achieved by leaching the aluminium or zinc
content from a precursor alloy (NiAl or NiZn) using a strong basic solution. The high
activity is due to the large increase in surface area versus smooth nickel. Interestingly
- the effect of the counter electrode in the morphology of the electrode deposit has
recently been observed by Gannon et al. as a result, Raney nickel deposits using both
the traditional Pt counter electrode, and a stainless steel counter electrode
6.3 Materials and methods

6.3.1 Materials for OER Catalyst
Nickel(ii) nitrate (Ni(NO3 )2 , Sigma Aldrich 99%), iron nitrate (Fe(NO3 )3 , Sigma
Aldrich 98%) and ammonium nitrate (NH4 NO3 ,Sigma Aldrich 98%) were used as
recieved. All aqueous solutions were prepared using deionised water (15 MΩ) from
a Merck Millipore water purification system. An Ag/AgCl (1M KCL) reference elec-
trode was used for all of the depositions and the electrochemical characterisation,
which were performed in a small beaker (volume 80 cm3 ), with a platinum mesh
counter electrode.
6.3.2 Materials for HER Catalyst

Nickel(ii) sulphate (Ni(SO4 )2 Sigma Aldrich 99%), nickel chloride (NiCl, Sigma
Aldrich 99%), zinc chloride (ZnCl2 , Sigma Aldrich 98%), boric acid (H3 BO3 ,Sigma
Aldrich 99.5%) and potassium hydroxice (KOH, Sigma Aldrich, 85%) were used
as received. All aqueous solutions were prepared using deionised water (15 MΩ)
from a Merck Millipore water purification system. An Ag/AgCl (1M KCl) reference
electrode was used for both the depositions and the electrochemical characterisation,
which were both performed in a small beaker (volume 80 cm3 ), with a platinum mesh
counter electrode, a stainless steel counter electrode was also used for one of the Raney
nickel depositions.
6.3.3 Deposition method for OER catalyst
Pieces of nickel mesh (Dexmet) were cut into 1 cm2 samples, sonicated for 20 min
each in acetone, ethanol and hydrochloric acid (5 M) before being rinsed thoroughly
in deionised water.
The cathodic deposition was made at 20 ± 2 ◦C, from a mixed nitrate solution
with a constant total of 18 mM of transition metals along with 25 mM ammounium
nitrate. The ratio of nickel to iron salts was varied from 100:0 to 50:50, and a constant
pH of 3 achieved using nitric acid where required. Galvanostatic electrodeposition
was used with a inital current density of -250 mAcm2 , based on previous methods.74
6.3.4 Deposition method for HER Catalyst
Pieces of nickel mesh (Dexmet) were cut into 1 cm2 samples, sonicated for 20 min
each in acetone, ethanol and hydrochloric acid (5 M) before being rinsed thoroughly
in deionised water.
The cathodic deposition was made at 50 ± 2 ◦C, from a modified watts bath
containing 330 g/l NiSO4 , 45 g/l NiCl, 37 g/l H3 BO3 , 20g/l ZnCl2 . Galvanostatic
electrodeposition was used at -50 mAcm2 for 3600 s.74
The electrode was then placed in a 200 cm3 PTFE beaker of 4M KOH, and held
at 50 ◦ C for 48 hours. The electrode was then rinsed thoroughly in deionised water.
6.3.5 Full cell testing

To test the catalysts in a working test cell, the deposition methods were scaled up.
Pieces of nickel mesh (Dexmet) were cut into 10 cm2 square samples, which under-
went the same preparation procedure of sonication for 20 min each in acetone, ethanol
and hydrochloric acid (5 M) before being rinsed thoroughly in deionised water.
The cathodic and anodic depositions were made using the same protocols as
above, with the depositions made in a larger 200 ml beaker, and appropriate adjust-
ments to the molar concentrations.
Flow plates made from nickel were fabricated, machining a 10 cm2 serpentine
flow field into the plates, with channels (1 mm deep, 1.5 mm wide) into plates (1.5
mm thick). The gas separator used was Zirfon Perl UTP 500 (Agfa). Silicone gaskets
of thickness 0.25 mm were used to seal the cell under compression. Electrolyte of 30%
wt. KOH made up, and the same test station was used as in chapter 5. The electrolyte
was preheated in a Fisher FB15053 temperature bath, with the cell temperature of
40◦ C maintained by monitoring the cell temperature using a Elitech RC-5 temperature
probe, and adjusting the temperature bath as needed. Electrolyte flow was maintained
at 100 ml·min−1 throughout the experiment.
6.3.6 Characterisation methods

The surface morphology and composition of the deposited films was investigated us-
ing SEM and EDX. EDX was performed for at least 3 sites on the electrode, and the
numbers averaged to give the amount of material observed in the film.
Elecrochemical analysis was undertaken in a 200 cm3 PTFE beaker filled with 1
M NaOH electrolyte. A 3-electrode setup was used, with the electrode under investi-
gation used as the working electrode, an Ag/AgCl reference electrode and a Pt mesh
counter electrode. The Ag/AgCl reference electrode was regularly checked against a
saturated calomel electrode to ensure stability in the basic environment.
The Ohmic resistance of each setup (Rs ) was determined using EIS. A signal am-
plitude of 10 mA was applied onto a fixed current of 50 mA·cm−2 , across a frequency
range of 100 kHz to 1 Hz. Rs was equated to the impedance at high frequencies, when
the phase angle is closest to zero,129 and each investigation was IRS corrected in the
post processing of the results.
For the OER catalysts, linear sweep voltammetry was performed by scanning
from 0.1 V to 0.9 V vs Ag/AgCl at a scan rate of 5 mVs−1 . Tafel analysis was obtained
by holding fixed currents for 10 s with 5 currents per decade, from 1 mAcm−2 to
1000 mAcm−2 and recording the voltage. Chronopotentiometry was undertaken to
investigate any obvious corrosion of the electrodes performance and was done at 100
mA·cm−2 .
For the HER catalysts the linear sweep voltammetry were obtained by scanning
from -0.9 V to -1.4 V vs Ag/AgCl at a scan rate of 5 mVs−1 . Tafel analysis was
obtained by holding fixed currents for 10 s with 5 currents per decade, from -1 mA·cm2
to -1000 mA·cm2 and recording the voltage. Chronopotentiometry was undertaken to
investigate any obvious corrosion of the electrodes performance and was done at -100
mA·cm2 .
For the full cell, a Elektro Automatik Power supply (20A, 30V) was used, the
cell was run for 10 minutes at 0.5 A·cm2 before currents between 0 - 2 A·cm2 were
chosen.
6.4.1 Oxygen evolution catalysts
The SEM images in Figure 6.1 show clear differences in the morphology of the de-
positions depending on the ratio of nickel and iron in the deposition bath - higher
amounts of nickel correlate to smaller islands of deposit, and better adherence to the
mesh substrate. The deposit from the 100 % nickel bath shows the smallest islands of
deposition. The growth of islands matches well with previous work,114 with multiple
layers of deposit seen in the coatings from the bath containing small amounts of iron.
Figure 6.1: SEM images (x300) for nickel-iron depositions from a bath of a) 100:0,
b) 90:10, c) 80:20, d) 70:30, e) 60:40 and f) 50:50
Further magnification of the SEM images in Figure 6.2 shows more clearly the
increase in size of the deposition ’flakes’, both deposits with 90:10 and 80:20 Ni:Fe
show layers of flakes on top of an adherent base film, whereas 50:50 Ni:Fe shows poor
adherence of the large deposition flakes.
Bath composition Film composition

Overpotential at Tafel Slope
(%) (%)
10mA·cm−2 (mV) (mV·dec−1 )
Nickel Iron Nickel Iron
100 0 100 0 402 55
90 10 73 27 280 37
80 20 67 33 292 43
70 30 52 49 296 44
60 40 34 66 303 40
50 50 25 72 317 39
Pure nickel N/A 398 54
Table 6.1: Composition and performance of Ni-Fe electrode
Figure 6.2: SEM images (x1800) for nickel-iron depositions from a bath of a)
100:0, b) 90:10, c) 80:20, d) 70:30, e) 60:40 and f) 50:50
Elemental analysis of the deposited coating by EDX showed that the elemental
makeup of the coatings, as shown in Table 6.1. Even small amounts of iron in the
deposition bath resulted in significant amounts of iron in the film composition. When
the bath compostion was 50:50, the film contained over 70 % iron.
Linear sweep voltammetry of the substrates plotted in Figure 6.3 shows a clear
increase in performance when iron is incorporated into the film, and that too much
iron in the film causes the performance to decrease somewhat; optimal performance
I-V Characteristics of Ni-Fe Coated Electrodes
0.1 NiFe(OH) - 90:10

2
NiFe(OH) 2 - 80:20
Current Density (A·cm -2)

NiFe(OH) 2 - 70:30
0.08
NiFe(OH) 2 - 60:40
NiFe(OH) 2 - 50:50
0.06 Ni(OH)
2
Pure Nickel
0.04
0.02
0
0 0.1 0.2 0.3 0.4 0.5
Overpotential / V
Figure 6.3: Linear sweep voltammetry recorded in 1 M NaOH at 20 ◦ for mixed

Ni-Fe(OH)2 layers deposited from a nitrate solution containing the ration of nickel
and iron salts as specified
appears to be for the deposit from a bath of 90:10 Ni:Fe, corresponding to a film ratio
of 73:27. There is an observed redox peak just before the onset of oxygen generation,
and this is accompanied by a change in colour of the deposit from a grey/red brown
to a black. This corresponds to a change in the oxidation state of nickel from Ni2+ to
Ni3+ .
The Tafel slopes of the substrates were calculated from the region of 10 mA·cm−2
to 100mA·cm−2 and are all in the range 37-44 mV·dec−1 for all deposits containing
iron, and increasing to 54 for the samples with just nickel. This is in line with the
theory which, states that Tafel slopes for the oxygen evolution reaction are expected
to be 30, 40 or 120 mV·dec−1 depending on the rate determining electron transfer
reaction. The Tafel slopes were calculated between 10 mA·cm−2 to 100 mA·cm−2 to
minimise the effect of gas evolution on the figures. As seen in Figure 6.4, there is an
increase of all the Tafel slopes towards 1 A·cm−2 , and this is attributed to the increased
rate of gas evolution causing an increase in Ohmic resistance between the working and
reference electrode, rather than a significant change to the real Tafel slopes.
Tafel Plot for Ni-Fe Electrodes of Varying Compositions
NiFe(OH) - 90:10
2
0.6
NiFe(OH) 2 - 80:20
0.55 NiFe(OH) 2 - 70:30
Overpotential / V
NiFe(OH) 2 - 60:40
0.5 NiFe(OH) 2 - 50:50
Ni(OH)
2
0.45
Pure Nickel
0.4
0.35
0.3
0.25
10-3 10-2 10-1 100
Current Density (A·cm -2)
Figure 6.4: Tafel plot of the deposited electrodes, the addition of iron to nickel
hydroxide demonstrates increased performance, with an optimal bath ratio of Ni:Fe
appearing to be 90:10
Chronopotentiometry was employed to investigate the stability of the substrates

under real life electrolysis conditions. The short scan time allowed the materials to be
quickly assessed for suitability. As seen in Figure 6.5, all deposits show good stability,
with the Ni(OH)2 deposition actually improving in performance across the scan, this is
likely due to the incorporation of ’incidental iron’ into the surface layer. The increase
in overpotential of the nickel sample can be observed, and such short term changes
can have a large impact on not only the performance of the electrode, but also on the
ability to perform precise studies on the effects of operating parameters on the full
cell. Overall, the Ni-Fe deposits show good stability, and are considered suitable for
use is a zero gap cell working at high current densities.
Chronopotentiometry of Ni-Fe Coated Electrodes

0.5
0.45
0.4
Overpotential (V) 0.35

NiFe(OH) 2 - 90:10
0.3
NiFe(OH) - 80:20
2
0.25 NiFe(OH) - 70:30

2
NiFe(OH) 2 - 60:40
0.2 NiFe(OH) 2 - 50:50
Ni(OH) 2
0.15
Pure Nickel
0.1
0 50 100 150 200 250 300
Time (s)
Figure 6.5: Chronopotentiometry was performed at 0.1 A·cm−2
Overall, the higher the amount of iron in the deposition, the larger the deposit,
and the worse the adhesion. This, coupled with the poor electrochemical performance,
shows that too much iron has a negative effect on the performance of the catalyst,
equally, the deposit with 100 % nickel hydroxide showed performance similar to pure
nickel. This investigation demonstrates that a film composition ration of 73:27 is the
highest performing coating, with good adhesion, stability and high electrocatalytic ac-
tivity making it a good choice for the anodic catalyst in a zero gap alkaline electrolysis
cell.
6.4.2 HER Catalyst

The two depositions of Raney nickel are compared with stainless steel and nickel
electrodes, which are still used in commercial systems. For simplicity, the two Raney
nickel deposits will herein be labelled as follows: The deposit made using a stainless
steel counter electrode will be referred to as Raney nickel 1, and the deposit made
using a platinum counter electrode will be referred to as Raney nickel 2. Figure 6.7
shows the substantial performance benefit of using high surface area nickel coatings,
with voltage savings of over 200 mV at 100 mA· cm−2 when compared to pure nickel.
Figure 6.6: SEM images at a) x1800 magnification, b) x300 for the Raney nickel
deposition with stainless steel counter electrode and at c) x1800, d) x300 for the
Raney nickel deposition with platinum counter electrode, and at e) x1800 and f)
x300 for pure nickel
The difference between the individual Raney nickel coatings is surprising, with the
small difference of counter electrode having a large effect on the electrochemical per-
formance. The use of the stainless steel counter electrodes gives a further voltage
saving of 40 mV at 100 mA· cm−2 when compared to using a platinum counter elec-
trode.
The high magnification SEM imaging in Figure 6.6 shows that the morphology
of the Raney nickel 1 has high surface area spherical features, this is compared with
the Raney nickel 2 deposit which has cracks and crevices but a more uniform surface
structure. Pure nickel provides a good reference, with its low surface area an obvious
reason for its poor performance. The lower magnification images in the same figure
again show finer features for the Raney nickel 1, with particular difference seen at the
edges of the mesh, where high surface area deposits have grown.
Table 6.2 shows that the EDX results taken from the sample deposited with the
stainless steel counter electrode have a relatively high integration of chromium into
I-V Characteristics of Ni-Fe Coated Electrodes

0
Pure Nickel
-0.01 SS 304
Raney Nickel
-0.02
Current Density (A·cm-2)
-0.03
-0.04
-0.05
-0.06
-0.07
-0.08
-0.09
-0.1
-0.5 -0.4 -0.3 -0.2 -0.1 0
Overpotential / V
Figure 6.7: Linear sweep voltammetry recorded in 1 M NaOH at 20 ◦ for the hy-
drogen evolution reaction on Raney nickel, pure nickel and stainless steel electrodes
Coating Ni Zn Cr Fe
Raney nickel 1 72.2 25.7 2.1 0
Raney nickel 2 59.1 40.0 0 0.9
Table 6.2: EDX results for Raney nickel coatings, the incorporation of chromium
into the surface layer is clear for the coating using a stainless steel counter electrode.
the sample - with approximately 2 % of the surface layer. This is compared with the
sample deposited using a platinum counter electrode which had only a small contam-
ination of iron, and no chromium integration. This incidental iron is a similar finding
to other reports as an incidental finding in other studies.133 SEM results shown in
Figure 6.6 show the difference in morphology of the two depositions. This suggests
that the chromium affects the morphology of the deposit. This finding matches well
with a recent investigation by Schalenchach - finding that the high activity of Raney
nickel-molybdenum coatings were down to the morphology of the deposit rather than
any intrinsic activity benefit added by the molybdenum.132
Figure 6.8 shows that the Raney nickel coatings have lower Tafel slopes than
both pure nickel and stainless steel electrodes, with Raney nickel 1 having a Tafel
Tafel Plot for Nickel Electrodes (HER)

0.1
-0.1
-0.2
Overpotential / V
-0.3
-0.4
-0.5
-0.6
Pure Nickel
-0.7 Raney Nickel
SS 304
-0.8
10 -4 10 -3 10 -2 10 -1 10 0
Current Density (A·cm-2)
Figure 6.8: Tafel plot of the Raney nickel, pure nickel and stainless steel electrodes
Tafel Slope
Material Overpotential at 10 mA·cm−2 (mV)
( mV·dec−1 )
Raney nickel 1 93 62
Raney nickel 1 135 66
Pure nickel 316 97
Stainless steel 520 120
Table 6.3: Performance of Raney nickel electrode compared with baseline materi-
als
slope of 62 mV·dec−1 , and Raney nickel 2 of 66 mV·dec−1 . All Tafel slopes are fairly
consistent across the whole current density range, although the rapid convergence
of the Raney nickel slopes with zero overpotential suggests that there could still be
some electrochemical contributions from the zinc that was still evident in the deposit
on the EDX results. Nevertheless, the results for the Raney nickel deposits match
well with literature values, with overpotentials of 92 mV and 135 mV repectively
at 10 mA·cm−2 , this is compared with pure nickel and stainless steel which have
overpotentials of 316 mV and 520 mV at the same current density. Table 6.3 shows
the performance metrics for the four catalysts, with the wide variation of the Tafel
slopes associated with the high surface area of the two Raney nickel electrodes, which
have more reaction sites, allowing higher reaction rates for the same geometric surface
area.
Chronopotentiometry of HER Electrodes

0
Raney Nickel
Pure Nickel
-0.1
SS 304
Overpotential (V)
-0.2
-0.3
-0.4
-0.5
-0.6
0 50 100 150 200 250 300
Time (s)
Figure 6.9: Chronopotentiometry was performed at -0.1 A·cm−2 for 300s to check
for the stability of the materials in such conditions.
Chronopotentiometry results in Figure 6.9 show that nickel and stainless steel
have an initial further decrease in performance, this decrease was seen to reverse when
the cell was turned off and experiment repeated. The Raney nickel deposits show good
stability in the same conditions, although there is a decrease in performance of Raney
nickel 1 by 9 mV over the 300s timeframe, the overpotential of Raney nickel 2 stays
consistent across the same timeframe.
Overall the employment of Raney nickel electrodes show a great reduction in

overpotential when compared to basic choices such as nickel and stainless steel. The
effect of chromium in the Raney nickel deposition appears to have positive effects
in terms of the morphology and electrochemical performance of the deposit. The
hydrogen evolution catalysts show far lower overpotentials than the oxygen evolution
counterparts characterised in the previous section, and it is believed this is due to the
faster kinetics for the 2 electron transfer hydrogen evolution reaction when compared
to more complex 4 electron transfer of the oxygen evolution.
6.5. OVERALL CELL MODEL 101
6.5 Overall cell model

Creating an electrochemical model of the zero gap cell offers an insight into the rela-
tive contributions towards the cell voltage and will provide a useful tool when select-
ing key areas to target when aiming to increase the efficiency of a zero gap alkaline
electrolysis cell working at 2 A·cm−2 .
The performance characteristics obtained from the catalytic materials in this

chapter can be combined with the membrane resistance investigations in the previ-
ous chapter to provide the necessary variables to complete a basic model.
In chapter 2, the actual cell voltage was demonstrated as:
V = 1.23 + ηO2 + ηO2 + I · RCell + ηMT (6.3)
As the overpotential associated with mass transport is still poorly understood, the
model will exclude contributions from this, such that the simplified equation is:
V = 1.23 + ηO2 + ηO2 + I · RCell (6.4)
Bringing together the information on Tafel slopes gathered earlier, the cell volt-
age from the model is
I I
V = 1.23 + bH2 · log10 ( ) + bO2 · log10 ( ) + I · RMembrane (6.5)
I0H2 I0H2
where bH2 and bO2 are the Tafel slopes of the cathodic and anodic electrocata-
lysts, I0H2 and I0O2 are the exchange current densities of the cathodic and anodic elec-
trocatalysts, and RMembrane is the resistance of Zirfon. I is the applied current. The
polarisation data for the model is displayed in figure 6.10.
6.5.1 Comparison with experimental data

The modelled cell voltage is based on a cell operating at 40◦ C and using 30% wt.
KOH as the electrolyte. The properties of the cell are listed in Table 6.4, taken from
Membrane Tafel Slope (mV·dec−1 ) I0 (A·cm−2 ) RMembrane (Ω· cm−2 )

Raney nickel 62 6x10−4 n/a
Ni-Fe(OH)2 37 9x10−10 n/a
Zirfon n/a n/a 0.43
Table 6.4: Electrochemical properties of the materials used in the zero gap alkaline
cell, assuming 40◦ C and 30 % wt. KOH
the properties experimentally obtained earlier in this thesis. The modelled cell perfor-
mance is shown in Figure 6.10. The modelled cell reaches 10 mA·cm−2 at a voltage
of 1.59 V, and 100 mA·cm−2 at 1.73 V which is considered to be strong performance
when compared to literature values. At 500 mA·cm−2 , the modelled cell voltage will
be 1.96 V, and 1 A·cm−2 at 2.21 V - operating a cell at above 2.2 V constitutes a
voltage efficiency of less than 70 % based on the HHV of hydrogen, which towards
the lower end of acceptable.
A 10 cm2 zero gap cell was assembled based on the cell from chapter 5, with
the Raney nickel and nickel-iron oxyhydroxide catalysts deposited onto the larger
substrates. The polarisation performance of the cell was characterised at 40◦ C, and
it can be seen in figure 6.10 that the performance of the test is notably worse than
the model, although it is still within 150 mV at 1 A·cm2 . The difference between the
modelled and actual data can be attributed to three main reasons, firstly, the scaling up
of the electrode depositions is unlikely to produce homogeneous deposits across the
larger area of the electrode, leading to a decreased electrochemical performance when
compared to the small area sample that the model is based on. This would be most
evident at low current densities, and it can be seen that even at 100 mA·cm2 , the real
cell voltage is higher than the model by 60 mV. Secondly, the model treats the Ohmic
resistance as a constant value, independent of current density, but as seen in chapter 5,
the Ohmic resistance of the cell is not a constant value, and it is likely that increased
gas evolution leads to higher Ohmic resistance. Finally, at high current densities, it is
likely that there will be a noticeable contribution to the cell voltage associated with
mass transport limitations, however, due to the lack of available data on this topic, this
contribution was not included in the model.

IV cell performance
2000
Full cell performance
Current density (mA·cm -2) Full cell model
1500
1000
500
0
1 1.5 2 2.5 3
Cell voltage (V)
Figure 6.10: The cell model is compared to the experimental results for a zero
gap cell with Raney nickel cathode, Ni-Fe(OH)2 anode, and a Zirfon membrane,
operating at 40 ◦ C and with an electrolyte of 30 % KOH
6.5.2 Cell voltage contributions

To find out the relative contributions to the cell voltage, and identify key contributions
to the high cell voltage at 1 A·cm−2 , visualisation of the relative contributions to
the cell voltage is made in Figure 6.11. At low current densities, the contribution
from the anodic overpotential dominates the cell voltage, with the cathodic equivalent
contributing less than 200 mV below 0.5 A·cm−2 . The Ohmic contribution to the cell
voltage from the membrane becomes the largest at 0.8 A·cm−2 , and contributes 45%
of the voltage at 1 A·cm−2 . If the cell was to be run to 2 A·cm−2 , then the Ohmic
overpotential would dominate the cell voltage with a 60% contribution, these results
demonstrate the importance of reducing the Ohmic resistance of the cell if the target
of running efficient alkaline eletrolysis cells above 1 A·cm−2 is going to be realised.
6.6. C ONCLUSION 104
Cell voltage contributions

1
Membrane
Cathodic overpotential
0.8 Anodic overpotential
Voltage contribution (V)
0.6
0.4
0.2
0
0 500 1000 1500 2000
-2
Current density (mA·cm )
Figure 6.11: The relative contributions to the cell voltage from the modelled elec-
trocatalytic and Ohmic overpotentials are shown from 0-2 A·cm−2
6.6 Conclusion
In this chapter, electrocatalysts for both the hydrogen evolution and oxygen evolution
reactions have been developed, displaying high electrocatalytic performance, good
adhesion to the substrate and good electrochemical stability at current densities ap-
propriate for alklaine electrolysis.
The use of nickel-iron oxyhydroxide has been shown to be an effective oxygen

evolution electrocatalyst, with an overpotential as low as 280 mV at 10 mA·cm−2 for
a deposit containing 73 % nickel and 27 % iron, which appears to be the optimal ratio.
The deposit had a Tafel slope of just 37 mV·dec−1 .
The use of Raney nickel has been shown to be an effective hydrogen evolution
electrocatalyst, with an overpotential of 93 mV at 10 mA·cm−2 . The effect of the
counter electrode on the morphology of the deposit has been demonstrated, with the
appearence of chromium in the deposit appearing to cause a ball shapes on the surface
of the deposit, thought to provide extra surface area. The Raney nickel electode has a
Tafel slope of 62 mV·dec−1 .
6.6. C ONCLUSION 105
The incorporation of these catalysts into a full zero gap cell allows the cell to
reach 1 A·cm−2 at 2.38 V using a 30 % wt KOH electrolyte and a cell temperature
of 40◦ , with the cell voltage remaining below 3 V at 2 A·cm−2 , this cell provides a
good baseline performance for which to improve on. The performance of the cell is
compared the the expected performance, modelled using the performance character-
istics of the electrocatalysts and membrane characterised earlier in the chapter, with
the performance of the working cell notably inferior to the modelled data. The model
shows that above 800 mA·cm−2 , the Zirfon membrane dominates the contributions to
the cell voltage, and presents a key challenge for working at current densities of 1-2
A·cm−2 . The visualisation of the cell contributions demonstrates the key areas to be
targeted when aiming to increase the efficiency of the zero gap cell at high current
densities, and this serves as a good tool to guide the rest of the thesis.
Chapter 7
Characterisation of two phase flow
7.1 Abstract
Optical imaging and mathematical modelling techniques are used to characterise the
two phase flow at the cathode of a small zero gap alkaline electrolysis test cell. Two
separate mathematical models predict the flow regime changes, and high speed pho-
tography is used to verify the models with experimental data.
The bubbly and slug flow regimes are visualised using bespoke flowfield designs
and a high specification camera to take images at 1000 frames per second. An inno-
vative approach to image processing is made by using Matlab to distinguish between
the bubbly and slug flow regimes and provide quantitative information on the transfor-
mation between the two flow regimes. A comparison between serpentine and parallel
flowfields is made with regards to their effect on the flow regime inside the cells.
Furthermore, the effect of gas evolution and electrolyte flow on the Ohmic re-
sistance of the cell is investigated. It is shown that gas evolution has a small but
significant effect on the Ohmic resistance of a zero gap cell.
7.2 Introduction
The importance of the development of electrolysers capable of operating at high cur-
rent densities has been highlighted in chapters 5 and 6, with the aim of lowering the
CAPEX of the stack, as well as maximising power to weight and power to volume
ratios. PEM electrolysers have been reported with current densities of 2 A·cm−2 at
1.8 V, with current densities as high as 19 A·cm−2 at a voltage below 3 V.134 The em-
ployment of zero gap cell design has helped alkaline electrolysis cells move towards
current densities of 1 A·cm−2 at voltages of approximately 2V, but to aid the improve-
ment of the alkaline technology beyond this value, the effects of these high current
densities and associated high rates of gas evolution on the alkaline cell must be un-
derstood.21, 23, 65, 103 To understand the effect of gas evolution on the electrochemical
performance of the alkaline cell, a fundamental understanding of the transport asso-
ciated with two phase flow inside the electrolyser must be achieved, and the effects
on the cell efficiency isolated. This area of electrolysis research lacks information in
the literature - particularly when compared to catalyst development, and it is for this
reason that the research will be pursued further.103
Studies into two phase flow have been centred around PEM electrolysers, due to
their low cell resistance permitting cells to be run at high current densities. The simi-
larities between PEM cell design and that of zero gap alkaline electrolysers means that
almost all of this information can be applied to this technology. AEM electrolysers
run in a similar manner and can also apply the information gained from these stud-
ies. A key difference between the technologies exists, however, since liquid alkaline
electrolysers have liquid feeds to both the anode and cathode, meaning that investiga-
tions regarding two-phase flow must also be extended to hydrogen evolution. Since
hydrogen is produced at a molar ratio of 2:1 with oxygen, a liquid alkaline cell will
experience the same effects as PEM from oxygen evolution at the anode, as well as
double the effect from hydrogen evolution at the cathode.
Investigations into the effects of gas evolution in a zero gap alkaline electrolysis
cell are yet to feature in the literature. This chapter seeks to characterise gas evolu-
tion at the cathode of an alkaline cell in a number of ways. Firstly, two mathematical
models will be employed to predict the two phase flow quality inside the flow chan-
nel of a zero gap cell. Secondly, high speed photography will be used to validate the
two models, particularly with regards to the transition between flow regimes. Thirdly,
Matlab will be utilised to process the images from the photography, and to provide
quantitative data on the flow quality across the full length of the flow channel. Fi-
nally the effect of gas evolution on the Ohmic resistance of the zero gap cell will be
investigated.
7.2.1 Background to Two Phase Flow
Two phase flow is defined as the simultaneous flow of a gas and a liquid. The flow
regime of a two phase flow depends on the flow quality, mass flux and superficial
velocities of the gas and liquid phases. Figure 7.1 shows the four main flow regimes;
bubbly flow, slug flow, churn flow, and annular flow. The first flow regime transition
to occur is that from bubbly to slug flow, and this will be the focus of the research.
Figure 7.1: The two phase flow regimes, the transition between the bubbly and slug
regimes will be the focus of this study135 ,136
Initial investigations into two phase flow was done by Radovcich and Moissis,
postulating that the initialisation of slug flow was due to the collision and agglomer-
ation of small bubbles associated with bubbly flow.137 A fraction of these collisions
result in the coalescence of the bubbles into a cap bubble, which grows with further
collisions until it reaches a similar diameter to the channel in which it is flowing.
Radovcich and Moissis showed quantatively that the probability of collision becomes
very large at a void fraction of 0.25-0.3, where they observed a flow regime transi-
tion from bubbly to slug flow. The effect of the width of the channel was introduced
as a key parameter in the void fraction at which the bubbly to slug flow transition
occurs.138
Mishima and Ishii investigated two phase flow in a circular channel, demonstrat-
ing that the flow regime transitions could be more accurately categorised by consid-
ering the velocity of each phase as if there was no other phase present, and these
velocities were labelled the superficial air and liquid velocities.139 Mishima charted
the transitions from bubbly to slug as well as slug to annular and slug to churn. Ito et
al. applied this fundamental two phase flow theory to the flow channel inside an elec-
trolysis cell, comparing the flow regime at the exit of the cell to the regime predicted
by Mishima, reporting the accuracy of the predictions.122, 123
7.2.2 Electrochemical Effects

The quality of the two phase flow is certain to have impacts on two areas of effi-
ciency losses in an electrolysis cell: Ohmic resistance and mass transport resistance.
Increases in Ohmic resistance can occur when the membrane is not fully hydrated,
or when bubbles fill areas within the interelectrode gap. Mass transport limitations
are due to the starvation of reagents at catalytic sites, and are likely to occur at high
current densities and low flow rates.
Ito et al. applied this fundamental knowledge to the flow channel of a PEM
electrolyser, and demonstrated that the choice of flowfield has a large impact on the
flow regime across the current density spectrum. It was noted that a parallel flow field
had favourable characteristics to inhibiting the transition from bubbly to slug flow.122
Favourable cell polarisation performance was also found for the parallel flow field,
and attributed to this effect, although the mass transport resistance was not isolated to
confirm this. Indeed, isolating the mass transport resistance proves to be an unresolved
issue, with multiple methods being proposed, but no universally accepted route to
achieving a value. Dedigama et al. used an optically transparent flow channel to
observe the flow properties inside a PEM electrolysis cell. It was proposed that using
the low frequency spectrum of EIS, the mass transport resistance of the cell could
be isolated, and results suggested that enhanced mass transport was observed in the
presence of slug flow. Van de Merve et al. also suggested that using the low frequency
tail on EIS spectra demonstrated the presence of mass transport limitation at high
current densities, however, despite these assertions, no quantified value was attributed
to the mass transport resistance.106
Indeed, during preliminary testing for this chapter, the mass transport resistance
of the cell was investigated by performing EIS and assessing the low frequency do-
main of the Nyquist plot, as suggested the researchers above. It was seen that there
was substantial noise in the spectrum below 5Hz. This is a feature of the characteristic
frequency of bubble release from the the surface of the electrode, and makes getting
meaningful information regarding a quantantive value difficult. It was observed that
there was relationship between the surface roughness and the frequency at which the
noise featured on the EIS spectrum, and this is believed to be due to the fact that bub-
bles are released at smaller volumes from the rough surface, increasing the frequency
at which bubble release occurs.
Seweryn et al. used neutron scattering to quantify water to gas ratio in each chan-
nel of the flowfield, showing increased presence of gas at lower water flow rates.140
Selamet coupled neutron scattering with visual analysis to analyse a multi-mesh flow
field, observing the build up of gas at locations of stagnant flow.141 Sun et al. in-
vestigated the performance of a PEM cell under water starvation, showing that at low
water stoichiometry, the performance of the cell is affected due to water starvation
mass transport resistance is noted alongside and obvious increase in Ohmic resistance,
attributed to the drying out of the membrane.142
Babic et al. highlighted the lack of reports targeted at understanding the two
phase flow of electrolysers.103 This differs substantially from equivalent research for
PEM fuel cells, where a large bank of publications is available using visual imaging,
neutron scattering and MRI imaging.143–145 It is observed that a key drawback of

visualisation cells is the lack of quantitative information available, as the true volumes
of films, slugs and bubbles cannot be extracted.
Certainly, there is still debate as to whether slug flow has a positive or negative
effect on the cells efficiency. Modelling by Aubras et al. back up the suggestion that
slug flow is associated with enhanced mass transport inside the cell, whereas Ito et
al. were straightforward in their assertion that there was increased electrochemical
performance of a parallel flow plate over it serpentine counterpart. This was attributed
to the inhibition of slug flow in the parallel flow field from the low superficial liquid
velocities.146 The bubble effects in the cell are not isolated to the two phase flow
in the flow channel, recent investigations by Mo et al. visualised the microfluidics
associated with the pores of the gas diffusion layer for PEM electrolysis.147 The
interaction of the macroscopic two phase flow and the microfluidics of the pores is
still poorly understood, and further work is needed to guide optimisation of operating
parameters associated with these phenomena.
This study aims to verify the criteria for the bubbly to slug flow transition by
using a zero gap alkaline visualisation cell, and use data from high speed image anal-
ysis to shed light on how the transition proceeds for hydrogen evolution. The effects
of flow rate and current density will be investigated, and compared to the theoretical
values proposed by Mishima and Ishii. Furthermore, the effects of current density and
flow rate on the Ohmic resistance of cell will be investigated.
In this study, both serpentine and parallel flowfields are chosen for study, as
shown in figure 7.3, in line with the types of flowfield reported in the literature. The
serpentine arrangement offers the simplicity of a single channel, with multiple 180◦
turns, whereas the parallel flowfield has been associated with improved cell perfor-
mance.122
7.3. M ETHODS 112
7.3 Methods
7.3.1 Visualisation cell
To visualise the gas interactions inside the cell, a electrolysis cell with optical access
was developed by machining both serpentine and parallel flowfields into 1.5 mm thick
nickel sheet as seen in figure 7.3. Both flowfields have an active cell area of 10 cm2 ,
with channels 1.5 mm wide cut through the full thickness of the sheet, leaving islands 1
mm wide, and a channel cross sectional area of 2.25mm2 . The parallel flowfield had an
area engraved to a depth of 1 mm between the electrolyte inlet and the flow channels
to promote uniform flow spreading. As shown in figure 7.2, a transparent PMMA
endplate was compressed onto the flowfield, allowing sufficient cell compression, as
well as optical access.
Figure 7.2: The visualisation cell allows optical access to the two phase mixture in
the flow channels, in this case, the serpentine flow field is setup for study.
Expanded nickel mesh (Dexmet) was cut to squares with an area of 10 cm2 , and
were used as the two porous electrodes, Zirfon UTP was used as the gas separator,
and 0.25 mm silicone gaskets were used to seal the cell. The electrolyte used was 1 M
NaOH and the temperature was maintained at 20 ◦ C. The electrolyte was supplied to
the anodic and cathodic chambers from separate feeds using an Anself 12 V DC pump
7.3. M ETHODS 113
and Dwyer variable area flow meters to control flow speed. The cell was compressed
using coach bolts, tightened using a torque wrench to 9 Nm.
Figure 7.3: The two visualisation flowfields include the serpentine flowfield on the
left, and parallel flowfield on the right. Both flowfields show the flow channels cut
through the full thickness of the metal plate, allowing visual access to the cell
7.3.2 Image capture and processing

A Photron Fastcam 1024-PCI camera was used with a resolution of 1024 by 1024
pixels, and images were recorded at 1000 frames per second. Two PhotoSel LES600
lights were used to illuminate the flowfield and facilitate the acquisition of high quality
images at high speed. The setup was assembled as shown in figure 7.4.
The images were prepared for image processing as shown in figure 7.5. Firstly
a mask was used to isolate the flow channel as the region of interest. A threshold
was then applied to each pixel in the image from a greyscale, with each pixel whiter
than the threshold classed as white, and each pixel darker than the threshold classed
as black. This method output a binary back/white image and as seen in figure 7.5d,
the pixels that are black can be classed as slug flow, and white pixels as bubbly flow.
7.3. M ETHODS 114
Figure 7.4: The high speed camera was focused onto the flowfield, with the speed
of the electrolyte controlled by variable area flow meters.
The flow channel were subdivided into discrete channels, so that more informa-
tion regarding the flow regimes could be gained. Noise was a key consideration when
undertaking the image analysis, as variations in light intensity across the image, as
well as effects from the white membrane background both affected the quality of the
data. To mitigate these issues, a gradient was applied before image processing to
provide a uniform brightness across the full image, and the thresholding was chosen
based on a visual comparison between the raw and binary image.
7.3.3 Electrochemical characteristation

The Ohmic resistance of the cell (Rs ) was determined using galvanostatic EIS using
an Ivium-n-stat (Max current 10A). A signal amplitude of 5 % of the applied current
and a frequency range (100 kHz to 1 Hz). Rs is equated to the impedance at high
frequencies, when the phase angle is closest to zero.129
A Elektro-Automatik EA-PS 2042-20B power supply (Max Current 20A, 30V)
was used to supply the current for the visualisation experiment, with currents supplied
7.3. M ETHODS 115
(a) Raw Image (b) Masked channel
(c) Live flow (d) Binary image
Figure 7.5: The process for taking a raw image and turning it into a binary state,
where white refers to bubbly flow, and black refers to slug flow
up to 2 A·cm−2 .
7.4. M ATHEMATICAL MODELS 116
7.4 Mathematical models

There are two principal models for determining the composition of a two phase mix-
ture. The first utilises the void fraction as the main flow regime indicator. The second
uses the superficial velocities of the gas and liquid to predict the flow quality inside the
cell. The models are applied to a 10 cm2 zero gap cell operating at current densities
of 0 to 2 A·cm−2 , with flow rates of 0 to 200 ml·min−1 . The properties of the cell and
the materials used are listed in Table 7.1.
Property Magnitude Units

ρH2 0.082 kg·m−3
ρKOH 1055 kg·m−3
MH2 0.002015 kg·mol−1
MKOH 0.056 kg·mol−1
A 0.001 m2
a 0.000001 m2
F 96500 C
Table 7.1: Flow regime model properties
7.4.1 Void Fraction Model

The composition of the two-phase mixture in the cell can be viewed in terms of the
volumetric flux of liquid and gas passing a defined point of the flow channel. At any
chosen point, the ratio of gas and liquid is called the void fraction. This is cited as a
key feature in the transition from bubbly to slug flow regimes. Radovcich and Moissis
stated that the transition occurs at a void fraction of 0.2 to 0.3, depending on the size
of the channel.
The void fraction of gas passing a point x along the flow channel is expressed as:
VGas
VF = (7.1)
VElec +VGas
Where VGas is the volume flux of gas, and VElec the volume flux of electrolyte.
Making assumptions that the current density is uniform across the full area of the
electrode, the volume flux of gas is passing a point x is written as:
GGas iAMGas 1
VGas = = · ·x (7.2)
ρGas nF ρGas
Where GGas is the mass flux of the gas, ρGas is the density of the gas, i is the
current density, A is the surface area of the electrode, MGas is the molar mass of the
gas, F is Faradays constant, n is the number of electrons transferred in the reaction.
The variable x corresponds to the distance along the channel that is being studied,
used in fractional form 0 < x < 1, where the inlet is considered to be 0, and where
the channel leaves the cell to be 1. The volume flux of electrolyte depends on the
volumetric flow rate of the electrolyte supplied,VFlow , and the volume of electrolyte
consumed by the electrolysis reaction, VCons , such that:
VElec = VFlow −VCons (7.3)
The rate of electrolyte consumption is proportional to the current density:
iAMElec 1
VCons = · ·x (7.4)
nF ρElec
Where MElec is the molar mass of the electrolyte and ρElec is the density of the
electrolyte.
These equations give us a continuous plot of volume fraction across the full
length of the channel. It is notable that the volume fraction it is independent of chan-
nel size or number of channels. Using observed data from Moissis, a void fraction of
0.25 is point at which the flow regime is expected to go from bubbly to slug flow.137
7.4.2 Superfical gas and liquid velocities model

The simplicity associated with the void fraction model was addressed by Mishima and
Ishii, who proposed using the superficial velocities of gas and liquid to provide a more
accurate description of the transition from bubbly to slug flow.138, 139 The superficial
velocities of the gas and liquid ( jGas and jElec ) are used as the key parameters that
guide phase regime changes. The superficial velocity of a phase is defined as the flow
velocity as if the given phase was the only one flowing in a defined cross sectional
area.
The mass flux of the gas is proportional to the current density, and is calculated a
point x on the flowfield, where 0 < x < 1 and 0 is the start of the flowfied, 1 is the exit
of the cell:
iAMGas
GGas = ·x (7.5)
nFNChannels
Where i is the current density, A is the surface area of the electrode, MGas is
the molar mass of the gas, F is the Faraday constant and NChannels are the number of
parallel channels that make up for flowfield.
The mass flux of liquid is related to the flow rate:
GElec = GFlow − GCons (7.6)
where GCons is the mass of liquid consumed by the electrolysis reaction. GCons is
also proportional to i, and at a point x is calculated as:
iAMElec
GCons = ·x (7.7)
nF
The volume flux of gas at a point x is given by the mass flux of gas passing point
x, divided by the density of the electrolyte:
GGas
VGas = (7.8)
ρGas
where ρGas is the density of the gas being produced.

The volume flux of electrolyte at a point x are given by the total mass flux of
electrolyte passing point x, divided by the density of the electrolyte:
GElec
VElec = (7.9)
ρElec
where ρElec is the density of the electrolyte, and GElec is the total mass flux of liquid,
as defined in equation 7.6.
Superficial velocities of gas and liquid are used as key parameters to determine
changes in flow regimes. These depend strongly on the flow field used, and are calcu-
lated as:
VGas
JGas = (7.10)
a
VElec
JElec = (7.11)
a
Where a is the cross sectional area of the flow channel.

The outline of the void fraction model in section 7.4.1 demonstrates the interplay be-
tween current density, flow rate, and cumulative area passed by the channel. Using a
test cell defined in section 7.3.1, current densities associated with commercial electrol-
ysers (0 to 2 A·cm−2 ) and a range of flow rates quoted in academic and commercial
publications (10 to 200 ml·min−1 ) are modelled, and plots of the void fraction are
displayed in figure 7.6.65, 109
This model allows the effect of current density and flow rate on the void fraction
in the flow channel to be understood. This provides a useful tool when planning the
design and operating conditions of an electrolyser and its systems.
By mapping out the void fraction for the experimental conditions, the flow regime
transitions can be predicted in terms of position in the channel, as well as providing
information about the flow regime at the exit of the cell.
Figures 7.6a and 7.6b show that at flow rates of 10 and 50 ml·min−1 and high cur-
rent densities, the void fraction quickly surpasses the 0.25 criteria for the bubbly/slug
flow transition. This occurs within 1/10 of the flow channel length at the slowest flow
rate and highest currents, such that almost the entire active cell area is subjected to
slug flow. At 2 A·cm−2 , both cells quickly reach slug flow.
7.5.1 Void fraction model

Void Fraction for Flow Rate 10 ml/min Void Fraction for Flow Rate 50 ml/min
1 1
0.1 A/cm2 0.1 A/cm2
0.8 0.5 A/cm2 0.8 0.5 A/cm2
1 A/cm2 1 A/cm2
Void Fraction
Void Fraction
2 A/cm2 2 A/cm2
0.6 Bubbly to Slug Flow Transition
0.4 0.4
0.2 0.2
0 0
0 2 4 6 8 10 0 2 4 6 8 10
2 2
Cumulative Area Passed by Flow Channel (cm ) Cumulative Area Passed by Flow Channel (cm )
(a) Electrolyte flow rate of 10 ml· min−1 (b) Electrolyte flow rate of 50 ml· min−1
Void Fraction for Flow Rate 100 ml/min Void Fraction for Flow Rate 200 ml/min
1 1
0.1 A/cm2 0.1 A/cm2
0.8 0.5 A/cm2 0.8 0.5 A/cm2
1 A/cm2 1 A/cm2
Void Fraction
Void Fraction
2 A/cm2 2 A/cm2
0.4 0.4
0.2 0.2
0 0
0 2 4 6 8 10 0 2 4 6 8 10
Cumulative Area Passed by Flow Channel (cm 2 ) Cumulative Area Passed by Flow Channel (cm 2 )
(c) Electrolyte flow rate of 100 ml· min−1 (d) Electrolyte flow rate of 200 ml· min−1
Figure 7.6: Void fraction plots for a range of flow rates and current densities. Ob-
served data is overlayed for comparison
Figure 7.6c shows that at 100 ml·min−1 the void fraction only reaches slug flow
inside the cell for the current densities above 1 A·cm−2 , and this only occurs half way
along channel. Figure 7.6d shows that with the electrolyte flow rate at 200 ml·min−1 ,
slug flow only occurs at 2 A·cm−2 , and right at the end of the cell. Each figure shows
Void Fraction for increasing Current Densities and Flow Rates
0.9
1 0.8
0.7
0.8
Void Fraction
0.6
0.6
Void Fraction
0.5
0.4
0.4
0.2
0.3
0
2 0.2
1.5 200
1 150 0.1
100
0.5
50 0
0 0
-1
Flow Rate (ml·min )
2
Current Density (A·cm )
Figure 7.7: Schematic showing the interplay between current density, flow rate,
and cumulative area passed by the channel
the large impact that electrolyte flow rate has on the two phase properties of the cell.
High speed photography was used to identify the position that the bubbly to slug
transition occurs in terms of the position along the flow field, and this transition point
is plotted on figure 7.6 to make a comparison between experimental and theoretical
results. It was stated by Moissis that the bubbly to slug transition will occur at a void
fraction of 0.25.137 These results show a wide range of void fractions at which the
transition is seen to occur, with values ranging from 0.18 to 0.47, and is particularly
wide ranging at the low flow rate. This demonstrates the limitations of the model, and
drives the motivation to refine the predictions further.
Figure 7.7 shows an alternative demonstration of the interplay between void frac-
tion, electrolyte flow rate and current density. The void fraction defined in this case
as the fraction taken at electrolyte exit from the cell. Figure 7.7 demonstrates that a
high void fraction can be expected for even relatively small current densities when low
rates of electrolyte flow are used. Higher flow rates keep the void fraction down, but
the combination of a high flow rate and low current density is the only combination
that allows for a low void fraction. For a cell running at 2 A·cm−2 , even an electrolyte
flow rate of 200 ml·min−1 cannot stop the void fraction at the exit of the cell exceeding
0.25. This method of visualisation allows operating conditions to be quickly assessed,
and the interplay between the three main parameters neatly observed.
Overall, this simple model allows quick and easy estimations of transition be-
tween the bubbly and slug flow regimes. The point at which the transition between the
two occurs, was compared to experimental data, and the match between the model and
validation data was far from optimal. Indeed it was these shortcomings that promote
the use of an alternate model based on the superficial velocities of the gas and liquid.
7.5.2 Superficial velocities model

Using Mishima and Ishiis data for gas/liquid mixtures in a vertical tube, the theoret-
ical transitions between flow regimes can be plotted for superficial flow velocities.
As shown in figure 7.8, the proposed boundaries between the four flow regimes, as
observed by Mishima in a vertical tube of diameter 2mm.137
Based on section 7.5.1, the flow rates and current densities relevant for alkaline
electrolysers were again applied, with the windows for flow rate and current density
chosen to be 10 to 200 ml·min−1 at current densities of 0 to 2 A·cm−2 .
Figure 7.9 shows the superficial gas and liquid velocities at the exit of the cell, as
predicted by the theory for a 10 cm2 cell. It can be seen that the serpentine flow field
experiences both greater superficial gas and liquid velocities, which predicts that this
cell is more likely to experience slug flow. The channel is predicted to demonstrate
churn flow at the lowest flow rate. The cell with the parallel flowfield is not expected
to transition from bubbly to slug flow, except at the lowest flow rates and highest
currents.
By visualising the flow fields, the progression of the bubbly/slug/annular flow

Figure 7.8: The transitions between flow regimes are mapped in terms of the su-
perficial liquid and superficial gas velocities, as based on observations by past re-
searchers.
regimes can be mapped and quantified, with the aim of providing useful data with
regards to the percentage of the channel which has bubble and slug flow. Figure 7.10
shows the comparison between the predicted flow regimes, and those observed using
the high speed photography, with the model predicting the flow transitions well at high
flow rates for the both flowfields, with most results matching the predictions. Low flow
rates show discrepancies between the data and predictions. Slug flow is observed for
the 10 ml·min−1 flow rate in the serpentine flow field, even when not predicted by the
model. The same was observed in the parallel flowfield. The experimental data shows
that at low flow rates, the mixture is more prone to transitioning to the slug flow than
suggested by the model. This could be attributed to the fact the Mishima and Ishii
were using a circular flow channel, rather than the rectangular channel used in this
study, and that the channel used in their study was slightly larger than the one used
here.
Flow regime transitions for hydrogen

10 1
Serpentine
200 ml/min
10 0
100 ml/min
Superficial liquid velocity
Parallel 50 ml/min
200 ml/min
10 -1
100 ml/min 10 ml/min
50 ml/min
-2
10 0.1 A·cm -2
0.5 A·cm -2
10 ml/min 1 A·cm -2
2 A·cm -2
10 -3
10 -3 10 -2 10 -1 10 0 10 1 10 2
Superficial gas velocity
Figure 7.9: This figure shows the theoretical flow regimes of the cell at a series of
flow rates and current densities, it can be seen that cells employing the serpentine
flowfield experience slug flow before those employing a parallel equivalent.
Despite the slight discrepancies, the model provides an accurate and useful tool,
and can be effectively utilised to establish the flow regime inside the electrolysis flow
channel. The two main limitations of the model are as follows: the boundaries be-
tween flow regimes are treated as hard borders, where there is a binary change between
bubbly/slug or slug/annular flow. Observations made during the experiment showed
that the flow classed as slug flow near the bubble/slug boundary still contained a large
proportion of bubbly flow in the channel. Secondly the information only relates to the
flow regime at the exit of the cell, whereas the visualisation shows us that the flow
regime that occurs at the exit of the cell is not representative of the full length of the
channel.
To provide more resolution on the change of flow regime, the model was used
to output the flow quality at selected positions along the flow channel. In this case,
the length of the channel was divided into fifths, providing information about the flow
Flow Regime Transitions for Hydrogen Evolution

10 1
Serpentine
200 ml/min
10 0
100 ml/min
Superficial Liquid Velocity
50 ml/min
Parallel
200 ml/min
10 -1
10 ml/min
100 ml/min
50 ml/min
10 -2
Bubbly Flow
10 ml/min Slug Flow
Annular Flow
10 -3
10 -3 10 -2 10 -1 10 0 10 1 10 2
Superficial Gas Velocity
Figure 7.10: This figure shows the observed flow regimes, with the results match-
ing the theory closely, and the parallel flowfield only experiencing slug flow at very
low flow rates, or current densities of 2 A·cm2
137
quality at intervals along the flow channel, rather than just at the exit. This provides a
step forward when compared to what has been done so far in the literature, and allows
information such as the percentage area of the channel under each flow regime to be
investigated.
The data in Figure 7.11 shows that for a flow rate of 100 ml·min−1 , although
in the Figure 7.9 the cell is shown to fall into the slug flow category, actually only
2/5 ths of the cell has a slug flow regime. This is compared to the cell running with
an electrolyte flow of 30 ml·min−1 , when almost the full length on the channel will
experience slug flow. This is particularly interesting when comparing to other liter-
ature data, notably by Aubras et al. and Ito et al. who try to elucidate the effect of
bubbly and slug flow on cell performance.122, 146 This data shows that further care
must be taken when trying to investigate the effects of bubble or slug flow, and that
Flow Regime Transitions

10 1
Serpentine
Increasing distance along channel
(flow area divided into fifths)
10 0
Superficial Liquid Velocity
100 ml/min
30 ml/min
Parallel
10 -1
100 ml/min
30 ml/min
-2
10
10 -3
10 -3 10 -2 10 -1 10 0 10 1 10 2
Superficial Gas Velocity
Figure 7.11: Schematic showing the flow regimes predicted at various locations
along the flow channel, this emphasises that the flow regime at the exit of the cell is
not representative of the full channel.
the boundary between flow regimes should not be treated as a hard border.
7.5.3 Matlab visualisation results

Figure 7.12 shows the percentage of the flow channel covered by slug flow, as assessed
by the Matlab image processing program defined in section 7.3.2. The flowfield is
again divided into fifths in terms of channel length, and the percentage coverage is
calculated for each individual fifth of the full flowfield. Data was processed for flow
rates of 30 ml·min1 and 100 ml·min1 and for both the serpentine and parallel flow-
fields. The use of the parallel flowfield is again shown to slow the transition from
bubbly to slug flow when compared to the serpentine equivalent. Interestingly, it can
be seen that slug flow begins within the first fifth of the serpentine channel for the
lower flow rate. The slug coverage reaches over 80% of the flow channel in the final
section. This is consistent with predictions in Figure 7.11. This compares to the par-
allel flowfield with the same conditions, where the percentage coverage of slug flow
reaches just over 40 %.
A Comparison of Slug Coverage for Different Flow Channels

90
100ml·min -1 at 2 A·cm-2 (Serpentine)
Percentage Coverage of Slug Flow

80
30ml·min -1 at 2 A·cm-2 (Serpentine)
70 100ml·min -1 at 2 A·cm-2 (Parallel)
60 100ml·min -1 at 2 A·cm-2 (Parallel)
50
40
30
20
10
0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fraction of Channel
Figure 7.12: Schematic the percentage of the channel covered by slug flow
At the higher flow rate, only the very end of parallel flow field with have slug
flow with just of 20 % coverage of slug flow in the final section. This compares to
over 40% coverage of slug flow in the same section for the comparable serpentine
flowfield. This data gives us a higher resolution of the flow regime inside the cell,
with an insight into the relative percentages of slug flow and bubbly flow. This dif-
fers from the traditional method of flow regime definition, where the flow regime is
attributed to the regime observed at the exit of the cell. To address the shortcomings
associated with the treatment of the regime changes shown in Figure 7.8, the tran-
sitions here are parameterised as follows; the boundary from bubbly to slug flow is
where there are greater than 0% coverage of slug flow, and the boundary between slug
flow and annular flow is defined as the point at which the percentage coverage of slug
flow reaches 100%. Currently, when the point falls between the two boundaries, it is
characterised as having slug flow, however it is clear that there will be a percentage
of slug and percentage of bubbly flow, which will change as the flow quality moves
towards the annular flow boundary.
Figure 7.13 demonstrates that the overall coverage of slug flow across the whole
cell is time independent. This shows that for the test cell with a flow rate of 30
ml·min−1 slug flow covers almost 50 % of the full flow channel. This is compared
to the parallel flowfield with the same conditions, where the overall slug coverage is
less than 20 %. This again addresses the issue in the first set of results that give the
cell a binary setting of the flow regime.
The imaging technique matches well with the theoretical predictions shown in
Figure 7.11, confirming that the percentage of slug flow increases as the flow mixture
approaches the end of the channel. The results also agree with previous results from
Ito et al. that the parallel flowfield demonstrates a lower percentage of slug flow
compared with the serpentine flowfield in identical conditions.122
Time dependent slug coverage

100
30 ml·min-1 at 2 A·cm -2 (Serpentine)
Overall coverage of slug flow (%)
30 ml·min-1 at 2 A·cm -2 (Parallel)

80
60
40
20
0
0 0.1 0.2 0.3 0.4 0.5
Time (s)
There were a number of errors associated with this technique of attaining a quan-
tified percentage of coverage of slugs. The white Zirfon membrane, which can be seen
behind the mesh in Figure 7.3(a), simulates the appearance of bubbly flow - as seen
in the upper channels of 7.3(d). The edges of bubbly flow also appearing as black,
which is defined by the program as slug flow, and this appears to be due to interaction
with the channel wall. The effect of this was mitigated by taking the flow quality in-
formation from a sample on the channel, with the template created slightly away from
the edge of the channel. When observed in video form, some of the slugs could be
observed cutting out the top two channels and going straight to the exit between the
flowfield and the mesh - this can be attributed to inhomogeneous compression of the
cell. These results flag up a key issue to be addressed in the form of homogeneous
compression being a key factor to ensure that electrolyte flow takes the path defined
by the flowfield. This is more likely at high flow rates, due to the higher pressure in
the flow system. Using the percentage coverage observed in Figure 7.12 for the fast
flow rates, the image processing technique outputs a percentage slug coverage of an
average of 6% in the initial fractions of the flow channel. It was clear from analysis
of the images by eye that the channel was entirely filled with bubbly flow, and this is
taken as the error from the image processing technique.
7.5.4 In-situ visualisation of flow regimes

This study has so far presented a method for the quantification of the transitions be-
tween flow regimes in an alkaline electrolysis cell. As one of the first studies to un-
dertake a visualisation study for an alkaline electrolysis cell, it is also useful to use the
qualitative visual data obtained during the study. High speed photography allows the
flow regimes to be seen, when in real time it is difficult to study due to the speed of
the flow mixture.
It is firstly noted that the three photos of Figure 7.14b, 7.14c and 7.14d with
current densities of 0.5 A·cm−2 , 1 A·cm−2 and 2 A·cm−2 would all be categorised
into the slug section of the graph by Mishima and Ishii. The images show that all
three have significantly differing conditions in terms of the ratio of bubbly to slug
flow.
The images in Figure 7.14 show the effect of current density on flow regime
makeup of the cell at a flowrate of 50 ml·min1 . Figure 7.14a shows that at 0.1 A·cm−2
there is bubbly flow throughout the full channel, with the initial channels appearing
(a) 0.1 A·cm−2 , 50 ml/min (b) 0.5 A·cm−2 , 50 ml/min
(c) 1 A·cm−2 , 50 ml/min (d) 2 A·cm−2 , 50 ml/min
Figure 7.14: Imaging of the flow regimes inside the serpentine cathodic flowfield
at 50 ml/min for four current densities
almost clear, indicating limited gas bubbles in the solution. Figure 7.14b demonstrates
that at 0.5 A·cm−2 , the initial signs of the transition to slug flow can be seen, with
larger bubbles spanning the full height of the channel appearing. At 1 A·cm−2 larger
slugs can be seen clearly in Figure 7.14c, with their makeup of a thin film of electrolyte
encasing a large gas bubble allowing clear visual access to the porous electrode. At
2 A·cm−2 the transition from bubbly to slug flow can be seen as early as the 3rd
channel, and Figure 7.14d shows that the frequency and sizes of the slugs increases as
the electrolyte progresses upwards through the flow channel.
(a) 0.1 A·cm−2 (b) 0.5 A·cm−2
(c) 1 A·cm−2 (d) 2 A·cm−2
Figure 7.15: Imaging of the flow regimes inside the parallel cathodic flowfield at
50 ml·min−1 for four current densities
The corresponding visualisation with the parallel flowfield is seen in Figure 7.15.
Figures 7.15a, 7.15b and 7.15c show that for current densities up to 1 A·cm−2 there
is limited evidence of slug flow. At 2 A·cm−2 , Figure 7.15d demonstrates that there
is slug flow at the top of all flow channels. There is an asymmetric nature to the
appearance of the channels, with the channels of the right hand side having a higher
percentage coverage of slug flow, this corresponds to the side closest to the electrolyte
flow outlet. This effect can be attributed to inertial effect of fluid flow from a dividing
manifold into multiple T-Junctions, as suggested by Majasan et al..148 This is an
important consideration for parallel flowfield designs, and the importance of effective
flow spreading at the inlet is key for minimising this effect and maintaining uniform
conditions in the cell.
7.5.5 Effect of flow rate on the Ohmic resistance of the cell
The minimum value of Ohmic resistance expected for the zero gap cell (at fixed tem-
perature) is when there is no current flowing between the electrodes, due to the ab-
sence gas bubbles in the electrode-membrane interface. PEM electrolysers report a
very small increase in the Ohmic resistance of cells, even at high current densities.
As such the Ohmic resistance of the cell was measured at a range of current densities
and electrolyte flow rates, to establish the possible effects on the performance of the
system, and understand their origins, and possible mitigation strategies.
Increase in Ohmic Resistance of Cell vs Current Density

10
200ml·min-1
Increase in Ohmic Resistance (mOhm·cm2 )
150ml·min-1
8
100ml·min-1
50ml·min-1
25ml·min-1
6
0
0 0.2 0.4 0.6 0.8 1
-2
Current Density (A·cm )
As seen in Figure 7.16, there is an increase of Ohmic resistance with current den-
sity for all of the applied flow rates, with a similar magnitude of increase throughout.
The effect is particularly large at the lower current densities, suggesting that the initial
bubbles cause an increase in resistance within the inter-electrode gap. Inhomogeneous
compression and bubbles forming at the electrode membrane are both thought to con-
tribute to this effect. At high current densities, the added rate of gas production does
not affect the Ohmic resistance so greatly. It is possible that once the easy bubble
entrapment areas have been filled, any additional bubbles leave the area easily, or per-
haps the turbulence caused by the added number of bubbles assist in the extraction
of gas from the membrane electrode area. The flow rate is seen to have a small but
definite effect on the Ohmic resistance, with the highest flow rate offering the best
results, this appear to be in line with previous assumptions that the gas bubbles will
be extracted from the cell faster at the increased flow rates.
7.6 Conclusions
The modeling undertaken in this chapter shows that the void fraction model provides
a simple method for evaluating to gas/liquid ratios - but is limited as the void fraction
becomes large. An improvement on this model uses the superficial flow velocities of
the gas and liquid phases to provide more accurate predictions of the transition from
bubbly to slug flow.
Using high speed photography, the bubbly and slug flow regimes have been visu-
alised using bespoke flowfield designs. An innovative approach to image processing
has been used to provide quantitative on the bubbly and slug flow regimes, and the
transition between the two. The transition of bubbly-slug is shown to closely match
the theoretical values, and the Matlab code was used to predict the flow regime at
various points inside the cell.
The effect of the electrolyte flow rate on the Ohmic resistance of the cell was
investigated. An increase in resistance was observed for increasing current density for
all flow rates, and this is attributed to the effect of gas buildup in the interelectrode
gap. The positive effect of flow rate on the Ohmic resistance is demonstrated, with the
higher flow rates shown to minimise the trapping of gas between the electrode and gas
separator. At electrolyte flow rates of 200 ml·min−1 , the Ohmic resistance of the cell
increases by just 5 mΩ·cm2 .
Chapter 8
Zero gap electrolysis with an anion

exchange membrane
8.1 Abstract
The electrochemical and gas separation properties of the anion exchange membrane
Sustainion are investigated and compared to the most used alkaline gas separator,
Zirfon. Sustainion is found to offer acceptable rates of gas cross permeation, with a
significant improvement in electrochemical performance over Zirfon.
The properties of the Zirfon membrane are found to include outstanding gas sep-
aration performance at a substantial electrochemical cost, so a model is used to of-
fer insight into the viability of a thinner version of the membrane to allow increased
electrochemical performance whilst maintaining acceptable levels of gas mixing. To
achieve this, the mechanisms for gas cross permeation are introduced and investigated,
with alternative strategies to minimise gas mixing also suggested.
Furthermore, the effect of current density on the rate of gas mixing was investi-
gated and is demonstrated to affect the rate of gas mixing through the porous Zirfon
membrane more than Sustainion. This suggests that Zirfon is more susceptible to
small pressure differences inside the cell.
8.2 Introduction
This thesis has been focused on increasing the efficiency and current density of a liq-
uid alkaline electrolysis cell as a means to reduce the capital expenditure and running
costs of the system. Earlier results in this thesis concluded that the key factor currently
holding back this development is the high resistance of the membranes used for alka-
line electrolysis. The most widely employed membrane is currently Zirfon, which has
been shown in previous chapters to contribute to an area specific resistance of the cell
of >0.4 Ω·cm2 at 40◦ C, and this contribution limited the zero gap cell to ∼ 2.35 V at
1 A·cm2 . The Zirfon membrane was the main contribution to the cell voltage above
0.8 A·cm2 . In line with results already achievable with PEM technology, performance
targets of 2 A·cm2 at a voltage of 1.9 V are realistic within the next decade.102 It is
clear that to achieve this high performance, a step change the membrane performance
is necessary to achieve this.
There are two main routes to reducing the membrane contribution to the Ohmic
resistance. Firstly, the resistance of the membrane can be reduced through the fab-
rication of a thinner version of the same material. This,however, comes at a cost in
terms of the cross permeation of gases through the membrane. Finding an optimal
balance between resistance and gas cross permeation is key to achieving maximal
cell performance using this method. The second route is through the development of
new membrane materials; this allows for new membrane properties, with the aim of
maximising the gas separation and minimising the resistance. As outlined in the lit-
erature review, there are a number of new membranes currently under development,
with approaches based on thin polymeric anion exchange membranes identified as an
attractive option.149
Gas mixing causes a number of negative impacts on the system; not limited to
the decrease in gas quality, it also reduces the overall electrical efficiency of the sys-
tem. The effects of foreign gas on the degradation of the electrode materials have been
reported in PEM electrolysis, with similar deterioration expected in the alkaline envi-
ronment. The importance of safety criteria mean that industrial electrolysers are shut
down as soon as 2 % of foreign gas is detected in the exhaust stream; the explosion
limits of an hydrogen and oxygen mixture are 4 % molar H2 and 95.2 % molar H2 for
the lower and upper explosive limits at 20◦ C and 1 bar.150
Haug et al. showed the effect that electrolyte management has on the gas purity
of an alkaline system, Mixed electrolyte cycles were shown to exhibit substantially
higher rates of gas mixing when compared with cells using separated and partially
separated electrolyte feeds. This issue with electrolyte management appears in multi-
ple studies as the principal reason for gas crossover, effective system design to min-
imise the interaction of the two electrolyte streams is key to keeping a low rate of gas
cross permeation.109, 151
Trinke et al. compared the PEM and liquid alkaline technologies in terms of
gas mixing, finding that the PEM technology is far more resilient to differential pres-
sure.152 Despite this, the PEM electrolysis cell experienced a rate of gas mixing one
order of magnitude higher than alkaline due to the high solubility of hydrogen and
oxygen in water when compared to concentrated potassium hydroxide solution.102
Schalenbach suggested that the cross permeation of gases in liquid alkaline elec-
trolysis was greatly affected by differential pressure, with small absolute pressures
leading to this being the principal method of gas mixing when porous Zirfon mem-
branes were used.153
The past decade has seen a number of innovations related to anion exchange
membranes which have led to these thin membranes acting as a solid polymer elec-
trolyte, permitting the use of reduced concentrations of alkaline electrolyte. These
developments open up possibilities of working at differential pressures, which saves
energy in the post-compression of the produced hydrogen. Using a membrane with
a lower resistance and comparable gas separation abilities when compared to Zirfon
appears to be key for the progression of the alkaline technology. The current limiting
factor of AEMs is their stability in the alkaline environment, with degradation of the
materials often evident within 1000 hours of operation.154
A recent development in AEM technology has seen the production of Sustainion

class TZ by Dioxide Materials, USA. This has led to a situation where the performance
of the membrane suggests that it may be a potential replacement to Zirfon, and open
up possibilities for the widespread commercialisation of AEM electrolysers.
8.2.1 Zirfon
Zirfon is produced by a fim casting technique, where the casting dope contains a sus-
pension of zirconinium oxide in a polysulfone solution. N-Methyl 2-pyrrolidone is
used as the solvent and the nonsolvent being isopropyl alcohol. The ZrO2 powder is
mixed with the polysulfone solution to obtain a homogeneous suspension. This sus-
pension is poured onto a surface where the membrane is formed via the extraction of
the solvent through immersion in the non-solvent. The loading of ZrO2 in the poly-
sulfone matrix has been optimised by Agfa to provide key performance characteristics
appropriate for gas separation in an alkaline environment. Zirfon in 1M KOH at room
temperature has a tensile strength of 3.8 MPa. The porosity of the membrane is 56%,
and the pore size of the matrix is 200 Angstroms. Zirfon shows excellent gas separa-
tion characteristics, long term stability, and a low ionic resistance of 0.3 Ω·cm2 (30◦C,
30 wt. % KOH).82 Gas purities for both H2 and O2 of greater than 99.9% at 70◦ C 5.3
M KOH have been reported at current densities between 100 and 400 mA·cm−2 .
8.2.2 Sustainion
Dioxide materials developed a novel imidazole-functionalized membranes using a

polystyrene-based backbone that are stable in strong alkaline solutions. The anion
exchange Sustanion membranes are based on a cheap and abundant, but more impor-
tantly, alkaline stable polystyrene backbone. The synthesis of the membrane involves
a two-step process of copolymerization followed by subsequent functionalization. The
membrane can be cast as a film or as a reinforced membrane using various reinforce-

ment materials such as Zirconium dioxide (ZrO2). A copolymer of styrene and vinyl
benzyl chloride is prepared by free radical addition polymerization. The copolymer
product is then washed and precipitated in ethanol. Afterwards, it is filtered, dried
and subsequently functionalized with 1,2,4,5-tetramethylimidazole in a Dowanol PM
(1-methoxy-2 propanol) solvent. Divinylbenzene is also added as a crosslinker to
help improve membrane strength. The cast membrane is formed in its chloride form,
which is then activated for 816 h in a 1M potassium hydroxide solution for conversion
to the hydroxide form. The membrane has an impressive ion exchange capacity of 1.1
mmol/g, and even more impressive is its area specific resistance in 1M KOH, which at
60◦C is just 0.045 Ω · cm2 . The membrane undergoes moderate swelling ( 50%) dur-
ing its conversion to the hydroxide form. This is believed to offer an explanation of
the low ionic resistance, as water absorption is seen as a key mechanism for hydroxide
ion transport through the membrane.
The imidazole-based anion exchange membrane has a PTFE support, and is

re-inforced with zirconium oxide for greater compression strength. This presents a
promising solution to the previously discussed limitations of alkaline based electrol-
ysis technologies. Due to its recent development, basic studies of the membrane are
lacking, and independent evaluations of its properties are still yet to feature in the
literature.108
This chapter aims to characterise the Sustainion membrane in terms of its electro-
chemical performance, as well as investigating its gas separation abilities, and making
comparisons to the current state of the art Zirfon gas separator. Furthermore, the
mechanisms for gas crossover will be studied, and strategies for reducing the rate of
gas mixing in a working cell will be discussed.
8.3. T HEORY OF GAS MIXING 140
8.3 Theory of gas mixing

The Faradic efficency of an electrolysis cell using effective catalysts can be assumed
to be 100 %. The largest impact on the current efficiency is from the cross-permeation
of hydrogen and oxygen through the gas separator. The mixing of the two gases is
typically characterised separately as the anodic hydrogen content and the cathodic
oxygen content.
Assuming 100 % Faradic efficiency, the molar production rate density is:
j
τ= (8.1)
nF
Where j is the current density, F is Faraday’s constant and n is the number of

electrons in the reaction (2 for hydrogen evolution, 4 for oxygen evolution).
The anodic hydrogen content is equal to the percentage of hydrogen that has
permeated through the membrane into the stream of oxygen being generated at the
anode. This constitutes the complete loss of this produced hydrogen, and thus reduces
the efficiency of the cell. This hydrogen is not oxidised at the anodic catalyst and
thus remains in the oxygen stream that is generally vented to the atmosphere. As the
hydrogen can be released into the atmosphere, the only penalty for the system is the
energy content of that quantity of gas.
Oxygen that mixes into the cathodic gas stream impacts the purity of the produced
hydrogen, which presents problems due to the high hydrogen purity demands of fuel
cells and other uses of hydrogen; this oxygen is generally removed by recombination
at a separate catalyst. This recombination uses hydrogen at a molar ratio of 2:1 to
oxygen, and so cathodic oxygen impacts the efficiency of the cell at twice that of
anodic hydrogen.
Therefore the current loss due to gas mixing can be defined as:
jloss = 2FΦH2 + 4FΦO2 (8.2)

where Φ represents the molar mixing flux density of the hydrogen and oxygen respec-
tively.
In terms of hydrogen purity at its point of use, these purity requirements have
been set out in a series of international standards (ISO 14687). Standards differ for
fuel cells for road vehicles and fuel cells for stationary applications, but both require a
hydrogen purity of >99.97 % and an oxygen presence of less than 5ppm. Hence it can
be seen that any oxygen impurity in the hydrogen stream will need to be removed in
post processing of the exhaust gas, with the energy required for its removal impacting
the efficiency of the system.
The anodic hydrogen content is used as a measure of the flux of hydrogen cross-
permeation.
ΦH2
AnodicHydrogenContent = · 100mol% (8.3)
τO2 + ΦH2
The two principal driving forces for gas mixing are diffusion (caused by the dif-
ferences in concentration of the dissolved gases immediately either side of the mem-
brane) and differential pressure (the gradient between the absolute anodic and cathodic
pressures).
8.3.1 Diffusion
The cross-permeation due to diffusion is described by Fick’s law of diffusion, with
units of mol·cm−2 s−1 and is denoted as:
∆c
ΦFick = −D (8.4)
d
where D is the diffusion coefficient of the dissolved gas in the membrane, ∆c is

the concentration difference across d, the thickness of the membrane. For the diffusion
of hydrogen towards the anodic compartment, this can be expressed as:
DH2 cat
ΦH2 Fick = − (cH2 − cano
H2 ) (8.5)
d
where DH2 is the effective gas diffusion coefficient in the seperator, d is the thick-
ness of the seperator, ccat
H2 is the concentration of dissolved hydrogen at the cathode and
cano
H2 is the concentration of dissolved hydrogen at the anode.
It can be assumed that the concentration of dissolved hydrogen on the anodic side
is negligible and ccat
H2 is related to the partial pressure, p, and solubility of the gas, S,
as follows
cgas = pgas · Sgas (8.6)
Therefore the cross permeation can be re-written as follows:
DH2 · SH2 cat

ΦH2 Fick = − (pH2 − pano
H2 ) (8.7)
d
Hydrogen diffusivity is defined as the multiple of the solubility of hydrogen in

the solution and the diffusion coefficient of hydrogen in the gas seperator:
ε H2 Fick = DH2 · SH2 (8.8)
such that:
εH2 ano
ΦH2 Fick = − (pH2 − pcat
H2 ) (8.9)
d
the exact solution can be simplified treating pcat ano ano cat
H2 >> pH2 , such that pH2 − pH2 ≈
−pcat
H2 , this simplifies to give:
Fick
εH2 · pcat
H2
ΦH2 = (8.10)
d
This equation shows that the rate of cross permeation of hydrogen into the anode
stream is equal to hydrogen diffusivity of the gas separator divided by its thickness,
multiplied by the partial pressure of hydrogen at the cathodic edge of the gas sepa-
rator. The partial pressure of hydrogen at the cathode is calculated from the absolute
pressure, p, and the partial pressure of saturated water vapour in the KOH solution,
Property Zirfon (1M KOH) Zirfon (30% wt. KOH) Units

SH2 5 0.75 10−7 mol cm−3 bar−1
εHFick
2
0.2 0.025 10−11 mol cm−1 bar−1 s−1
pwv 0.023 0.016 bar
εHDarcy
2
40 1 mol s−1 bar−1
Table 8.1: The properties of Zirfon from literature in different electrolyte concen-
trations at 20 ◦ C and atmospheric pressure.153
pwv using Daltons law such that:
pH2 = p − pwv (8.11)
The partial pressure of saturated water vapour in potassium hydroxide is taken

from the data provided by Balej et al. and is shown in Table 8.1.155
For experimental purposes, the hydrogen diffusivity of a material can be calcu-

lated by measuring the rate of cross permeation of hydrogen into the anode stream, as
well as knowing the thickness of the membrane and the partial pressure of hydrogen
at the cathode such that:
ΦH2 Fick · d
εH2 = (8.12)
pcat
H2
For practical use, once the hydrogen diffusivity of a material is known, the rate
of hydrogen crossover based on diffusion can be calculated. Literature values exist for
Zirfon from other studies and are shown in table 8.1. It is notable that the diffusivity
of hydrogen in Zirfon depends strongly on the electrolyte. This is due to the reduced
solubility of hydrogen in potassium hydroxide at increased electrolyte concentrations.
8.3.2 Differential pressure

Another key mechanism for gas mixing occurs when pressure gradients are formed
across the membrane, and electrolyte, saturated with dissolved gases, is forced through
the membrane.
As described by Darcy’s law, the molar permeation flux density is:

∆p
ΦH2 Darcy = −εHDarcy · (8.13)
2 d
where ∆p is the absolute pressure difference between the cathodic and anodic
compartments. εHDarcy
2
is the hydrogen permeability of the membrane, and d is the
thickness of the membrane. If the absolute pressure is negative, then oxygen from the
anode will be forced through towards the cathode.
Table 8.1 displays the literature values for the hydrogen permeability of Zirfon,
with the large difference between the values for 30 % wt KOH and 1M KOH due to
the increased viscosity of the electrolyte at increased electrolyte concentrations. Less
viscous liquid flows more easily through the pores of the Zirfon membrane.
8.3.3 Overall gas flux
For the overall gas flux:
ΦH2 = ΦH2 Fick + ΦH2 Darcy (8.14)
For atmospheric pressure, Schalenbach noted that the pressure control of the an-
odic and cathodic compartments are normally regulated within 0.01 bar that the con-
tributions, which meant that for an alkaline cell with an electrolyte of 30 % wt KOH,
that the contributions from ΦH2 Fick and ΦH2 Darcy were approximately the same.
The anodic hydrogen content is usually the main figure quoted for gas crossover
as it is the main factor to be considered regarding gas safety and explosion limits.
The flux of gas cross-permeation is generally similar for hydrogen and oxygen, and as
there is twice the amount of hydrogen being produced in the cell when compared to
oxygen, the ratio of oxygen in the hydrogen stream is expected to be lower.
8.4. E XPERIMENTAL METHODS 145
8.3.4 Electrolyte mixing
From the fundamentals of alkaline electrolysis, it can be seen that water is consumed
at the cathode, and produced at the anode, creating a change in electrolyte concen-
tration between the anodic and cathodic electrolyte cycles. Therefore, historically the
two electrolytes supplies have been mixed in a general supply tank to minimise this
concentration gradient. Research by Haug et al. demonstrated that the mixing of elec-
trolyte also caused the mixing of the dissolved gas species in the general electrolyte
tank, which was then supplied back to the cell, and stated that this was, in fact, a
key driver for electrolyte mixing.151 The separation of the electrolyte cycles could be
considered, but the shift in electrolyte concentrations would cause other inefficiencies
in the system over extended time periods. As a result, the idea of a partially mixed
electrolyte feed was proposed by Haug, where the electrolyte is separated into two
separate tanks, connected by a small diameter pipe, that allows concentration gradi-
ents to dissipate whilst minimising gas mixing by convection.
8.4 Experimental methods

A Agilent Technologies 7820A gas chromotographer was used as to measure the puri-
ties of the two exhaust gases. The equipment was calibrated by sampling test hydrogen
and air separately with argon acting as the carrier gas. Hydrogen was calibrated at 0.8
and 2.5 % in argon, and oxygen and nitrogen calibrated using air at nitrogen con-
centrations of 0.850 % and 2.53 % respectively. The gas chromotographer was setup
using the operating conditions described in Table 8.2.
Carrier gas Argon

Valve temperature 150◦ C
Column flow 6 ml·min−1
Split ratio 20:1
Oven temperature 270◦ C
Detector TCD
Table 8.2: Key operating conditions for the gas chromatograph

8.4. E XPERIMENTAL METHODS 146
8.4.1 Gas mixing cell setup
Nickel flow plates were used as in Chapter 6, employing a 10 cm2 serpentine flow field
with channels 1.5 mm deep and 1.5 mm wide. 10 cm2 pure nickel mesh electrodes
were used (Dexmet), and silicone gaskets of thickness 0.25 mm were used to seal
the cell under compression. For the electrochemical characterisation of the two cells,
depositions of the catalysts were made as outlined in Chapter 6, with the Raney nickel
coating used as the cathodic catalyst and NiFe(OH)2 used as the anodic catalyst.
The membranes were prepared by immersing in the relevant electrolyte for 48

hours before use in the cell. Two separate Zirfon Perl UTP 500 (Agfa) gas separators
were used, in 30 % wt. KOH and 1M KOH electrolyte respectively, and Sustainion
was used with 1M KOH.
Electrolyte flow was driven by Anself 12V pumps, with the flow rate fixed by two
sets of Dwyer variable area flow meters, with a range of 20-200 ml·min−1 . For the gas
purity analysis experiment, electrolyte flow to the anodic and cathodic sides of the cell
were done at 20◦ C using the partially mixed electrolyte management strategy, the two
electrolyte reservoirs were connected by a 4 mm piece of tubing, which prevented a
build up of electrolyte on one side due to water consumption, and minimised pressure
gradients between the two sides of the cell. The setup is visualised in Figure 8.1.
For electrochemical analysis, a mixed electrolyte reservoir was used. The elec-
trolyte was preheated in a Fisher FB15053 temperature bath, with the cell temperature
of 40◦ C maintained by monitoring the cell temperature using a Elitech RC-5 tem-
perature probe, and adjusting the temperature bath as needed. Electrolyte flow was
maintained at 100 ml·min−1 throughout the characterisation.
The Ohmic resistance of each setup (Rs ) was determined using EIS. A signal
amplitude of 5% of the applied current was applied onto the current densities of 10 to
1000 mA·cm−2 , across a frequency range of 100 kHz to 1 Hz. Rs was equated to the
impedance at high frequencies when the phase angle is closest to zero.129
Figure 8.1: The equipment setup for gas analysis in the gas chromotographer. Par-
tially mixed electrolyte cycles were used, as shown by the connecting tube between
the two chambers. Gas is sampled from the top of the electrolyte chamber.
Power was supplied to the cell from Elektro-Automatik PS 2042-20B power sup-
ply, with a current range of 0-20 A. EIS was acquired using an IviumStat.h with a
current range of 0-10 A.
8.5.1 Choice of Membrane
The gas separation properties of Sustainion and Zirfon in 1M KOH are compared to
Zirfon in 30 % wt. KOH. In a working electrolyser, Zirfon is used in 30 % wt. KOH
due to the advantages in terms of the reduced ionic resistance, whereas Sustainion
permits the use of less concentrate KOH (the manufacturer recommends 1 M) due to
its reduced thickness and resistance, but also the reduced KOH concentration is re-
quired due to the lack of chemical stability of the materials in a very strongly basic
environment. To provide a meaningful comparison between the two setups, their prop-
erties are also compared to Zirfon in 1M KOH, to allow the effect of the electrolyte
concentration to be accounted for.
Figure 8.2: Comparison of Zirfon UTP and Sustainion membranes at room temper-
ature and with flow rates of 100 ml/min for both the anodic and cathodic electrolyte
feeds.
Figure 8.2 shows that all membranes demonstrate a reduction in anodic hydrogen
content as current density increases, caused by the increasing rate of oxygen evolution
diluting the hydrogen that is cross permeating into the anodic compartment, therefore
reducing its fraction. The anodic hydrogen content for Zirfon using 30 % wt. KOH
is substantially lower than that of both Zirfon and Sustainion using 1M KOH. There
was less than 0.2% hydrogen detected in the oxygen stream at 100 mA·cm−2 when
compared to 0.85 and 2.97 % respectively for the Zirfon and Sustainion using 1M
KOH. This improved performance in concentrate potassium hydroxide is due to the
lower relative solubility of hydrogen when compared to the less concentrate solution.
Zirfon in 30 % wt. KOH has less than 0.025 % H2 in O2 at current densities above
1000 mA·cm−2 . These results match well with the literature values.152
Both Zirfon and Sustainion in 1M KOH have higher rates of mixing, but still
demonstrate levels of gas mixing below the explosive limit of 4% across the current
density range 100-2000 mA·cm−2 . At 500 mA·cm−2 , Sustainion records gas mixing
of less than 0.75% H2 in O2 , with the two membranes reaching a similar baseline of
around 0.25 % at current densities above 1000 mA·cm−2 . A safety limit is usually
applied to commercial electrolysers of 2% foreign gas content, and Sustainion fulfills
this criteria between 250 - 2000 mA·cm−2 .
Gas crossover comparison between membranes

0.015
Zirfon (1M KOH)
Gas crossover rate (µmol·cm-2s -1)
Zirfon (30% wt. KOH)

Sustanion (1M KOH)
0.01
0.005
0
0 500 1000 1500 2000
-2
Figure 8.3: Comparison of Zirfon UTP and Sustainion membranes at room temper-
ature and with flow rates of 100 ml/min for both the anodic and cathodic electrolyte
feeds.
Figure 8.3 shows the rates of gas crossover of the three membranes; Zirfon in the
strong electrolyte has an initial crossover rate of less than 0.1 µmol·cm−2 s−1 , more
than five times smaller than the comparative rate for Zirfon in 1M KOH. Gas crossover
at low current densities can be assumed to be almost entirely from diffusion, and so the
reduced hydrogen solubility of the concentrate KOH brings down the hydrogen diffu-
sivity of the gas separator, as seen by the relation in equation 8.8, reducing the rate of
gas mixing. Sustainion has the highest initial rate of gas crossover, three times higher
than Zirfon in the same electrolyte. Figure 8.3 shows that the rate of gas mixing in the
cell with Sustainion increases by less than 20 % across the full current range, whereas
the rate for Zirfon in the same electrolyte increases by more than 500 %. The rate of
gas cross permeation through Zirfon in 1M KOH is sensitive to current density, and
demonstrates a nearly linear increase in the rate of gas mixing across the current range,
such that at 2000 mA·cm−2 , Zirfon in 1M KOH has the highest rate of gas mixing.
In comparison, Zirfon in 30% KOH increases by approximately 150 % for the same
current density window. As the Zirfon is a microporous gas separator, it is suscepti-
ble to differential pressures. It appears that even small differential pressures building
up inside the cell have a substantial effect on the gas mixing when compared to the
solid polymer Sustainion, which is less permeable to these differential pressures. The
increased concentration of KOH increases the viscosity of the liquid, making Zirfon
in the concentrate electrolyte less susceptible to differential pressures than the same
material in the less concentrate electrolyte. This is consistent with observations in
Table 8.1 which showed that the rate of gas mixing due to differential pressure varies
by fourty times between the two electrolyte concentrations. These results were taken
from research by Schalenbach et al., who showed that even small discrepancies in the
absolute pressure difference between the two electrolyte chambers result in increased
gas mixing. System design will play a key role in regulating these pressure differ-
ences, but it is clear that use of a solid polymer electrolyte allows for a more robust
arrangement as it is less susceptible to gas mixing from small changes in differential
pressure.
8.5.2 Cathodic oxygen content

Although the anodic hydrogen content provides the biggest threat to safe running of
the electrolyser, knowing the cathodic oxygen content is also of interest with regards
to the performance of the cell, as well as to the post processing of the hydrogen before
it is stored and used in the desired application. Hydrogen purity is of key importance
to its use in fuel cells, and the extraction of the oxygen from the cathodic stream,
either by a recombination catalyst or another method, must be factored into the system
design.
Foreign gas content for Sustanion

4
Hydrogen in oxygen stream
3.5 Oxygen in hydrogen stream
Foreign gas content (%) 3
2.5
1.5
0.5
0
0 500 1000 1500 2000
-2
Figure 8.4: Anodic hydrogen and cathodic oxygen gas content for Sustainion
Figure 8.4 shows that the cathodic oxygen content is less than half that of the
anodic hydrogen content for the lower current densities. This is because the molar
production rate of hydrogen is double that of oxygen and so it dilutes the oxygen that
permeates through the membrane. The cathodic oxygen content is much less than 2
% even at 100 mA·cm−2 , and so provides no threat to the safe running of the system,
even at low current densities. The oxygen content falls below 0.1 % at current densities
above 1000 mA·cm−2 , and this high purity exhaust gas will reduce the post processing
cost of the gas, providing advantages of running at high current densities.
The rate of gas mixing is displayed in Figure 8.5, and shows that the oxygen
appears to have a lower rate of mixing than hydrogen at low current densities. Oxygen
that permeates from the anode to the cathode can be catalytically recombined to water
at the cathodic electrocatalyst, and this would reduce the apparent rate of gas mixing,
which could explain this discrepancy between the two rates of cross permeation. At
high current densities, the rates of gas mixing become similar. Both gases show a
similar increase in the rate of gas cross permeation across the full current density
window.
Foreign gas crossover rate for Sustanion

0.02
Gas crossover rate(µmol·cm-2·s -1)

Hydrogen in oxygen stream
Oxygen in hydrogen stream
0.015
0.01
0.005
0
0 500 1000 1500 2000
-2
Figure 8.5: The rate of cross permeation of both hydrogen and oxygen through the
Sustainion membrane
8.5.3 Comparison to literature

Due to the operation of the cell at atmospheric pressure, the theory states that the
gas mixing of the cell will follow Fick’s law of diffusion. To assess how closely the
experimental results match the theory, the two results are compared in Figure 8.6 for
Zirfon in 30% wt KOH.
The literature value for the ex-situ measurement of the diffusivity of hydrogen
in Zirfon employing 30 % wt KOH was 0.025 10−11 mol·cm−1 bar−1 s−1 , whereas for
the data collected here the hydrogen diffusivity in the same conditions was 1.77 10−10
mol·cm−1 bar−1 s−1 . Despite the order of magnitude difference between the two val-
ues, the gas separation performance of Zirfon in this study was consistent with other
research using it inside an alkaline electrolyser.152 A number of factors contribute to
the higher level of gas mixing observed in this cell, notably the use of the partially sep-
arated electrolyte cycles, where dissolved gases can still mix to some extent through
the connecting tubing. Secondly, due to the non-equilibrium conditions of electrolyte
gas evolution, the phenomenon of supersaturation occurs in the electrolyte, where the
Gas crossover comparison with mathematical model

0.2
Observed data
Ficks model
Anodic hydrogen content (%) 0.15
0.1
0.05
0
0 500 1000 1500 2000
-2
Figure 8.6: A comparison of the experimental and predicted data from literature
values, and assuming gas crossover solely by diffusion
concentration of the evolved gases in their dissolved states are more than under normal
equilibrium conditions. Finally, the lack of accurate pressure regulation inside the cell
allows gas mixing by differential pressures to occur within the cell.
8.5.4 Turndown Ratio
The idea of using a turndown ratio to assess the working window of membranes was
first proposed by Babic, this display emphasises the useful range of current densities
that membranes will function in. This is achieved setting criteria for the maximum
and minimum current densities, and Babic et al. set the upper limit to be a cell voltage
of 2 V, and the minimum current density to be the point at which a foreign gas content
of 2% H2 in O2 is reached. This idea is adapted for the alkaline system, with the upper
current density limit set in this case to be the point at which the voltage contribution
from the membrane exceeds 0.3 V. This allows the comparison of membranes to factor
in the key relationship between the resistance of a membrane and its gas separation
abilities. Using the figures supplied by the manufacturers regarding the resistances of
the two membranes used in this study, coupled with the gas separation results from
earlier in this chapter, a comparison of turndown ratio’s between Zirfon and Sustainion
can be made. Zirfon is reported to have a resistance of 0.3 Ω·cm2 , and Sustainion 0.09
Ω·cm2 .
Figure 8.7 shows that the Sustainion membrane has a current density range from
175 - 2000 mA·cm−2 , whereas the Zirfon UTP membrane is workable from 25 - 1000
mA·cm−2 . This demonstrates how the use of Sustainion opens up the potential to
work between 1000 - 2000 mA·cm−2 , although this is at the expense of lower current
densities (25 - 175 mA·cm−2 ). The advantages associated with increasing the current
density range from a CAPEX point of view make this an attractive option.
Figure 8.7: Turndown ratio comparison between Zirfon and Sustainion, the lower
limit is where there is greater than 2% H2 in O2 , and the upper limit where the
voltage contribution from the membrane resistance exceeds 0.3V.
8.6. C ELL P ERFORMANCE 155
8.6 Cell Performance
8.6.1 I-V Curves
The electrochemical performance of both zero gap cells employing Sustainion and
Zirfon membranes was characterised across a current range of 0 to 2 A·cm−2 at 40◦ C,
with 1M KOH used for the cell employing Sustainion and 30% wt KOH for the cell
using Zirfon. Figure 8.8 shows that the cell employing the Sustainion membrane hav-
ing better performance, particularly above 500 mA·cm−2 . The cells reach 1 A·cm−2
at 2.17 V and 2.39 V respectively. Both cells use the same electrodes, and so the prin-
cipal reason for the difference is the lower resistance associated with the Sustainion
cell. At 2 A·cm−2 , the Sustainion cell has a voltage of 2.48 V, nearly 400 mV lower
than that of the cell using Zirfon. These results demonstrate the electrochemical per-
formance improvement that can be achieved by utilising improved membrane in an
alkaline cell.
IV cell performance
2000
Cell with Sustainion
Cell with Zirfon
Current density (mA·cm -2)
1500
1000
500
0
1 1.5 2 2.5 3
Cell voltage (V)
Figure 8.8: The polarisation data for zero gap cells with Raney nickel cathodes
and Ni-Fe(OH)2 anodes, employing Sustainion and Zirfon membranes at 40◦ in
1M KOH and 30% KOH respectively.
8.6. C ELL P ERFORMANCE 156
8.6.2 Resistance of Membranes

Figure 8.9 shows that initially the resistance of the Sustainion cell is 0.24 Ω·cm2 ,
compared to 0.43 Ω·cm2 for the Zirfon cell, across the full current density range, the
Sustainion cell resistance increases by over 50 mΩ·cm2 , compared to just 22 mΩ·cm2
for the Zirfon cell. It is not clear why there is such a difference between the two, but
as the Zirfon membrane is more robust, it may allow more robust compression of the
cell, and tolerate a higher level of inhomogeneous compression, whereas the Sustain-
ion cell is more susceptible to short-circuiting when inhomogeneous compression is
applied, and as such gas may become trapped in any gaps between the electrode and
the membrane. Both values of Ohmic resistance are above the values quoted by the
manufacturers, of 0.09 and 0.3 Ω·cm2 respectively, with part of this due mesh elec-
trodes not providing the shortest path of ion conduction across the whole membrane.
This cell specific difference appears to be associated with the type of mesh used for
the electrode substrate.
At high current densities, the losses associated with Ohmic resistance are con-
verted into heat energy, and consequently cause a temperature gradient across the
course of the flow channel. By measuring the electrolyte temperature at both ther en-
trance and exit of the flow channel, this temperature was characterised as 11degC for
the cell employing Sustainion, and 16degC for the cell employing Zirfon. The differ-
ence between the two temperatures is due to the increased losses associated with the
increased resistance of the Zirfon membrane.
Overall the Sustainion membrane offers a significant step forward in cell perfor-
mance, attributed to the resistance savings from employing the thin anion exchange
membrane.
8.7. M ODEL 157
Ohmic resistance of cell

0.6
Cell with Sustainion
Cell with Zirfon
0.5
Cell resistance (Ω·cm 2 )
0.4
0.3
0.2
0.1
0
0 200 400 600 800 1000
-2
Figure 8.9: A comparison of the cell resistances, both cells see an increase in cell
resistance across the current density range. The change in temperature from the
entrance to the exit of the cell is notable at high temperatures, with ∆T = 11 and
16◦ C respectively for the Sustanion and Zirfon cells at 1 A·cm2 .
8.7 Model
The results from this investigation demonstrate the excellent gas separation properties
of the Zirfon membrane, however comparison with the Sustainion AEM show that
despite a decrease in gas separation abilities when compared to Zirfon, Sustainion
offers significant advantages to the electrochemical performance of the system due
to its reduced Ohmic resistance. In a proposed cell aiming to run at 2 A·cm−2 , as
demonstrated by the turndown ratio in Figure 8.7, the Sustainion would be the chosen
membrane between the two. It is clear that Zirfon could afford to sacrifice its gas
separation abilities to bring down its Ohmic resistance, in an attempt to open up a
wider range of realistic working current densities. To assess the potential of this, a
mathematical model is set up to investigate the trade off between membrane thickness,
Ohmic resistance and gas separation abilities.
The aim of the mathematical model is to assess the impact of reducing the thick-
ness of Zirfon on the gas crossover inside the cell. The model will compare the rate
8.7. M ODEL 158
d (µm) ROhmic (Ω · cm2 ) Voltage drop at 2 A · cm−2 (mV) ΦFick −1 −2

H2 (mol·s ·cm )
500 0.300 600 3.44 x 10−10
250 0.211 422 6.88 x 10−10
100 0.156 312 17.20 x 10−10
75 0.146 292 22.93 x 10−10
50 0.137 274 34.40 x 10−10
Table 8.3: The effect of the thickness of Zirfon on the resistance and gas crossover
properties of the membrane in 30 % wt. KOH
of hydrogen cross permeation using the current 500 µm thick Zirfon with a range of
thinner versions down to a thickness of 50 µm.
With a well designed system that made contributions from differential pressure
negligible, the crossover of gases would follow Ficks law of diffusion. Using Ficks
law as shown in equation 8.10, it can be seen that the rate of cross permeation of gas
is constant, and independent of the rate of gas production, provided that the solution is
saturated with dissolved gas, which it can be assumed happens at very low production
rates. Experimental results in section 8.5.1 revealed that the rate of gas cross perme-
ation was significantly higher than that calculated ex-situ.153 In this model the exper-
imental value from this study for the rate of hydrogen cross permeation is used, to
help provide realistic results for this system. The model parameters are listed in Table
8.3, with the Ohmic resistance for Zirfon using 30 % KOH electrolyte taken from the
manufacturers specifications. The rate of gas permeation is assumed to vary linearly
with thickness, with equation 8.10 showing that a decrease in membrane thickness by
a half will correspond to a doubling in the rate of hydrogen cross permeation.
The Ohmic resistance of the membrane cannot be treated as varying linearly with
thickness, work by Vogt et al. treated a similar polysulfone based membrane as having
two layers with distinct properties - the top layer offering the main gas separation
characteristics, and the bulk layer offering structural integrity.156 The resistivity of the
membrane is dependent on the resistance of the two layers.
Using the relation of resistivity and resistance:

8.7. M ODEL 159
RA
ρ= (8.15)
L
The resistivity of Zirfon can be defined as:
A ρt Lt + ρb (LZir f on − Lt ) Lt
ρZir f on = (Rt + Rb ) = = ρb + (ρt − ρb ) (8.16)
L LZir f on LZir f on
Where Rt and Rb are the resistance of the top and bulk layers of the membrane,
ρt and ρb are the resistivity of the respective layers, and Lt and Lb are the thicknesses
of the layers. Previous research demonstrated the difference of resistivity between the
top layer and the bulk of two orders of magnitude. Using the properties reported by
Vogt et al. for their 0.5mm polysulfone membrane, the combined thickness of the top
layers are assumed to be 5 µm thick, with a resistivity of 242 Ω·cm, and the bulk layer
to be 495 µm thick, with a resistivity of 3.69 Ω·cm.156 Applying these properties to
the equations above, the resistivity of the membrane is 6 Ω·cm, which matches closely
with that of Zirfon.
Figure 8.10: The Zirfon membrane is represented by two distinct layers, the top
layer of thickness Lt , and the bulk layer Lb
The calculations of the Ohmic resistance for reduced thickness membranes is

based around the membrane retaining the same thickness top layer throughout, but
reducing the thickness of the bulk, as shown in Figure 8.10. The calculated resistivity’s
for the membranes decreases with reduced membrane thickness using equation 8.16,
and are shown in Table 8.3. The results show that a membrane half as thick as the
8.7. M ODEL 160
current version of Zirfon would save early 90 mΩ·cm2 , and a version of Zirfon ten
times thinner would result in a membrane resistance of 0.137 Ω·cm2 .
Modelled gas crossover of Zirfon

1.5
500 µm
250 µm
Anodic hydrogen content (%)
100 µm
75 µm
1 50 µm
0.5
0
0 500 1000 1500 2000
-2
Figure 8.11: The effect of the thickness of Zirfon on the gas crossover inside the
cell using Ficks law of diffusion
As seen in Figure 8.11, the advantages of operating at high current densities is

clearly seen, due to the relative dilution of the foreign gas. It can also be seen that
current Zirfon membrane (500 µm) has excellent gas separation properties, with less
than 0.15 % H2 in O2 even at current densities as low 100 mA·cm−2 , and less than
0.1 % mixing at 2000 mA·cm−2 . Halving the thickness of the membrane increases
the gas mixing, but there is still less than 0.3% H2 in O2 at 100 mA·cm−2 , and similar
small quantities of mixing at high current densities.
For the thinnest membrane modelled, at a tenth of the current commercial thick-
ness, the gas mixing is still below 1.5 % H2 in O2 at 100 mA·cm−2 and under 0.3 %
at 500 mA·cm−2 , with further improvements with increasing current density.
These results demonstrate that the remarkable gas separation properties of Zirfon
could allow for a next generation version of the membrane facilitating the efficient
running of an alkaline electrolyser at current densities of 1-2 A·cm−2 . For compar-

ison with the PEM technology, the current thickness of Zirfon demonstrates a rate
of gas mixing eight times smaller than that seen in a PEM cell using a Nafion 117
membrane of thickness 180 µm. This difference is attributed to the reduced solubility
of hydrogen in concentrated KOH when compared to the pure water used for PEM
electrolysis. The relatively high resistance of Zirfon identified in Chapters 5 and 6,
combined with the excellent gas separation identified in this chapter shows a trade off
exists between the two. This model shows that a cell containing Zirfon could tolerate
a larger amount of gas mixing in a trade off to permit a lower membrane resistance,
and in turn more efficient cell.
8.8 Conclusions
This study demonstrated that Sustainion using 1M KOH electrolyte has good gas sep-
aration capabilities, with less than 1% H2 in O2 at 500 mA·cm−2 . The small variation
of the rate of gas crossover with current density demonstrates that Sustainion is re-
sistant to differential pressure build ups inside the cell. At 2 A·cm−2 , the Sustainion
membrane has a lower rate of gas mixing than Zirfon in the same electrolyte.
When compared to a regular liquid alkaline cell using 30 % wt. KOH electrolyte
and a Zirfon membrane, Zirfon is seen to have significantly better gas separation ca-
pabilities, with less than 0.2% H2 in O2 at 100 mA·cm−2 . By investigating the perfor-
mance of Zirfon in 1M KOH and comparing the results with the other two membranes
investigated, it is observed that the majority of the tenfold reduction in the rate of gas
mixing between Zirfon in 30 % wt. KOH and Sustainion in 1M KOH is due to the
decreased solubility of H2 in O2 .
The electrochemical performance of Sustainion is significantly better than the

current state of the art cell using a Zirfon membrane, attributed to the 44% reduction
in the Ohmic resistance of the cell using Sustainion at 1 A·cm−2 and 40◦ . Modelling
shows that thinner versions of Zirfon could reduce its high contribution to the cell
voltage from the Ohmic resistance, whilst maintaining tolerable levels of gas mixing.
A Zirfon membrane of thickness 100 µm would have an Ohmic resistance almost half
of the resistance associated with the current version, whilst maintaining a H2 in O2
percentage of less than 1% at 100 mA·cm−2 .
Overall, a high performance alkaline electrolysis cell is demonstrated, running at
current densities up to 2 A·cm−2 below 2.5 V. Gas separation characterisation is made
for the new to market Sustainion membrane, and suggest that this membrane opens up
the possibility of running alkaline cells up to 2 A·cm−2 at credible cell voltages.
Chapter 9
Conclusions and outlook
9.1 Conclusions
This thesis investigated alkaline electrolysis as a pathway towards cost-effective and
sustainable hydrogen production. A small zero gap test cell was developed, with the
efficiency of the system broken down into its component parts of reaction overpo-
tentials and Ohmic losses. The characterisation of high performing electrocatalysts
deposited onto porous electrode substrates was made, and, using a new to market an-
ion exchange membrane, the test cell reached 2 A·cm2 at 2.48 V in 1M KOH and
at 40◦ C. This is one of the first reported alkaline based cells that has been reported
working at such high current densities, and represents the new phase for the technol-
ogy, where a push towards high current densities is made with the aim of reducing the
capital costs of the system.
Quantifying the benefit of employing zero gap cell design in terms of cell per-
formance was a key step in informing future cell design. A 30 % lower initial Ohmic
resistance was demonstrated for the zero gap cell when compared to the traditional
finite gap setup; this cell design allows the alkaline cell to achieve higher efficiencies.
The cell design also demonstrates a smaller change in resistance when the current
density of the cell is increased, attributed to reduced bubble effects. This reveals the
potential of the cell design to be run at higher current densities than the finite gap cell.
The use of porous mesh electrodes allowed easy design, fabrication and assembly of
the cell, and the results show that the existence of the optimal electrode gap exhibited
by Nagai et al. can be overcome by employing a zero gap cell design.117
Electrocatalysts based on earth abundant metals were developed with high elec-
trochemical performance, good adhesion to the substrate and good electrochemical
stability at current densities appropriate for alkaline electrolysis. These coatings re-
duced the overpotentials at the anode and cathode by 0.11 V and 0.32 V respectively
when compared to a system based on pure nickel. The use of a nickel-iron oxyhydrox-
ide electrocatalyst for oxygen evolution, with an overpotential as low as 280 mV at 10
mA·cm−2 and a Tafel slope of just 37 mV·dec−1 demonstrated performance compa-
rable to the highest in the literature. The use of Raney nickel as a hydrogen evolving
electrocatalyst demonstrated an overpotential of 93 mV at 10 mA·cm−2 and a Tafel
slope of 62 mV·dec−1 .
A cell model of the zero gap electrolysis cell identified the Zirfon membrane as
the key limiting step to running an alkaline cell at current densities above 1 A·cm−2 .
The unacceptable high Ohmic resistance contributed to over 60 % of the cell voltage
at 2 A·cm−2 . The overpotential from the oxygen evolution reaction is seen to dom-
inate at low current densities, and demonstrates that further development of both the
membrane technology and anodic catalysts are key to increasing the efficiencies of the
alkaline system further
High speed photography was used to visualise the bubbly, slug and churn flow
regimes in two bespoke flowfield designs. This was the first of its kind for alkaline
electrolysis, and provided insight into the conditions at which phases changes occur.
By coupling mathematical modelling and the experimental data, a clear pathway to
calculating the flow regime of the liquid/gas mixture was outlined, with the mathe-
matical model based on the superficial velocities of the gas and liquid found to best
match the experimental data. An innovative method of image processing using Matlab
was developed to provide quantitative information on the flow regime occurring in the
flow channel, and results demonstrated that the transition from bubbly to slug flow oc-
curred slowly as the rate of gas evolution was increased, and a method for describing
9.2. O UTLOOK 165
the state of slug flow based on its percentage coverage was suggested.
Finally, the employment of the anion exchange membrane, Sustainion, showed

the potential to reduce the Ohmic resistance of the alkaline cell, with a resistance of
just 0.24 Ω· cm2 compared to 0.45 Ω· cm2 for Zirfon at 40 ◦ C. The gas separation
properties of Sustainion were characterised using online gas chromotography, with
less than 0.4 % H2 in the O2 stream at 1 A·cm−2 . The rate of gas crossover for the
membrane was barely affected by increased current density of the cell, there was less
than 30 % increase in the rate across a current density range of 0 - 2 A·cm−2 .
The characterisation of the exhaust gas streams for Zirfon demonstrated its excel-
lent gas separation abilities, with less than 0.2% H2 in O2 at 100 mA·cm−2 . This was
principally attributed to the fact that Zirfon works in 30% wt KOH, where hydrogen
solubility is very low. Modelling of the membrane suggested that thinner versions of
the membrane could enable a much lower Ohmic resistance, whilst still maintaining
high levels of gas separation.
Overall, this thesis addressed some key gaps in the literature regarding zero gap
electrolysis cells, and methods to achieve higher current densities and efficiencies
were explored to allow the technology to become competitive with its PEM counter-
part.
9.2 Outlook
The increasing pace of the energy transition heightens the need for energy storage.
Alkaline electrolysis has the potential to fulfill a large part in this unfolding shift to
renewable energy sources, with the potential for hydrogen to provide long term energy
storage, as well as an energy dense source of heat. There are still key challenges,
however, that the technology must overcome to provide meaningful competition to
the PEM technology which is currently the commercial technology of choice. Indeed,
during the three years studying for this thesis, significant advances have been made
with regards to the alkaline technology, and an increase in focus on this research topic
9.2. O UTLOOK 166
has been observed.

Firstly, zero gap cell design is an essential strategy for the transition, with finite
gap systems encountering unacceptably high contributions to the Ohmic resistance
from gas evolution at high current densities.
Although the catalyst coatings in this thesis show good performance compared
with current literature, further research must be focused on bringing down the overpo-
tentials associated with the hydrogen and oxygen evolution reactions, with a particular
focus on the large cell contribution from the anodic electrocatalyst.
The mass transport resistance inside all electrolysis cells is still poorly under-
stood, recent improvements suggest that progress is being made, but this is the last
contribution to the cell voltage that is currently left unaccounted for.107
The membrane is a key component to reduce the large contribution to the cell
voltage at high current densities. Sustainion offers a promising candidate to achieve
this, although further studies into the degradation of the membrane in working con-
ditions must be undertaken. Other anion exchange membranes are being developed,
and rigorous testing of the resistance, gas separation ability and durability of these
membranes must be undertaken to validate them for commercial applications. This
appears to be where the key breakthrough will take place to increase the performance
and accelerate the roll out of alkaline electrolysers.
The long term stability of electrodes at high current densities in alkaline con-
ditions has been poorly reported in the literature, and must be investigated further.
Investigations should be made with an emphasis on current densities at and above 1
A·cm2 and accelerated degradation protocols should be defined to simulate years of
operation in tough operating conditions.
Overall, the future is bright for alkaline electrolysis, the increasing cell efficien-
cies, coupled with rapidly increasing demand for energy storage, will make the next
ten years an exciting time for the technology.
Appendix A
List of Publications
1. Zero gap alkaline electrolysis cell design for renewable energy storage as
hydrogen gas
Robert Phillips and Charles W. Dunnill
RSC Advances, 2016, 6, 100643-100651
2. Minimising the Ohmic resistance of an alkaline electrolysis cell through

effective cell design
Robert Phillips, Adam Edwards, Bertrand Rome, Daniel R. Jones, Charles W.

Dunnill
International Journal of Hydrogen Energy, 2017, 42, 38, 23986-23994

Bibliography
1 S. Bilgen. Structure and environmental impact of global energy consumption. Re-

newable and Sustainable Energy Reviews, 38:890–902, 2014.
2P Nejat, F Jomehzadeh, M M Taheri, M Gohari, and M Majid. A global review

of energy consumption, co2 emissions and policy in the residential sector (with
an overview of the top ten co2 emitting countries). Renewable and Sustainable
Energy Reviews, 43:843–862, 2015.
3J Chow, R J. Kopp, and P R. Portney. Energy resources and global development.

Science, 302(5650):1528, 2003.
4J Hickman, D Hassel, R Joumard, Z Samaras, and S Sorenson. Methodology for

calculating transport emissions and energy consumption. 1999.
5P Crompton and Y Wu. Energy consumption in china: past trends and future
directions. Energy economics, 27(1):195–208, 2005.
6T Ekholm, V Krey, S Pachauri, and K Riahi. Determinants of household energy

consumption in india. Energy Policy, 38(10):5696–5707, 2010.
7 J E Jonson, J Borken-Kleefeld, D Simpson, A Nyri, M Posch, and C Heyes. Impact

of excess nox emissions from diesel cars on air quality, public health and eu-
trophication in europe. Environmental Research Letters, 12(9):094017, 2017.
8E J Hutchinson and P JG Pearson. An evaluation of the environmental and health

effects of vehicle exhaust catalysts in the uk. Environmental health perspec-
tives, 112(2):132, 2004.
9S Solomon, G Plattner, R Knutti, and P Friedlingstein. Irreversible climate change

B IBLIOGRAPHY 169
due to carbon dioxide emissions. Proceedings of the national academy of sci-

ences, 106(6):1704–1709, 2009.
10 S V Krupa and RN Kickert. The greenhouse effect: impacts of ultraviolet-b (uv-b)

radiation, carbon dioxide (co2 ), and ozone (o3 ) on vegetation. Environmental
Pollution, 61(4):263–393, 1989.
11 J Rogelj, A Popp, K V Calvin, G Luderer, J Emmerling, D Gernaat, S Fujimori,

J Strefler, T Hasegawa, and G Marangoni. Scenarios towards limiting global
mean temperature increase below 1.5◦ c. Nature Climate Change, 8(4):325,
2018.
12 Lazard. Lazard’s levelized cost of energy analysis version 11.0, 2018.
13 International Energy Agency Renewable Enegy Technology Department. Compar-

ative analysis of international offshore wind energy development, 2018.
14 Energy Department for Business and Industrial Strategy − UK Government. Uk

energy statistics. 2018.
15 Scottish Government. The future of energy in scotland: Scottish energy strategy,

2017.
16 IA G Wilson, A JR Rennie, Y Ding, P C Eames, P J Hall, and N J Kelly. Historical

daily gas and electrical energy flows through great britain’s transmission net-
works and the decarbonisation of domestic heat. Energy Policy, 61:301–305,
2013.
17 A Zahedi. Maximizing solar pv energy penetration using energy storage technol-

ogy. Renewable and Sustainable Energy Reviews, 15(1):866–870, 2011.
18 S Davis, N S. Lewis, M Shaner, S Aggarwal, D Arent, I Azevedo, S M. Benson,

T Bradley, J Brouwer, Y Chiang, C Clack, A Cohen, S Doig, J Edmonds, P Fen-
nell, C Field, B Hannegan, B Hodge, M Hoffert, and K Caldeira. Net−zero
emissions energy systems. Science, 360, 2018.
B IBLIOGRAPHY 170
19 C T. M. Clack, S A. Qvist, Jay Apt, M Bazilian, A R. Brandt, K Caldeira, S J.

Davis, V Diakov, M A. Handschy, P D. H. Hines, P Jaramillo, D M. Kammen,
J C. S. Long, M. G Morgan, A Reed, V Sivaram, J Sweeney, G R. Tynan, D G.
Victor, J P. Weyant, and J F. Whitacre. Evaluation of a proposal for reliable
low-cost grid power with 100% wind, water, and solar. Proceedings of the
national academy of sciences, 114(26):6722–6727, 2017.
20 G Berckmans, M Messagie, J Smekens, N Omar, L Vanhaverbeke, and

J Van Mierlo. Cost projection of state of the art lithium-ion batteries for electric
vehicles up to 2030. Energies, 10(9):1314, 2017.
21 A Hales, L B. Diaz, M W. Marzook, Y Zhao, Y Patel, and G Offer. The cell cool-
ing coefficient: A standard to define heat rejection from lithium-ion batteries.
Journal of The Electrochemical Society, 166(12):A2383–A2395, 2019.
22 A Hales and X Jiang. Optimisation of low energy cooling through phase variation
between adjacent piezoelectric fan blades. International Journal of Heat and
Mass Transfer, 139:362 – 372, 2019.
23 M R Kraglund. Alkaline membrane water electrolysis with non-noble catalysts.

Department of Energy Conversion and Storage, Technical University of Den-
mark, 2017.
24 H Ibrahim, A Ilinca, and J Perron. Energy storage systems−characteristics and

comparisons. Renewable and Sustainable Energy Reviews, 12(5):1221–1250,
2008.
25 A Ozarslan. Large-scale hydrogen energy storage in salt caverns. International

Journal of Hydrogen Energy, 37(19):14265–14277, 2012.
26 F Crotogino, S Donadei, U Bnger, and H Landinger. Large-scale hydrogen under-

ground storage for securing future energy supplies. In 18th World hydrogen
energy conference, volume 78, pages 37–45.
27 S H Jensen, P H Larsen, and M Mogensen. Hydrogen and synthetic fuel production

B IBLIOGRAPHY 171
from renewable energy sources. International Journal of Hydrogen Energy,

32(15):3253–3257, 2007.
28 D R Jones, W A Al-Masry, and C W Dunnill. Hydrogen-enriched natural gas as a

domestic fuel: an analysis based on flash-back and blow-off limits for domestic
natural gas appliances within the uk. Sustainable Energy & Fuels, 2(4):710–
723, 2018.
29 Z Heydarzadeh, D McVay, R Flores, C Thai, and J Brouwer. Dynamic modeling of

california grid-scale hydrogen energy storage. ECS Transactions, 86(13):245–
258, 2018.
30 M Gtz, J Lefebvre, F Mrs, A M. Koch, F Graf, S Bajohr, R Reimert, and T Kolb.

Renewable power-to-gas: A technological and economic review. Renewable
energy, 85:1371–1390, 2016.
31 Scottish Government. Decarbonisation of heat and the leeds city gate project,
2018.
32 RI Cox. Hydrogen: Its Technology and Implication: Production Technology, vol-

ume 1. CRC press, 2018.
33 A Hauch, S H Jensen, S Ramousse, and M Mogensen. Performance and durabil-

ity of solid oxide electrolysis cells. Journal of The Electrochemical Society,
153(9):A1741–A1747, 2006.
34 M Carmo, D L. Fritz, J Mergel, and D Stolten. A comprehensive review on pem

water electrolysis. International Journal of Hydrogen Energy, 38(12):4901–
4934, 2013.
35 R Phillips and C W. Dunnill. Zero gap alkaline electrolysis cell design for renew-
able energy storage as hydrogen gas. RSC Advances, 6(102):100643–100651,
2016.
36 O Schmidt, A Gambhir, I Staffell, A Hawkes, J Nelson, and S Few. Future cost and
B IBLIOGRAPHY 172
performance of water electrolysis: An expert elicitation study. international

journal of hydrogen energy, 42(52):30470–30492, 2017.
37 C Graves, S D Ebbesen, M Mogensen, and K S Lackner. Sustainable hydrocarbon
fuels by recycling co2 and h2 o with renewable or nuclear energy. Renewable
and Sustainable Energy Reviews, 15(1):1–23, 2011.
38 P Millet. Fundamentals of water electrolysis. Hydrogen Production: Electrolysis,
pages 33–62, 2015.
39 K Zeng and D Zhang. Recent progress in alkaline water electrolysis for hydro-
gen production and applications. Progress in Energy and Combustion Science,
36(3):307–326, 2010.
40 N Guillet and P Millet. Alkaline Water Electrolysis. 2015.
41 R Phillips, A Edwards, B Rome, D R Jones, and C W Dunnill. Minimising the
ohmic resistance of an alkaline electrolysis cell through effective cell design.
International Journal of Hydrogen Energy, 42(38):23986–23994, 2017.
42 D Pletcher, X Li, and S Wang. A comparison of cathodes for zero gap alkaline wa-
ter electrolysers for hydrogen production. International Journal of Hydrogen
Energy, 37(9):7429–7435, 2012.
43 D Pletcher and X Li. Prospects for alkaline zero gap water electrolysers for hy-
drogen production. International Journal of Hydrogen Energy, 36(23):15089–
15104, 2011.
44 D Bessarabov, H Wang, H Li, and N Zhao. PEM electrolysis for hydrogen produc-
tion: principles and applications. CRC press, 2016.
45 O. Schmidt, A. Gambhir, I. Staffell, A. Hawkes, J. Nelson, and S. Few. Future
cost and performance of water electrolysis: An expert elicitation study. Inter-
national Journal of Hydrogen Energy, 42(52):30470–30492, 2017.
46 R. J. Marshall and F. C. Walsh. A review of some recent electrolytic cell designs.
Surface Technology, 24(1):45–77, 1985.
B IBLIOGRAPHY 173
47 S. A. Grigoriev, V. I. Porembsky, and V. N. Fateev. Pure hydrogen production

by pem electrolysis for hydrogen energy. International Journal of Hydrogen
Energy, 31(2):171–175, 2006.
48 S Siracusano, V Baglio, N Van Dijk, L Merlo, and A S Aric. Enhanced perfor-

mance and durability of low catalyst loading pem water electrolyser based on
a short-side chain perfluorosulfonic ionomer. Applied Energy, 192:477–489,
2017.
49 M Bernt, A Siebel, and H A. Gasteiger. Analysis of voltage losses in pem water

electrolyzers with low platinum group metal loadings. Journal of The Electro-
chemical Society, 165(5):F305–F314, 2018.
50 E Price. Durability and degradation issues in pem electrolysis cells and its com-
ponents. Johnson Mattheys international journal of research exploring science
and technology in industrial applications, page 47, 2017.
51 D MF Santos, C AC Sequeira, and J L Figueiredo. Hydrogen production by alka-

line water electrolysis. Qumica Nova, 36(8):1176–1193, 2013.
52 RJ Gilliam, JW Graydon, DW Kirk, and SJ Thorpe. A review of specific conduc-

tivities of potassium hydroxide solutions for various concentrations and tem-
peratures. International Journal of Hydrogen Energy, 32(3):359–364, 2007.
53 A Lasia. Hydrogen evolution reaction. Handbook of fuel cells, 2, 2010.
54 J K Nrskov, T Bligaard, A Logadottir, JR Kitchin, J G Chen, S Pandelov, and

U Stimming. Trends in the exchange current for hydrogen evolution. Journal
of The Electrochemical Society, 152(3):J23–J26, 2005.
55 Y Li, H Wang, L Xie, Y Liang, G Hong, and H Dai. Mos2 nanoparticles grown on
graphene: an advanced catalyst for the hydrogen evolution reaction. Journal of
the American Chemical Society, 133(19):7296–7299, 2011.
56 Y Surendranath, M W. Kanan, and D G. Nocera. Mechanistic studies of the oxygen

B IBLIOGRAPHY 174
evolution reaction by a cobalt-phosphate catalyst at neutral ph. Journal of the

American Chemical Society, 132(46):16501–16509, 2010.
57 D. E. Hall. Ni(oh)2 impregnated anodes for alkaline water electrolysis. Journal of
The Electrochemical Society, 130(2):317–321, 1983.
58 D. E. Hall. Alkaline water electrolysis anode materials. Journal of The Electro-
chemical Society, 132(2):41C–48C, 1985.
59 J. Divisek, P. Malinowski, J. Mergel, and H. Schmitz. Improved components for
advanced alkaline water electrolysis. International Journal of Hydrogen En-
ergy, 13(3):141–150, 1988.
60 H. B. Suffredini, J. L. Cerne, F. C. Crnkovic, S. A. S. Machado, and L. A. Avaca.
Recent developments in electrode materials for water electrolysis. Interna-
tional Journal of Hydrogen Energy, 25(5):415–423, 2000.
61 I A Raj. Nickel based composite electrolytic surface coatings as electrocatalysts
for the cathodes in the energy efficient industrial production of hydrogen from
alkaline water electrolytic cells. International Journal of Hydrogen Energy,
17(6):413–421, 1992.
62 A Y Faid, A O Barnett, F Seland, and S Sunde. Highly active nickel-based catalyst
for hydrogen evolution in anion exchange membrane electrolysis. 2018.
63 X Tang, L Xiao, C Yang, J Lu, and L Zhuang. Noble fabrication of nimo cathode
for alkaline water electrolysis and alkaline polymer electrolyte water electrol-
ysis. International Journal of Hydrogen Energy, 39(7):3055–3060, 2014.
64 W Hu. Electrocatalytic properties of new electrocatalysts for hydrogen evolu-
tion in alkaline water electrolysis. International Journal of Hydrogen Energy,
25(2):111–118, 2000.
65 M Schalenbach, O Kasian, and K J. J. Mayrhofer. An alkaline water electrolyzer
with nickel electrodes enables efficient high current density operation. Inter-
B IBLIOGRAPHY 175
66 J Luo, J Im, M T Mayer, M Schreier, M K Nazeeruddin, N Park, S D Tilley,

H J Fan, and M Grtzel. Water photolysis at 12.3% efficiency via perovskite
photovoltaics and earth-abundant catalysts. Science, 345(6204):1593–1596,
2014.
67 G Chen, T Ma, Z Liu, N Li, Y Su, K Davey, and S Qiao. Efficient and stable bifunc-
tional electrocatalysts ni/nix my (m= p, s) for overall water splitting. Advanced
Functional Materials, 26(19):3314–3323, 2016.
68 Y Jin, X Yue, C Shu, S Huang, and P Shen. Three-dimensional porous moni4

networks constructed by nanosheets as bifunctional electrocatalysts for overall
water splitting. Journal of Materials Chemistry A, 5(6):2508–2513, 2017.
69 S Tanaka, N Hirose, T Tanaki, and Y H. Ogata. The effect of tin ingredients on

electrocatalytic activity of raney-ni prepared by mechanical alloying. Interna-
tional Journal of Hydrogen Energy, 26(1):47–53, 2001.
70 M. H. Miles, Y. H. Huang, and S. Srinivasan. The oxygen electrode reaction in

alkaline solutions on oxide electrodes prepared by the thermal decomposition
method. Journal of The Electrochemical Society, 125(12):1931–1934, 1978.
71 D A. Corrigan. The catalysis of the oxygen evolution reaction by iron impurities

in thin film nickel oxide electrodes. Journal of The Electrochemical Society,
134(2):377–384, 1987.
72 M Gong and H Dai. A mini review of nife-based materials as highly active oxygen
evolution reaction electrocatalysts. Nano Research, 8(1):23–39, 2015.
73 J. M. Gras and P. Spiteri. Corrosion of stainless steels and nickel-based alloys

for alkaline water electrolysis. International Journal of Hydrogen Energy,
18(7):561–566, 1993.
74 X Li, F C. Walsh, and D Pletcher. Nickel based electrocatalysts for oxygen evo-
lution in high current density, alkaline water electrolysers. Physical Chemistry
Chemical Physics, 13(3):1162–1167, 2011.
B IBLIOGRAPHY 176
75 X Lu and C Zhao. Electrodeposition of hierarchically structured three-dimensional

nickeliron electrodes for efficient oxygen evolution at high current densities.
Nat Commun, 6, 2015.
76 R. N. Singh, J. P. Pandey, N. K. Singh, B. Lal, P. Chartier, and J. F. Koenig. Sol-gel

derived spinel mxco3 xo4 (m=ni, cu; 0x1) films and oxygen evolution. Elec-
trochimica Acta, 45(12):1911–1919, 2000.
77 J Jiang, A Zhang, L Li, and L Ai. Nickel−cobalt layered double hydroxide

nanosheets as high-performance electrocatalyst for oxygen evolution reaction.
Journal of Power Sources, 278:445–451, 2015.
78 Y Xiao, L Feng, C Hu, V Fateev, C Liu, and W Xing. Nico2 o4 3−dimensional

nanosheet as effective and robust catalyst for oxygen evolution reaction. RSC
Advances, 5(76):61900–61905, 2015.
79 V. M. Rosa, M. B. F. Santos, and E. P. da Silva. New materials for water electrolysis

diaphragms. International Journal of Hydrogen Energy, 20(9):697–700, 1995.
80 Ph Vermeiren, W Adriansens, JP Moreels, and R Leysen. The Composite Zirfon

Separator for Alkaline Water Electrolysis, pages 179–184. Springer, 1998.
81 J. Kerres, G. Eigenberger, S. Reichle, V. Schramm, K. Hetzel, W. Schnurnberger,

and I. Seybold. Advanced alkaline electrolysis with porous polymeric di-
aphragms. Desalination, 104(1):47–57, 1996.
82 Ph Vermeiren, W. Adriansens, J. P. Moreels, and R. Leysen. Evaluation of the

zirfon separator for use in alkaline water electrolysis and ni-h2 batteries. Inter-
83 P Vermeiren, JP Moreels, and R Leysen. Porosity in composite zirfon membranes.

Journal of Porous Materials, 3(1):33–40, 1996.
84 H Wendt and H Hofmann. Ceramic diaphragms for advanced alkaline water elec-
trolysis. Journal of Applied Electrochemistry, 19(4):605–610, 1989.
B IBLIOGRAPHY 177
85 W Ju, M VF Heinz, L Pusterla, M Hofer, B Fumey, R Castiglioni, M Pagani,

C Battaglia, and U F Vogt. Lab-scale alkaline water electrolyzer for bridging
material fundamentals with realistic operation. ACS Sustainable Chemistry &
Engineering, 6(4):4829–4837, 2018.
86 Y Leng, G Chen, A J Mendoza, T B Tighe, M A Hickner, and C Wang. Solid-

state water electrolysis with an alkaline membrane. Journal of the American
Chemical Society, 134(22):9054–9057, 2012.
87 X Wu and K Scott. A polymethacrylate-based quaternary ammonium oh ionomer

binder for non-precious metal alkaline anion exchange membrane water elec-
trolysers. Journal of Power Sources, 214:124–129, 2012.
88 Z Liu, S D Sajjad, Y Gao, J Kaczur, and R Masel. An alkaline water electrolyzer

with sustainion membranes: 1 a·cm2 at 1.9 v with base metal catalysts. ECS
Transactions, 77(9):71–73, 2017.
89 Z Liu, S Sajjad, Y Gao, H Yang, J J Kaczur, and R I Masel. The effect of membrane
on an alkaline water electrolyzer. International Journal of Hydrogen Energy,
42(50):29661–29665, 2017.
90 MI Gillespie and RJ Kriek. Scalable hydrogen production from a mono-circular

filter press divergent electrode-flow-through alkaline electrolysis stack. Jour-
nal of Power Sources, 397:204–213, 2018.
91 MI Gillespie, F van der Merwe, and RJ Kriek. Performance evaluation of a mem-

braneless divergent electrode-flow-through (deft) alkaline electrolyser based on
optimisation of electrolytic flow and electrode gap. Journal of Power Sources,
293:228–235, 2015.
92 M Kraglund, D Aili, K Jankova, E Christensen, Q Li, and J Jensen. Zero-

gap alkaline water electrolysis using ion-solvating polymer electrolyte mem-
branes at reduced koh concentrations. Journal of The Electrochemical Society,
163(11):F3125–F3131, 2016.
B IBLIOGRAPHY 178
93 M R Kraglund, D Aili, K Jankova, E Christensen, Q Li, and J O. Jensen. Zero-

gap alkaline water electrolysis using ion-solvating polymer electrolyte mem-
branes at reduced koh concentrations. Journal of The Electrochemical Society,
163(11):F3125–F3131, 2016.
94 S Karimi, N Fraser, B Roberts, and F R Foulkes. A review of metallic bipolar plates

for proton exchange membrane fuel cells: materials and fabrication methods.
Advances in Materials Science and Engineering, 2012, 2012.
95 G. Schiller, R. Henne, and V. Borck. Vacuum plasma spraying of high-performance

electrodes for alkaline water electrolysis. Journal of Thermal Spray Technol-
ogy, 4(2):185–194, 1995.
96 S Ahn and B Lee. Development of a membrane electrode assembly for alkaline

water electrolysis by direct electrodeposition of nickel on carbon papers. Ap-
plied Catalysis B: Environmental, 154:197–205, 2014.
97 L Xiao, S Zhang, J Pan, C Yang, M He, L Zhuang, and J Lu. First implementation
of alkaline polymer electrolyte water electrolysis working only with pure water.
Energy and Environmental Science, 5(7):7869–7871, 2012.
98 J Kim, J Lee, C Yoo, K Lee, and W Lee. Low-cost and energy-efficient asym-
metric nickel electrode for alkaline water electrolysis. International Journal of
Hydrogen Energy, 40(34):10720–10725, 2015.
99 F Allebrod, C Chatzichristodoulou, and M B. Mogensen. Alkaline electrolysis

cell at high temperature and pressure of 250◦ c and 42 bar. Journal of Power
Sources, 229:22–31, 2013.
100 C. Chatzichristodoulou, F. Allebrod, and M. B. Mogensen. High temperature alka-

line electrolysis cells with metal foam based gas diffusion electrodes. Journal
of The Electrochemical Society, 163(11):3036–3040, 2016.
B IBLIOGRAPHY 179
101 T Pandiarajan, L John Berchmans, and S Ravichandran. Fabrication of spinel fer-

rite based alkaline anion exchange membrane water electrolysers for hydrogen
production. RSC Advances, 5(43):34100–34108, 2015.
102 M Schalenbach, G Tjarks, M Carmo, W Lueke, M Mueller, and D Stolten. Acidic

or alkaline? towards a new perspective on the efficiency of water electrolysis.
Journal of The Electrochemical Society, 163(11):F3197–F3208, 2016.
103 U Babic, M Suermann, F Bchi, L Gubler, and T J. Schmidt. Critical review iden-
tifying critical gaps for polymer electrolyte water electrolysis development.
Journal of The Electrochemical Society, 164(4):387–399, 2017.
104 I Dedigama, P Angeli, N van Dijk, J Millichamp, D Tsaoulidis, P R. Shearing,

and D J. L. Brett. Current density mapping and optical flow visualisation of a
polymer electrolyte membrane water electrolyser. Journal of Power Sources,
265:97–103, 2014.
105 I. Dedigama, P. Angeli, K. Ayers, J. B. Robinson, P. R. Shearing, D. Tsaoulidis,

and D. J. L. Brett. In situ diagnostic techniques for characterisation of poly-
mer electrolyte membrane water electrolysers flow visualisation and electro-
chemical impedance spectroscopy. International Journal of Hydrogen Energy,
39(9):4468–4482, 2014.
106 J van der Merwe, K Uren, G van Schoor, and D Bessarabov. Characterisation tools
development for pem electrolysers. International Journal of Hydrogen Energy,
39(26):14212–14221, 2014.
107 M Suermann, T J Schmidt, and F N Bchi. Investigation of mass transport losses in

polymer electrolyte electrolysis cells. ECS Transactions, 69(17):1141–1148,
2015.
108 R Masel, Z Liu, H Yang, S D. Sajjad, Y Gao, and J J. Kaczur. A high performing
zero gap alkaline electrolyzer. Meeting Abstracts, MA2017-02(21):1006, 2017.
109 P Haug, M Koj, and T Turek. Influence of process conditions on gas purity
B IBLIOGRAPHY 180
in alkaline water electrolysis. International Journal of Hydrogen Energy,

42(15):9406–9418, 2017.
110 G Passas and CW Dunnill. Water splitting test cell for renewable energy storage
as hydrogen gas. J Fundam Renewable Energy Appl 5:, (5):188, 2015. -.
111 M. A. Laguna-Bercero. Recent advances in high temperature electrolysis using
solid oxide fuel cells: A review. Journal of Power Sources, 203:4–16, 2012.
112 J Xu, G Liu, J Li, and X Wang. The electrocatalytic properties of an iro2 /sno2 cat-
alyst using sno2 as a support and an assisting reagent for the oxygen evolution
reaction. Electrochimica Acta, 59:105–112, 2012.
113 M Shalom, D Ressnig, X Yang, G Clavel, T P. Fellinger, and M Antonietti. Nickel
nitride as an efficient electrocatalyst for water splitting. Journal of Materials
Chemistry A, 3(15):8171–8177, 2015.
114 F. J. Perez-Alonso, C. Adn, S. Rojas, M. A. Pea, and J. L. G. Fierro. Ni/fe elec-
trodes prepared by electrodeposition method over different substrates for oxy-
gen evolution reaction in alkaline medium. International Journal of Hydrogen
Energy, 39(10):5204–5212, 2014.
115 S Marini, P Salvi, P Nelli, R Pesenti, M Villa, M Berrettoni, G Zangari, and
YohYannes Kiros. Advanced alkaline water electrolysis. Electrochimica Acta,
82:384–391, 2012.
116 D MF Santos, C Sequeira, and J Figueiredo. Hydrogen production by alkaline
water electrolysis. Quı́mica Nova, 36(8):1176–1193, 2013.
117 N. Nagai, M. Takeuchi, T. Kimura, and T. Oka. Existence of optimum space be-
tween electrodes on hydrogen production by water electrolysis. International
118 D Zhang and K Zeng. Evaluating the behavior of electrolytic gas bubbles and
their effect on the cell voltage in alkaline water electrolysis. Industrial and
Engineering Chemistry Research, 51(42):13825–13832, 2012.
B IBLIOGRAPHY 181
119 B. E. Bongenaar-Schlenter, L. J. J. Janssen, S. J. D. Van Stralen, and E. Baren-

drecht. The effect of the gas void distribution on the ohmic resistance during
water electrolytes. Journal of Applied Electrochemistry, 15(4):537–548, 1985.
120 Y Kobayashi, K Kosaka, T Yamamoto, Y Tachikawa, K Ito, and K Sasaki. A solid

polymer water electrolysis system utilizing natural circulation. International
121 I Vincent, A Kruger, and D Bessarabov. Development of efficient membrane elec-

trode assembly for low cost hydrogen production by anion exchange membrane
electrolysis. International Journal of Hydrogen Energy, 42(16):10752–10761,
2017.
122 H Ito, T Maeda, A Nakano, C Hwang, M Ishida, A Kato, and T Yoshida. Exper-
imental study on porous current collectors of pem electrolyzers. International
123 H Ito, T Maeda, A Nakano, A Kato, and T Yoshida. Influence of pore structural
properties of current collectors on the performance of proton exchange mem-
brane electrolyzer. Electrochimica Acta, 100:242–248, 2013.
124 R. L. LeRoy, M. B. I. Janjua, R. Renaud, and U. Leuenberger. Analysis of timevari-

ation effects in water electrolyzers. Journal of The Electrochemical Society,
126(10):1674–1682, 1979.
125 K. Zeng and D Zhang. Evaluating the effect of surface modifications on ni

based electrodes for alkaline water electrolysis. Fuel, 116:692–698, 2014.
10.1016/j.fuel.2013.08.070.
126 I. Herraiz-Cardona, C. Gonzlez-Buch, C. Valero-Vidal, E. Ortega, and V. Prez-

Herranz. Co-modification of ni-based type raney electrodeposits for hydro-
genevolution reaction in alkaline media. Journal of Power Sources, 240:698–
704, 2013.
127 I. Herraiz-Cardona, E. Ortega, J. G. Antn, and V. Prez-Herranz. Assessment of the

B IBLIOGRAPHY 182
roughness factor effect and the intrinsic catalytic activity for hydrogen evolu-
tion reaction on ni-based electrodeposits. Int. J. Hydrogen Energy, 36, 2011.
128 H Dong, T Lei, Y He, N Xu, B Huang, and C. T. Liu. Electrochemical performance
of porous ni3 al electrodes for hydrogen evolution reaction. International Jour-
nal of Hydrogen Energy, 36(19):12112–12120, 2011.
129 M B. Stevens, L J. Enman, A S. Batchellor, M R. Cosby, A E. Vise, C D. M.

Trang, and S W. Boettcher. Measurement techniques for the study of thin film
heterogeneous water oxidation electrocatalysts. Chemistry of Materials, 2016.
130 P. Lettenmeier, R. Wang, R. Abouatallah, B. Saruhan, O. Freitag, P. Gazdzicki,

T. Morawietz, R. Hiesgen, A. S. Gago, and K. A. Friedrich. Low-cost and
durable bipolar plates for proton exchange membrane electrolyzers. Scientific
Reports, 7:44035, 2017.
131 S Siracusano, V Baglio, N Briguglio, G Brunaccini, A Di Blasi, A Stassi, R Or-

nelas, E Trifoni, V Antonucci, and AS Aricò. An electrochemical study of a
pem stack for water electrolysis. International Journal of hydrogen energy,
37(2):1939–1946, 2012.
132 M Schalenbach, F D. Speck, M Ledendecker, O Kasian, D Goehl, A M.

Mingers, B Breitbach, H Springer, S Cherevko, and K J. Mayrhofer. Nickel-
molybdenum alloy catalysts for the hydrogen evolution reaction: Activity and
stability revised. Electrochimica Acta, 259:1154–1161, 2018.
133 L Trotochaud, S L. Young, J K. Ranney, and S W. Boettcher. Nickel−iron

oxyhydroxide oxygen-evolution electrocatalysts: The role of intentional and
incidental iron incorporation. Journal of the American Chemical Society,
136(18):6744–6753, 2014.
134 K A. Lewinski, D van der Vliet, and S M. Luopa. Nstf advances for pem electrol-
ysis - the effect of alloying on activity of nstf electrolyzer catalysts and per-
formance of nstf based pem electrolyzers. ECS Transactions, 69(17):893–917,
B IBLIOGRAPHY 183
2015.
135 Transporte y Almacenamiento de hidrocarburos Ingenieria de Produccion de

Petroleo. Flow regimes in horizontal and vertical pipes, 2013.
136 F. Zavareh, A. D. Hill, and A. Podio. Flow regimes in vertical and inclined
oil/water flow in pipes, 1988.
137 N A. Radovcich. The transition from two phase bubble flow to slug flow. PhD
thesis, 1962.
138 K Mishima and M Ishii. Flow regime transition criteria for upward two-phase flow
in vertical tubes. International Journal of Heat and Mass Transfer, 27(5):723–
737, 1984.
139 K Mishima and T Hibiki. Some characteristics of air-water two-phase flow in small
diameter vertical tubes. International journal of multiphase flow, 22(4):703–
712, 1996.
140 J Seweryn, J Biesdorf, T J Schmidt, and P Boillat. Communication−neutron ra-

diography of the water/gas distribution in the porous layers of an operating
electrolyser. Journal of The Electrochemical Society, 163(11):F3009–F3011,
2016.
141 O F Selamet, U Pasaogullari, D Spernjak, D S Hussey, D L Jacobson, and M Mat.

In situ two-phase flow investigation of proton exchange membrane (pem) elec-
trolyzer by simultaneous optical and neutron imaging. ECS Transactions,
41(1):349–362, 2011.
142 S Sun, Y Xiao, D Liang, Z Shao, H Yu, M Hou, and B Yi. Behaviors of a pro-
ton exchange membrane electrolyzer under water starvation. RSC Advances,
5(19):14506–14513, 2015.
143 T Ous and C Arcoumanis. Visualisation of water droplets during the operation of
pem fuel cells. Journal of Power Sources, 173(1):137–148, 2007.
B IBLIOGRAPHY 184
144 DJ Ludlow, CM Calebrese, SH Yu, CS Dannehy, DL Jacobson, DS Hussey,

Muhammad Arif, MK Jensen, and GA Eisman. Pem fuel cell membrane hydra-
tion measurement by neutron imaging. Journal of Power Sources, 162(1):271–
278, 2006.
145 K R Minard, V V Viswanathan, P D Majors, L Wang, and P C Rieke. Magnetic

resonance imaging (mri) of pem dehydration and gas manifold flooding during
continuous fuel cell operation. Journal of power sources, 161(2):856–863,
2006.
146 F. Aubras, J. Deseure, J. J. A. Kadjo, I. Dedigama, J. Majasan, B. Grondin-Perez,

J. P. Chabriat, and D. J. L. Brett. Two-dimensional model of low-pressure pem
electrolyser: Two-phase flow regime, electrochemical modelling and experi-
mental validation. International Journal of Hydrogen Energy, 42(42):26203–
26216, 2017.
147 J Mo. Fundamental studies of electrochemical reactions and microfluidics in pro-

ton exchange membrane electrolyzer cells. 2016.
148 J O. Majasan, J I. S. Cho, I Dedigama, D Tsaoulidis, P Shearing, and D J. L. Brett.

Two-phase flow behaviour and performance of polymer electrolyte membrane
electrolysers: Electrochemical and optical characterisation. International Jour-
nal of Hydrogen Energy, 43(33):15659–15672, 2018.
149 I Vincent, A Kruger, and D Bessarabov. Hydrogen production by water electrolysis

with an ultrathin anion-exchange membrane (aem). Int. J. Electrochem. Sci,
13:11347–11358, 2018.
150 U Maas and J Warnatz. Ignition processes in hydrogen-oxygen mixtures. Com-

bustion and Flame, 74(1):53–69, 1988.
151 P Haug, B Kreitz, M Koj, and T Turek. Process modelling of an alkaline water
electrolyzer. International Journal of Hydrogen Energy, 42(24):15689–15707,
2017.
B IBLIOGRAPHY 185
152 P Trinke, P Haug, J Brauns, B Bensmann, R Hanke-Rauschenbach, and T Turek.

Hydrogen crossover in pem and alkaline water electrolysis: Mechanisms, di-
rect comparison and mitigation strategies. Journal of The Electrochemical So-
ciety, 165(7):F502–F513, 2018.
153 M Schalenbach, W Lueke, and D Stolten. Hydrogen diffusivity and electrolyte
permeability of the zirfon perl separator for alkaline water electrolysis. Journal
of The Electrochemical Society, 163(14):F1480–F1488, 2016.
154 John R. Varcoe, Plamen Atanassov, Dario R. Dekel, Andrew M. Herring,
Michael A. Hickner, Paul. A. Kohl, Anthony R. Kucernak, William E. Mustain,
Kitty Nijmeijer, Keith Scott, Tongwen Xu, and Lin Zhuang. Anion-exchange
membranes in electrochemical energy systems. Energy Environ. Sci., 7:3135–
3191, 2014.
155 J. Balej. Water vapour partial pressures and water activities in potassium and
sodium hydroxide solutions over wide concentration and temperature ranges.
International Journal of Hydrogen Energy, 10(4):233 – 243, 1985.
156 D Burnat, M Schlupp, A Wichser, B Lothenbach, M Gorbar, A Zttel, and U F.
Vogt. Composite membranes for alkaline electrolysis based on polysulfone
and mineral fillers. Journal of Power Sources, 291:163–172, 2015.

Robert Phillips PHD Submission

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Robert Phillips PHD Submission

Uploaded by

Copyright:

Available Formats

Z ERO GAP CELL DESIGN FOR

E NERGY S AFETY R ESEARCH I NSTITUTE

Date: February, 2019

Date: February, 2019

4.4 Four-terminal sensing . . . . . . . . . . . . . . . . . . . . . . . . . . 61

5 Minimising the Ohmic resistance of an alkaline electrolysis cell 63

7 Characterisation of two phase flow 106

8 Zero gap electrolysis with an anion exchange membrane 135

9 Conclusions and outlook 163

A List of Publications 167

1.1 2015 energy consumption of Scotland . . . . . . . . . . . . . . . . . 7

2.1 Alkaline electrolysis cell . . . . . . . . . . . . . . . . . . . . . . . . 14

3.1 Electrical circuit analogy . . . . . . . . . . . . . . . . . . . . . . . . 30

4.1 EIS signal and response . . . . . . . . . . . . . . . . . . . . . . . . . 52

5.1 Finite gap vs zero gap cell design . . . . . . . . . . . . . . . . . . . . 64

6.1 SEM, Ni-Fe electodes (x300) . . . . . . . . . . . . . . . . . . . . . . 92

7.1 Flow regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

8.1 Gas analysis setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

2.1 Cell configurations: advantages and disadvantages . . . . . . . . . . . 29

3.1 Electrocatalyst performance for HER . . . . . . . . . . . . . . . . . . 34

6.1 Composition and performance of Ni-Fe electrode . . . . . . . . . . . 93

7.1 Flow regime model properties . . . . . . . . . . . . . . . . . . . . . 116

8.1 Diffusivity of Zirfon . . . . . . . . . . . . . . . . . . . . . . . . . . 143

AEM Anion Exchange Membrane

CCM Catalyst Coated Membrane

CCS Catalyst Coated Substrate

EDX Energy-dispersive X-ray Spectroscopy

EIS Electrochemical Impedance Spectroscopy

HER Hydrogen Evolution Reaction

LSV Linear Sweep Voltammetry

MEA Membrane Electrode Assembly

OER Oxygen Evolution Reaction

PEM Proton Exchange Membrane

SEM Scanning Electron Microscopy

As the world moves towards an energy landscape that is dominated by renewable

Decreasing availability of fossil fuels, the competitive price of renewable energy

1.1 Energy source transition

1.2 Fossil fuel usage

1.3 Demand for renewable energy

Renewable energy production allows for potential energy independence for a

1.4 Decarbonisation of the energy sector

1.5 Energy storage and distribution

energy capacities and storage duration’s.

The ability to soak up large quantities of renewable generations, and be able to

1.6 Hydrogen production

1.7 Project aims

2. Characterise each contribution to the cell voltage using electrochemical tech-

2.1 Cell potential of water electrolysis cell

2H2 O(l)− > 2H2 (g) + O2 (g) (2.1)

This is achieved by connecting a metallic anode and cathode to a DC power

2H + + 2e− − > H2 (E 0 = 0.00V vsSHE) (2.2)

to those necessary to withstand electrolysis in an acidic system. The electrochemical

2H2 O + +2e− − > H2 + 2OH − (E 0 = −0.83V vsSHE) (2.4)

ERev = ERight − ELe f t = ECathode − EAnode (2.6)

It can therefore be seen that for both conditions, ERev = -1.23 V.

2.2 Alkaline electrolysis

Figure 2.1: An alkaline electrolyser requires two metallic plates separated by a

2.3 PEM electrolysis

The enthalpy change necessary to dissociate one mole of water is ∆H = 285.84 kj

∆Gd = −nFECell (2.8)

where F is the Faraday constant (96,485 C) and n is the number of electrons

2.5 Actual cell voltage

ECell = ERev + ηAnode + ηCathode + I · RCell + ηMT (2.9)