WEI Yin 2019

Sorbonne Université
ED 397 – Physique et Chimie des Matériaux
Collège de France
Chaire de Chimie du Solide et de l’Energie
Fundamental understanding of high-energy-density Li-ion batteries

and beyond: structure-property relationships and reaction
chemistries
Wei YIN
Doctor thesis in Chemistry/Materials Science
Supervised by Jean-Marie TARASCON and Alexis GRIMAUD
To be presented and defended in public on 18 Octorber 2019 in front of the jury:
Dr. Fanny Bardé Program manager, IMEC, Belgium Referee

Prof. Betar M. Gallant Assistant Professor, Massachusetts Institute of Technology Referee
Prof. Mathieu Salanne Professor, Sorbonne Université, Paris Examiner

Prof. Ying Shirley Meng Professor, University of California, San Diego Examiner
Dr. Erik J. Berg Associate senior lecturer, Uppsala University, Sweden Examiner
Dr. Alexis Grimaud CNRS researcher, Collège de France, Paris Co-supervisor

Prof. Jean-Marie Tarascon Professor, Collège de France, Paris Supervisor
To my dear parents
and
to my dear sisters
Acknowledgements
First and foremost, I would like to express my sincerest gratitude to my supervisors Prof.
Jean-Marie Tarascon and Dr. Alexis Grimaud. Their passion to science and availability to students
will always be a source of inspiration and motivation for me. They are great mentors and I truly
appreciate their professional guidance, scientific support and kind encouragement during the last
four years.
I gratefully acknowledge the referees and examiners of my PhD committee Prof. Betar M.
Gallant, Dr. Fanny Bardé, Prof. Mathieu Salanne, Prof. Ying Shirley Meng and Dr. Erik J. Berg for
spending time reading my thesis, and for giving insightful comments and suggestions.
My appreciations also go to Dr. Florent Lepoivre for spending time during his thesis writing
to answer my questions and to teach me how to play with the pressure cell at the beginning of my
PhD, and to Dr. Daniel Alves Dalla Corte for sharing his expertise when designing the gas line and
OEMS cells. It was a great pleasure to work with both of them. I learned from them that
engineering can be elegant.
I would like to specially thank all the local and international collaborators for their technical
supports and fruitful discussions: Dr. Daniel Alves Dalla and Dr. Arash Jamali for SEM imaging;
Dr. Grégory Gachot for GC-MS measurements; Prof. Gwenaëlle Rousse for analyzing the
synchrotron XRD and neutron powder diffraction patterns; Prof. Artem Abakumov for TEM studies;
Dr. Leiting Zhang and Dr. Sigita Trabesinger for OEMS measurements; Dr. Antonella Iadecola for
XAS experiments; Dr. Dominique Foix for XPS experiments and Dr. Domitille Giaume for ICP-
OES experiments.
Throughout the four-year journey, I have worked with many talented and supportive
colleagues. I sincerely appreciated the valuable discussions with Dr. Sathiya Mariyappan, Dr. Iban
Azcarrate, Dr. Lukas Lutz, Mr. Nicolas Dubouis, Dr. Guochun Yan, Dr. Chunzhen Yang, Dr. Jiwei
Ma, Dr. Gaurav Assat, Dr. Quentin Jacquet, Dr. Ronghuan Zhang, Dr. Forencia Marchini, Dr. Biao
Li, and Dr. Jiaqiang Huang. I would like also to thank Mr. Thomas Marchandier for his help during
the two-day XAS experiment at Soleil.
Specially, I would like to thank my office mates Ms. Laura albero blanquer, Dr. Forencia
Marchini, Mr. Jean Vergnet, Mr. Thomas Dargon and Ms. Zeqi Zhou for creating such an enjoyable
atmosphere, I cannot image better office mates than you. Many thanks to Dr. Vanessa Perreira
Pimenta, Dr. Meiling Sun, Dr. Yan Duan, Dr. Biao Li, Dr. Jiaqiang Huang, Dr. Ronghuan Zhang,
Dr. Chunzhen Yang, Dr. Guochun Yan, Dr. Qing Wang, Dr. Leiting Zhang, Dr. Yinghui Yin, Dr.
Huijun Zhang, Dr. Yang Xia, and to my “hiking team” for the great time and moments we have
shared.
Finally, I would like to express my utmost gratitude to my parents and my family for their
patience and unconditional support.
Table of contents
Table of contents ...................................................................................................... 1
Broader context and thesis outline ......................................................................... 4
Outline of thesis ..................................................................................................... 6

Chapter I General introduction.............................................................................. 8
I. 1 Basic concepts of Li-based batteries and insertion-type electrodes .......... 8

I. 1. 1 Intercalation chemistry and the advent of Li battery ...................................................... 8
I. 1. 2 From Li battery to Li-ion battery.................................................................................... 9
I. 1. 3 Positive electrode materials for Li-ion batteries and their energy limitation ............... 10
I. 2 Increase the energy density of positive electrode materials for Li-ion

batteries ................................................................................................................ 12
I. 2. 1 M/M’ chemical substitution ......................................................................................... 12
I. 2. 2 Li substitution and the participation of oxygen redox .................................................. 14
I. 2. 3 Positive electrode materials with elevated (de)intercalation voltage ........................... 17
I. 3 Searching for new chemistries with potentially higher energy storage .. 19

I. 3. 1 Conversion chemistry ................................................................................................... 20
I. 3. 2 Li-O2 battery ................................................................................................................. 22
I. 4 Conclusions ................................................................................................... 28
Chapter II Monitor the oxygen release at high potentials in Li-rich layered
oxides ....................................................................................................................... 30
II. 1 Background and motivation ...................................................................... 30

II. 2 Description of the in-situ gas analysis techniques.................................... 32
II. 2. 1 In-situ pressure cell ..................................................................................................... 32
II. 2. 2 Online electrochemical mass spectrometry ................................................................. 33
II. 2. 3 Rotating ring disk electrode voltammetry ................................................................... 34
1
II. 3 Results and discussion ................................................................................ 37
II. 4 Chapter conclusion and outlook................................................................ 50
Chapter III Revisiting the structural evolutions and electrochemical properties
in the first cycle of Li-rich NMC .......................................................................... 51
III. 1 Background and motivation..................................................................... 51

III. 2 Results and discussion .............................................................................. 53
III. 2. 1 Phase transition and O2 release on deep oxidation .................................................... 53
III. 2. 2 Detailed structural characterization of phase A’ ........................................................ 60
III. 2. 3 The impact of phase transition on the electrochemical properties ............................. 70
III. 2. 4 Mn migration triggered by oxygen redox and its consequence of extra low-voltage
electrochemical activities ........................................................................................................ 74
III. 3 Discussion and chapter conclusion .......................................................... 89

Chapter IV Li-CO2 battery: a new system for energy storage and CO2
conversion ............................................................................................................... 94
IV. 1 Background and motivation ..................................................................... 94

IV. 2 Li−O2/CO2 batteries: CO2 conversion via electrogenerated superoxide
(O2●¯) .................................................................................................................... 96
IV. 2. 1 Introduction ............................................................................................................... 96
IV. 2. 2 Results and discussion ............................................................................................... 99
IV. 2. 3 Discussion and conclusion....................................................................................... 118
IV. 3 Li−CO2 batteries: CO2 conversion mediated by quinone derivatives 121

IV. 3. 1 Introduction ............................................................................................................. 121
IV. 3. 2 Results and discussion ............................................................................................. 122
IV. 3. 3 Conclusion ............................................................................................................... 136
Chapter V General conclusion and outlook ...................................................... 137
Appendix ............................................................................................................... 136
2
A. 1 Design of experimental equipment .......................................................... 136
A. 1.1 Gas filling station ....................................................................................................... 136
A. 1. 2 Description of OEMS systems .................................................................................. 136
A. 2 Supporting information for Chapter II .................................................. 138

A. 2. 1 Methods..................................................................................................................... 138
A. 2. 2 Supporting figures ..................................................................................................... 141
A. 2. 3 Calculation of the quantity of released O2 gas .......................................................... 144
A. 3 Supporting information for Chapter III................................................. 146

A. 3. 1 Methods..................................................................................................................... 146
A. 4 Supporting information for Chapter IV ................................................. 164

A. 4. 1 Methods..................................................................................................................... 164
Bibliography ......................................................................................................... 174
3
Broader context and thesis outline
For thousands of years, humankind has relied on fossil fuels to generate energy. The wide-
scale use of fossil fuels, coal at first and petroleum later, to fire steam engines has enabled the first
industrial revolution which marked a major turning point in history. In particular, the population
began to experience an unprecedented growth, alike for the living standards that kept increasing.
Nevertheless, despite of the critical role that fossil fuels have played in the development of most
industrial nations, several practical problems related to the use of fossil fuels arise: i) fossil fuels are
finite energy sources and take typically millions of years to be formed from natural process, the
known viable reserves are being depleted much faster than new ones are being produced; ii)
combustion of fossil fuels produce CO2 which accounts for 75 % of the greenhouse gas (GHG)
emissions and causes the global temperature to rise; iii) burning of fossil fuels also releases other air
pollutants such as NO2 and SO2 which contribute to smog, acid rain and the formation of fine
particulate matter; iv) radioactive elements, mainly uranium and thorium, are also released into the
atmosphere from burning fossil fuels. More importantly, these global issues are becoming more
serious than ever as the ever-rising energy consumption due to the overall population growth and
prosperity increases (Figure 1). There is thus a strong need to transform the way we generate,
distribute and use energy.
Figure 1 World energy consumption rises 28 % between 2015 and 2040 with most of the
increases in energy demand coming from non-OECD (Organisation for Economic Co-operation and
Development) countries. Adapted from: IEA World Energy Outlook 2017.1
4
Within this context, gradually shifting the world’s energy sources from fossil fuels to lower-
carbon fuels such as renewable energies (biofuels, geothermal, hydro, wind, solar sources, etc.) is
considered to be a viable alternative to fulfil our energy demand with lower environment effects
(Figure 2). To realize this global energy transition, several renewable technologies are being
developed to broaden the access to more abundant and cleaner energy supplies. Despite renewable
technologies have overcame various technological pitfalls, the major challenge resides in how to
balance power supply and demand, with one example being how to tackle the intermittency issue
caused by the fluctuations of wind and solar powered systems. Energy storage plays an important
role in this balancing act and helps to ensure more flexible and steady power supply.
Figure 2 World energy consumption by energy source: energy sources shift gradually
toward lower-carbon fuels such as renewables and natural gas, although fossil fuels (coal,
petroleum and other liquids) continue to meet much of world’s energy demand. Adapted from: IEA
World Energy Outlook 2017.1
Energy storage technologies have continued to evolve over the last century, striving to meet
the changing energy requirements and potential applications. These technologies can be broadly
divided into four categories: mechanical, thermal, electromagnetic and chemical energy storage.
Among them, chemical energy storage technologies, which can store and convert chemical energy
into electricity, have experienced by far the greatest diversity of research and commercial products.
So far, the prevailing chemical storage systems are batteries, especially those based on lithium-
intercalation materials, i.e., Li-ion batteries. Although currently Li-ion batteries are the dominant
power sources for portable electronic devices and are rapidly expanding into the electric vehicle
5
markets, they still do not compete with fossil fuels in terms of the energy storage capability.
Therefore, the major challenge of today’s battery research is to maximize the energy that can be
stored in batteries. Towards this goal, intense research activities have been placed on continuously
improving the electrochemical performance of the state-of-art Li-ion batteries while searching for
new chemistries with potentially higher energy storage.
Outline of thesis
This thesis enrolls into the global push for designing batteries with better electrochemical
performances and is parted between i) the fundamental understanding of the electrochemical-
structural relationship in the state-of-art Li-rich layered oxides, seeking to unlock the full potential
of Li-ion batteries at the material level, and ii) the investigation of the underlying reaction
mechanisms for Li-CO2 battery, an alternative to Li-ion battery which has the advantages of both
energy storage and the conversion of the detrimental greenhouse gas CO2 into value-added energy
product like Li2CO3.
This work is thus divided into five chapters:
Chapter I start with a general introduction about the basic concepts of Lithium-based
batteries and their energy limitation on the positive electrode side, prior to provide an overview of
the research efforts that the battery community has made to pursue positive electrode materials with
increased energy densities. Alternative battery chemistries that could potentially move beyond the
energy-storage capability of Li-ion batteries will also be discussed.
Chapter II addresses the degradation phenomenon of Li-rich positive electrode materials

via in-situ gas/pressure analyses. More specifically, rotating ring disk electrode voltammetry will be
developed to monitor the irreversible O2 release from Li-rich layered oxides on their initial charges.
As a proof-of concept, the obtained results will be systematically compared with those obtained by
two other well-established techniques, i.e., in-situ pressure cell and online electrochemical mass
spectrometry. A general discussion concerning the pros and cons of these three pressure/gas
analytical techniques will be also presented.
6
Based on the knowledge from the previous chapter about the structural instability of Li-rich
layered oxides associated with oxygen redox, Chapter III explores deeper the dynamic structural
change during the first electrochemical cycle in a practically important Li-rich positive electrode
with a nominal formula of Li1.2Ni0.13Mn0.54Co0.13O2. By cycling the material under various oxidative
conditions as well as enlarging the discharge voltage window, we will demonstrate how the Li+
extraction coupled with anionic oxidation will results in O2 evolution, cation migration, and
eventually a bulk phase densification. In addition, we will discuss how the cation migration
progress upon cycling and how it will impact on the electrochemical properties. At last, we will
reconsider in this chapter the origin of its undesirable first-cycle capacity loss.
Going beyond the horizon of Li-ion batteries, Chapter IV is specifically dedicated to a

newly-born battery concept: Li-CO2 batteries. The first part of this chapter will reveal how the
electrolyte properties dictate the discharge processes of O2-assisted Li-CO2 batteries by controlling
the superoxide intermediate. In the second part of this chapter, we will explore the potential use of
quinone-based homogenous catalysts to promote the electrochemical reduction of CO2 in “real” Li-
CO2 batteries. The practical potential of these two CO2-battery systems will also be discussed.
The main results of this Ph.D. work will be summarized in Chapter V, their impacts and
implications on further battery researches will be discussed along with the remaining questions.
7
Chapter I General introduction
I. 1 Basic concepts of Li-based batteries and insertion-type electrodes
I. 1. 1 Intercalation chemistry and the advent of Li battery
Intercalation chemistry refers to the reversible insertion of guest species into an open host
structure with the maintenance of major structural features of the host. In 1970s, the discovery of
highly reversible Li intercalation/de-intercalation in layered titanium disulfide, according to the
reaction x Li + TiS2 ↔ LixTiS2, has enabled the birth of Li battery.2,3 In a typical Li battery, two
electrodes, the positive electrode and the negative electrode, are separated by an electrolyte (Figure
I. 1a). The electrolyte allows the reversible shuttling of Li ions between the two electrodes while the
external circuit allows for electron transport. The battery voltage (V) is determined by the
equilibrium voltage difference between the positive electrode and the negative electrode, more
specifically the difference between the Li chemical potentials in both electrodes (Equation I. 1).
Cat An
μ −μ
𝐿𝑖 𝐿𝑖
𝑉=− (Equation I. 1)
𝑒
Cat An
where μ is the Li chemical potential of the cathode, and μ is the Li chemical
𝐿𝑖 𝐿𝑖
potential of the anode.
The energy storage capability of an electrode (Q) is defined as the specific capacity (Ah kg-
1
), which is dependent on the number of Li ions that can be reversibly intercalated into the electrode
without causing irreversible degradation of its structure (Equation I. 2)
96485 n(𝐿𝑖 + )
𝑄 = (Equation I. 2)
3600  𝑀
where n(𝐿𝑖 + ) is the number of Li ions, and M is the molecular weight of the electrode
material.
8
The energy storage capability of a battery (E), termed as energy density (Wh L-1) or specific
energy (Wh kg-1), is determined by the voltage (V) and the specific capacity (Q) of the positive
electrode and the negative electrode (Equation I. 3)
𝐸 = 𝑄  𝑉 (Equation I. 3)
Therefore, since the first demonstration of Li battery, intensive research activities have been
placed on screening/designing suitable materials for the two electrodes that would provide the
highest battery voltage as well as the greatest energy density for the system. A brief historical
review will be presented next regarding the developments of materials for Li batteries.
I. 1. 2 From Li battery to Li-ion battery
Starting with the negative electrode, metallic Li was first used owing to its high theoretical
specific capacity (3860 mAh g-1 or 2061 mAh cm-3) and lowest voltage (-3.04 V vs standard
hydrogen electrode) as compared to all other candidates. In the late 1980s, Moli Energy
commercialized the first Li battery using a MoS2 positive electrode paired with excess metallic Li.
However, soon this device had to be withdrawn from the market due to explosion hazards caused by
uneven Li growth (dendrite) as Li was stripped/re-plated upon each charge/discharge. NEC and
Mitsui have conducted reliability tests on over 500,000 Li metal batteries but failed to resolve the
safety concern. Initial attempt to avoid the dendrite formation was thus made to substitute Li metal
with LiAl alloy. However, such replacement would jeopardize the cycling performances of the
battery because of large volume expansion associated with the alloying process.4,5 The second
approach to circumvent this safety issues aimed at replacing Li metal by another insertion-type
material. Inherently safer than Li battery, this new concept is termed as Li-ion battery (or rocking-
chair battery) (Figure I. 1b). Subsequently, with the discovery of low voltage and highly reversible
Li intercalation/de-intercalation in carbonaceous material, Sony finally commercialized the first Li-
ion battery in 1991. Carbon negative electrode, generally graphite, is still commonly used in today’s
commercial Li-ion batteries.
However, the capacity of carbon negative electrode is ~ 10 times less than that of Li metal
(theoretical capacity of 372 mAh g-1 for graphite - LiC6). This huge capacity difference is indeed an
impetus for researchers to seek chemical or physical modifications for improving the carbon
9
negative electrode performances,6,7 as well as to search for other promising carbon alternatives.8,9 In
parallel, intensive research activities have also been devoted to improve the understanding of Li
metal chemistry via developing both characterization techniques10,11 and Li dendrite growth
theory12,13, as well as to stabilize Li metal negative electrode through interface engineering and/or
electrolyte engineering.14-17
Figure I. 1 Two different Li-based batteries with negative electrodes of: (a) metallic Li and
(b) insertion anode (graphite). Adapted from: PhD thesis of Arnaud Perez, 2017.
I. 1. 3 Positive electrode materials for Li-ion batteries and their energy

limitation
Turning to the positive electrode side, it is worth firstly recalling that the positive electrode
materials for Li batteries can be lithium free because of the use of Li metal negative electrode,
while those for Li-ion batteries must act as the lithium source since the carbonaceous negative
electrode contains no Li.
Soon after the discovery of TiS2, layered oxide LiCoO2 was introduced by Goodenough and
coworkers as a new Li intercalation host.18 Layered oxides have a higher intercalation potential than
sulfides in general (Figure I. 2), owning to the more pronounced ionic character of ‘M-O’ bonds as
compared to ‘M-S’ bonds. Research interests have thus quickly shifted from layered chalcogenides
to oxides that are still widely used in today’s Li-ion batteries. However, the drawback of LiCoO2 is
that only one-half of Li ions (corresponding to 150 mAh g-1 capacity) can be reversibly cycled
without causing severe polarization and capacity loss (Figure I. 2d), owing to the abrupt structural
10
changes caused by the strong electrostatic repulsions of the negatively charged CoO2 layers in
LixCoO2 (0  x  0.5).19 More specially, on deep delithiation, the original O3 hexagonal phase
converts to a hybridized O1-O3 hexagonal phase (also denoted as the H1-3 phase), where O
represents octahedral sites, 3 is the stacking sequence of oxygen layers ABCABC and 1 is ABAB
oxygen stacking sequence, as defined using the conventional nomenclature developed by Delmas
and coworkers.20 Meanwhile, gliding of the layers along with partial collapse of the O3 stacking as
well as loss of oxygen further add up to the serious performances degradation of LiCoO2 at high
voltages and jeopardize its practical energy density.
Figure I. 2 Crystal structure of TiS2 (a) and LiCoO2 (c). TiS2 can insert 1 Li per formula unit
at an average voltage of 2.2 V (b) and LiCoO2 can release 0.5 Li per formula unit at an average
voltage of 3.9 V (d). Adapted from: PhD thesis of Gaurav Assat, 2018.
Aware of the energy limitation of LiCoO2, material scientists have embarked on a journey to
improve the performances of positive electrode materials which currently limit the energy density
of commercial Li-ion batteries. Various research directions were pursued: i) push the 0.5 Li+ limit
11
of LiCoO2 towards its theoretical capacity via chemical substitution, aiming to stabilize the layered
framework; ii) further substitute transition metal with Li+ (termed as Li-rich layered oxides) to
increase the Li+ content available for (de)intercalation; and iii) screen new intercalation-type
positive electrodes and tailor their (de)intercalation voltage by selecting the electrochemical redox
center. Directions i) and ii) target at improving the material capacity, while direction iii) focuses on
increasing the material voltage.
Before going into more details in the next subchapter, let us first recall that, in addition to
show a high voltage and be able to accommodate large number of Li+ while maintaining its
structural integrity, an ideal positive electrode should also possess a high electronic conductivity as
well as fast Li ion diffusion, have low cost and toxicity, be chemically and thermally stable, be
environmentally benign and easy to synthesize.
I. 2 Increase the energy density of positive electrode materials for Li-

ion batteries
I. 2. 1 M/M’ chemical substitution
To circumvent the capacity limitation of LiCoO2 along with the high cost and ethical issues
associated with the use of Co, the scientific community has initially attempted to replace Co with
low-cost and more abundant transition metals such as Ni and Mn by ion exchange from -NaMnO2.
However, isostructural LiNiO2 is unable to compete with LiCoO2 as it shows comparable capacity
but lower (de)intercalation voltage, and was further overlooked because of its poor thermal stability
in the delithiated form.21,22 In contrast to the rhombohedral (R3̅m) -NaFeO2 structure of LiCoO2
and LiNiO2, layered LiMnO2 predominantly crystallizes in monoclinic (C2/m) unit cell with minor
concentration of lithiated-spinel Li2Mn2O4 (𝐹41 /𝑑𝑑𝑚 ), and readily transforms into the spinel
LiMn2O4 upon cycling.23
Mixed/partial substitution was then proposed to stabilize the layered structure as well as to
tailor the (de)intercalation voltage. Numerous LiMO2-type layered compounds were synthesized
thereafter and can be broadly divided into binary Li[MM’]O2 and ternary Li[MM’M’’]O2 systems.
Among them, the most practically important layered oxides for Li-ion batteries are Li[NixCoyAlz]O2
12
(NCA) and Li[NixMnyCoz]O2 (NMC) series. NCA can be intuitively viewed as a partial substitution
of Ni by Co in LiNiO2, together with a small amount of Al doping. Co helps to reduce the degree of
Li/Ni disorder and effectively stabilizes the structure, while Al is electrochemically inactive but
improves the thermal stability. Thanks to the synergistic effects of Co and Al co-substitution,
Li[Ni0.8Co0.15Al0.05]O2 could achieve an attractive practical discharge capacity of ~ 200 mAh g-1
(Figure I. 3), and is widely used in today’s electrical vehicles including Tesla Model S.24,25
However, NCA suffers from capacity fading, particularly at high temperatures and at high current
rates.
Figure I. 3 Typical discharge voltage profiles of different intercalation-type positive

electrodes. Adapted from Nitta et al.24
Similarly, NMC series can be prepared by partially substituting Co with Ni and Mn in

LiCoO2 (Figure I. 4). In doing so, Ni provides the capacity, Co the structural stability and Mn the
thermal stability,26 and Li[Ni1/3Mn1/3Co1/3]O2 (also referred to as NMC 111) is found to deliver
promising electrochemical performances with reversible capacity of ~ 200 mAh g-1 when charged
up to 4.6 V (Figure I. 3) and is currently used in BMW i3, Chevy Bolt and the new Nissan Leaf. By
steadily increasing the Ni content (decreasing the Co content), researchers further enhanced the
capacity of NMC materials and prepared a series of Ni-rich NMCs termed as NMC 433, NMC 532,
NMC 622 and NMC 811. However, raising the Ni-content result in penalties of reduced thermal
13
stability, safety issues owing to the release of O2 and CO2, along with poor capacity retention
caused by the irreversible Li+ loss associated with the progressive development of a surface film
triggered by the fracturing of particles induced by large volumetric changes.27-29 Extensive efforts
seeking to mitigate these issues have been put into surface coating and modifications, morphology
and mesostructured design, as well as electrolyte engineering.30 Despite of the substantial
progresses achieved so far, the most effective approach to further push the energy density of
positive electrode materials towards their upper limits resides in playing with the chemical
composition, as will be discussed next.
Figure I. 4 Push the capacity of layered oxide LiCoO2 by 1) partially substitute Co (pink)
with Ni (green) and Mn (blue) leading to the NMC phase and 2) further substitute with Li (yellow)
to give Li-rich NMC phase. Adapted from Rozier et al.31
I. 2. 2 Li substitution and the participation of oxygen redox
In an attempt to achieve higher capacities than LiMO2-type layered compounds, Li-rich

Li2MO3-type materials (alternatively described as Li[Li1/3M2/3]O2) with excess Li in the transition
metal layers were also heavily investigated since late 1980s.32-35 As expressed in their general
formula, Li[Li1/3M2/3]O2 and LiMO2-type compounds share similar structures with one notable
difference: all octahedral sites in the metal layers are occupied by M3+ cations in LiMO2 structure,
whereas one third of those sites are occupied by Li+ and two third by M4+ in Li[Li1/3M2/3]O2
structure. Pioneering works on Li[Li1/3M2/3]O2-type materials involved the detection of
14
electrochemical activity in Li2MnO3 which is rather abnormal since the octahedral Mn4+ cannot be
further oxdized to release Li+ during charge.36 Through acid-leaching, this unexpected
electrochemical activity in Li2MnO3 was attributed to the Li2O removal with “MnO2-like”
activation.33 In parallel, binary phase Li[Li(1/3-2x/3)M(2/3-x/3)M’x]O2 prepared by partial transition
metal substitution were also explored.37-40 Among them, Li[Li(1/3-2x/3)Mn(2/3-x/3)Nix]O2 with x = ¼
delivers reversible capacities of nearly 230 mAh g-1 (corresponding to almost 1 Li per formula unit)
when charged up to 4.8 V.41 The improved capacity obtained for Li[Li(1/3-2x/3) Mn(2/3-x/3)Nix]O2 when
compared to classical layered oxides created an impetus for further exploration of these Li-rich
phases, before to eventually lead to the discovery of Li[LixNiyMnzCo(1-x-y-z)]O2 (also termed as Li-
rich NMC) that could offer staggering discharge capacities of more than 270 mAh g-1 (Figure I.
4),34,42,43 a high value which cannot be explained by the conventional view of transition metal redox.
Markable differences are also observed in the voltage-composition profiles of Li-rich layered
compounds. Typically, Li-rich oxides exhibit a two-step staircase charge voltage profile with a long
high-voltage plateau (ca. 4.5 V) followed by a sloping S-shaped voltage profile on discharge. This
S-shape profile pertains on subsequent cycles, and differs from classsical Li-stoichiometric oxides
whose charge and discharge voltage-composition profiles mirror each other nicely.
Such high capacities and the unusual long charge plateau were initially explained by the Li-
driven activation of oxygen redox (simultanous removal of Li+ and oxidation of lattice O2¯ to O2,
also referred to as Li2MnO3 activation) based on Rietveld refinement and further demonstrated with
in-situ differential electrochemical mass spectrometry (DEMS).41,44,45 However, controversies arise
as the onset of O2 evolution does not exactly match with the beginning of the high-voltage plateau.46
Furthermore, quantitative gas analysis reveals that the amount of released O2 (amount of O2¯
oxidation) is far too little to be responsible for the capacity obtained during the high-voltage plateau
(the detected O2 amount can only account for ~ 6.4 mAh g−1 of charge capacity, which is negligible
compared to the overall extra charge capacity of ~ 200 mAh g−1 (the part of capacity that cannot be
explained by cationic redox)).47 Finally, a global “bulk-shell model” was suggested in order to
rationalize the oxygen redox mechanisms: reversible oxygen redox (2n O2¯↔ (O2)n¯+ 2n e¯ (n <
2)) in the bulk which explains the superior reversible capacities obtained with these Li-rich type
materials and irreversible oxygen release at the near-surface region (2 O2¯ → O2 + 2 e¯ ).47-50
Aware of the diffculty of establishing the oxygen redox mechanism in the complex Li-rich NMC
systems which include three cationic redox centers (Ni2+, Mn4+, Co3+), Tarascon and co-workers
15
designed a model system Li2Ru1-ySnyO3 which is isostructural with Li-rich NMCs but with only one
single cationic redox center Ru4+, and demonstrated direct evidence on the formation of (O2)n¯
species (n < 2).51 This observation confirmed the aformentioned speculation of reversible
participation of oxygen redox in Li-rich oxides, and unambiguously explained their high capacities
as arising from cumulative cationic (Mn+ ↔ M(n+1)+ + e¯) and anionic (2n O2¯ ↔ (O2)n¯+ 2n e¯(n
< 2)) redox processes. Years later, the reversible participation of lattice oxygen redox in Li-rich
NMC was also experimentally proved using 18O-labelled online electrochemical mass spectrometry
by quantifying the amount of latticed oxygen released as O2 and CO2.46
Through intensive experimental and theoretical studies, it is now generally understood that
the origin of oxygen redox in Li-rich layered oxides is the O (2p) lone-pair states which are close
enough to the Fermi level to be oxidized.52-54 Those lone-pair states are preferentially formed for
the Li-O-Li configurations in which the Li (1s) orbitals of Li+ from both the transition metal and Li
layers cannot overlap with O (2p) orbitals. This explains well the different redox mechanisms
observed for conventional and for Li-rich layered oxides. Indeed, for conventional LiMO2 layered
oxides, each oxygen ions are coordinated by three transition metal ions so that all the O (2p)
orbitals overlap with transition metal d orbitals, leading to zero O (2p) lone-pair state. Accordingly,
their electrochemical reactions are dominated by cationic redox. Note that although the presence of
excess Li+ in transition metal layers plays an important role in activating oxygen redox activity, it is
not an essential requirement as O (2p) lone-pair states also occur in environments containning Mg2+
(stemming from Mg2+ 3s interaction with O 2p) or cationic vacancies. This fact explains the
reversible participation of oxygen redox in Na-based layered oxides without Na excess such as
Na2/3Mg0.28Mn0.72O2 and Na4/7[1/7Mn6/7]O2 ( being Mn vaccancy).55,56
Increasing the O/M (or Li/M) ratio thus offers the opportunity to push the oxygen redox
capacity by increasing the number of O (2p) lone-pair states. Following this strategy, Tarascon and
co-workers reported a Li3IrO4 phase which could reversibly uptake and release 3.5 Li+ per iridium
ion, the highest value ever reported among all the intercalation-type positive electrode materials. In
this phase, two Li+ could be reversibly (de)intercalated from Li3IrO4 with the sole participation of
oxygen redox, while the remaining insertion of 1.5 Li+ into Li3IrO4 is counter-balanced by the
reduction of Ir5+.57 Nevertheless, the large oxygen gas release and structural amorphization
observed at highly charged states of Li3-xIrO4 clearly suggest that, pushing the oxygen redox
16
activity in Li-rich materials comes with the risk of increasing the structural instability. Several
approaches have thereafter been suggested to stabilize the oxygen network including chemical
substitutions,58 increasing M-O bond covalency,59 and altering the dimensinaltiy of the the M-O
network.60 Also worth mentioning is a recent theoretical work which propose that the charge-
transfer gap (∆CT) of Li-rich layered oxides and the number of holes per oxygen (h0) can serve as
descriptors to quantify the achievable oxygen redox capacity on charge and its reversibility on
discharge.61
Overall, the discovery of reversible oxygen redox activity in Li-rich oxides is appealing as it
provides an opportunity to raise the energy limit of current positive electrode materials by combing
cationic and anionic redox processes within the same compound. In those compounds, oxygen can
effectivly serve as an extra electron source when more Li+ needs to be extracted than the transition
metal redox center can offer. Moreover, the demonstration of oxygen redox in 4d and 5d transition
metal-based Li-rich oxides opens a wide playground for material scientists to design new high-
capacity electrodes including a new class of cation-disordered oxides,58 and to bring new
understanding about the oxygen redox process. Nevertheless, the extra capacity offered by the
oxygen redox activity generally associated with structural instabilities (i.e., irreversible O2 gas loss
(further addressed in Chapter II), cation migrations (further explored in Chapter III)), and
unfavorable electrochemical properties (i.e., large first cycle irreversibility, voltage decay and
capacity fading upon cycling).
I. 2. 3 Positive electrode materials with elevated (de)intercalation voltage
Given the Equation I. 3, an alternative approach to enhance the energy density of Li-ion
batteries is to increase the output cell voltage without compromising the specific capacity. Since the
potential of graphite negative electrode commonly used in modern Li-ion battery is less than 0.25 V
vs Li+/Li (for LiyC6, 0.1 < y < 1), the cell voltage is obviously dictated by the Li+ (de)intercalation
potential of the positive electrodes.
Other than playing with the transition metal redox center (e.g., slightly higher operation
potential can be obtained by moving from Ni3+/Ni4+ to Co3+/Co4+ redox couple), the redox potential
of positive electrodes can be alternatively tuned by using more electronegative anions or anionic
ligands. Along this line, moving from transition metal sulfides to transition metal oxides would be
17
advantageous as already discussed in Chapter I. 1. 3. Further replacing oxygen ions with polyanion
units (XO4)n¯(X = S, P, Si, etc.) lead to the discovery of a class of poly-anion compounds with
higher (de)intercalation voltage. The strong covalent bonding within the poly-anion units reduces
the separation between the bonding and antibonding orbitals, shifting the antibonding orbitals,
which are pinned at the Fermi level and responsible for the electrochemical activity of the material,
to lower energy thereby increasing the (de)intercalation voltage (the so-called “inductive
effect”).62,63 Following this strategy, introducing fluorine as the anion to form M-F bond would
further raise the (de)intercalation voltage by lowering the energy of the antibonding d-orbital of the
transition metal.64 Despite offering the potential to extend the (de)intercalation voltage up to 5 V
and even beyond,65 researches dedicated to poly-anion compounds are now leveling off because of
the energy and the power penalties associated with their heavy formula weight, poor
electronic/ionic conductivity and low packing density.
For a material with a given chemical composition, its (de)intercalation voltage can be tuned
by playing with the crystal structure, introducing defects or changing the local chemical
environment. As the Li+ (de)intercalation in electrode materials involves simultaneous addition
(removal) of lithium ions and electrons, the redox voltage of electrode materials is directly related
with the energy required to add (remove) lithium ions into (from) the host crystal lattice and to add
(remove) electrons into (from) the d orbitals of the transition metals. Different positions in a given
crystal structure exhibit different Li+ (de)intercalation voltage depending on their site energy, i.e.,
the main component of the Gibbs free energy, defined as the enthalpy change during Li+
(de)intercalation. For instance, in spinel LixMn2O4, Li+ ions are intercalated into smaller tetrahedral
sites at 4.3 V (x  1) and bigger octahedral sites at 3.1 V (x  1), despite of operating on the same
Mn3+/Mn4+ redox couple within the spinel framework.66 Introducing disorder of atomic arrangement
can also affect the (de)intercalation voltage. For example, spinel LiNi0.5Mn1.5O4 with the disordered
face-centered Fd3 m structure exhibits lower de-intercalation voltage than that of LiNi0.5Mn1.5O4
with ordered primitive simple cubic structure (P4332) (i.e., 4.69 V and 4.75 V compared to 4.74 V
and 4.77 V).67 Up to now, the most successful high-voltage cathode material is the Ni-doped Mn
spinel LiNi0.5Mn1.5O4 showing an operation voltage in the range of 4.7 – 4.8 V with reasonable
cyclability even at high rate.68 Although LiNi0.5Mn1.5O4-based spinel materials cannot compete with
Li-rich layered oxides in terms of energy density because of their moderate capacities, they hold
18
promise for high-power applications due to fast Li+ diffusion enabled by the interconnected
diffusion channels in their 3D structures.
Nevertheless, the practical application of these LiNi0.5Mn1.5O4-based spinel electrodes poses

severe challenges to the electrochemical stability of electrolytes against oxidation as their operation
voltages entirely or partially exceed the electrochemical window of the commonly used organic
electrolytes. Major issues are associated with the electrode/electrolyte interface where electrolyte
(both solvent and salt) decomposition is promoted by high-valence transition metals in the charged
electrodes, resulting in the surface dissolution of transition metals, inter-reaction of bulk electrolyte
components, “electrochemical dialogues” between positive and negative electrodes, and corrosion
of other cell components (e.g. aluminum current collector).69-72 Additionally, the effect of these
parasitic reactions becomes even more severe at elevated temperatures and in full cell configuration
where the Li+ content is limited. Mastering the surface/interface chemistry of LiNi0.5Mn1.5O4 is
therefore critical for its successful commercialization. Various approaches have been explored to
stabilize the electrode/electrolyte interface, including i) modifying the crystal chemistry through
cationic or anionic substitution;68 ii) surface treatment;73 iii) designing novel electrolytes that are
intrinsically more stable such as ionic liquids, polymer and inorganic solid electrolytes;74 iv)
introducing electrolyte additives aiming to form a thin but robust protective layer on the electrode
surface;75 v) adding HF scavenger (HF is generated from the hydrolysis reaction of LiPF6 salt and is
responsible for most of the aforementioned parasitic reactions);76 vi) developing new lithium salts
that can potentially overcome the two shortcomings of the stellar salt LiPF6 (i.e., chemical and
thermal instability);77 and vii) applying concentrated electrolytes (electrolytes with high salt/solvent
ratio) with improved oxidative stability.78
I. 3 Searching for new chemistries with potentially higher energy

storage
Despite all the outstanding developments achieved so far for the Li-ion battery technology,
their energy storage capability based on intercalation chemistry is insufficient for the long-term
needs of modern society. This mismatch spurred intensive researches on alternative chemistries
holding the potential to surpass the energy limitations of Li-ion batteries, among which several
representatives will be introduced next.
19
I. 3. 1 Conversion chemistry
In contrast to intercalation chemistry, conversion chemistry involves the cleavage and

formation of new chemical bonds during Li+ insertion/extraction. So far two types of conversion
reactions have been identified for Li chemistries:
Type I : MXz + y Li0  M0 + z Li(y/z)X
Type II: y Li0 + X0  LiyX
where M are typically transition metal cations and X are anions.
Type I conversion reactions, in contrary to classical intercalation reactions that are often
limited to 1 e¯ per transition metal, can offer 2 e¯ or even more per transition metal, therefore
enabling staggering capacities. Moreover, their redox potentials can be tuned by playing with the
electronegativity of anions and transition metals (e.g., the redox potential ranges from 0.4 to above
3.0 V as moving from covalent phosphides to ionic fluorides), hence offering the opportunity to use
those conversion electrodes as both positive and negative electrodes. The concept of “conversion
reaction” was considered as a breakthrough but still remains a laboratory curiosity because of its
intrinsic drawbacks: i) the formation of metal nanoparticles actively decompose the Li(y/z)X matrix
in which they are embedded; ii) the strong structural re-organization during phase transformation
induces large volume change and iii) the existence of large voltage hysteresis. Moreover, being
lithium-free compounds, MXz require an additional chemical lithiation step prior to be used as
positive electrodes. Along that line, worth mentioning is a recent work which address this limitation
by blending nano-sized MnO with LiF that function as a Li source.79 The MnO-LiF composite is
activated at high potentials (~ 4.5 V) to form a new Mn-O-F-like surface that can reversibly
uptake/release Li+ (LiF + MnIIO  Li+ + e¯ + MnIIIO–F). Such “surface conversion reaction” has
the merit of maintaining the high-valence states of transition metals (MII/MIII) as they are housed by
the anions on both charge (by O2¯ and F¯) and discharge states (by O2¯), hence maintaining the
operation cell voltage. Although this work opens a new route to design positive electrodes, cautions
must be exercised in the role of LiF and side reactions brought by the use of nanocomposites
(increased exposure surface), as suggested by Tarascon and coworkers, without LiF, the
20
electrochemical activation of MnO to Mn-O-F-like phase can equally occur via the decomposition
of LiPF6 salt.80
Type II conversion reaction is also quoted as “chemical transformation reaction” because it

involves the transformation of one single phase into another single phase. Typical examples include
the lithiation of chalcogens and halogens (i.e., S, Se, Te, Br, I). Amongst the lithiation of S (referred
as Li-S battery) has been studied the most because of its greater specific energy, since S is light
element and one S atom can accept two electrons (uptake two Li+) to form lithium sulfide (2 Li+ + 2
e¯ + S  Li2S), amounting to a theoretical capacity one magnitude higher than that of transition-
metal oxide at the expense of a lower operating voltage (ca. 2.15 V). Moreover, being the fifth most
abundant element on earth, S has a cost advantage. On the negative side, the S/Li2S conversion
reaction is accompanied by a large volume change of about 80 %. In addition, the poor electrical
conductivity of S and Li2S necessitates the incorporation of conducting additives (e.g., carbon) to
construct composite S electrode, which markedly compromises the volumetric energy density.
Another major issue is associated with the soluble intermediate polysulfides (Li2Sx) generated at the
positive electrode during the discharge. Those Li2Sx species diffuse to the negative electrode (i.e.,
metallic lithium) where they are chemically reduced to lower polysulfides or insoluble Li2S2 and
Li2S, giving rise to the so-called “shuttle” effect which induces the loss of active material,
impedance build-up and capacity fading. Several approaches have been suggested to tackle the
dissolution of polysulfides: i) play with surface chemistry or solvation properties of electrolytes
aiming to promote the chemical interactions between the host material and the dissolved
polysulfides; ii) design smart electrode architectures to allow encapsulation of polysulfides, iii)
utilize polymer or solid electrolytes instead of liquid ones to physically separate the two electrodes;
and iV) introduce electrolyte additives, such as the “magic” LiNO3, to enhance the passivation of
metallic lithium. Although the battery community has witnessed in the last four decades Li-S
battery with greatly improved electrochemical performance and fundamental understanding, step-
changes are awaited in many aspects before Li–S chemistry can be successfully employed in
practical systems: i) accommodating the large volume expansion and contraction on full S/Li 2S
conversion; ii) realizing high-S-loading electrode with reasonable conductivity to ensure both the
energy density and power rate; iii) controlled Li2S deposition/dissolution and iV) improving the
metallic lithium electrode, a task that has in common with Li battery as introduced in Chapter I.1
whist with an additional requirement in its chemical stability against polysulfides.
21
Another important class of type-II conversion reaction is the O2/Li2O2 conversion in Li-O2
battery with a high theoretical specific energy rivals that of gasoline. Because gaseous oxygen is
involved in the electrochemical reaction enlisting a solid-gas-solution reaction, Li-O2 batteries
possess intrinsically different merits and challenges, which will be introduced in the next
subchapter.
I. 3. 2 Li-O2 battery
Li-O2 batteries can operate under both aprotic and aqueous conditions, respectively forming
Li2O2 (2 Li + O2 ↔ Li2O2) and LiOH (4 Li + O2 + 2 H2O ↔ 4 LiOH) as the final discharge product.
Their cell configurations and working principles are depicted in Figure I. 5. In both cases, metallic
Li is used as the negative electrode, O2 gas as the reactant along with a porous carbon matrix as the
positive electrode to accommodate the discharge product. Because water readily reacts with
metallic Li, a Li conducting while electronic insulating layer or a polymer membrane must be
incorporated into the aqueous system to protect Li. During discharge, the metallic Li negative
electrode is oxidized to release Li+ into the electrolyte. At the positive electrode, the reactant (O2)
undergoes reduction before to combine with Li+ to form lithium peroxide (Li2O2) in non-aqueous
system; or to produce LiOH in aqueous system since water also participates in the reaction. The
reverse reactions are expected to occur on charge. Nevertheless, the decomposition of Li2O2 or
LiOH is unfavorable due to the difficulty of off-stoichiometric Li2−xO2 formation or O-H bond
cleavage, jeopardizing the reversibility and cyclability of the system. Because of the more complex
cell-design required for aqueous system and safety concerns, more attention has been paid to non-
aqueous system until a publication from Grey and coworker in 2015 revived interests and debates in
the aqueous systems.81
At this stage, it is clear that the alluring theoretical energy density of Li-O2 battery (3436
Wh L-1) arises from three aspects: i) metallic Li which offers the highest specific energy is used as
the negative electrode; ii) the reactant O2 gas is not contained within the system but comes from the
atmosphere; iii) the energy storage product Li2O2 (or LiOH) has a much lower molecular weight to
electron ratio than typical intercalation transition metal oxides used in Li-ion batteries (i.e., 23
g/(mol*e¯) for Li2O2 and 24 g/(mol*e¯) for LiOH versus 98 g/(mol*e¯) for LiCoO2, amounting to
specific energy of 3505 Wh Kg-1 and 3582 Wh Kg-1 for aprotic and aqueous Li-O2 batteries, and
387 Wh Kg-1 for Li-ion batteries). However, the practically accessible specific energy of Li-O2
22
batteries is less attractive for two reasons: i) the utilization of oxygen-containing pressure vessel or
oxygen-selective membrane is mandatory in order to cycle under pure O2 atmosphere, which will
inevitably jeopardize the energy density and increase the cost at the system level; ii) the loss of
electronic conductivity as arising from the deposition of insulating discharge products onto the
electrode surface.
Figure I. 5 Schematic illustration of (a) non-aqueous and (b) aqueous Li-O2 batteries. Note
that the yellow arrow indicates the discharge process, whereas the green arrow indicates the charge
process.
To overcome the challenges at the positive electrode, essentially one needs to understand the
electrochemical mechanisms of O2 reduction in Li+-containing non-aqueous electrolytes. Now it is
well understood that the first step involves the one electron reduction of O2 and the subsequent
combination with Li+ to form LiO2 as the intermediate (Li+ + e¯ + O2 → LiO2). Successively, two
different reaction paths have been envisioned: i) LiO2 adsorbed on the electrode surface (denoted as
LiO2*) to be converted into Li2O2 via a second-electron reduction (LiO2* + e¯ + Li+ → Li2O2), the
so-called “surface mechanism” or ii) LiO2 dissolved in the electrolyte (referred to LiO2(sol)) where it
undergoes a disproportionation reaction (2 LiO2(sol) → Li2O2 + O2), also known as “solution
mechanism” (Figure I. 6).82 Notably, these two mechanisms lead to a marked difference in the
morphologies of discharge products: surface mechanism generally induces film-like Li2O2 on the
electrode surface, whereas solution mechanism results in a precipitation of toroidal-shape Li2O2
particles. The morphology difference also suggests that the solution mechanism is more desirable
than the surface mechanism for maximizing the discharge capacity and the energy density: the
formation of insulating Li2O2 film on the electrode surface rapidly increases the charge-transfer
resistance thereby causes early cell death,83 whereas the solution-mediated growth of large toroidal
23
particles generally leads to larger discharge capacity. Because the competition between the surface
and the solution mechanism depends on the equilibrium between LiO2* and LiO2(sol), increasing the
solubility of LiO2 should offer opportunities to boost the discharge performance of Li-O2 batteries.
Researchers have suggested several strategies to enhance the solubility of LiO2 so as to promote the
solution-driven growth of Li2O2 including i) selecting electrolyte solvents with higher Gutmann
donor number82 or acceptor number;84 ii) utilizing salt anions with increased Li+ solvation
strength;85 iii) using protic additives;86 or iv) facet-engineering of the positive electrode.87
Figure I. 6 Schematic illustration of O2 reduction in Li+-containing non-aqueous electrolytes

following (a) the surface and (b) the solution mechanism.
Although the solution mechanism alleviates the surface passivation of positive electrode and
improves the discharge capacities of Li-O2 batteries, the electrolytes and additives that support this
mechanism also introduce additional parasitic reactions. For instance, strongly solvating solvents
are often more vulnerable to nucleophilic attack or proton abstraction by the reactive LiO2(sol)
intermediate.88 Further exploration is thus needed to reach an optimized combination between
solvating properties and stability. Alternatively, worth mentioning is the use of a molecular additive,
named 2, 5-di-tert-butyl-1,4-benzoquinone (DBBQ), which can promote solution-phase discharge
in weakly solvating electrolytes that are more stable towards LiO2(sol).89 This work dodges the
“double-edge sword” (high capacity but poor stability) and provides a new route to enable the
solution growth of Li2O2.
The decomposition reaction of Li2O2 during charge is another critical issue. As Li2O2 is a
poor electronic conductor with a large band gap of ~ 5 eV, a huge overpotential (0.5 ~ 1.3 V) is
typically observed on charge due to the sluggish electron transport. The situation is even worse for
the Li2O2 particles that are away from the electrode surface, for instance, those formed via solution
24
mechanism. Charging at high potentials not only undermines the round-trip efficiency, but also
causes carbon corrosion and electrolyte decomposition. Therefore, controlling the defect chemistry
of Li2O2 emerged as a potential strategy to improve the electronic conductivity in bulk Li2O2
including doping,90 creating off-stoichiometric (O-rich) Li2-xO291 and Li2O2 with surface/grain
boundaries.92 Along that line, tuning the crystal structure of Li2O2 through heterogeneous catalysts
like noble metal palladium was shown to lower the charge potential via the formation of amorphous
Li2O2 which has a better electronic conductivity than crystalline Li2O2.93,94 Nevertheless, caution
must be exercised in this strategy since the use of metallic particles also promotes the degradation
of organic electrolytes. Adding soluble catalysts (denoted as redox mediators hereinafter) to the
electrolyte that shuttle electrons between “distant” Li2O2 particles and the electrode surface has
been heavily explored as an alternative strategy to facilitate the Li2O2 oxidation. The fascinating
aspect of this approach is that the use of a redox mediator shifts the Li 2O2 oxidation from an
electrochemical pathway to a chemical pathway such that the battery will be charged at the redox
potential of the redox mediator, bypassing the high over-potential resulted from the electron transfer
limitation in bulky Li2O2 particles (Figure I. 7). Nevertheless, this intriguing approach has not yet
led to a breakthrough because of four considerations: i) the chemical stability of redox mediators
themselves, ii) degradation of other cell components induced by redox mediators, iii) once oxidized,
redox mediators might diffuse to the negative electrode (metallic Li) side, rather than diffuse to a
Li2O2 particle, creating an electron shuttle between the negative and the positive electrode, iv) other
insulating side products deposited on the electrode surface block the oxidation of redox mediators.
Overall, as compared to the discharge process, the charge of Li-O2 batteries is more
problematic and less understood. Several key puzzling questions still remain unanswered: i) what is
the charge transfer mechanism in Li2O2? ii) what is the origin for the differences in the shape of the
charge voltage profiles? and more importantly, 3) what is the decomposition reaction mechanism of
Li2O2?
25
Figure I. 7 Schematic illustration of (a) direct electrochemical Li2O2 oxidation and (b)
chemical Li2O2 oxidation mediated by redox mediators. Red “X” mark shows that Li2O2 particles
deposited away from the electrode surface fails to be oxidized because they are electronically
isolated from the electrode surface.
The last, and perhaps the most significant challenge, pertains to understand how the
constituents of ambient air affect the Li-O2 electrochemistry. Ideally, in a Li-O2 battery, O2 can be
fed directly from ambient air with a filtering system to ensure the removal of all other air
components. Nevertheless, such approach induces penalties on the energy density which
significantly drops from 3436 Wh L-1 at the material level to 411 Wh L-1 at the system level.95 Even
worse is that, practical-wise, trace amount of ambient impurities are inevitable. Amongst two
common impurities, CO2 and H2O will be considered in the following paragraphs.
The effects of H2O in Li-O2 electrochemistry are very complex and seem to vary with H2O
concentration, electrolytes and additives. Several possibilities were suggested in pioneering works: i)
small amount of H2O promotes LiO2 dissolution, leading to the formation of toroid Li2O2 through
the solution-mediated mechanism and therefore the increase of discharge capacity;86 ii) H2O slows
down the nucleation rate of Li2O2 on the electrode surface and thereby triggers the solution
mechanism;96 iii) increasing the acidity of H2O by specific catalysts like lithium iodide leads to the
formation of LiOH as the dominating discharge product instead of Li2O2;81,97 iv) H2O modifies the
discharge intermediate from superoxide to hydrosuperoxide;98 and v) H2O stabilizes the redox
mediators in their reduced form, shifting the discharge potential to higher values.99 In addition, the
presence of H2O also induces more parasitic reactions: i) H2O degrades metallic Li negative
electrode; ii) H2O may chemically react with Li2O2 to give LiOH that can hardly decompose during
26
charge; and iii) the stabilization of nucleophilic LiO2 species by H2O induces proton/hydrogen
abstraction from non-aqueous solvents, undermining the solvent stability.88 Overall, how the
addition of H2O will modify the morphology, the nucleation rate, the formation pathway and even
the chemical nature of discharge products are still open questions. Unlike for the discharge process,
so far little is known about the effect of H2O on the charge process.
Equally important is to understand the influence of CO2 contamination in Li-O2 batteries.

Coupled galvanostatic charge/discharge measurements and density function theory (DFT)
calculations suggest that, even a low concentration of CO2 (1 %) could effectively blocks the
nucleation sites of Li2O2 and alters its shape and growth directions due to Li2CO3 formation,
resulting in an increased discharge capacity at the expense of a raise in charge potentials.100 Apart
from increasing the charge over-potential thus undermining the round-trip efficiency, the presence
of Li2CO3 as discharge product deteriorates the rechargeability of Li-O2 batteries in another three
aspects: i) the portion of Li2CO3 which fails to be decomposed on subsequent charges concentrates
at the electrode surface leading to early cell death; ii) other cell components (i.e., porous carbon,
binder, solvents, salts…) are likely unstable at the very high charge potentials induced by the high
Li2CO3 concentration on the electrode surface; and iii) the decomposition of Li2CO3 releases
reactive superoxide radicals (2 Li2CO3 → 2 CO2 + 4 Li+ + O2●¯ + 3 e¯) which are detrimental for
the stability of organic solvents.101 Moreover, CO2 was shown to react with O2●¯in Li+-containing
non-aqueous electrolytes to form Li2CO3 via an electrochemical pathway.102 This potential reaction
pathway further underscores the importance of examining how the discharge reactions and the final
discharge products will be altered by the existence of CO2. This question will be addressed more in
detail in Chaper IV.
In light of the severe energy inefficiency associated with the phase change between solid
Li2O2 and gaseous O2, solid-phase conversion (Li2O(s)  Li2O2(s)  LiO2(s)) avoiding O2 gas
uptake and release was alternatively explored with nanolithia composite electrodes.103,104 Because
the reaction proceeds via the redox of light element oxygen akin to Li-O2 battery, this
Li2O/Li2O2/LiO2 chemistry preserves the merit of high energy storage (1000 Wh kg-1) while
significantly reduces the charge overpotential from 0.5 ~ 1.3 V to ~ 0.24 V. Nevertheless, further
studies are awaited to overcome the critic barrier for advancing this new energy storage system: i)
the low electrochemical activity and poor electronic conductivity of Li2O; ii) possible O2 gas release
27
on further oxidation due to the metastability of the delithiated Li2O; and iii) parastic reactions
associated with LiO2 which readily reacts with organic electrolytes, as often observed in Li-O2
batteries. Transition metal doping and the use of nanosized electrode were adopted to promote this
electrochemical reactions.105 In parallel, an internal charge shuttle may be incorporated to maintain
the charge voltage below the thermodynamic potential for O2 to be evolved from Li2O2 (i.e., 2.96 V),
offering the chance to prevent O2 gas evolution.104
I. 4 Conclusions
The energy density of Li-ion batteries has been tripled since their initial commercialization
(from 80 Wh kg‒1 in 1991 (Sony) to 250 Wh kg‒1 in 2017 (Tesla)), thanks to the continuous
improvement of the layered oxide chemistry (LiCoO2) via chemical substitutions. In the recent
years, we have seen the emergence of Li-rich layered oxides showing oxygen redox activity aside
from the conventional cationic one, with great potential for application. Indeed, oxygen redox is an
intriguing concept to move beyond the outlook of the classical positive electrode materials. In order
to explore the full potential of this concept, it is crucial whilst challenging to maximize the
reversible anionic redox participation while maintaining the structural integrity especially at high
state of charge. Tridimensional structures should be advantageous over layered structures in this
regards.60 On the other hand, most of the Li-rich oxides reported so far contain heavy and expensive
4d or 5d transition metal ions such as Nb, Mo, Ru and Ir, hence designing oxygen redox based
positive electrodes with light and cost-effective 3d transition metals would be beneficial practical-
wise.
In parallel, numerous new chemistries are actively pursued hoping to surpass the energy
limit of conventional Li-ion intercalation chemistry. Marked improvements have been achieved
regarding the fundamental understanding of the charge/discharge mechanisms, the interfacial
chemistry/electrochemistry and critical parasitic reactions. Interesting concepts/strategies continue
to emerge but are often accompanied with new challenges. Overall, each chemistry newly proposed
has its own distinct technical challenges to overcome and fundamental questions to answer. It is so
far unclear whether or when any of those new battery technologies will reach mass
commercialization. Nevertheless, even if their practical applications may not be witnessed in the
near future, investigating those different whist relevant battery systems has been a rich learning
28
playground. Notably, it leads to a better understanding about how to stabilize the reactive
intermediates via playing with solvation-desolvation kinetics so as to elaborate the desired reaction
pathway.
29
Chapter II Monitor the oxygen release at high
potentials in Li-rich layered oxides
Chapter II Monitor the oxygen release at high potentials in Li-rich layered oxides
Chapter II Monitor the oxygen release at high potentials in Li-

rich layered oxides
II. 1 Background and motivation
As mentioned in Chapter I, the recent discovery of anionic redox has effectively unlocked
the energy storage limitation of positive electrode materials, which was previously believed to be
pinned by the transition metal cations (how much their oxidation state can change). Despite the fact
that combined redox activity of cations (transition metals) and anion (oxygen) can push the positive
electrode capacity well beyond what the sole cationic redox can offer,26,43 most of these Li-rich
layered oxides studied so far exhibit irreversible O2 gas release when charged to high potentials, the
extent of which depends on the nature of the material (chemical composition, band structure,
surface chemistry, dimensions, etc.).44,57,106 Combined experimental and theoretical calculations
have shown that such lattice oxygen loss which deteriorates the materials’ performances
(irreversibility, parasitic surface reactions, voltage fade, and safety) is unfortunately the largest for
the 3d metals which present the greatest interest practical-wise.46,51-54,59 Aware of these limitations,
and searching for high-energy-density, cost-competitive and stable positive electrode materials,
intense research activities have therefore been devoted to the operando detection and reliable
quantification of possible O2 evolution.107-111 The direct probe of O2 gas release is of special
significance not only to provide insights about the irreversible part of anionic redox activities, but
also to help in understanding the origins of bulk structure reorganization, surface morphology
change as well as battery performance deterioration.28,47,50,106,112-115
A straightforward approach to probe the gas evolution phenomenon is to employ operando

gas/pressure analytical techniques which can directly analyze the gaseous products formed during
cycling. When this work was started, two of such techniques have been heavily implemented in the
battery field. The first one is the differential (or online) electrochemical mass spectrometry, also
termed as DEMS (or OEMS), which allows the identification and quantification of various gaseous
species evolved/consumed upon battery operation. This technique has been extensively used in Li-
O2 battery research to quantify O2 consumption/evolution,116-118 and has been implemented in Li-
ion battery field to follow the gas evolution caused by the degradation of battery components since
30
1990s (salt, electrolyte, positive electrode material, negative electrode material…).119-124 Despite
OEMS providing valuable insights toward the complex electrochemical processes occurring upon
battery cycling, the disadvantages related to the utilization of OEMS are nested in its complexity,
high cost and sensitive operation, which currently hamper its day-to-day use. Such limitations have
led to the development of another technique termed as in-situ pressure cell, which can be easily
implemented to enable long-time gas monitoring.125-127 However, the main drawback of this rather
user-friendly approach being that it only measures the pressure change, and thus does not allow for
determining the nature of the evolved gases. These limitations naturally motivate us to search for
simple and accurate alternate that can be used to rapidly screen new positive electrode materials
regarding their electrochemical stability against O2 release.
Within this context, the rotating ring disk electrode (RRDE) voltammetry has caught our
attention. RRDE is an electrochemical technique which offer unique opportunity to investigate the
short-lived intermediates generated in steady-state measurements.128-130 Benefiting from the laminar
electrolyte flow created by the mechanical rotation and the existence of double concentric working
electrodes (the disk and the ring electrode) separated by a non-conductive barrier, intermediates
generated at the disk electrode can be swept across the surrounding ring electrode along with the
electrolyte flow, and be electrochemically observed (detected) at the ring. In fact, this concept has
much in common with the “generation-collection” experiments in scanning electrochemical
microscopy (SECM).131 If the intermediate is stable in the transit time (i.e., the average time
required for the intermediate to traverse the gap between the disk and the ring electrode, which is a
function of the gap distance and the rotation rate), then the collection efficiency given by the ratio
of the disk current and the ring current is only a function of the geometry of the electrode. Giving
RRDE experiments in which the ring electrode is maintained at a sufficient positive (negative)
potential such that any intermediate reaching the ring is rapidly oxidized (reduced), knowledge
about species produced at the disk electrode surface can be obtained by simply measuring the
magnitude of the ring currents. Therefore, since its first introduction in 1958 by Alexander
Frumkin,130 RRDE has been widely applied to study the reaction kinetics and transport properties
for the oxygen reduction reaction (ORR) relevant to metal-air batteries and proton exchange
membrane (PEM) fuel cell,132-138 as well as to probe the intermediates of lithium-sulfur redox
reactions,139 to study the manganese dissolution of spinel-based positive electrodes,140 to survey the
dissolution of metal ions during the corrosion processes,141,142 and to determine the faradaic
31
143,144
efficiency of oxygen evolution reaction (OER) electrocatalysts for water splitting, among
other usages.
In this chapter, we will describe our methodology of using RRDE voltammetry to monitor
the O2 release phenomenon in positive electrode materials.145 To this end, two families of well-
studied Li-based layered oxides (Li-stoichiometric and Li-rich NMCs as well as Li2Ru1-yTiyO3
phases) were chosen for initial evaluation. As a proof-of-concept, the results obtained by our RRDE
method will be further compared with those obtained by the existing pressure/gas analysis
techniques (in-situ pressure cell and OEMS). Lastly, the merits as well as the limitations of this new
RRDE approach for O2 gas detection will be evaluated, and an outlook on its future optimization
and possible extension will be provided.
We will start in the next section with a brief description of the three gas analysis techniques
used in this work: in-situ pressure cell, OEMS and the newly proposed RRDE voltammetry.
II. 2 Description of the in-situ gas analysis techniques
II. 2. 1 In-situ pressure cell
Figure II. 1a presents a photo as well as a corresponding schematic of the in-situ pressure
cell developed in our laboratory by Dr. F. Lepoivre and employed throughout this thesis. The
pressure cell is derived from the Swagelok-type cell, and is composed of a gas reservoir of 9.7 cm3,
a gas outlet/inlet that can be connected to the gas filling station (a schematic of the gas filling
station is given in Appendix Figure A. 1. 1) for purging/refilling the cell with desired gases,
together with a pressure sensor which measures the pressure evolution inside the cell. More detailed
descriptions can be found in the Ph.D. work of Dr. F. Lepoivre.146 Figure II. 1b presents the
pressure (blue solid line) and voltage profile (black solid line) typically measured by in-situ
pressure cell, from which the specific voltage where the cell starts to degas and the quantity of the
evolved gas can be deduced.
32
Figure II. 1 (a) Photograph (left side) and schematic (right side) of the in-situ pressure cell.
Adapted from: Ph.D. thesis of Florent Lepoivre, 2016. (b) A representative in-situ pressure profile
during the cycling of a Li2Ru0.75Ti0.25O3 half-cell.
II. 2. 2 Online electrochemical mass spectrometry
Online electrochemical mass spectrometry (OEMS) is composed of an electrochemical cell

with both a vacuum system and a quadrupole mass spectrometer which analyzes the volatile species
sampled in the head space of the electrochemical cell.118,147 Figure A. 1. 2 in Appendix shows a
schematic view of the whole OEMS system used in this work, which is a closed system allowing
for the continuous gas sampling from the electrochemical cell to the mass spectrometer (MS)
through a thin capillary. More detailed description of the system can be found in Appendix section
A. 1. 2. Figure II. 2 present the electrochemical cell used for the OEMS experiments. The cell is
derived from the in-situ pressure cell with one single modification: the quick connector for gas
inlet/outlet in the pressure cell has been replaced by a Swagelok adaptor which can be connected to
either a connector to the inlet capillary of MS or to the gas filling station.
33
Figure II. 2 Picture (left side) and schematic (right side) of the electrochemical cell
dedicated to the closed OEMS system.
II. 2. 3 Rotating ring disk electrode voltammetry
As illustrated in Figure II. 3a, RRDE tip is constructed by embedding in the cylindrical
insulator (polytetrafluoroethylene (PTFE) is typically used owing to its great chemical stability and
inertness) a central disk electrode and a surrounding ring electrode. Depending on the target usage,
different materials and sizes can be selected for the electrodes as well as the insulator. Platinum ring
is chosen in this work due to its remarkable catalytic activity for oxygen reduction reaction, as
witnessed in the fields of Li-O2 battery and fuel cell.148,149 The disk and the ring electrode are
electrically separated by an insulating barrier (a PTFE U-cup as marked with blue arrow in Figure II.
3a). The RRDE experiments are carried out with a bipotentiostat, which controls independently the
potentials of the disk and the ring electrode with respect to the reference electrode and measures
separately the current going through each electrode (Figures II. 3b & 4). The RRDE tip is also
connected with a motor controller through a shaft which allows for rotating both the two electrodes
at the same rate (Figure II. 4). A schematic of the whole RRDE system together with a four-
electrode electrochemical cell is presented in Figure II. 4 and more details about the set-up can be
found in Appendix Section A. 2. 1.
34
Figure II. 3 (a) Picture of the RRDE tip used in this work and (b) block diagram of the
bipotentiostat (CE stand for counter electrode, Ref stands for reference electrode), adapted from
Bard et al.150
Figure II. 4 (a) Schematic of RRDE system and the four-electrode cell configuration, WE1,
WE2, CE and RE stand for working electrode 1 (disk electrode), working electrode 2 (ring
electrode), counter electrode and reference electrode, respectively. (b) A magnified view of the
RRDE tip and its near-electrolyte region depicts the transport and redox processes occurring in the
RRDE system designed to assess the O2 release phenomenon in Li-rich layered oxides.
In this work, we are interested in studying the O2 gas generation from Li-rich layered oxides.
To this end, the oxides of interest were firstly deposited onto the disk electrode and linear scan
35
voltammetry, cyclic voltammetry or galvanostatic techniques were performed to charge the oxides
until high potentials where O2 gas release likely occurs. Simultaneously, the ring electrode is
maintained at a sufficiently negative potential for oxygen generated at the disc electrode to be
rapidly reduced. To determine the required ring potential, we performed initial oxygen reduction
measurements in LP30 electrolyte (1 M LiPF6 dissolved in 1:1 v/v Ethylene carbonate/Dimethyl
carbonate) with platinum rotating disk electrode at a rotation speed of 1600 rpm (Figure II. 5a). The
gradual decrease in current densities observed after 2.4 V vs Li+/Li shown that further increasing
the overpotential would drastically decrease the active catalytic sites owing to surface passivation
by Li2O2 (the ultimate O2 reduction product in Li+-containing electrolytes).
Figure II. 5 (a) Potentiodynamic O2 reduction current densities on a platinum rotating disk
electrode at a rotation speed of 1600 rpm in LP30 electrolyte with a scan rate of 10 mV s-1. Note
that the dashed square indicates that the current density decreases as the overpotential increases,
owing to the passivation of Pt electrode surface. (b) RRDE profiles obtained with Li2Ru0.5Ti0.5O3 at
a rotation speed of 1600 rpm in LP30 electrolyte by holding the ring electrode at different potentials:
2.3 V (blue curve), 2.4 V (yellow curve), and 2.5 V (black curve). Linear scan voltammetry was
applied at the disk electrode to delithiate Li2Ru0.5Ti0.5O3. Note that the dashed square marks the
increase of reductive currents due to O2 reduction. The black arrow reveals that similar onset
potential is obtained for O2 reduction, irrespective of the applied ring potential.
We then performed preliminary RRDE measurements with a Li-rich compound which

known to release O2, namely Li2Ru0.5Ti0.5O3, by holding the ring at various potentials (2.3 V, 2.4 V
and 2.5V), similar onset potential were observed for O2 reduction (detection) irrespective of the
36
applied ring potential (Figure II. 5b). Therefore, the ring potential was selected to be 2.5 V vs Li+/Li,
a value that is sufficient to allow for O2 reduction (detection) while relatively positive to minimize
reductive parasitic reactions. Note that all the results reported in this work were at least duplicated
two times but not corrected for the background ring current, the magnitude of which being typically
less than 3 µA. The background ring currents show a tendency to decrease as the concentration of
residual oxygen and/or impurity in the electrolyte decreases.
II. 3 Results and discussion
Despite the fact that combining cationic (transition metals) and anionic (oxygen) redox
activity can enhance the capacity well beyond what the sole cationic redox can offer with capacities
exceeding 280 mAh g-1 for Li1.2Ni0.13Mn0.54Co0.13O2 (Li-rich NMC) as compared to solely 180 mAh
g-1 for LiNi1/3Mn1/3Co1/3O2,26,43 classical Li(NixMnyCoz)O2 denoted as NMC which are presently
used in commercial Li-ion batteries do not show O2 gas release upon oxidation in contrast to the Li-
rich phases.28,44,46,107,108,124 As a proof of concept to demonstrate the usage of RRDE for O2 gas
detection, we therefore decided to firstly investigate these two NMC phases: the stoichiometric
LiNi1/3Mn1/3Co1/3O2 (NMC 111) and Li1.2Ni0.13Mn0.54Co0.13O2 (Li-rich NMC).
Figures II. 6a & 6b present their voltage and internal pressure change as a function of
capacity. The cell with Li-rich NMC displays a discernable increase of pressure during the high-
voltage charging plateau (from a to b) followed by a rapid pressure growth (from b to the end of
charge) (bottom panel in Figure II. 6a). The two different pressure increments (from a to b, and
from b to the end of charge) during the charge period indicate the presence of at least two different
phenomena governing the release of gases, which will be further discussed below. Unlike for Li-
rich NMC, only a minor increase of the pressure was measured for NMC111 throughout the charge
(from f, bottom panel in Figure II. 6b), suggesting negligible gas evolution. In order to discern the
chemical nature of the evolved gases, OEMS measurements were further carried out. Figures II. 6c
& 6d show the OEMS data obtained with NMC 111 and Li-rich NMC during their first charges
followed by an open circuit voltage (OCV) step for 0.5 h. Note that only O2 and CO2 are
considered here as they represent for the major constitutes of the sampled volatile species. Starting
with the Li-rich NMC (Figure II. 6c), the evolution of O2 was measured only after the high-voltage
charging plateau (starting from d). Unlike for O2, the CO2 evolution starts at the beginning of the
37
high-voltage charging plateau (from c), before to dramatically increase at the end of charge
simultaneously with the evolution of O2 (from d). These results are in good agreement with
previous reports.47 Moreover, they explain the results previously obtained by pressure analysis
(Figure II. 6a) with the initial slight increase of pressure (from a to b) originating solely from CO2
evolution, while both O2 and CO2 evolutions contribute to the second rapid pressure build-up (from
b to the end of charge). In the case of NMC 111 cell (Figure II. 6d), CO2 was the only detected gas,
accounting for the minor pressure build-up previously observed in pressure cell, whereas no O2
evolution was observed during both the charge and OCV period.
38
Figures II. 6 Galvanostatic charge curves (top panels) and their respective pressure change
(bottom panels) for Li-rich NMC (a) and NMC 111 (b) in in-situ pressure cells. Potential (top
panels) and gas evolution rate (bottom panels) as a function of time (and capacity) for Li-rich NMC
(c) and NMC 111 (d) measured by OEMS. RRDE profiles measured for Li-rich NMC (e) and NMC
111 (f) in LP30 electrolyte. Top panels present the galvanostatic charge curves obtained in the disk
electrode. Bottom panels illustrate their corresponding ring current response. All the cells are
charged at a current rate of 0.3 C to 4.8 V vs Li+/Li. Gas evolution onsets are defined by dash lines
labeled from a to g as guide for the eye.
39
Knowing that O2 is evolved at the end of charge for Li-rich NMC but not for NMC 111,
RRDE measurements were then performed for both materials. The respective voltage-capacity
profiles are presented in Figures II. 6e & 6f (top panels) and are in agreement with those obtained
with pressure and OEMS cells. Turning to the ring current, an increase of the reduction current was
measured for Li-rich NMC close to the end of charge (from e, Figure II. 6e), alike the potential at
which O2 release was previously detected by OEMS and pressure measurements. In contrast, no
such cathodic current increase was observed (bottom panel in Figure II. 6f) for NMC 111 in
agreement with the absence of O2 release as deduced from OEMS. The obvious correlation of the
cathodic current increase and the oxygen evolution for Li-rich NMC, as well as the absence of
cathodic current increase and oxygen release for NMC 111, allows us to attribute the increase of
reduction current to the electrochemical reduction of O2 evolved from Li-rich NMC. Since the
evolution of O2 is accompanied by the evolution of CO2, as shown by our OEMS measurement with
Li-rich NMC (Figure II. 6c), the effect of CO2 generation on the detection of O2 at the ring
electrode was then considered. As previously mentioned, the release of CO2 begins for Li-rich
NMC at an earlier stage of the charge (before the plateau, from c, Figure II. 6c) compared to the
onset potential for O2 generation (after the plateau, from d, Figure II. 6c). However, an increase of
the reduction current was only recorded, at the ring electrode, after the plateau (from e, Figure II. 6e)
where O2 starts to release. Therefore, one can conclude that RRDE specifically detects O2 while
being silent to CO2 when holding the ring potential at 2.5 V vs Li+/Li. This conclusion is further
supported by performing RRDE measurements with a mixture of NMC 111 and Li2CO3 (Li2CO3 is
known to evolve CO2 but not O2 within the charge potential range)101,151 where no modification for
the ring current response is observed compared to pure NMC 111 (Figure II. 7). Bearing in mind
that the ring electrode detects every soluble product that could be reduced/oxidized at the ring
potential (impurities, dissolved metal ions from oxides and/or counter electrodes, electrolyte
degradation products, etc.), we then evaluate whether species other than O2 could contribute to the
increase of the cathodic ring current upon charging Li-rich NMC. By measuring bare and Csp-
loaded glassy carbon disk electrodes in cyclic voltammetry mode up to 4.8 V vs Li+/Li, no increase
of the reduction current was observed at the ring electrode (Figure II. 8). Therefore, the detection of
species originating from the carbon additive, binder and/or direct electrolyte oxidation could be
excluded. Finally, to demonstrate the versatility of the proposed RRDE method for O2 detection,
linear scan voltammetry mode was also applied at the disk electrode instead of galvanostatic
charging, giving a good agreement between RRDE and pressure cell measurements (Figure A. 2. 3).
40
Hence, by comparing with in-situ pressure and OEMS measurements, we could demonstrate that
the RRDE method can indeed selectively detect O2 generated from Li-rich NMC.
Figures II. 7 RRDE profile for 95 % NMC 111 + 5 % Li2CO3 measured at 1600 rpm in
LP30 electrolyte. The bottom panel presents the galvanostatic charge curve obtained in the disk
electrode at a current rate of 0.3 C to 4.8 V vs Li+/Li and the top panel illustrates its corresponding
ring current response.
Figures II. 8 RRDE profiles on bare (yellow line) and Csp-casted (blue line) glassy carbon
disk electrodes cycled at a scan rate of 0.5 mV s-1 between 2.4 and 4.8 V vs Li+/Li.
To assess if the RRDE method can be utilized to detect oxygen release for other Li-rich
compounds, we further extended our work to the Li2MO3 family,51,152,153 with a special attention
41
paid to the Li2Ru1-yTiyO3 series which was previously shown to evolve O2 at high potentials.154
Applying the same methodology, we first monitored with in-situ pressure cell the gas release upon
charging Li2Ru0.5Ti0.5O3 (Figure II. 9a). A slight increase of the internal cell pressure was recorded
at the beginning of the high-voltage charging plateau (from a to b, Figure II. 9a), followed by a
rapid growth (from b to the end of charge). Subsequently, OEMS measurements show that the
evolution of CO2 begins at around 4.0 V vs Li+/Li (line c, Figure II. 9b), which correlates well with
the onset of pressure increase detected by pressure cell (line a, Figure II. 9a). Thereafter, O2 is
found to evolve at a relatively higher rate in comparison to the CO2 evolution (from d, Figure II. 9b).
In conclusion, upon charging Li2Ru0.5Ti0.5O3, CO2 first evolves resulting in a small pressure
increase, then O2 evolution turns on at a higher rate accompanied with the evolution of CO2 gas
leading to a rapid growth of pressure. Now switching to the RRDE technique (Figure II. 9c), an
increase of the cathodic current was recorded at the ring electrode (from e, bottom panel), similarly
to Li-rich NMC phase. The onset of this ring current increase (~ 4.0 V vs Li+/Li) is in good
agreement with the onset of O2 evolution measured by OEMS (~ 4.06 V vs Li+/Li), which indicates
that O2 is evolved from Li2Ru0.5Ti0.5O3 and is detected at the ring electrode via its electrochemical
reduction. It is worth mentioning that the capacity obtained in charge with the RRDE technique was
~ 20% smaller compared to that obtained with pressure and OEMS cells. This can be explained by
the different cell configurations and by the fact that a small portion of the oxide remains inactive
due to the lack of stacking pressure in RRDE set-up. Unlike for Li2Ru0.5Ti0.5O3, similar capacities
were obtained for NMCs materials by RRDE, pressure and OEMS cells, suggesting their better
electronic conductivity as compared to the Li2Ru1-yTiyO3 series.
42
Figures II. 9 (a) Galvanostatic charge curve (top panel) and its respective pressure change
(bottom panel) for Li2Ru0.5Ti0.5O3 in in-situ pressure cell. (b) Potential (top panel) and gas evolution
rate (bottom panel) as a function of capacity for Li2Ru0.5Ti0.5O3 measured by OEMS. (c) RRDE
profiles for Li2Ru0.5Ti0.5O3 measured in LP30 electrolyte. Top panel shows the galvanostatic charge
curve obtained in the disk electrode. Bottom panel illustrates its corresponding ring current
response. All the cells are charged at a current rate of 0.3 C to 4.5 V vs Li+/Li. Gas evolution onsets
are defined by dash lines labeled from a to e as guide for the eye.
43
The effect of titanium substitution in Li2Ru1-yTiyO3 series on O2 evolution was then studied
by measuring Li2Ru0.75Ti0.25O3. As seen in Figures II. 10a & 10b, the evolution of O2 begins at
around 4.1 V vs Li+/Li (line d), leading to a rapid growth in internal cell pressure (from b to the end
of charge). As expected, an increase of ring reduction current was recorded by RRDE at this
potential (from e, Figure II. 10c), indicating the reduction (detection) of O2. This result clearly
confirms that the proposed RRDE method is capable of probing the release of O2 in various Li-rich
materials. Finally, we could demonstrate the reproducibility of these results by performing
repetitive RRDE experiments with both Li2Ru0.75Ti0.25O3 and Li2Ru0.5Ti0.5O3 phases. As seen in
Figure A. 2. 4, nearly identical disk electrochemical performances (showing the charge of the oxide)
and ring current responses (corresponding to the detection of O2) were obtained.
44
Figures II. 10 (a) Galvanostatic charge curve (top panel) and its respective pressure change
(bottom panel) for Li2Ru0.75Ti0.25O3 in in-situ pressure cell. (b) Potential (top panel) and gas
evolution rate (bottom panel) as a function of capacity for Li2Ru0.75Ti0.25O3 measured by OEMS. (c)
RRDE profiles for Li2Ru0.75Ti0.25O3 measured in LP30 electrolyte. Top panel shows the
galvanostatic charge curve obtained in the disk electrode. Bottom panel illustrates its corresponding
ring current response. All the cells are charged at a current rate of 0.3 C to 4.5 V vs Li+/Li. Gas
evolution onsets are defined by dash lines labeled from a to e as guide for the eye.
45
At this stage of the study, the sensitivity of the RRDE technique is worth discussing. As
seen in Figures II. 11, when using the same amount of active material for pressure analysis as for
the RRDE study (~ 100 µg), the pressure measured upon charging Li2Ru0.75Ti0.25O3 remained nearly
constant when using a scan rate of 0.5 mV s-1 (green curve) or 1.0 mV s-1 (blue curve). Only when a
slower scan rate was applied (0.1 mV s-1, red curve) to allow longer time for gas accumulation in
the headspace and most likely larger amount of O2 gas release, a pressure increase was then
recorded. In fact, the mass loading of active materials used for pressure analyses and OEMS
experiments were respectively ≈ 70 and 130 times greater than that used for the RRDE
measurements. The capability of RRDE method in detecting O2 at a significant lower mass loading
(lower O2 concentrations) suggests a better sensitivity. Therefore, the proposed RRDE method can
be expected to be a useful tool for studying materials from which relatively smaller amount of O2 is
released, which may be below the detection limit of the pressure analysis or OEMS measurement.
We should note that all the results reported herein were not normalized to the loading of active
materials, therefore the magnitude of pressure build-up, O2 mass spectrometry signal and ring
currents increment should not be considered for quantitative comparison or sensitivity indicators.
Figures II. 11 Cyclic voltammograms (bottom panels) recorded in in-situ pressure cells at
various scan rates: 0.1 mV s-1 (yellow line), 0.5 mV s-1 (black line), and 1.0 mV s-1 (blue line)
cycled between 2.4 and 4.5 V vs Li+/Li and their respective pressure changes (top panels) for
Li2Ru0.75Ti0.25O3 with the same mass loading of active material as used in RRDE measurements.
46
To further evaluate the potential use of RRDE for accurately determining the state of charge
at which oxygen is released from Li-rich compounds, cyclic voltammetry experiments were carried
out for Li2Ru0.75Ti0.25O3 at various scan rates using both pressure cell and the RRDE set-up. During
the RRDE tests, the onset potential for O2 release was constantly determined at ~ 4.03 V vs Li+/Li at
every scan rate (Figure II. 12a), while for pressure measurements the measured onset potentials for
gas release vary with the scan rates (Figure II. 12b).
Figures II. 12 Comparison of RRDE profiles (a) and in-situ pressure analysis (b) for
Li2Ru0.75Ti0.25O3 recorded at various scan rates: 0.1 mV s-1 (yellow line), 0.5 mV s-1 (blue line), and
1.0 mV s-1 (black line). The cyclic voltammograms are presented in the bottom panels and their
respective ring current response (a) and pressure change (b) are shown in the top panels.
This different behavior is in fact not surprising when considering the different working
principles of these two techniques. In pressure measurements, the gaseous species generated at the
electrode must first diffuse through the electrolyte and accumulate in the headspace of the
electrochemical cell before being detected by the pressure sensor. Unlike for pressure
measurements, with the RRDE setup, the transport of O2 is controlled by the convection movement,
i.e. the laminar electrolyte flow created by mechanical rotation, which is faster than the diffusion of
O2 in electrolytes. Aside from the faster O2 transportation rate in RRDE set-up, the unique ring-disk
geometry also provides a shorter O2 transport length than in a pressure cell. Besides, the sensitivity
of pressure measurement (and also OEMS) is jeopardized by the amount of gas dissolved into the
electrolyte as well as the dead volume of the electrochemical cell, which is not the case for RRDE
method where O2 can be detected once it is generated at the disk electrode surface and swept
47
outwards the ring electrode. In addition, pressure cell and OEMS measure both the gas released
from the material itself and the gases resulted from electrode surface reactions (e.g., electrolyte
reactions at the electrode surface), RRDE can minimize the contribution of the latter and more
selectively detect O2 released from the material by shortening the time needed for detection.
Considering all the pros and cons, we could conclude that RRDE is a complementary technique to
the in-situ pressure cell and the OEMS, and can be employed to accurately determine the onset
potential for oxygen release.
After demonstrating that RRDE is capable to detect the oxygen loss from Li-rich
compounds during their initial charges, the electrochemical behavior for Li2Ru0.5Ti0.5O3 and
Li2Ru0.75Ti0.25O3 compounds during the second voltammetric cycles were recorded (Figure II. 13).
The results reveal a very limited increase of the ring currents arising from the reduction of O 2
generated on the surface of these two compounds, which is consistent with the small irreversibility
previously observed during galvanostatic cycling after the first charge.154
Figures II. 13 RRDE profiles for Li2Ru0.5Ti0.5O3 (a) and Li2Ru0.75Ti0.25O3 (b) during their
second cycles at a scan rate of 0.5 mV s-1, with the cyclic voltammograms presented in the bottom
panels and their respective ring current evolution shown in the top panels.
The question then comes to the quantitative analysis of the amounts of O2 generated from
the Li-rich materials which can theoretically be deduced from the integration of the cathodic ring
current increase. Comparing the results obtained for Li2Ru0.5Ti0.5O3 (Figure II. 14a) and
Li2Ru0.75Ti0.25O3 (Figure II. 14b), the quantity of O2 released from Li2Ru0.5Ti0.5O3 was calculated to
48
be larger than that from Li2Ru0.75Ti0.25O3 (1.036 % and 0.62 % loss of the total oxygen content for
Li2Ru0.5Ti0.5O3 and Li2Ru0.75Ti0.25O3, respectively, see Appendix Section A. 2. 3 for detailed
calculation). This trend correlates well with the electrochemical performance for which a greater
irreversibility was observed during the first cycle with increased titanium substitution in the Li 2Ru1-
xTixO3 series.154 However, discrepancy arises when comparing their final compositions after the
initial charge calculated based on RRDE and pressure measurements. For Li 2Ru0.5Ti0.5O3, a
composition of Li0.65Ru0.5Ti0.5O2.74 was obtained from the pressure increase by assuming that all the
detected gas is oxygen, whereas a composition of LiRu0.5Ti0.5O2.97 is calculated from the RRDE
results assuming a collection efficiency of 25 % (Table A. 2. 1). Even though the contribution from
other gas components such as CO2 to the pressure evolution has not been taken into account, the
expected value for oxygen loss measured by RRDE should be higher since O2 is the main gas
evolved upon charging Li2Ru0.5Ti0.5O3, as deduced from OEMS measurements (Figure II. 9b). Such
discrepancy could likely arise from the lower collection efficiency for O2 at the ring electrode
compared to the theoretical value given by the geometrical parameters of the RRDE tip (25.05 %
0.15 with the geometry used in this study). Indeed, this theoretical value is only achieved for
soluble species such as ferrocene (Figure A. 2. 2), while for dissolved gas like O2, limitations
associated with a limited O2 solubility in the supporting electrolytes as well as possible bubble
formation from the evolved O2, as discussed elsewhere,143,155 can significantly lower the O2
detection efficiency. Besides, a precise quantification of the amount of released oxygen is further
complexed by the fact that small errors in the ring current and the mass loading can lead to a
relatively large deviation. This is especially true considering the small amounts of active materials
used for the RRDE measurements. Therefore, we believe that the amounts of O2 measured from the
ring current increase can be used for qualitative comparison among different materials, but caution
must be exercised when implement such comparison. The values obtained by RRDE should be
complemented by other techniques such as OEMS and/or pressure cells for quantification purposes.
49
Figures II. 14 Comparison of RRDE profiles for Li2Ru0.5Ti0.5O3 (a) and Li2Ru0.75Ti0.25O3 (b)
recorded at a scan rate of 0.5 mV s-1 with the cyclic voltammograms presented in the bottom panels
and their respective ring current evolution shown in the top panels. The blue shaded area in the top
panels represents the increment of cathodic ring current due to the reduction of detected O2.
II. 4 Chapter conclusion and outlook
In this chapter, we propose the utilization of rotating ring disc electrode (RRDE)
voltammetry for identifying the O2 gas release phenomenon in Li-rich layered oxides. This simple
method includes two sequential steps: (i) the deposition of the oxides of interest onto the disk
electrode surface via a drop-casting method; (ii) application at the disk electrode of a standard
electrochemical technique (i.e., linear scan voltammetry, cyclic voltammetry or galvanostatic
cycling) to control the delithiation/lithiation of the oxides, while concomitantly maintaining the ring
electrode at a potential which oxygen reduction reaction could effectively takes place. By rotating
the whole electrode assembly, O2 gas released from the oxides can reach and be electrochemically
detected (reduced) in the ring electrode. We have validated the proposed methodology by
conducting RRDE measurements using various Li-rich layered oxides and found that our results
nicely compare with those obtained by OEMS and in-situ pressure cell. Through this comparative
study, we could demonstrated that RRDE i) selectively detect oxygen while being silent to other
species potentially generated at high potentials, ii) show better sensitivity and reliability than the
aforementioned techniques allowing for the detection of O2 at relatively lower concentrations, iii)
enable a precise determination of the onset potential for O2 gas release, and 4) can be used to
50
rapidly screen the O2 release phenomena in a wide variety of new materials owing to its wide-
spread in the fields of batteries and electrocatalysis. We believe that the proposed RRDE approach
which offers high sensitivity while being fast and simple could greatly help to advance the
understanding of oxygen redox processes and promote the design of new oxygen-redox-based
materials. However, caution must be exercised when tentatively quantifying with RRDE the amount
of O2 generated from the materials. Overall, the RRDE method comes as a valuable complement to
the existing pressure/gas analysis techniques which usually do not allow for either quickly
screening multiple materials (i.e., OEMS) or for specifically detecting O2 (i.e., in-situ pressure cell).
In principle, one can expect this proposed method to be extended to detect other soluble
species or other type of gases such as CO2. To this end, the selection of the adequate ring electrode
will be crucial since it should selectively reduce or oxidize the target gas or species of interest (e.g.,
copper ring for CO2 detection) without jeopardizing the electrochemical/chemical stability of the
whole system. Another possible extension of this method is to modify the operation mode of RRDE
voltammetry. This approach is helpful to screen a series of products formed at the disk electrode
upon charging the oxides to certain potentials by using the ring potential as an indicator.
51
Chapter III Revisiting the structural evolutions
and electrochemical properties in the first cycle
of Li-rich NMC
Chapter III Revisiting the structural evolutions and electrochemical properties in the first cycle of Li-rich NMC
Chapter III Revisiting the structural evolutions and

electrochemical properties in the first cycle of Li-rich NMC
III. 1 Background and motivation
As mentioned in Chapter I, Li-rich layered oxides represent one of the few leading
candidates for achieving greater energy storage. Their archetypical composition
-1
Li1.2Ni0.13Mn0.54Co0.13O2 (Li-rich NMC) delivers capacities as high as ~ 300 mAh g due to the
cumulative contribution of both transition metal and oxygen redox.46,156 However, extra capacities
offered by oxygen redox come with practical drawbacks, such as voltage fade, large voltage
hysteresis, and poor rate capability. Although quite remarkable progresses have been made in
exploring the underlying science governing these Li-rich materials, we are far from reaching a
sufficient understanding for their practical applications. Thus the need persists for revisiting Li-rich
NMC with a fresh perspective on a few remaining questions that, despite intense discussions, have
not reached a consensus.
Specifically, one of the questions is: what are the mechanistic details of the staircase vs S-
shape potential evolution between the first charge and discharge? The origin of this voltage profile
change remains ill understood, but shall be deeply rooted in the irreversible structural change during
the first charge and/or the inherently different charge/discharge reaction paths. To address this
question, the feasibility of eliminating structural disorder and restoring cation ordering, thereby
recovering the staircase charge profile and voltage, by simply heating a discharged sample is of
paramount importance,112 since it highlights how the cation disordering which is triggered during
the first charge, more specifically, during the high voltage plateau at which the anionic-redox takes
place, impacts on the various aspects of the electrochemical performances of Li-rich phases. Going
deeper into this question, Yu et al. have recently showed the appearance of a core-shell structure
within particles at potentials greater than 4.4 V (during and above the high voltage plateau) using
X-ray fingerprint associated with an additional Bragger peak next to the (003) one belonging to the
main phase.157 This peak was proposed to be associated with a disordered-spinel structure, growing
at the surface of the particle, which reversibly converts back to a rocksalt one upon lithiation.
Nevertheless, doubts regarding the nature of this new structure still remain because long-range
51
techniques like X-ray diffraction (XRD) should be “blind” to surface structures of ≈ 2 - 3

nanometers.49 Indeed, Meng and co-workers suggested the formation of a second layered phase in
the high voltage region.158-160 Unfortunately, difficulties reside in unambiguously solving the
structure of this new phase because two independent phases may be present while only one extra
peak is clearly observed in the whole pattern. Moreover, neither the dynamics of this phase
conversion was studied nor any means to promote it was proposed.
The second question, which naturally arises, is: why do Li-rich NMCs display larger first-
cycle irreversibility (~ 25 %) as compared to classical insertion materials? Early works proposed
irreversible oxygen loss,46,47 the migration of transition metals into lithium vacancies,48,161 and the
formation of a surface spinel phase158 to be responsible for the first-cycle irreversible capacity. In
addition, Shunmugasundaram et al. shown that “Li-rich oxides” containing metal site vacancies in
the transition metal layer (Li[□qM(1−q)]O2, M being a combination of Ni, Mn, and Co,) exhibit very
low irreversible capacity (~ 6.5 %), as a potential consequence of enhanced atomic diffusion
provided by vacancies.162 It is most likely that the first-cycle irreversible capacity has multi-fact
origins, namely a combination of irreversible oxygen loss, oxidative electrolyte decomposition,
solid electrolyte interface formation, kinetic limitation, inactive domain, etc. Nevertheless, the
irreversible capacity caused by the sluggish lithium diffusion which might be recovered at elevated
temperature, at slower C-rates, or by applying a constant voltage step alike what has been suggested
for classical layered oxides, namely LiNi1/3Mn1/3Co1/3O2163 and LiNi0.8Mn0.1Co0.1O2,164 should be
differentiate from the capacity loss arising from irreversible processes. Only this complete
understanding can pave the way to effectively mitigate the large first-cycle irreversibility.
Therefore, we decided to revisit the electrochemical-structural relationship of

Li1.2Ni0.13Mn0.54Co0.13O2 under severe oxidative-reductive conditions. Firstly, we show the
feasibility to stabilize for the first time, the aforementioned new phase by pushing the high-voltage
oxidation, entailing the possibility of detailed structural characterization using an arsenal of
techniques like transmission electron microscopy (TEM), synchrotron XRD and neutron powder
diffraction (NPD). We also revealed the underlying origins of the new phase formation and its
impacts on electrochemical properties. Secondly, we demonstrated that a major part (~ 70 %) of the
first-cycle capacity loss is induced by kinetic limitation, and can be recovered by applying a
constant potential step at ~ 2.0 V. Thirdly, we showed that the oxygen redox activity on charge
triggers an extra low voltage electrochemical activity, which has never been reported before. In the
52
end of the chapter, we further discuss these findings by putting them into the context of a vast
number of literature reports on this topic, and providing a more holistic comprehension of the
electrochemical-driven structural evolution in Li-rich NMC materials.
III. 2 Results and discussion
III. 2. 1 Phase transition and O2 release on deep oxidation
A pure Li1.2Ni0.13Mn0.54Co0.13O2 phase, as deduced from Rietveld refinement of the

synchrotron XRD pattern (Figure III. 1 and Table III. 1), was prepared by combined co-
precipitation and solid state synthesis.43
Figure III. 1 Rietveld refinement of the synchrotron XRD pattern of the pristine
Li1.2Ni0.13Mn0.54Co0.13O2 phase. The red line, black line and bottom green line represent the
observed, calculated XRD patterns and the difference, respectively. Vertical blue tick bars mark the
Bragg positions in the R3̅m space group.
53
Table III. 1 Crystallographic data and atomic positions for the pristine
Li1.2Ni0.13Mn0.54Co0.13O2 phase as determined from Rietveld refinement of the synchrotron XRD
pattern.
Electrochemical tests were made in Swagelok cells against metallic lithium. As typically
reported for this material,43 the voltage-composition curves exhibit a two-step first-charge profile
that converts into a sloping S-shaped one on the following discharge. Afterwards, the S-shape
voltage profile is maintained, which is however tainted by capacity decay and voltage fade (Figure
III. 2).
54
Figure III. 2 (a) Potential-composition curves of Li1.2Ni0.13Mn0.54Co0.13O2 cycled within a

potential window of 4.8 - 2.0 V at a current rate of C/5. (b) Normalized discharge curves of Li-rich
NMC cycled within a potential window of 4.8 - 2.0 V at a current rate of C/5, i.e., the Li content (x)
were normalized such that the maximum Li content in each discharge is taken as unity (x = 1.01 and
0.84 are taken as unity for the first and fifteenth cycle, respectively.)
Since the first-cycle voltage profile significantly differs from those for the subsequent cycles,
suggesting that structure rearrangements mainly occur during the first electrochemical cycle,
especially during the first charge. Therefore, as an initial step in revisiting this topic, we explored
the structural evolution during the first charge under various oxidative conditions. For that purpose,
two in-situ XRD cells were charged at C/3 to 4.8 V and to 4.6 V, respectively, followed with an
open circuit voltage (OCV) step for 10 h. As shown in Figure III. 3, after charging to 4.8 V, in
accordance with previous reports,46,48,165 a shoulder peak starts to appear at the high-angle side of
the original (003) peak (denoted hereinafter phase A), indicative of the growth of a new phase
(denoted hereinafter phase A’). Interestingly, no appearance of phase A’ was observed when the
charge cutoff voltage was limited at 4.6 V. It is worth noting that the phase A → A’ transition only
appears during the OCV period when intentionally charging the cells at a relatively fast current rate
(i.e., C/3), suggesting the slow kinetics for such phase transition.
55
Figure III. 3 In-situ XRD study of the first-charge of Li1.2Ni0.13Mn0.54Co0.13O2 to 4.6 V (a)
and 4.8 V (b), followed with an OCV step for 10 h. Note that only the (003) Bragg peak is shown
because it represents the major change of the whole XRD pattern. Dash line in panel (a) suggests
that only phase A is formed after 4.6 V charge. On the contrary, a new phase A’ is grown at the
expense of phase A on further charging to 4.8 V, as indicated by the dashed arrows in panel (b).
In light of the observation that a new phase A’ is formed on charging

Li1.2Ni0.13Mn0.54Co0.13O2 to higher voltage (4.8 V), we continued to push the material to higher
oxidation state by clamping the voltage at 4.8 V for 5 h. Notably, the phase A’ grows at the expense
of the original phase A, and eventually becomes a single phase (Figure III. 4). This result clearly
indicates that the newly formed phase A’ is a bulk rather than a surface structure, in contrast with
previous literatures which interpreted it as either a disordered-spinel or a densified surface
layer.48,157
56
Figure III. 4 In-situ XRD study of the first-charge of Li1.2Ni0.13Mn0.54Co0.13O2 to 4.8 V at

C/3, followed with a constant voltage (CV) hold for 5 h. Dashed square reveals that the phase A
gradually transform into phase A’ during the voltage holding at 4.8 V.
Before understanding more about the structure of phase A’, we first considered the physical
origins of such phase A → A’ transition. In-situ and ex-situ XRD studies clearly shown that the
oxidation state of the material matters (Figures III. 3 & 4 and A. 3. 1). More specifically, the
removal of more Li+ creates more Li vacancies and further destabilizes the material’s structure,
hence facilitating the phase transition. In addition, O2 loss from the material is known to occur
during the first charge of Li-rich materials, creating oxygen vacancies and causing structure
rearrangements.44,45,49,166,167 Therefore, to explore the gas release of Li1.2Ni0.13Mn0.54Co0.13O2 under
various oxidative conditions, OEMS measurements were carried out by Dr. L. Zhang at Paul
Scherrer Institut (PSI), Switzerland. The onset of O2 release is located at ca. 4.6 V (Figures III. 5b
and A. 3. 2), where the phase transition also starts to appear. Furthermore, either an OCV step or a
CV hold was tentatively applied after charging to 4.8 V, since in-situ XRD demonstrated that within
this time period phase A → A’ transition occurs (Figures III. 3 & 4). Notably, O2 continues to
evolve after entering the OCV step and the CV hold, with its evolution rate being larger for the
latter case, ending with total amounts of 751 µmol g-1 and 383.4 µmol g-1 respectively for cells
charged with additional CV and OCV steps, as compared to 269.5 µmol g-1 for the cell directly
57
switched to discharge (Table III. 2). The additional O2 evolution during OCV, i.e., when no charge
is passed, has several possible origins: i) the delay between the O2 evolution from the material and
its detection in the mass spectrometer; ii) the kinetic hindrance related to the O2 diffusion across the
van der Waals gap; iii) structure rearrangement of the material or iv) chemical reactions (e.g.,
disproportionation) of the released oxidized oxygen species (On¯, 0 < n < 2) that are still unknown
in their chemical nature. Nevertheless, the O2 evolution quickly stops in the cell which directly
switched to discharge, demonstrating that the additional O2 release cannot be explained by the
former two possibilities. Overall, OEMS combined with in-situ XRD results show that the growth
of phase A’ solely occurs after the onset of O2 release, and is always accompanied by O2 release
during OCV and CV period, suggesting that the O2 release and the phase A → A’ transition are
strongly correlated.
58
Figure III. 5 OEMS gas analysis during the first constant current (CC) charge of
Li1.2Ni0.13Mn0.54Co0.13O2 followed by a CC discharge (black lines), an OCV step (blue lines) and a
CV hold at 4.8 V (red lines). O2 (m/z = 32) and CO2 (m/z = 44) ion currents were recorded and then
converted to gas evolution rates. (a) The voltage vs time curves (note that the curves are
superimposed up to the 4.8 V charge cutoff voltage indicating a good reproducibility of the OEMS
cells, and the dash black line illustrates partially the subsequent discharge for the sake of
comparison) and the evolved rates of O2 (b) and CO2 (c) in the units of μmol min-1g-1 with colors
quoted the same as in panel (a). The grey dash line indicates the onset of O 2 evolution. The red line
marks the 4.8 V cutoff voltage of the CC charge. The red dash arrow reveals that larger O2
evolution rate was observed in the CC - CV charged cell.
59
Regarding the CO2 evolution, it begins at ca. 4.2 V followed by a second release which
coincides with the onset of O2 evolution at ca. 4.6 V (Figure III. 5c). It is generally recognized that
the first CO2 evolution corresponds to the decomposition of surface Li2CO3 contaminants, while the
second CO2 evolution may have multiple contributions, ranging from continued Li2CO3
decomposition, reactive oxidized oxygen species attacking carbonate electrolyte, and high-voltage-
driven electrolyte decomposition, etc.111,168-170 A notable larger rate for the second CO2 evolution
was observed during the CV hold as compare to the OCV step and CC discharge, suggesting
possible electrochemically-driven oxidation of the electrolyte.
Further quantitatively exploiting the gas evolution in the CC-CV charged cell, we estimated
the total extraction of O atom from the material by 1) only considering the recorded O2 amounts,
and 2) assuming that CO2 evolved at high voltages (denoted as high V CO2) also comes from the
released O2 reacting with electrolyte, as suggested by Strehle et al.47 We could deduce a maximum
loss of ~ 0.15 O per formula unit (Table III. 2). This value, combined with the Li, Co, Ni and Mn
contents obtained by the inductively coupled plasma – optical emission spectrometry (ICP-OES)
measurements, leads to a theoretical chemical composition of Li0.05Ni0.13Mn0.54Co0.13O1.85 for the A’
phase.
Table III. 2 Estimation of the O content in charged Li1.2Ni0.13Mn0.54Co0.13O2 phases. The

evolved O2 and CO2 amounts are deduced from the same cells presented in Figure III. 5 based on
the ion currents recorded respectively for m/z = 32 and m/z = 44.
O2 amount CO2 amount y in LixTMOy y in LixTMOy (considering

(µmol g-1) (high V, µmol g-1) (considering O2) O2 + high V CO2)
CC 269.5 45.8 1.95 1.95
CC-OCV 383.4 40.7 1.93 1.93
CC-CV 751 126 1.87 1.85
III. 2. 2 Detailed structural characterization of phase A’
Next, the structure of the A’ phase was determined by combined transmission electron
microscopy (TEM), synchrotron XRD and NPD.
60
TEM data were collected by Dr. A. Abakumov at EMAT, Belgium. The electron diffraction
(ED) patterns reflect that the majority of the A’ phase crystallizes in the O3 𝑅3̅𝑚 structure (Figure
III. 6b). The [010] ED pattern demonstrates reflection splitting along the c* direction, which is the
typical signature of the mirror-twinned O3 𝑅3̅𝑚 structure with the (001) twin plane (Figures III. 6b
and 7). The nanosized twinned domains are evident in the low magnification high-angle annular
dark-field scanning transmission electron microscopy (HAADF-STEM) image (Figure III. 6b).
Besides sharp reflections of the O3 structure, diffuse spots are present in the [010] ED pattern of the
A’ phase at the positions characteristic of the O1 structure (marked with vertical arrowheads in
Figure III. 6b). In contrast to the pristine Li-rich NMC sample,43 the [110] ED pattern of the A’
phase shows only very faint diffuse intensity lines from the “honeycomb” Li-TM ordering (marked
with horizontal arrowheads in Figure III. 6b), indicating that this ordering is largely suppressed.
This justifies the validity of the simple O3 𝑅3̅𝑚 structure model used for further Rietveld
refinement of the A’ structure.
61
Figure III. 6 Electron diffraction (ED) patterns and low magnification HAADF-STEM
images of Li-rich NMC charged to 4.6 V (a, the A phase), 4.8 V (b, with 8 h constant voltage
holding to form the pure A’ phase) and discharged to 2.0 V (c). The ED patterns are indexed with
the O3 𝑅3̅𝑚 structure. Note the reflection splitting in the h0l, h  3n reciprocal lattice rows in all
[010] ED patterns which is caused by mirror twinning of the O3 𝑅3̅𝑚 structure with the (001) twin
plane (Figure III. 7). The primitive reciprocal unit cells of two twinned variants are outlined in red
and green in the [010] ED patterns. Twinned domains with few tens of nanometers in thickness
along the c-axis are clearly visible in low magnification HAADF-STEM images for all three states
of charge. Extra spots elongated along the c* axis appear between the pairs of bright reflections of
the O3 structure in the [010] ED pattern of the A’ phase (b, marked with vertical arrowheads).
These reflections originate from thin domains of the O1 structure, as evidenced with the HAADF-
62
STEM images in Figure III. 8. [110] ED pattern of the A phase (a) reveals faint diffuse intensity
lines (marked with horizontal arrowheads) from residual short-range “honeycomb” Li-M ordering.
These diffuse intensity lines become almost invisible in the [110] ED pattern of the A’ phase (b)
indicating significant suppression of the “honeycomb” Li-M ordering, but regain their brightness
after discharge to 2.0 V (c) due to restoration of this ordering. Weak spots between diffuse intensity
lines in the [110] ED patterns of Li-rich NMC charged to 4.6 V (a) and 4.8 V (b) come from the
near-surface M cation ordering analyzed in Figure III. 9.
Figure III. 7 Tentative model of twins in the O3 𝑅3̅𝑚 structure. Two twinned domains
viewed along [010] possess mirror-related orientation of the c-axis. At the twin plane the “cubic”
ccp close packing of the O atoms is interrupted by a single “hexagonal” h layer (marked with an
arrow). The M cations, Li cations and O atoms are shown as cyan, yellow and red spheres,
respectively. The MO6 octahedra are shaded.
63

Figure III. 8 Local structure of the A’ phase viewed with [010] HAADF-STEM images. a,
The near-surface O3 structure: the M layers are horizontally being traced with the rows of bright
dots of the M cationic columns. The “cubic” stacking sequence is evidenced with cumulative lateral
shift of the Li1−xM2+x layers (marked with a sloped line). The migration of the M cations to the
octahedral interlayer sites is visible as a nucleation of the {102}-oriented TM-cation layer at ~ 60o
to the {001}-oriented Li1−xM2+x layers (marked with arrowheads). Thin lamella of the O1 phase is
marked with a bracket. b, The near-surface structure of an extended O1 domain: the “hexagonal”
stacking sequence is evidenced by absence of the lateral displacements of the Li1−xM2+x layers
(marked with a vertical line). The migration of the M cations to the interlayer octahedral sites is
very pronounced seen by appearance of extra rows of dots between the Li1−xM2+x layers (look
brighter).

64

Figure III. 9 Near-surface locally ordered states in the A’ phase. a, b, HAADF-STEM
images of the ordering of M cations in the tetrahedral interstices between the M layers with
pertaining “honeycomb” M cation ordering (the scale bar is of 1 nm). c, d, Projected structures
corresponding to the cationic positions in the HAADF-STEM images a and b, respectively. The M
cations, Li cations and O atoms are shown as cyan, yellow and red spheres. e, The projected
structures, in fact, correspond to the same 3D structural arrangements viewed at different directions
rotated by 60. Note that the tetrahedral interstices are linked by common edge and cannot be
occupied simultaneously.
Rietveld refinements of synchrotron X-ray and neutron diffraction patterns were performed
by Dr. G. Rousse at Sorbonne Université, France. In agreement with the ED data, the synchrotron
XRD pattern can be indexed in an O3 𝑅3̅𝑚 structure with lattice parameters of a = 2.84194(11) Å,
c = 13.9722(14) Å (Table III. 3), As the distribution of five species with different scattering factors
(Li, Ni, Mn, Co and cation vacancy) among two crystallographic positions of the 𝑅3̅𝑚 structure
cannot be distinguished uniquely, combined neutron/synchrotron Rietveld refinement was carried
out with the oxygen composition and Li content fixed to Li0.05MO1.85 formula (as deduced from
ICP-OES for Li content and OEMS for O content), and assuming that only one of the three
transition metal (TM) cations may migrate. Indeed, neutron diffraction provides a unique contrast
between Mn (bMn = - 3.73 fm), Ni (bNi = 10.3 fm) and Co (bCo = 2.49 fm). The total occupancy of
each TM cation was imposed to the one expected from the pristine chemical formula. The best fit,
shown in Figure III. 10 with corresponding crystallographic parameters reported in Table III. 3, was
65
obtained by populating the vacant Li 3a sites with ~ 0.079(2) Mn, hence a chemical cationic
composition [Li0Mn0.08][Li0.05Ni0.13Mn0.46Co0.13]O1.85, where square brackets refer to atoms in the
octahedral 3a and 3b sites of the 𝑅3̅𝑚 structure, respectively. The migration of TM into octahedral
lithium site was also supported by difference-Fourier mapping (Figure A. 3. 3). Note that the Mn
migration to the vacant octahedral sites in the Li layers has already been experimentally observed
and theoretically proposed.171-174 However, one should note that there are two possible
interpretations of such a refinement. The first one corresponds to the above-written chemical
formula and involves the presence of oxygen vacancies and an under-coordination of metals. The
second interpretation is to rewrite the chemical formula as [Li0Mn0.08][Li0.05Ni0.14Mn0.50Co0.14]O2.
This corresponds to an O3 structure without any oxygen vacancies but “densified”, because the
TM/O ratio has increased compared to the pristine Li1.2Ni0.13Mn0.54Co0.13O2 (0.43 for A’ phase; 0.40
for pristine phase). The second interpretation is in full agreement with the fact that the unit cell
volume of the A’ phase is smaller compared to the pristine phase (97.730(11) Å3 for A’ phase;
100.545(2) Å3 for pristine phase), because of the shortening of the c lattice parameter (from
14.25381(15) Å for the pristine phase to 13.9722(14) Å for the A’ phase). These results clearly
indicate that “densification” is not a surface but rather a bulk effect. Lastly, it is worth mentioning
that the c/a aspect ratio in the A’ phase is very close to (but different from) that of an ideal cubic
lattice (c/a = 4√3/√2), which can explain why this phase may have been previously misinterpreted
as a spinel cubic phase in literatures.
66
Figure III. 10 Rietveld refinement of synchrotron (a) and neutron (b) powder diffraction for
A’ phase.
Table III. 3 Crystallographic data and atomic positions of the phase A’ determined from
Rietveld refinements of the combined synchrotron and neutron powder diffraction patterns.
To enforce this structural analysis, we further performed atomic-resolution HAADF-STEM

and annular bright-field scanning transmission electron microscopy (ABF-STEM) imaging. The
67
[010] HAADF-STEM image (Figure III. 8a) reflect the prevailing O3-type stacking of the TM
layers, albeit with thin domains of the O1-type stacking spanning over just few unit cells. More
extended O1 domains are also observed, as evidenced in Figure III. 8b. In the O3 structure, the
layered ordering of the TM cations is violated by the TM cation migration to the vacant Li sites
appearing as a nucleation of the {102}-oriented TM-cation layer at ~ 60o to the {001}-oriented
Li1−xTM2+x layers. The O1 domains demonstrate more pronounced occupation of the octahedral
interstices by the TM cations. The most remarkable picture of the structural inhomogeneity of the A’
phase related to TM migration is visualized with the [110] HAADF-STEM image (Figure III. 11a).
First inhomogeneity is related to the “honeycomb” Li-M ordering which pertains only in part of the
Li1−xM2+x layers whereas the layers with suppressed “honeycomb” ordering are frequently present
(both variants are marked with arrowheads in Figure III. 11a). Significant M cation migration
occurs over whole > 15 nm distance from the surface demonstrating a variety of local ordering,
where the M cations occupy either octahedral Li sites or tetrahedral interstices (illustrated with
inserts in Figure III. 11a). The enlarged images of this local ordering are provided in Figures III. 9
and A. 3. 4, along with tentative distribution of cations among the tetrahedral and octahedral sites.
These locally ordered sequences, which also manifest themselves in reciprocal space as very faint
reflections between the diffuse intensity lines in the [110] ED pattern (Figure III. 6b), can be
considered as a prerequisite to the formation of the spinel-type structure. Hence we conclude that
the A’ phase preferentially adopts the O3-type structure with M cation partially migrating to the
empty octahedral or tetrahedral sites at the surface as well as in the interior parts of the crystallites
causing bulk densification. This migration is associated with the formation of the A’ phase and it is
substantially reversible as comparatively illustrated with the [110] HAADF-STEM images of Li-
rich NMC charged to 4.6 V (the A phase), 4.8 V (the A’ phase) and discharged to 2.0 V (Figure III.
11b, 11c & 11d, respectively).
Overall, the TEM observations corroborate the Rietveld refinement results. Hence we
conclude that the A’ phase preferentially adopts the O3-type structure with TM cation partially
migrating to the empty octahedral sites in the interior parts of the crystallites causing bulk
densification, whereas the “honeycomb” Li-TM ordering maintains in the near-surface region along
with TM cations migrate into both octahedral and tetrahedral interlayer site (Figure III. 12).
68

Figure III. 11 Local structures of Li-rich NMCs viewed with HAADF-STEM images. a.,
Overview [110] HAADF-STEM image showing structural inhomogeneity of the A’ phase. The
Li1−xM2+x layers are visible as rows of brighter dots perpendicular to the c-axis. “Honeycomb” Li-M
ordering is visualized by pairs of bright dots with ~ 0.14 nm interdot separation (such as marked
with pairs of small black arrowheads). The Li1−xM2+x layers with pertaining “honeycomb” ordering
(such as marked with large white arrowheads) alternate with the layers where this ordering is
largely suppressed (such as marked with large black arrowheads). The interlayer spaces
demonstrate pronounced HAADF intensity indicating migration of the M cations to either
octahedral sites Mo (see insert 1) or to the tetrahedral interstices Mt (see insert 2, see Figure III. 9
for atomic position assignment). This migration occurs at the near-surface region as well at > 15 nm
away from the surface. Near-surface [110] HAADF-STEM images of Li-rich NMC charged to 4.6
V (b, the A phase), 4.8 V (c, the A’ phase) and discharged to 2.0 V (d). The M cation migration sets
in at very thin 1-2 nm surface layer in the A phase, then it becomes very pronounced in the A’
phase and finally gets almost fully suppressed after 2.0 V discharge. The “honeycomb” Li-M
ordering is largely restored in the discharged phase as demonstrated by the pattern of the bright dot
69
pairs, but some disorder still remains indicated by residual intensity between the dot pairs (marked
with black arrowheads in c).
Figure III. 12 Schematic of the bulk and surface structure of the newly formed A’ phase
along with the illustrative atomic-resolution HAADF-STEM images.
III. 2. 3 The impact of phase transition on the electrochemical properties
Next, we continued to explore how the phase transition impacts the electrochemical
behavior for this Li-rich phase. As previously demonstrated in the subchapter III. 2. 1, following the
4.8 V charge, phase A → A’ transition starts to occur after entering the OCV period, and gradually
becomes a single phase A’ with CV hold. Therefore, we tentatively cycled three identical
Li1.2Ni0.13Mn0.54Co0.13O2/Li cells between 4.8 V and 2.0 V, but with different first charge protocols:
one cell was directly switched to discharge (denoted as CC, black curve) with phase A forming at
the end of the first charge, whereas either an intermediate OCV step for 120 h (blue line, denoted as
CC - OCV) or CV hold for 8 h (red line, denoted as CC - CV) was imposed to the other two cells to
trigger the formation of phase A’. Figure III. 13 compare the first charge-discharge voltage curves,
and Figure III. 14 compare the retention for the average discharge voltage and the discharge
capacity upon cycling. Notably, phase A’ exhibits a larger first-cycle irreversible capacity and an
70
initial voltage drop as compare to phase A (Figure III. 13), whilst upon cycling, the persistence of
the voltage decay and capacity fading (Figure III. 14).
From an electrochemical perspective, the initial voltage drop can be explained by the
reduction of transition metal ions (Co and Ni) during OCV or CV step, as revealed by ex-situ X-ray
absorption spectroscopy (XAS) measurements (Figure III. 15). The observed capacity drop after
imposing the OCV or CV step is closely associated with the gradual phase A → A’ transition. More
specifically, the contraction of TM layers and the growing Mn occupation in the octahedral Li sites
upon A → A’ phase transition certainly would cause a kinetic hindrance in Li+ intercalation. In
addition, the formation of phase A’ is accompanied with O2 release, which are known to induce
particle cracking that could also contribute to the performance deterioration.
Figure III. 13 The first charge-discharge voltage curves of Li1.2Ni0.13Mn0.54Co0.13O2 cycled

with different protocols. One cell was charged to 4.8 V and then directly discharged to 2.0 V,
denoted as CC (black line), whereas an intermediate OCV step for 120 h (blue line, denoted as CC -
OCV) or CV hold for 8 h (red line, denoted as CC - CV) was respectively imposed to the other two
cells. All the cells were cycled at C/3.
71
Figure III. 14 Retention profiles of the average discharge voltage (a) and the discharge
capacity (b) of Li1.2Ni0.13Mn0.54Co0.13O2 cycled between 4.8 and 2.0 V with different first-charge
protocol: CC (black line), CC - OCV (blue line) and CC - CV (red line). Error bars represent
standard deviation of two cells.
72
a 2.0
4.8 V CC Ni K-edge
Normalized absorption
4.8 V CC-OCV
4.8 V CC-CV
1.5 4.8 V charged
LiNi0.33Mn0.33Co0.33O2
1.0
0.5
0.0
8330 8340 8350 8360
b 1.5
Energy (eV)
4.8 V CC Co K-edge
4.8 V CC-OCV
4.8 V CC-CV
4.8 V charged
1.0 LiNi0.33Mn0.33Co0.33O2
0.5
0.0
7710 7720 7730 7740
c 2.0
Energy (eV)
4.8 V CC Mn K-edge
4.8 V CC-CV
4.8 V CC-OCV
1.5 4.8 V charged
LiNi0.33Mn0.33Co0.33O2
1.0
0.5
0.0
6540 6550 6560 6570
Energy (eV)
Figure III. 15 Normalized XANES spectra at the (a) Ni K-edge, (b) Co K-edge and (c) Mn
K-edge for Li-rich NMC samples which were charged to 4.8 V (black line), and charged to 4.8 V
with an additional OCV step (grey line) or CV hold (blue line) for 5 h. The spectra for
LiNi0.33Mn0.33Co0.33O2 collected at 4.8 V charge state are given as references.
73
III. 2. 4 Mn migration triggered by oxygen redox and its consequence of extra

low-voltage electrochemical activities
As previously discussed, phase A’ forming under harsh oxidizing condition enlists Mn

migration, we further investigated the influence of depth of charge on the Mn migration. For that,
we prepared several Li1.2Ni0.13Mn0.54Co0.13O2/Li cells that were electrochemically oxidized till ∆x =
0.4, 0.6, 0.8, and 1.0 followed by 120 h OCV (∆x refers to the amount of Li+ removal on charge or
the amount of Li+ insertion on discharge), prior to sampling the electrodes for ex-situ synchrotron
X-ray diffraction measurements. By employing the same Rietveld methodology as before, we could
deduce that while the ∆x = 0.4 charged sample shows no Mn migration in interlayer octahedral sites
(defined as in [Li1.0-∆xMn][Li0.2Ni0.13Mn0.54-Co0.13]O2), a progressive increase of Mn migration ()
was observed with increasing ∆x from 0.6 up to 1.0 (Figure III. 16a). Interestingly, further
maintaining the ∆x = 1.0 charged sample at 4.8 V for 8 hours to obtain the A’ phase did not change
the amount of Mn migration () within the accuracy of this method, but slightly decreased the c
parameter from 14.2058(4) Å to 13.9722(14) Å (Tables III. 3 & A. 3. 1). This is consistent with the
decrease in Li+ content from 0.02 (forx = 1.0 charged sample) to 0.005 (for CC - 8 h CV charged
sample) according to ICP-OES, which leads to less charged oxygen ions thereby weaker oxygen-
oxygen repulsion. In parallel, X-ray photoemission spectroscopy (XPS) measurements were carried
out for the various charged samples to determine the fraction of oxygen involved in the anionic
redox process (Figure III. 16c). To no surprise, this fraction increases with increasing ∆x alike the
degree of Mn migration (), hence, implying their correlation. In contrast, note that neither Mn
migration (or the percentage of oxidized oxygen correlated with O2 release (Figure III. 16).
74
0.08 (a)
Li site ()
Mn in
0.04
0.00
750 (b)
O2 amount
(mol g-1)
500
250
% of oxidized 0
40.0 (c)
oxygen
20.0
0.0
0.0 0.4 0.6 0.8 1.0 1.15
(4.8 V CV)
Li+ removal on charge (x)
Figure III. 16 (a) Mn migration in interlayer octahedral Li sites () deduced from Rietveld
refinement of synchrotron XRD patterns collected for the various charged samples with the amount
of Li+ removal (∆x) ranging from 0.0 (pristine), 0.4, 0.6, 0.8, 1.0 to 1.15 (with 4.8 V CV hold for 8
h). (b) The amount of released O2 quantified by OEMS, and (c) the percentage of oxidized oxygen
(define as On¯/ (On¯+ O2¯) quantified from XPS are plotted as a function of ∆x.
Besides exploring the harsh oxidizing conditions, we next investigated the effect of harsh
reducing conditions on the electrochemical signatures of Li-rich NMC. At first, we fixed the
discharge cut-off potential to 2.0 V and monitored the effect of state of charge on the subsequent
discharge profile by opening the charge potential window stepwise (i.e., increasing x from 0.4 to
1.0 in steps of 0.2, the same as in the ex-situ synchrotron XRD study). We observed that the S-
shape discharge profiles become more pronounced, when the charge voltage reaches the high-
voltage plateau (Figure III. 17). Simultaneously, the degree of hysteresis and the amplitude of
irreversible capacity were found to increase with deeper oxidation on charge (inset in Figure III.
17).
75
Figure III. 17 Voltage-composition profiles of Li1.2Ni0.13Mn0.54Co0.13O2 obtained by varying

+
the Li removal on charge (∆x) from 0.4 (purple line), 0.6 (orange line), 0.8 (dark yellow line), to
1.0 (red line), while fixing the discharge cutoff voltage at 2.0 V. The inset shows the corresponding
irreversible capacity observed in discharge.
Nevertheless, when the cells were further discharged to 1.2 V, totally unexpected is the
gradual appearance of extra redox activities located within 1.5 ~ 1.2 V, whose capacity increases
with increasing ∆x on charge (Figures III. 18a and 18b). Strikingly, these extra capacities are not
present when the cell was directly started on reduction (∆x = 0.0) or when the charge was limited to
the cationic domain (∆x = 0.4). This suggests a straightforward correlation between the extra
capacities and the extent of the anionic redox process. A possible explanation could be rooted in the
Mn-migration () which is triggered during the charge once the anionic activity becomes part of the
oxidation process. To check this hypothesis, we revisited the Rietveld refinements of the
synchrotron XRD patterns collected for the various charged samples with ∆x = 0.4, 0.6, 0.8, and 1.0
(Figure A. 3. 7 and Tables A. 3. 1 ~ A. 3. 4), using the R3m space group and the same methodology
as before. The refinements indicate Mn-migration (), appears only when charging to the anionic
redox region, and progressively builds up with increasing state of charge. Moreover, no Mn-
migration is observed when the charge is limited to the cationic redox (∆x = 0.4), akin to the
pristine phase (∆x = 0.0) (Figure III. 16a). Rietveld refinement of the synchrotron XRD pattern
collected at 2.0 V discharged state after charging up to ∆x = 1.0 shows that such Mn-migration can
only be partially restored upon discharging to 2.0 V, since Mn migration () of ~ 0.0349(4) still
76
remains for a sample that was charged to 4.8 V and then discharged to 2.0 V (Figure A. 3. 8 and
Table A. 3. 5). From ED patterns and HAADF-STEM analysis of this discharged phase, we could
deduce that the microstructure of Li1.2Ni0.13Mn0.54Co0.13O2 is largely restored (Figure III. 6c). Hence,
such a discharged phase, which slightly deviates from the pristine one, is denoted as P’ hereinafter.
Altogether, these results suggest a robust correlation between the appearance of low-voltage redox
activity and the Mn-migration induced by anionic redox, which is to some extent irreversible.
77
Figure III. 18 (a) Voltage-composition profiles of Li1.2Ni0.13Mn0.54Co0.13O2 obtained by

varying ∆x on charge from 0.0 (pristine, black line), 0.4 (purple line), 0.6 (orange line), 0.8 (dark
yellow line), to 1.0 (red line) whilst fixing the discharge cutoff voltage at 1.2 V. The dashed grey
square indicates the gradual appearance of extra low-voltage (2.0 – 1.2 V) electrochemical activities
on pushing the extent of charge to trigger anionic redox. (b) A magnified view of the voltage-
composition profiles in low-voltage region (also marked with dashed grey square in panel a). The
red arrow indicates the voltage overshooting point in the discharge process after charging up to ∆x
= 1.0. The inset shows the amount of low-voltage discharge capacities as a function of ∆x. Note that
the colors in panel (b) are quoted the same as in panel (a).
Further scrutinizing the low-potential capacity region (below 2.0 V), enlists two different
domains separated by a voltage overshooting (marked by a red arrow in Figure III. 18b) which is
78
usually reminiscent of the nucleation-growth process of a new phase. To confirm this point, a
Li1.2Ni0.13Mn0.54Co0.13O2/Li in-situ XRD cell was charged to 4.8 V and then discharged to 3.0 V,
prior to collecting XRD patterns, while the discharge was pushed down to 1.2 V (Figure III. 19).
We observed a continuous initial shift of the P’ phase reflections (i.e., (003) reflection shifts toward
higher angles, whereas (111) and (131) reflections shift toward lower angles) with some decreasing
in intensity until near the voltage overshoot (~ 1.4 V), indicating a solid solution transformation
process. Then upon further increasing in x, some reflections disappear at the expense of new ones
that remain broad until becoming a single reflection, indicative of a new phase (denoted hereinafter
as P”), when the voltage reaches 1.2 V.
Figure III. 19 Structural evolutions during the low-voltage discharge process within 3.0 V –
1.2 V. (a) Selective galvanostaic discharge curve (left side), and selective regions ((003), (111) and
(131) reflections) of the corresponding in-situ XRD patterns (right side) collected upon discharging
Li1.2Ni0.13Mn0.54Co0.13O2 from 3.0 V down to 1.2 V, after charging to 4.8 V and discharging to 3.0 V
(see results for the complete discharge in Figure A. 3. 10). Note that the dashed blue arrow marks
the evolution of phase P’ and the dashed red arrow indicates the growth of phase P”, respectively.
79
Figure III. 20 Rietveld refinements of the synchrotron XRD patterns for the samples
collected after 1.4 V (phase P’) and 1.2 V (phase P”) discharge. The red line, black line and bottom
green line represent the observed, calculated XRD patterns and the difference, respectively. Vertical
blue tick bars mark the Bragg positions in the R3̅m space group.
Owing to the width of these reflections, we have collected high-resolution synchrotron XRD
patterns on powders recovered from cells that have been discharged respectively to 1.4 V (phase P’)
and 1.2 V (phase P”) (Figure III. 20). Rietveld refinements indicate that both phases remain in the
R3m space group with smaller lattice parameters for the P’ phase (a = 2.87386(2) Å, c = 14.3442(2)
Å, V = 102.598(2) Å3, Table A. 3. 6) as opposed to the larger values for the P” phase (a =
2.90026(5) Å, c = 14.3783(5) Å, V = 104.739(4) Å3, Table A. 3. 7). The amount of Mn-migration ()
was determined to be 0.057(2) for the P’ phase and 0.085(2) for the P” phase, respectively (Figure
III. 21). The increase in lattice parameters is consistent with the uptake of Li+ upon reduction to 1.2
V (Li1.14Ni0.13Mn0.54Co0.13O1.95 for the phase P’ and Li1.44Ni0.13Mn0.54Co0.13O1.95 for the phase P”, as
deduced by ICP-OES for Li content and OEMS for O content), but not clarifying the location of
these extra Li+ within the structure. Previous studies have shown that LiNiO2 upon reduction can
uptake one more Li+ leading to the Li2NiO2 structure, having all the Li+ in the tetrahedral sites175,176.
Therefore, we can postulate, at this stage, that our 1.2 V discharged sample (P” phase) could
contain some Li+ in the tetrahedral sites. This hypothesis is also consistent with the fact that all the
octahedral sites are fully occupied in the Li1.2Ni0.13Mn0.54Co0.13O2 phase, in another words, only the
80
tetrahedral sites are available to accommodate additional Li+ when the lithium content exceeds that
in the pristine phase. However, Li2NiO2 has a 1T structure (space group P3m1) with ABAB oxygen
stacking sequence, whereas P” phase shares the same O3 structure, i.e., the same ABCABC oxygen
stacking sequence, as the pristine Li1.2Ni0.13Mn0.54Co0.13O2. Therefore, another puzzle concerns why
the overlithiation of P’ phase to P” phase does not induce an O3 to 1T phase transition. Since
considerable amount of Mn (~ 0.057(2)) migrate to the interlayer octahedral sites in the phase P’,
we believe that these migrated Mn ions serve as “pillar” to stabilize the structure, preventing the
rearrangement of oxygen sublattice. This argument is also in good agreement with previous report,
showing that Li1.2Cr0.4Mn0.4O2 phase with Cr migration in the interlayer octahedral sites could retain
its original ccp oxygen stacking during overlithiation.177,178
Figure III. 21 Evolution of Mn migration () in the interlayer octahedral Li sites during the
first charge-discharge of Li1.2Ni0.13Mn0.54Co0.13O2.
To further verify our proposed mechanism that the Mn-migration associated with the
oxygen redox is responsible for the uptake of additional Li+ at low-voltage, we carried out a heat
treatment study of phase P’. This study is inspired by the previous report from A. Singer et al.,
showing that annealing the cycled Li-rich oxide material at its discharged state can recover the
voltage, and restore the superstructure peak which is reminiscent of the “honeycomb” Li-TM
ordering within the LiTM2 layers.112 One Li1.2Ni0.13Mn0.54Co0.13O2/Li cell was charged to 4.6 V and
then discharged to 2.0 V, prior to collect, and anneal the sample at the discharged state under Ar
atmosphere at 250 °C for 1 h. Indeed, the superstructure has been partially recovered after the
81
annealing, in agreement with the previous literature (Figure III. 22). More importantly, Rietveld
refinements of the synchrotron XRD patterns reveal that the degree of Mn migration () has been
decreased from 0.038(2) to nearly zero (0.007(2)) after annealing (Figure III. 23 and Tables A. 3. 8
& A. 3. 9).
Figure III. 22 Comparison of the superstructure peak in the pristine (marine line), the sample
collected at 2.0 V discharged state after 4.6 V charge before (denoted as “2.0 V dis”, black line) and
after annealing (denoted as “annealed 2.0 V dis”, red line).
Figure III. 23 Rietveld refinements of the synchrotron XRD patterns of the sample collected
at 2.0 V discharged state after 4.6 V charge before and after annealing. The red line, black line and
bottom green line represent the observed, calculated XRD patterns and the difference, respectively.
Vertical blue tick bars mark the Bragg positions in the R3̅m space group.
82
Further, coin-type cells were assembled with the two 2.0 V discharged samples, before and
after annealing, to evaluate how the recovery of Mn-migration impacts on the electrochemical
properties (Figure III. 24). Galvanostatic tests started directly on charge reveal that the charge
profile was restored to its original stair-case shape and its voltage was recovered after the annealing
(Figures III. 24 & A. 3. 12), in good agreement with the previous literature.112 Note that the stair-
case charge profile converts back to an S-shaped one during the subsequent discharge, as one can
expect that the Mn-migration would occur again when the charge reaching the oxygen redox regime
(Figure A. 3. 13). More importantly, the cells started directly on discharge show that the amount of
Li+ (∆x) inserted during the low-voltage discharge process was drastically decreased, by the
annealing, from ~ 0.4 Li+ to ~ 0.07 Li+. Strikingly, both the degree of Mn-migration () and the
low-voltage discharge capacity decrease in a similar manner after annealing, unambiguously
confirming that the extra lithiation at low voltages is triggered by the local cation disordering
associated with Mn migration.
Figure III. 24 Voltage-composition profiles of the 2.0 V discharged sample before (black
lines) and after (red lines) annealing. The cells were either directly started on charge to 4.6 V, or on
discharge to 1.2 V, as indicated by the grey arrows. The inset shows the amount of Mn-migration ()
in the 2.0 V discharged sample before (black column) and after (red column) annealing.
Having identified the source of this extra capacity, it pertains now to gain better insights into
the origin of the large voltage drop (occurs at ca. x = 1.0) needed to insert the additional Li+. The
83
possible electronic limitations across the conductive matrix in the electrode was excluded, since
increasing the electrode carbon content from 20 wt.% to 40 wt.% did not alter the amplitude of
voltage drop (Figure A. 3. 14). Galvanostatic intermittent titration technique (GITT) measurements,
which combine current pulses and open-circuit sequences, were further performed to study the
thermodynamic and kinetic features of the low-voltage discharge process (Figure III. 25). A large
polarization, concomitant with the increase of Ohmic resistance, occurs before the overshoot (marks
with black arrow in Figure III. 25), representing a slow kinetic process. On the contrary, a markedly
lower open circuit voltage and small polarization was observed after the overshoot, implying a
different reaction. The GITT results are in accordance with the previous in-situ XRD study, which
demonstrating that the low-voltage discharge process involves two distinct regions separated by the
voltage overshooting: first a solid-solution with a continuous evolution of the lattice parameters,
followed by the growth of a new phase P” at the expense of the original phase P’. In the P” phase,
which possesses higher Li content than that in the pristine material, Li+ can be inserted into the
tetrahedral sites in order to accommodate additional Li+. This shift of Li+ insertion site from
octahedral to tetrahedral sites can accounts for the lowering of Li+ insertion voltage.
Figure III. 25 Galvanostatic intermittent titration technique (GITT, every ∆x = 0.05) test of
Li1.2Ni0.13Mn0.54Co0.13O2 during the first discharge to 1.2 V after 4.8 V charge. The cell was cycled
at C/5 rate, and the relaxing process was controlled by either dV/dt 0.01 mV s-1 or time = 4 h. The
Ohmic resistance was deduced from the voltage jump in a short time (1 s) after the application of
the current pulse. The black arrow marks the voltage overshooting point.
84
The amplitude of the voltage drop can be reduced by either raising the operation temperature
or decreasing the discharge current rate (Figure III. 26a), providing further evidence of the kinetic
controlling voltage drop. Given that the part of low-voltage capacity before the overshoot is of
kinetic origin, it should be able to be recovered by an additional constant voltage (CV) step. Indeed,
by applying the CV hold at 2.0 V over 40 h until the current decays to nearly zero (~ 6 μA), we
could succeed in intercalating more Li+ until the Li content reaches ~ 1.14, the same value at which
the voltage overshoot appears (Figure III. 26b). As the possible limitation of electron transport has
already been excluded, and no phase transformation was observed by in-situ XRD before the
overshoot, we can conclude that the voltage drop to be mainly caused by the sluggish Li+ diffusion.
Altogether, these results suggest that the part of low-voltage capacity before the overshoot is due to
kinetic hindrance of the lithiation process, whereas the part after the overshoot is predominately of
thermodynamic origin.
Figure III. 26 Voltage-composition profiles of Li1.2Ni0.13Mn0.54Co0.13O2 in the low-voltage

discharge region under various conditions, followed by the first charge to 4.8 V. (a) The cells were
discharged with a CC protocol at various temperatures of 25 °C (black line), 55 °C (blue line) and
75°C (green line). The inset shows the cells discharged at current rates of C/5 (black line) and
C/500 (marine blue line). (b) The cells were discharged with either a constant current (CC, black
line) protocol, or a constant current-constant voltage (CC-CV) protocol with the CV hold applied at
2.0 V (red line) over 40 h until the current decays to 6 μA.
85
The part of low-voltage capacity before the overshoot can also be considered in a cell that is
cycled under normal condition, i.e., discharged to 2.0 V rather than to a voltage as low as 1.2 V. By
applying an additional CV hold at 2.0 V, ~ 0.12 Li+ could be recovered from the initial apparent Li+
loss of ~ 0.17 Li+, suggesting that a major part (ca. 70 %) of the first-cycle irreversible capacity
observed in Li1.2Ni0.13Mn0.54Co0.13O2 is due to kinetic inhibition and the ‘real’ (unrecoverable)
irreversible capacity accounts for only ~ 0.05 Li+ (Figure III. 27).
Figure III. 27 The first charge-discharge curves of Li1.2Ni0.13Mn0.54Co0.13O2 cycled between

4.8 V and 2.0 V with an additional CV step at 2.0 V over 40 h until the current decays to 6 μA.
During the CV step, ~ 0.12 Li+ could be recovered from the initial apparent Li+ loss of ~ 0.17 Li+.
Next, to examine the change in the oxidation states of Ni, Co, and Mn during the low-
voltage discharge process, ex-situ XAS measurements were carried out on the Rock beamline of
SOLEIL synchrotron, France. The X-ray absorption near edge structure (XANES) spectra was
normalized by Dr. A. Iadecola. Figure III. 28 compares the XANES spectra of the samples collected
after 2.0 V, 1.4 V and 1.2 V discharge at the three transition-metal (Ni, Co, and Mn) K-edge. Ni K-
edge XANES spectra of the three samples show almost no change (Figure III. 28a). In contrast, Mn
and Co K-edge XANES spectra for the 1.4 V and 1.2 V discharged sample were observed at lower
energies as compared to those for the 2.0 V discharged sample, indicating the concurrent reduction
of Mn and Co (Figures III. 28b & 28c). These results unambiguously reveal that the extra lithiation
at the low-voltage region is mainly charge compensated by cationic (Co and Mn) redox.
86
Figure III. 28 Normalized XANES spectra at the (a) Ni K-edge, (b) Co K-edge and (c) Mn
K-edge for the samples collected after 2.0 V (blue line), 1.4 V (cyan line) and 1.2 V (red line)
discharge. The spectra for Li1.2Ni0.13Mn0.54Co0.13O2, Co3O4, and Mn2O3 are given as references.
87
Lastly, the electrochemical performance of Li1.2Ni0.13Mn0.54Co0.13O2 involving the low-

voltage discharge process was evaluated by charging several cells identically to 4.8 V, but varying
their discharge cutoff voltage from 2.0 V to 1.4 V and to 1.2 V. As shown in Figure III. 29, ~ 1.3
Li+ (403 mAh g-1) can be reversibly uptaken and released upon discharging to 1.2 V, as compared
to ~ 1.0 Li+ (322 mAh g-1) for 1.4 V discharge, and ~ 0.86 Li+ (272 mAh g-1) for 2.0 V discharge.
Though most of the extra capacity is delivered at low potentials, the 48 % (for 1.2 V discharge) and
18 % (for 1.4 V discharge) gain in the discharge capacity represents 18 % and 7 % increase in the
specific energy. Nevertheless, a faster capacity fading was observed during the subsequent cycles
when discharging the cell to 1.2 V (Figure III. 30a). After 25 cycles, the discharge capacity
decreases to ~ 281 mAh g-1 for the 1.2 V discharged cell, as compared to ~ 260 mAh g-1 for the 2.0
V discharged cell, leading to a similar value of specific energy (Figure III. 30c). Overall these
results suggest that, application-wise, discharging to voltage as low as 1.2 V is not an effective
strategy to improve the energy density of Li1.2Ni0.13Mn0.54Co0.13O2 over the long term.
Figure III. 29 The first (solid lines) and second (dash lines) charge-discharge curves of
Li1.2Ni0.13Mn0.54Co0.13O2 charged to 4.8 V, while varying the discharge cutoff voltage from 2.0 V
(black lines), 1.4 V (blue lines) and 1.2 V (red lines).
88
Figure III. 30 Retention profiles of the discharge capacity (a), the average discharge voltage
(b), and the discharge energy density (c) of Li1.2Ni0.13Mn0.54Co0.13O2 discharged to 2.0 V (black
symbols) and 1.2 V (red symbols).
III. 3 Discussion and chapter conclusion
We have enlarged the cycling potential window for Li1.2Ni0.13Mn0.54Co0.13O2 electrode to

both high and low potentials, and unraveled novel structural evolution mechanisms that we discuss
next by putting them into the context of numerous literature reports on this topic, to provide a more
holistic comprehension of the electrochemically-driven structural evolution in Li-rich NMC
materials.
Firstly, by applying a constant voltage step at 4.8 V, we could, for the first time, obtain a
single fully charged phase A’, thereby enabling its detailed structural characterization. Combing
synchrotron X-ray and neutron diffraction, TEM, OEMS and ICP-OES, we have identified phase A’
as a densified O3 bulk structure enlisting Mn migration into the interlayer octahedral sites, in
contrast to previous reports in stating that it is a densified layer48,179 or spinel-like structure157 which
grows only at the surface. Moreover, we have revealed that the formation of phase A’ is caused by
the conjoint removal of Li+ and oxygen that creates both cationic and anionic vacancies. The
cationic vacancies are partially refilled by TM cations causing their appearance in the octahedral
interlayer sites and suppressing the “honeycomb” Li-M cation ordering. The anionic vacancies are
refilled by their long range migration from the bulk to the surface with subsequent annihilation.
Consequently, the M/O ratio increases leading to the structural densification.
89
Secondly, we demonstrated that a major part of the capacity loss during the first cycle is due
to kinetic hindrance, and can be recovered by either applying a constant potential step or
discharging to lower voltages. Note that the feasibility of recovering the irreversible capacity by
pushing the low-voltage reduction is not specific to Li1.2Ni0.13Mn0.54Co0.13O2. Indeed, by
investigating the first-cycle coulombic efficiency of LiNi1/3Mn1/3Co1/3O2, Kasnatscheew et al.
showed that a main part of the capacity loss is of kinetic origin and linked to sluggish Li+ diffusion,
therefore, they recommended applying a constant voltage step at the end of discharge to fill the
remaining Li vacancies.163 Similar conclusion has been reached by Zhou et al. when studying the
LiNi0.8Mn0.1Co0.1O2 electrode.180 Indeed, a drop of Li+ diffusion coefficient at high state of
discharge was commonly observed in many layered oxides,30,164 and may be explained by i) the
contraction of c lattice parameter typically occurring at high Li content (x > 0.6)165,181; and ii) the
inhomogeneous lithiation leading to very high Li concentration (not enough vacancy) at the surface
of the particles.182
Thirdly, using highly reductive conditions, we revealed the formation of a new discharged
phase P” that is responsible for an additional low-voltage capacity. By gradually varying the depth
of charge from x = 0.0 (the pristine phase) to 0.4 (cationic redox region), and eventually to 0.6,
0.8 and 1.0 (anionic redox region), we further demonstrated that the appearance of phase P’’ is
directly linked to the necessity of triggering the oxygen redox activity on charge. The feasibility to
inject additional Li+ at potentials below 2.0 V has been reported in other Li-rich177,178 and Li-
stoichiometric175,183 layered oxides. The overlithiation usually leads to the formation of a Li2MO2-
type phase with P3m1 structure, as confirmed by Robert et al.175 Within the context of this study,
we experimentally showed that some additional Li+ could also be injected into
Li1.0Ni0.33Mn0.33Co0.33O2, when discharging directly the pristine material to 1.2 V (Figure A. 3. 15).
On the contrary, what is unique about the Li1.2Ni0.13Mn0.54Co0.13O2 phase is that i) extra Li+ cannot
be injected into the pristine phase by lowering the discharge voltage to 1.2 V (Figure A. 3. 16), ii)
the low voltage electrochemical activity can only be activated when the compound has been
charged sufficiently to trigger the oxygen redox process, and iii) the overlithiation does not lead to a
O3 → 1T phase transition. At first, it could be argued that this specific behavior is associated with
the presence of Li+ ions within the MO2 layers in Li1.2Ni0.13Mn0.54Co0.13O2 that modifies the
respective stability of the P’ vs P” phase. However, this does not hold as other pristine Li-rich
layered oxides have equally shown the feasibility to reversibly uptake extra Li+ ions (Li2+xIrO3,
90
Li3+xIrO4),57,60 through an R3m to P3m1 structural transition that doubles the number of available
sites in the Li layers. A more reasonable explanation is rooted in the observed Mn-migration in
Li1.2Ni0.13Mn0.54Co0.13O2, triggered by the anionic redox activity, unbalancing the electrostatic
interactions between the MO2 layers together with Li inter-intra layer repartition. In addition, the
presence of Mn ions at the interlayer Li sites could act as “pillar” to prevent the gliding of the
layered planes which is needed to obtain the 1T structure as previously reported for Li2NiO2.176
Lastly, we designed a path to retrieve the Mn-migration in the discharged P’ phase by mild-
temperature annealing (> 250 °C), as inspired by the work from Singer et al.112 Indeed, the low-
voltage discharge capacity vanishes in the same manner as the degree of Mn-migration, further
confirming the correlation between the Mn-migration and the low-voltage redox activity. Moreover,
the recovery of superstructure and likewise the original stair-case charge profile after the heat
treatment, indicate that the cation disordering is decisive in transforming the stair-case charge to an
S-shaped sloping discharge during the first electrochemical cycle. An obvious future direction to
understand better the role of cation migration (structural disordering) on the electrochemical
properties of Li1.2Ni0.13Mn0.54Co0.13O2 resides in assembling solid state Li batteries that can be
operated at various high temperatures.
Application-wise, although we have showed the positive effects of recovering part of the
first-cycle irreversible capacity loss, neither reaching the full formation of the P’ phase nor the
exploitation of the low voltage activity provides substantial improvement regarding the energy
density, whilst retains the voltage fade and hysteresis pertaining to Li1.2Ni0.13Mn0.54Co0.13O2.
Returning to the electrochemically-driven structural evolution of Li1.2Ni0.13Mn0.54Co0.13O2,

we propose the following structure evolution sequence for the first full cycle of
Li1.2Ni0.13Mn0.54Co0.13O2 (Figure III. 31). First, the oxidation of the Ni2+ and Co3+ cations is
occurring up to x of 0.4, followed by the oxidation of the oxygen anions that triggers Mn-
migration until 4.6 V (x < 1.0). Afterwards, some Li+ can still be removed concomitantly with O2
evolution to reach nearly the fully delithiated phase, creating both cationic and anionic vacancies
that are filled by migrated Mn ions, and eventually leading to a bulk densified phase A’ (x = 1.15).
Densification implies that the oxygen vacancies formed during the O2 evolution stage migrate from
bulk to the surface with subsequent annihilation that causes increasing of the M/O atomic ratio. A
91
pure phase A’ can only be obtained when sufficient amount of Li+ is removed and sufficient time is
given for the migration of transition metals, which might explain why the phase A’ was never
reported before in a pure form. Upon discharge, the anionic and cationic network of the phase A’
are conjointly reduced with only part of the migrated cations return to their initial positions,
explaining that the discharge phase P’ differs from the pristine phase, even after reaching a nearly
full reduction by either applying a constant voltage step or discharging to lower voltages.
Figure III. 31 Schematic of the A → A’→ P’ phase transition. Delithiation and oxygen
evolution create vacancies in both cationic and anionic sublattices upon first charge above 4.6 V.
The oxygen vacancies are refilled implying their long range migration from the bulk to the surface
with subsequent annihilation. The vacant Li sites are partially refilled with the TM cations causing
their appearance in the octahedral interlayer sites and suppressing the “honeycomb” Li-TM cation
ordering. As a result, the TM/O ratio increases that is considered as “densification”. During
subsequent discharge, the TM cations partially move back to their original positions and the vacant
cationic sites are filled with Li+.
During the subsequent charge, phase P’ releases Li+ through a solid-solution process (Figure
A. 3. 17), which differs from the two-phase process (A → A’ phase transition) on the first charge,
leading to the formation of phase A’ at the end of second CC - CV charge. As was demonstrated by
OEMS analysis, no O2 evolves throughout the second CC - CV charge (Figure III. 32b), suggesting
that although O2 evolution is critical to the initial formation of phase A’, once it forms, it can
reversibly uptake and release Li+ up to 26 cycles (the maximum number of cycle we have tried,
Figure III. 32a) without the need for having more anionic vacancies, hence O2 release.
92
Figure III. 32 (a) Laboratory XRD patterns of Li1.2Ni0.13Mn0.54Co0.13O2 collected at various

state of charge in in-situ XRD cell which were cycled between 4.8 V and 2.0 V, with an
intermediate CV step at 4.8 V for 5 h. The results reveal that phase A’ was formed at the end of
each CC-CV charge until the 26th cycle (the maximum number of cycle we have tried). (b) OEMS
gas analysis during the first and second cycle of Li1.2Ni0.13Mn0.54Co0.13O2. The cell has been cycled
between 4.8 V and 2.0 V, with a CV step at 4.8 V for 8 h. O2 (m/z = 32) and CO2 (m/z = 44) ion
currents were recorded and then converted to gas evolution rates. The voltage vs time curves is
presented in the upper panel, and the evolved rates of O2 (red line) and CO2 (black line) in the units
of μmol min-1g-1 are shown in the bottom panel. O2 evolution was only observed at the first charge,
whereas CO2 continues to evolve at the second cycle.
In summary, the results demonstrate the fundamental benefit of operating a cell under harsh
electrochemical conditions to deepen our understanding of the Li-rich NMC in terms of O2 release,
phase evolution, cation migration, and oxygen activity with the establishment of a few key
correlations that could guide the design of Li-rich phases having greater stability against O2 release
or cation migration. However, a cure to circumvent the voltage decay is still waiting to be found.
93
Chapter IV Li-CO2 battery: a new system for
energy storage and CO2 conversion
Chapter IV Li-CO2 battery: a new system for energy storage and CO2 conversion
Chapter IV Li-CO2 battery: a new system for energy storage

and CO2 conversion
IV. 1 Background and motivation
CO2 has been recognized as one of the major greenhouse gas (GHG) that responsible for the
global temperature increase. To alleviate the carbon emission, research is presently parted between
two major focuses: i) CO2 capture and sequestration in which major findings were obtained by
mimicking natural photosynthetic processes or introducing new techniques, such as physical
absorption and biological capture by microalgae184,185 and ii) CO2 utilization either by
electrochemically reducing CO2 into carbon-based fuels and feedstocks (e.g. carbon monoxide,
ethylene and methanol)186-188 or by transforming CO2 into other useful compounds via chemical or
electrochemical processes (e.g. methane, formic acid, C2-C4 olefins, C5-C11 hydrocarbons and
oxalate).189-193 Following the second approach, the activation of CO2 is unfortunately a challenging
task because of the inertness of CO2. The difficulties thus reside in finding catalysts that are
selective and are capable of mediating multiple electron and proton transfer at relatively low
overpotentials (i.e., low energy input). Within this context, the battery community has recently
proposed an alternative strategy for CO2 conversion by introducing CO2 into battery systems,
aiming to transform this detrimental greenhouse gas into value-added Li2CO3 which can then be
used, for instance, as the precursor for synthesizing Li-ion positive electrode materials. The concept
of CO2 battery has the potential not just to mitigate the atmospheric CO2 concentration back to its
pre-industrial level, but more importantly to use CO2 as a renewable energy carrier. In such a
scheme, CO2 either serves as a gas additive, i.e. Li-O2/CO2 battery which involves the chemical
reaction between CO2 and superoxide resulting from the electrochemical reduction of oxygen;102 or
used as the sole reactant, i.e., the “real” Li-CO2 battery.194
On the other hand, as previously mentioned in Chapter I. 3, despite the significant

development achieved so far in Li-ion batteries, their energy storage capacities are far from
satisfying the energy demands of model society. In fact, even with the state-of-art Li-rich NMC
positive electrodes, possibilities for Li-ion battery to achieve the 2025 targets of automotive market
rely on further improvements in cyclability, positive electrode loading and negative electrode
94
performance.25 Therefore, new energy storage systems such as Li-O2 battery has been intensively
studied since the first report by Jiang et al. in 1996, owing to its superior theoretical energy
density.195,196 However, multiple issues still prevent Li-O2 battery from making substantive
progresses. One critical barrier is that Li-O2 battery is very sensitive to atmospheric pollutions such
as CO2 or moisture traces leading to Li2CO3 or LiOH deposition.81,100,197,198 Another important issue
concerns the nucleophilic superoxide radicals, the main intermediates during discharge, which tends
to attack organic solvents199-203 and carbon electrodes204-206 to form Li2CO3, resulting in the
passivation of the electrode surface and hence the premature cell death. The decomposition of
Li2CO3 is incomplete on subsequent charge and only occurs at very high potentials leading to
relevant decomposition of cell components (E°= 3.82 V vs Li/Li+ for the reaction Li2CO3 → CO2 +
½ O2 + 2 Li+ + 2 e¯, 207,208 and with a practical potential of > 4.3 vs Li/Li+ as observed in the
literatures and computed by first-principles calculation).209,210 Moreover, the CO2 released from
Li2CO3 decomposition chemically reacts with Li2O2, the ultimate discharge product of Li-O2 battery,
in turn to form more Li2CO3 and further deteriorates the cycling performance. Last but not least, the
chemical reactivity of superoxide radical with CO2 dissolved in aprotic solvents has also been
reported to potentially alter the discharge products.211,212 In light of the multifaceted role played by
CO2 on Li-O2 electrochemistry, it is therefore essential to provide a comprehensive understanding
about the interplay between CO2 and all the intermediates or final products formed in Li-O2 battery
(i.e., Li+, superoxide radical, Li2O2).
This chapter will be divided into two parts. The first part (IV. 2) will focus on the
fundamental understanding about the formation mechanism of Li2CO3 in Li-O2/CO2 batteries and
how the electrolyte properties dictate the reaction pathway, the morphology of Li2CO3 and the
battery performances. The second part (IV. 3) will demonstrate a mediated Li-CO2 battery based on
quinone derivatives. A series of quinone derivatives will be surveyed in order to reveal the
correlation between the molecular structure of quinones and their affinity towards CO2. In addition,
the effect of cations and solvents on the catalytic reduction of CO2 by quinone derivatives will be
discussed.
95
IV. 2 Li−O2/CO2 batteries: CO2 conversion via electrogenerated

superoxide (O2●¯)
IV. 2. 1 Introduction
Similar to Li-O2 battery, a typical Li-O2/CO2 battery which can also be intuitively viewed as
Li-O2 battery with CO2 as gas additive consists of a metallic lithium negative electrode, a separator
soaked with organic electrolyte and a porous carbon matrix for accommodating the discharge
product (Figure IV. 1). As their names suggest, the notable difference between Li-O2 battery and
Li-O2/CO2 battery is the gaseous reactant, resulting in different discharge products with Li2O2 for
the former while Li2CO3 is formed for the latter.
Figure IV. 1 Schematic illustration of (a) Li-O2 battery and (b) Li-O2/CO2 battery.
First proposed by Albertus et al. in 2010,165 the Li-O2/CO2 battery is still in its infancy with
several fundamental questions remaining unanswered. Primary Li-O2/CO2 (1:1 v/v) batteries were
demonstrated using 1 M lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) in a mixture of
ethylene carbonate and diethyl carbonate (3:7 v/v) as electrolytes, while Liu et al. argued that Li-
O2/CO2 (2:1 v/v) batteries can operate reversibly when lithium triflate-tetraglyme was used as
electrolytes.213 This controversy led to the first question: what is the role of the electrolyte and why
is it so critical for the rechargeability of Li-CO2/O2 battery? Lim et al. further suggested that the
solvents can effectively alter the discharge reaction pathways and eventually change the final
discharge product from pure Li2CO3 in high DN solvent (DMSO) to a mixture of Li2O2 and Li2CO3
96
in low DN solvent (DME).214 Moreover, the solvent seems to affect the discharge capacity of Li-
O2/CO2 battery. Starting from Takechi et al.’s work which showed an increased discharge capacity
in the presence of CO2 using carbonate electrolytes.102 Different results were later reported with
comparable discharge capacity obtained when using DMSO electrolyte while a smaller discharge
capacity was obtained using DME electrolyte.214 Although several possibilities are still largely
debated, such as i) the low concentrations of CO2 which can block the surface-active nucleation
sites of Li2O2100, ii) the increased electronic conductivity of Li2CO3 through O2-induced
morphology change,198 and iii) increased electronic conductivity of the discharge product due to the
favorable electron polaron hopping within the Li2O2 part of the Li2O2@Li2CO3 interface,215 these
rather contradicting results emphasized the important role played by the solvents.214 Another critical
issue concerns the discharge reaction mechanism, more specifically, whether there is a direct
electrochemical reduction of CO2 in Li-O2/CO2 battery. While one may expect a straightforward
answer to this question by cycling batteries in pure CO2 atmosphere (termed as Li-CO2 battery
hereinafter), nevertheless, discrepancy exists in the literatures with Takechi et al. showing very
limited discharge capacity of 66 mAh g-1 for Li-CO2 battery,102 whereas Liu et al. reported
reversible Li-CO2 battery with a discharge capacity of ~ 1000 mAh g-1 and attributed this
improvement to the use of glyme electrolyte.213
Aware of the important role played by the electrolytes in the electrochemical properties of
Li-O2/CO2 batteries, we therefore decided to embark into a detailed electrochemical study on the
discharge processes for Li-O2/CO2 (7:3 v/v) batteries using various solvents. Doing so, we aim (1)
to provide an overall mechanistic image about the discharge reaction mechanism, (2) to identify the
dominant electrolyte property which governs the reaction route, (3) to reveal how the capacity and
the morphology of final product will be modified depending on the solvent used and also (4) to
evaluate the rechargeability of Li-O2/CO2 battery. Dimethyl sulfoxide (DMSO), 1, 2-
dimethoxyethane (DME) and acetonitrile (MeCN) were chosen for the initial evaluation, since they
have different DN but DMSO and MeCN have similar acceptor number (AN) and different
dielectric strength (ε) while DME alone has significantly smaller ε (Table IV. 1). Thus comparing
the results obtained with these solvents should help us in identifying which parameter of the solvent,
DN, AN, or ε, dictates the discharge reaction routes.
97
Table IV. 1 Summary of the intrinsic physical and chemical properties of all the solvents
investigated in this work. Donor number, acceptor number, solubility data and diffusion coefficients
of oxygen and carbon dioxide were collected from literatures84,216-218 and the conductivity values
were experimentally determined using 0.1 M LiClO4 as the supporting salt.
ɛ DN AN O2 CO2 105*Do2 105*Dco2 Conductivity Viscosity

(25℃) solubility solubility 2
[cm /s] 2
[cm /s] [mS/cm] η 217
[mmol] [mmol]
DMSO 48.0 29.8 19.3 2.1 125 2.2±0.4 1.02±0.1 2.72 1.948
DME 7.2 24.0 10.2 10.1±0.3 NA 6.0±0.5 NA 0.22 0.46
MeCN 36.64 14.1 18.9 8.1 280 11±1.2 38.3±1.1 9.88 0.361
Considering the complexity of the Li-O2/CO2 system with all the possible interactions
among its reactants, electrolytes and intermediates, for the sake of clarity, various possible
discharge paths based on the previous knowledge in the literatures102,211,217,219,220 as well as their
establishments in this work, are shown in Scheme IV. 1.151
98
Scheme IV. 1 Survey of the most likely reaction paths from the various scenarios formerly
proposed in the literatures and their establishment in the present work.
Possible reaction pathways:
IV. 2. 2 Results and discussion
Aware of the fact that CO2 is more soluble in organic solvents than O2 (125 ± 13 mM for
CO2 and 2.1 mM for O2 in DMSO, for instance),217 together with the aforementioned controversy
over direct electrochemical reduction of CO2 in Li-CO2 battery and its possible dependence on the
nature of solvents and electrodes used, we therefore first explore the electrochemical activity of Li+
with CO2 alone on a glassy carbon electrode (GCE) before to add O2 to the system. Similar cyclic
voltammograms were obtained for CO2- and with Ar- (inert gas) saturated electrolytes (Figure IV.
2), suggesting that CO2 cannot be electrochemically reduced in the potential range of Li-CO2
battery, namely 2 – 4.5 V vs Li+/Li, irrespective of the solvents used (DME, DMSO or MeCN).
These results are in good agreement with the fact that, injecting an electron into the lowest
unoccupied molecular orbital (LUMO) of the linear CO2 molecule to form CO2•¯ requires a very
low potential (high energy) in aprotic solvents without efficient catalyst.217,221 Hence, the direct
99
electrochemical reduction of CO2 can practically be neglected in this work. To further confirm this,
Li-CO2 cells were tested with 0.1 M LiClO4 in both DMSO- and DME- based electrolytes using
carbon super P (Csp) electrodes. In agreement with the previous CV results, when using a cut-off
voltage of 2 V vs Li+/Li, CO2 cells delivered very small capacity (Figures IV. 3a & 3c, ~ 100 mAh
g-1 DMSO, ~ 10 mAh g-1 in DME) associated with very limited gas consumption (Figures IV. 3b &
3d). Further decreasing the cut-off voltage to 1 V (a voltage in which DMSO is unstable as can be
seen from the cells discharged in argon (Figures A. 4. 1e & 1f) as well as observing the
characteristic “yellow” color found in the separators (Figure A. 4. 2)), a gas consumption
corresponding to 4 e¯ / mole of gas is measured for DMSO under pure CO2 (Figure A. 4. 1d).
Knowing that the S=O double bond of DMSO can be reduced; one can tentatively assign this
reaction to the nucleophilic attack of reduced DMSO to CO2. Moreover, bearing in mind that
successful Na-CO2 cells were also been reported in the literature,222 we discharged cells in pure
CO2 with various Na+-based electrolytes in order to examine the cation effect. Once again, a very
limited capacity (less than 5 mAh g-1) was obtained, either with Csp or with gas diffusion layer
(GDL) electrodes (Figure A. 4. 3). These results are surprisingly in contradiction with previous
works213,222-224 and indicate that, in the absence of O2 and under the conditions that allows to
reversibly cycle Li-O2 cells at limited capacity of 1000 mAh g-1 or to discharge Na-O2 cells,127,225
CO2 cannot be electrochemically reduced and Li-CO2 cells deliver almost no discharge capacity.
100
Figure IV. 2 Cyclic voltammograms at 50 mV s-1 for CO2- (blue curves) and Ar- (black
curves) saturated (a) DMSO, (b) DME and (c) MeCN. 0.1 M LiClO4 was used as supporting salt in
all the solvents to increase the conductivity.
101
Figure IV. 3 (a) Discharge curves and (b) pressure change of Li-CO2 batteries using 0.1 M
LiClO4/DMSO as electrolyte, at a current density of 100 mA g-1; and using 0.1 M LiClO4/DME as
electrolyte, at a current density of 50 mA g-1 in (c) and (d); black curve and blue curve are
representative of two cells measured under the same condition.
Having established that CO2 is inert when limiting the potential range and under these
conditions, CVs were then performed with TBA+ salt to explore the electrochemical behavior of
O2/CO2 gas mixture and compared with pure O2. As seen in Figure IV. 4, a one electron reduction
of oxygen to superoxide species (cathodic scan) followed by the reoxidation of the latter (anodic
scan) is observed in pure O2-saturated DMSO, DME and MeCN (black curves), as already largely
reported in the literatures.82,84,216 Switching to O2/CO2-saturated electrolytes, the same reduction
peak is observed (blue curves) regardless of the solvents, demonstrating that the first step of the
reaction is the formation of superoxide species (Scheme IV. 1, equation (1.1)). It should be noted
that the reoxidation peak of superoxide was totally ‘suppressed’ (titrated) under O2/CO2,
102
demonstrating that all the generated superoxide species had fully reacted with CO2 to form the
CO4•¯ intermediates that cannot be oxidized back.211 It is noteworthy that this strong reactivity
between O2•¯ and CO2 was observed in all the three solvents herein studied. This result indicates that
the stability of the TBA+-O2•¯ complex formed upon reduction is not sufficient to stabilize the
superoxide species and to prevent its nucleophilic attack on CO2 to form CO4•¯.
103
Figure IV. 4 Cyclic voltammetry curves at 50 mV s-1 for O2-saturated (black curves) and O2/CO2-
saturated (blue curves) (a) 0.1 M TBAPF6/DMSO, (b) 0.1 M TBATFSI/DME and (c) 0.1 M
TBAPF6/MeCN.
104
After observing that all the superoxide species generated during reduction in O2/CO2-
saturated electrolytes readily react with CO2 when using TBA+ as a salt, a legitimate question
concerns the relative reactivity of the superoxide species towards Li+ compared to CO2. For that,
CV measurements were performed with O2/CO2-saturated electrolytes using Li+ salt and were
compared with the previously discussed results obtained with TBA+ salt (Figure IV. 5). This is the
first report of O2/CO2 interaction in presence of Li+ to the best of our knowledge. Beginning with
DMSO (Figure IV. 5a), the onset reduction potential in O2/CO2 mixture is ~ 0.1 V higher than the
onset potential measured in pure O2. This observation indicates that the reaction of CO4•¯ with Li+
forms an intermediate which is thermodynamically more stable than lithium superoxide
intermediates in DMSO. Thus it is reasonable to write the first two reaction steps in DMSO as
equation (1.2) in Scheme IV. 1 followed by the reaction of CO4•¯ with Li+ (Scheme IV. 1, equation
(1.3)). A positive shift was also observed when comparing CV results obtained in O2/CO2-saturated
Li+- and TBA+-based electrolytes (Figure IV. 6a), demonstrating that the formation of LiCO4 is
thermodynamically more favorable and governs the reduction potential in DMSO. In light of this
observation, it is interesting to reexamine the results obtained in O2-saturated DMSO where no
significant shift from TBA+ to Li+ was observed (Figure A. 4. 4). We can therefore conclude that
the stabilization of Li+ by DMSO is sufficiently strong to prevent the early formation of lithium
superoxide (Scheme IV. 1, equation (1.4)), hence enabling the formation of LiCO4 which is
thermodynamically more favored (Scheme IV. 1, equation (1.3). During the positive scan, no
current response was detected until at very high potential in O2/CO2-saturated DMSO in contrary to
pure O2 (Figure IV. 5a). This different oxidation behavior suggests that, instead of Li2O2, Li2CO3
which shows an oxidation potential of about 3.82 V vs Li+/Li is probably formed from a subsequent
reduction of the LiCO4 intermediate during the reduction process.209 Moreover, the greater
reactivity of CO4•¯ intermediate when compared to O2•¯ is further seen by the fact that it cannot be
reversibly oxidized in TBA+-containing solution, indicating a possible reaction with the solvent
(Figures IV. 4a & 6a). Therefore, in light of the increasing instability of high DN solvent towards
O2•¯ as it has been seen for Li-O2 batteries, one can expect even more limited stability of high DN
solvents towards CO4•¯ radicals when used in Li-O2/CO2 batteries.226,227 Switching to DME and
MeCN, notably similar onset reduction potentials were observed in both Li+-O2 and Li+-O2/CO2
systems (Figure IV. 5b & 5c), implying that, unlike for DMSO, the step that governs the onset
potential is the same in O2/CO2 mixture as in pure O2, namely the formation of lithium superoxide
(Scheme IV. 1, equation (1.4)). Note that significantly different O2•¯ reaction pathways were found
105
for DMSO and MeCN, despite MeCN share the similar values of AN and ε with DMSO, we can
therefore excludes AN and ε but identify DN as the most important parameter governing the
reaction routes in Li-O2/CO2 batteries. The effect of DN is further supported by the fact that similar
reduction potentials were measured for other low DN solvents such as diethylene glycol dimethyl
ether (DGME, DN = 18) and tetraethylene glycol dimethyl ether (TGME, DN = 12) glymes (Figure
A. 4. 5).
106
Figure IV. 5 Cyclic voltammetry curves at 50 mV s-1 for O2-saturated (black curves) and
O2/CO2-saturated (blue curves) (a) 0.1 M LiClO4/DMSO, (b) 0.1 M LiClO4/DME and (c) 0.1 M
LiClO4/MeCN. Inset figure in (b) is the zoom image for the cyclic voltammetry curve of O2/CO2-
saturated 0.1 M LiClO4/DME above 2.8 V vs Li/Li+.
107
Our results highlight the clear difference in reactivity of O2•¯ towards CO2 and Li+ in DMSO
(Scheme IV. 1, equation (1.2)) and in DME and MeCN (Scheme IV. 1, equation (1.4)), which is
consistent with the DFT calculations reported by Lim et al..214 Specifically, a strong coordination
shell will form between Li+ and the solvent molecules in a high DN solvent like DMSO,82,85,214,228
pushing the reaction towards the nucleophilic attack of O2•¯ with CO2 instead of Li+. In contrary, the
Li+ stabilization is weaker for DME and MeCN and therefore O2•¯ reacts preferentially with Li+
rather than with the neutral dissolved CO2. Finally, even though DMSO possesses large acceptor
and donor numbers compared to DME (AN = 19.3 and DN = 29.8 for DMSO, AN = 10.2 and DN =
24 for DME), it is not enough to sufficiently stabilize CO4•¯ and to prevent its reaction with Li+.
These different pathways are summarized in Scheme IV. 1. However, at this stage of the study, one
cannot distinguish for DME between two different pathways: one involving the electrochemical
formation of Li2CO3 from lithium superoxide (pathway 2 in Scheme IV. 1) and one involving the
chemical reaction of Li2O2 with CO2 (pathway 3 in Scheme IV. 1), this will be discussed later.
Another interesting result was obtained when comparing the reduction currents measured in
O2/CO2-saturated Li+- and TBA+-based electrolytes. For DMSO, only slightly smaller reduction
currents were measured in Li+- with respect to TBA+ - based electrolyte (Figure IV. 6a). On the
contrary, significantly smaller reduction currents were measured in Li+-based DME and MeCN
electrolytes compared to their TBA+ counterparts (Figure IV. 6b & 6c), suggesting a faster blockage
of electrode surface in Li+-containing electrolytes of DME and MeCN than in DMSO. This point
will also be discussed in more details later on in this chapter.
108
Figure IV. 6 Cyclic voltammetry curves at 50 mV s-1 for O2/CO2-saturated TBA+- (dash
curves) and Li+- (solid curves) based (a) DMSO, (b) DME and (c) MeCN.
109
To definitively conclude on the discharge reaction mechanisms, cells were discharged at

various current densities using 0.1 M LiClO4 in DMSO and DME as electrolytes in both pure O2
and O2/CO2 mixture and their resulting discharge products were analyzed (Figures IV. 7 & 8). In
agreement with our cyclic voltammetry results previously discussed in Figure IV. 5a, the discharge
potential for O2/CO2 cells with DMSO is consistently higher (> 0.1 V) than the discharge potential
measured for O2 cells at all investigated current densities (Figures IV. 7a, 7c & 7e), which confirms
a different electrochemical pathway for O2/CO2 cells from O2 cells. The electrode fully discharged
in O2/CO2-saturated electrolyte showed by XRD the formation of highly crystalline Li2CO3, while
no Li2O2 was detected (Figure IV. 8a), indicating that Li2CO3 was the ultimate discharge product as
expected from the mechanism uncovered by our CV study. Since there can be numerous complex
intermediate steps to form Li2CO3 from LiCO4 and which cannot be distinguished by CVs , we
simply wrote the overall reaction as equation (1.5) (Scheme IV. 1). Overall, we can conclude that
the discharge reaction of Li-O2/CO2 battery in DMSO-based electrolyte is following the equation
(1.6) (Scheme IV. 1). This conclusion was further confirmed by measuring the number of electrons
consumed per mol of gas deduced from pressure measurements (Figures IV. 7b, 7d & 7f).
Independently of the current density, every O2/CO2 cell consumed ~ 1.33 ± 0.1 e¯ per mole of gas
during discharge compared to ~ 2 ± 0.3 e¯ per mole of gas measured for O2 cells, in agreement with
the overall reaction 4 Li+ + 4 e¯ + O2 + 2 CO2 → 2 Li2CO3 (Scheme IV. 1, equation (1.6) ). This
assumption is also further supported by the fact that Li2CO3 formed in DMSO cells shows a sea
urchin-like shape (Figure IV. 8b), very different from that of Li2O2 particles formed under the same
conditions in pure O2 systems (Figure A. 4. 6a).229-232 Another interesting observation is that
O2/CO2 cells delivered larger capacity than O2 cells at lower current densities of 50 mA g-1 and 100
mA g-1 (Figures IV. 7a, 7c), while the reverse trend being observed at higher current density of 200
mA g-1 (Figures IV. 7e). Even though the physical origin for this capacity difference is not fully
understood, based on previous reports on the impact of discharge product morphology on the Li-O2
cells capacity, 229,230,233 we can suspect at this point that morphology can also play a role for Li-
O2/CO2 cells. At higher current density, the nucleation rate will be faster than the growth rate, 234
leading to the formation of smaller Li2CO3 particles (Figure A. 4. 6d) and eventually to the
formation of a film covering the electrode, alike what has been observed for Li-O2 cells. Also
considering that Li2CO3 is a wider band gap insulator than Li2O2 (5.03 eV for Li2O2 and 8.01 eV for
Li2CO3, as calculated previously),215 we can postulate that a shorter electron tunneling distance is
expected for Li2CO3 film compared to Li2O2, hence leading to a premature cell death.
110
Figure IV. 7 (a, c, e) Galvanostatic discharge curves and (b, d, f) the corresponding pressure
changes in 0.1 M LiClO4/DMSO electrolyte under pure O2 (black curves) and under O2/CO2 (blue
curves) at various current densities of 50 mA g-1 (a, b); 100 mA g-1 (c, d) and 200 mA g-1 (e, f). For
each measurement, two cells measured under the same conditions are reported to show the data
reproducibility.
111
Figure IV. 8 (a) X-ray diffraction pattern and (b) SEM image of electrode after fully
discharged under O2/CO2 using 0.1 M LiClO4/DMSO electrolyte at a current density of 50 mA g-1.
As we observed by CVs , O2•¯ reacts first with Li+ in DME in O2/CO2 atmosphere, in a
similar way to what is observed in pure O2. To further test and verify this observation, Li-O2/CO2
cells were assembled and discharged in DME-based electrolyte (Figure IV. 9). In agreement with
the aforementioned CV results (Figure IV. 5b), the same discharge potential is measured in O2 and
O2/CO2 at all studied current densities (Figures IV. 9a, 9c & 9e), indicating that the potential is
governed by the same equilibrium corresponding to the lithium superoxide formation (Scheme IV.
1, equation (1.4)). Interestingly, the pressure analysis shows that the e¯ / mole of gas consumption is
~ 1.33 ± 0.2 during discharge for O2/CO2 cells, compared to ~ 2 ± 0.8 e¯ / mole of gas for O2 cells
(Figures IV. 9b, 9d & 9f). This strongly suggests that Li2CO3 is the discharge product, which is
confirmed by XRD (Figure IV. 10a). SEM images show that similar morphology of particles was
found in both O2 and O2/CO2 cells (Figures IV. 10b & A. 4. 6b). Reexamining the CV results in Li+-
O2/CO2 system in DME in the potential range of 2.8 - 4.5 V vs Li/Li+, an oxidation peak, although
very weak, can be noticed at the same potential as the oxidation peak measured in pure O2 (inset
figure of Figure IV. 5b). As previously discussed, while our CV study clearly demonstrates that
equation (1.1) and equation (1.4) (Scheme IV. 1) initially occurred, two subsequent pathways for
the Li2CO3 formation can be envisioned: pathway 2 (equations (1.7) and (1.5)) and pathway 3
(equations (1.8) and (1.9)) (Scheme IV. 1). At this stage, the question is to know if the chemical
reaction of lithium superoxide with CO2 is faster than its disproportionation into Li2O2. Knowing
112
that the Li2O2 morphology observed after discharge in Li-O2 cells was found to be influenced by the
lithium superoxide stability,37 we examined the morphology of Li2CO3 formed under
electrochemical or chemical conditions to determine which pathway is the most likely to occur in
DME-based Li-O2/CO2 cells. Comparing the SEM images obtained by chemical reaction of Li2O2
with CO2 with those obtained after discharge in O2/CO2 atmosphere (Figure A. 4. 7), a similar
morphology with small and round shaped particles was observed in both cases. Besides, based on
the findings by Lim et al. suggesting that Li2O2 tends to primarily form in DME30 and on the
observation by McCloskey’s group that DME-based cells discharged in CO2-O2 with a 10-90 ration
show a discharge potential and a e¯ / mole of gas ratio similar to O2 cells, 25 we postulate that
lithium superoxide first disproportionates to form Li2O2, and then spontaneously reacts with CO2 to
form Li2CO3, following the pathway 3 (Scheme IV. 1, equations (1.8) and (1.9)). However, since
lithium superoxide is known to be very reactive, we should not neglect the possibility for pathway 2
(Scheme IV. 1, equations (1.7) and (1.5)) to occur. Another observation that can be made from the
discharge curves is that the O2/CO2 cells deliver larger capacity than O2 cells at lower current
densities (50 mA g-1 and 100 mA g-1, Figures IV. 9a & 9c); but the trend was reversed when
increasing the current density to 200 mA g-1 (Figure IV. 9e). From the SEM images obtained for
electrodes discharged at different current densities in O2/CO2 (Figures IV. 10b, 10c & 10d), we
noticed that particles grow bigger in size and number when decreasing the current density from 100
mA g-1 to 50 mA g-1 (Figures IV. 10b & 10c). However, at relatively high current density of 200
mA g-1, a continuous film covering the Super P electrode surface was found in agreement with an
enhanced nucleation rate at high current density (Figure 10d). This change in morphology could be
attributed to the previously discussed conductivity difference between Li2CO3 and Li2O2 films and
which might be the reason for the current density dependence of the discharge capacity measured
for DME-based O2/CO2 cells when compared with pure O2 cells.
113
Figure IV. 9 (a, c, e) Galvanostatic discharge curves and (b, d, f) the corresponding pressure
changes in 0.1 M LiClO4/DME electrolyte under pure O2 (black curves) and under O2/CO2 (blue
curves) at various current densities of 50 mA g-1 (a, b); 100 mA g-1 (c, d) and 200 mA g-1 (e, f). For
each measurement, two cells measured under the same conditions are reported to show the data
reproducibility.
114
Figure IV. 10 (a) X-ray diffraction pattern of electrode after fully discharged under O2/CO2
using 0.1 M LiClO4/DME electrolyte at current densities of 50 mA g-1. SEM images of electrodes
after fully discharged under O2/CO2 using 0.1 M LiClO4/DME electrolyte at current densities of 50
mA g-1 (b), 100 mA g-1 (c) and 200 mA g-1 (d).
Finally, bearing in mind that previous works reported rechargeable Li-O2/CO2 batteries, 27, 30
we investigated the cycling behavior of Li-O2/CO2 cells. For that, we compared two cells: one with
a capacity limitation of 1000 mA g-1 (corresponding to an absolute capacity of 0.36 mAh) and one
without capacity limitation. At first glance, our results could naively lead us to conclude that Li-
O2/CO2 cells are rechargeable, but it is not the case (Figure IV. 11a & 11b). Indeed, the gas
evolution measurements for both cells clearly indicate drastic side reactions (dashed areas in Figure
115
IV. 11c and in inset of Figure IV. 11d). More specifically, the gas evolution in charge doesn’t
exhibit the ratio of ~ 1.33 e¯ / mol of gas which is measured in discharge, instead, a much higher
value of ~ 2.26 ±0.3 e¯ / mol of gas was observed from the very beginning of the charge process for
both cells (stage I in Figure IV. 11c and inset of Figure IV. 11d). This observation evidences that
although a Li-O2/CO2 cell can be cycled for few tens of cycles when drastically limiting the
capacity, a similar situation that has been encountered for many years and erroneously associated to
the reversibility of Li2O2 rather than electrolyte decomposition in Li-O2 cells. Therefore, cycling
cells with limited capacity and decreasing the electrode loading so as to obtain cycling
performances are not ways to achieve cleaner chemistry, and one can conclude that Li-O2/CO2 cells
are not rechargeable under these conditions (electrode, electrolyte, and current density) owing to the
very high potential needed to oxidize Li2CO3. Efforts must be made to find redox mediators or solid
catalysts so as to decrease the oxidation potential closer to its theoretical value of 3.82 V vs Li/Li+.18,
19
However, as the discharge potential is pinned in DMSO by the formation of LiCO4 at about 2.8 V
vs Li/Li+ that governs the discharge potential, a poor round trip efficiency can already be foreseen
for this system, which should therefore be more certainly seen as a primary battery. Moreover,
looking in details to the cell cycled with no capacity limitation (Figure IV. 11c), a negative pressure
evolution of ~ 3.26 e¯ / mol of gas at high potential range (stage II) was observed following the
positive gas evolution of 2.26 ± 0.3 e¯ / mol of gas measured at stage I. Knowing that the capacity
corresponding to stage I matches with the discharge capacity of about 10000 mAh g-1, one can
postulate that the reaction in stage I is associated with the oxidation of Li2CO3 discharge product
which presumably forms solid products instead of releasing gaseous O2 and CO2 as expected from
equation (1.6). Using gas analysis carried out with online electrochemical mass spectrometry
(Figure IV. 12) as well as previous report,52 we found that CO2 is released upon charge while almost
no gaseous O2 is evolved, explaining the ratio of ~ 2.26 ± 0.3 e¯ / mol of gas found by pressure
measurements in Figure IV. 11c. Moreover, knowing that no gas corresponding to DMSO
decomposition was found by mass spectrometry, we therefore suspected that oxidized oxygen
species react with DMSO and/or the electrode to form a solid product. Moreover, this reactivity
certainly leads to different product formation than the direct oxidation of the electrolyte measured at
stage II, as seen by the different pressure evolutions measured for both processes in Figure IV. 11c
as well as the different CO2 evolution rates found by mass spectroscopy (Figure IV. 12).
116
Figure IV. 11 (a) Galvanostatic discharge-charge profiles and (c) the corresponding pressure
changes of Li-O2/CO2 cells with a limited potential range, and with a limited capacity of 1000 mAh
g-1 in (b) and (d). Both cells are cycled at a current density of 100 mA g-1, using 0.1 M
LiClO4/DMSO as electrolyte.
117
Figure IV. 12 (a) Charge profile of a Li-O2/CO2 cell at 25 °C using 0.1 M LiClO4/DMSO at
a current density of 100 mA g-1; and (b) partial pressure curve for CO2 (m/z = 44) and O2 (m/z = 32)
during charge, the cell was vacuumed and refilled with argon before mass spectroscopy
measurement.
IV. 2. 3 Discussion and conclusion
Firstly, our findings reveal that CO2 cannot be directly reduced in aprotic solvents within
their electrochemical window (DME, DMSO and MeCN are investigated in this work) using carbon
Super P as positive electrode. Correspondingly, no electrochemical activity in Li-CO2 batteries was
observed in contrasts with many literatures. Worth mentioning is a recent study which demonstrated
that, despite Li-CO2 battery exhibiting negligible discharge capacity (with positive electrodes of
bare and monodispersed Ru nanoparticles functionalized graphene nanosheet), the introduction of 2 %
O2 (at volume ratio) can lead to a high capacity of 4742 mAh g-1 (termed as O2-assisted Li-CO2
118
battery).235 In general, Li-CO2 battery is still in its infancy, and new insights are necessary to
explain this discrepancy in electrochemical activity.
Secondly, we demonstrate that different discharge reaction mechanisms in Li-O2/CO2

batteries are indeed at play, depending on the Li+ solvation strength by the electrolyte solvents. As
depicted in Scheme IV. 2, although the initial discharge step either in DMSO or DME, is the
electrochemical reduction of oxygen into superoxide, afterwards, superoxide favorably reacts with
CO2 in high DN solvent DMSO due to the strong solvation of Li+ by DMSO while it reacts with Li+
in low DN solvent DME to form lithium superoxide. Consequently, Li2CO3 is electrochemically
formed in DMSO (denoted as the electrochemical pathway). Whereas in DME, Li2O2 is first formed
alike in Li-O2 batteries before to chemically react with CO2 to form Li2CO3 (denoted as the
chemical reaction pathway). Despite the different superoxide reactivity that was highlighted in this
work, Li2CO3 is the only final product for Li-O2/CO2 batteries, regardless of the solvents. Note that,
following our work, similar reaction scenario was also established in trace O2-assisted Li-CO2
battery (which can be intuitively viewed as Li-O2/CO2 battery with very high CO2/O2 ratio) based
on in-situ surface enhanced Raman spectroscopy.236 Although the abovementioned scenario
rationalizes the major difference in the discharge of Li-O2/CO2 batteries, the following reaction
steps in DMSO, i.e., after the superoxide reaction with CO2, remain to be identified (the reaction
paths from superoxide to form Li2O2 in DME have already been investigated in detail within the
context of Li-O2 batteries). Along this line, Qiao et al. later on emphasized the importance of
CO2/O2 ratio in dictating the reaction of superoxide and CO2 based on fitted Raman spectra.237
Specifically, peroxodicarbonate (C2O62¯) tends to form under conditions with a high CO2/O2 ratio
while peroxomonocarbonate (CO42¯) is the dominant reaction intermediate at low CO2/O2 ratio.
119
Scheme IV. 2 Schematic illustration of the Li2CO3 formation in Li-O2/CO2 batteries with DMSO-
and DME-based electrolytes, following either the electrochemical pathway or the chemical pathway.
Thirdly, different reaction pathways result in significantly different Li2CO3 morphology (sea
urchin-like shape in DMSO compared to particle shape in DME). In addition, the capacity
difference between Li-O2/CO2 batteries and Li-O2 batteries depends on the discharge current
density, possibly resulting from the competition between nucleation and growth, as well as the
charge-transfer difference between Li2CO3 and Li2O2.
Finally, we demonstrate that Li-O2/CO2 batteries may not be rechargeable batteries based on
two facts: i) only CO2 is evolved but not O2 during the oxidation of Li2CO3 on charge; and ii) the
very high potentials required for Li2CO3 oxidation lead to drastic side reactions. To overcome these
limitations, one interesting future direction is to alter the discharge reaction thus to avoid the
“troublemaker” Li2CO3. Following this strategy, Qiao et al. demonstrated a Li-O2/CO2 battery
which operates based on peroxodicarbonate at a low charge potential of 3.5 V by stabilizing the
peroxodicarbonate intermediates in concentrated electrolytes.237
120
IV. 3 Li−CO2 batteries: CO2 conversion mediated by quinone

derivatives
IV. 3. 1 Introduction
In contrast to the O2-based battery systems, where the electrochemical reduction of O2 into
O2•¯ in aprotic solvents is thermodynamically more favored (-0.85 V vs SCE in DMF),238 the
electrochemical reduction of CO2 into CO2•¯requires a highly negative potential (-2.21 V vs SCE in
DMF),239,240 because of the energy required to rearrange a linear molecule into a bent radical anion.
This realization i) adds difficulty to trigger the electrochemical activity in Li-CO2 battery (as
discussed in Chapter IV. 2) and ii) induces copious parasitic reactions (e.g., electrolytes and other
cell components are generally unstable at such low potentials). It is therefore essential to promote
the electrochemical reduction of CO2 so as to improve the electrochemical performance of Li-CO2
battery. Starting from the first report on Li-CO2 battery, Archer et al. showed that the
electrochemical reduction of CO2 can be thermally activated at elevated temperatures based on the
reaction of 4 Li + 3 CO2 → 2 Li2CO3 + C.241 Later on, research efforts have been placed in
exploring heterogeneous electrocatalysts (e.g., carboneous materials with improved electronic
conductivity,194,223,224,242,243 noble metals, 207,208 etc.), seeking to facilitate the electron transfer to
CO2 at room temperature. Nevertheless, the efficiency of this research direction remains limited
since the surface deposition of Li2CO3 will block the catalytic sites. Alternatively, an electron can
be injected into the CO2 molecule indirectly via a solution chemical reaction, with the hope to evoke
solution formation of Li2CO3. Another fascinating aspect of this approach is to avoid the energy-
hungry step in classical electrochemical reduction of CO2, namely the formation of CO2•¯, thereby
effectively increasing the discharge voltage of Li-CO2 battery. One example supporting this view is
provided by introducing O2 into Li-CO2 battery system (denoted as Li-O2/CO2 battery, as detailed
investigated in Chapter IV. 2). Consequently, CO2 reduction is preceded through an initial
electrochemical reduction of O2 to O2•¯ which then chemically reacts with CO2 to eventually form
Li2CO3.102,151,214 Therefore, the discharge voltage of Li-O2/CO2 battery is now given by the
electrochemical reduction of O2 rather than that of CO2, a process thermodynamically more
favorable than the latter. Nevertheless, the major issue of this system resides in charge where the
oxidation of Li2CO3 is irreversible,101,151 namely CO2 is released but almost no O2 is given
back,102,151 entailing O2 as a scavenger instead of a catalyst.
121
Aside from the battery community, the electrochemical reduction of CO2 in aprotic solvents
is also heavily investigated by the electrocatalysis community. Numerous chemical catalysts were
proposed aiming to bypass the formation of CO2•¯intermediate along the reaction pathway to final
products.244,245 Note that, in contrast to redox catalysts (also referred as “soluble catalyst” or “redox
mediator”) which only shuttle electrons from the electrode surface to the target reactant,74,214-218
chemical catalysts involve multi-step CO2 reduction through the formation/ decomposition of
adducts formed between CO2 and the catalyst in their reduced form. More specifically, the catalyst
first undergoes electrochemical reduction prior to chemically react with CO2 to give the final
products and to regenerate the parent catalyst.246-248 Worth mentioning is that the concept of
chemical catalyst has also been extended to Li-O2 battery field to promote the electrochemical
reduction of O2 and the solution growth of Li2O2.89
Quinones are well-known for their high binding affinities toward CO2 in their reduced form
and were used to concentrate and selectively separate CO2.249-251 In addition, the synergistic effect
of Lewis acid cation such as Li+ on the electrochemical reduction of CO2 by chemical catalyst, i.e.,
iron(0) porphyrins, has been previously identified.252 Encouraged by the aforementioned findings,
we decided to investigate quinones as potential chemical catalysts for the CO2 in Li-CO2 batteries.
IV. 3. 2 Results and discussion
Quinones represent a class of organic compounds with versatile chemical/electronic

structures, redox potentials and affinities toward CO2 as well as Li+.253-255 Therefore, three quinones
with different chemical structures ─ 9, 10-anthraquinone (AQ), 9, 10-phenanthrenequinone (PAQ),
and 2, 5-di-tert-butyl-1, 4-benzoquinone (DBBQ) ─ have been surveyed aiming to examine the
possible correlation between structure and catalytic properties (Figure IV. 13).
Figure IV. 13 Chemical structures of the herein investigated quinones.
122
To spot the electrochemical behavior of quinones in the absence of CO2 and Li+,
voltammetry studies were first carried out in argon-saturated electrolytes using
+
tetrabutylammonium (TBA ) as the supporting cation. Compared to the voltammograms recorded
without quinone showing almost no current response (green curves in Figure IV. 14), two well-
defined redox waves was observed in the presence of quinone, irrespective of the solvent (orange
curves in Figure IV. 14) as well as their chemical structures (red curves in Figure IV. 15a).
Figure IV. 14 Cyclic voltammograms recorded under argon atmosphere in solutions of 0.1
M LiClO4 in DMF (a) and MeCN (b) with (orange curves) and without (green curves) DBBQ. Scan
rate: 100 mV s-1.
Consistent with previous findings, these two waves correspond successively to the
formation/oxidation of quinone anion radicals Q•¯and dianions Q2¯(Q stands for quinones):251,256
Q + e ¯ ↔ Q•¯ (1)
Q•¯ + e ¯ ↔ Q2¯ (2)
The reactivity of quinones towards CO2 was then studied (Figure IV. 15a), and no
substantial change in both the onset potentials and peak currents of the first reduction waves was
observed, indicating no CO2 reactivity of Q•¯independent of the chemistry of quinones. Instead, all
the second reduction waves shifted to more positive voltages in the presence of CO2, suggesting a
strong interaction between CO2 and Q2¯. Note that similar voltammetric behavior was previously
123
reported and indeed these positive shifts were attributed to the coupling of CO2 to Q2¯following the
reactions:257
Q + e ¯ → Q•¯ (3)
Q•¯ + e ¯ → Q2¯ (4)
Q2¯+ pCO2 → Q (CO2) p 2¯ (5)
The reaction of Q2¯with CO2 (eqn. (5)) is further supported by the disappearance of the Q2¯
oxidation peak, indicating the formation of a different reduction product rather than Q2¯. Overall,
the observed voltammogram is similar to that of an EC reaction, i.e., electrochemical reduction
followed by a fast chemical step.258
Figure IV. 15 Cyclic voltammograms of quinones (2.5 mM AQ (top), 5 mM PAQ (middle)

and 5 mM DBBQ (bottom)) in Ar - (red curves) and CO2 - (blue curves) saturated DMF containing
0.1 M TBAClO4 (a), and 0.1 M LiClO4 (b), respectively. Scan rate: 100 mV s-1. 2.5 mM AQ was
used due to the solubility limit.
The positive shifts measured for the second reduction waves (∆E2) following the chemical
reactions of Q2¯ with CO2 (eqn. (5)) are reported in Table A. 4. 1. As previously established, a
higher ∆E2 value is associated with greater association constant and stoichiometric constant (p) for
CO2 reaction.257 Among all the three quinones studied in this work, the highest ∆E2 value is
measured for DBBQ, indicating its greatest CO2 reactivity (affinity) which can be explained by the
124
better stability of DBBQ-CO2 adduct (the term DBBQ-CO2 adduct was used here for the sake of
simplicity, although it was formed from DBBQ dianion and CO2) and its high aromaticity.
Moreover, the ∆E2 value for PAQ (ortho-quinone) was smaller than those for AQ and DBBQ (para-
quinones) possibly due to steric hindrance and/or coulombic repulsion induced by the two oxygens
in ortho positions of PAQ dianion.
To gain deeper insights into the reaction rate of Q2¯and CO2, voltammetry measurements at
various scan rates were performed (Figure IV. 16). Even at scan rate as high as 20 V s-1, no
oxidation peak could be recovered for the reverse reaction of Q•¯ + e¯ → Q2¯(eqn. (4)), indicating
a very fast chemical reaction between Q2¯and CO2. Unfortunately, the very fast nature of this step
prevents the estimation of its kinetic rate constant.
Figure IV. 16 Cyclic voltammograms in the presence of CO2 at various sweep rates for 5
mM DBBQ in 0.1 M TBAClO4/DMF. For all the measurements, the potential was held at -1.5 V vs
Ag/Ag+ for 20 s to intentionally produce the same amount of Q•¯.
Next, we went on to investigate the effects of adding a Lewis acid, namely Li+, on the
reduction of quinones. Let us first recall that Li+ was demonstrated, both experimentally and
theoretically, to form ion-pairs with both Q•¯ and Q2¯ as evidenced by shifting their redox
potentials to more positive values.259,260 Therefore, we first compared the voltammograms obtained
with argon-saturated electrolytes containing Li+ and TBA+, as bulky TBA+ does not form stable
chelates with organic compounds upon reduction. Similar to what is observed in TBA+-based
electrolytes, all quinones exhibit successively two reduction waves in the presence of Li+ (Figure IV.
125
17). However, when comparing their reduction potentials (Table A. 4. 2), positive shifts were
observed for the first redox waves for DBBQ and PAQ, whereas little change was found for AQ.
The difference in the extent of positive shift is related to the different strengths of ion-ion
interactions between Li+ and Q•¯.259,261 Indeed, the most pronounced positive shift measured for
PAQ is in agreement with its high Li-binding energy, as computationally studied elsewhere.253
Turning to the second reduction waves, positive shifts were found for all three quinones (Figure IV.
17). The extent of this shift is in general greater than those for the first waves, suggesting a stronger
Li+ interaction with Q2¯than with Q•¯. Alike what was observed for the first reduction wave, PAQ
dianion shows the strongest interaction with Li+ (Table A. 4. 2). Overall, we have shown that Li+
shifts the reduction potentials of quinones for each electron transfer via ion-ion interactions,
following the reactions:
Q + e ¯+ Li+ ↔ QLi (6)
QLi + e ¯+ Li+ ↔ QLi2 (7)

and 5 mM DBBQ (bottom)) in Ar- saturated DMF (a) and MeCN (b) containing 0.1 M TBAClO4
(orange curves), and 0.1 M LiClO4 (green curves). Scan rate: 100 mV s-1.
Having understood the interaction of Li+ and quinones, we finally explored the CO2 reaction
with quinones in the presence of Li+ (Figure IV. 15b). The onset reduction potentials and the peak
currents for the first reduction waves remained unaltered as compared to the results obtained under
argon, suggesting that CO2 does not react with Q•¯ even with the addition of a Lewis acid.
126
Nevertheless, positive shifts were found for the second reduction waves for AQ and DBBQ, alike
what have been observed in TBA+-based electrolyte (Figure IV. 15a). After demonstrating that no
direct CO2 reduction occurs in the presence of Li+ without quinone (Figure A. 4. 8), we can
conclude that such positive shifts are indeed related to the chemical reaction between Q2¯-Li+
complex and CO2. This reaction led to the formation of a new stable adduct, as suggested by the
disappearance of the oxidation peak for Q2¯-Li+ complex and the appearance of an oxidation peak
at more positive potential (especially in the case of DBBQ). Step-wise voltammetry further
confirmed that the new oxidation peak appeared at -0.55 V vs Ag+/Ag (in the case of DBBQ) was
indeed related to the reaction between Q2¯-Li+ complex and CO2 (Figure IV. 18). Notably, very
little change in the voltammogram was observed for PAQ after adding CO2 (Figure IV. 15b),
indicating a weak interaction between PAQ2¯-Li+ complex and CO2, likely due to the very strong
interaction between PAQ2¯and Li+.
Figure IV. 18 Stepwise cyclic voltammograms of 5 mM DBBQ in 0.1 M LiClO4/ DMF in

the presence of CO2. Scan rate: 100 mV s-1.
Until now, we have demonstrated that AQ and DBBQ are capable to react with Li + and CO2
during their second electron transfer and results in a new adduct formation. Worth mentioning is
that similar reductive currents were measured at the end of anodic scan in both CO2- and argon-
saturated electrolytes, suggesting negligible “surface blocking” effect (negligible surface deposition
of insulating products).
127
In light of the important role played by the interaction between Li+ and quinone dianions on
the CO2 reaction, we carried out similar study in MeCN, which is known to have weaker Li+ and
slightly stronger anion solvation ability (Table IV. 2), as suggested from their donor number (DN)
and acceptor number (AN), respectively.
Table IV. 2 Summary of the intrinsic properties of DMF and MeCN.
Solvent DN AN ε CO2 solubility 105 * DCO2 Viscosity
(Kcal mol-1)84 (Kcal mol-1)84 25 ℃ (mM)217 (cm2 s-1)217 Ƞ (cP)
DMF 26.6 16 36.7 198 3.56±0.2 0.92
MeCN 14.1 18.9 36.64216 280 38.3±1.1 0.361216
We first considered the influence of the solvents on the reduction of quinones in TBA +-
based electrolytes (Figure IV. 19a). A shift towards more positive potentials for the two reduction
waves typical for quinones were observed in MeCN as compared to DMF (Table A. 4. 3), which
can likely be explained by the fact that MeCN solvates stronger, and therefore stabilize more, the
anions.262 In CO2-saturated solutions (blue curves in Figure IV. 19a), little change was found for the
onset reduction potentials and peak intensities for the first reduction waves, but positive shifts were
observed for the second reduction waves, revealing the chemical reaction of Q2¯ with CO2. The
extent of positive shifts were in general more pronounced in MeCN than in DMF, indicating that a
more stable adduct of Q2¯ and CO2 was formed in MeCN (Table A. 4. 3). Notably for PAQ and
DBBQ, the addition of CO2 causes the disappearance of the oxidation peaks for Q2¯and the growth
of new oxidation peaks at more positive potentials, suggesting the formation of quinone─CO 2
adducts. By comparing the potential of such new oxidation peaks, which can be seen as an indicator
of the stability of the quinone─CO2 adducts, DBBQ dianion was determined to exhibit the strongest
interaction with CO2 as deduced from its highest oxidation potential. Whereas for AQ, the constant
existence of Q2¯ oxidation peak at a more positive potential implies that the interaction of AQ
dianion with CO2 is the weakest.
128

and 5 mM DBBQ (bottom)) in Ar- (red curves) and CO2- (blue curves) saturated MeCN containing
0.1 M TBAClO4 (a), and 0.1 M LiClO4 (b), respectively. Scan rate: 100 mV s-1.
When comparing our results with those previously reported by Wrighton et al. for PAQ in
MeCN/0.1 M TBABF4 solution after the addition of 0.24 M CO2, differences arise. That is, only
one redox wave with increased reduction current density was observed in their study, which was
interpreted as the change from two successive one-electron processes to a two-electron process in
the presence of CO2.250 However, we found that such variations might result from the presence of
water in the electrolyte (Figure IV. 20), as similar results were obtained when MeCN was not
further dried and in which the water content is relatively high (without drying: 400 ppm, after
drying: < 20 ppm). The presence of water as a proton source shifts the second reduction wave to a
potential identical to that for the first electron transfer because of the strong interaction of Q2¯with
proton, i.e., strong H-bonding, as established elsewhere.263
129
Figure IV. 20 Cyclic voltammograms of 5 mM PAQ under Ar (red curves) and CO2 (blue
curves) in MeCN with a water content of 400 ppm (a) and below 20 ppm (b), respectively. Scan
rate: 20 mV s-1.
As previously stated, Li+ tends to interact with quinone anions through ion-ion pairing.
Given the weaker Li+ solvation strength by MeCN, a stronger Li+ pairing with quinone anions can
already be foreseen. Indeed, we observed in MeCN a greater influence of Li + on the redox behavior
of quinones (green lines in Figure IV. 17b). In DMF-based electrolytes, the two successive
reduction waves preserved with only their onset potentials shifting to more positive values (green
lines in Figure IV. 17a). In contrary, in MeCN-based electrolytes, all quinones lost their typical
feature of two well-defined redox waves. More specifically, the second reduction wave for AQ
shifted positively to such a large extent that the two redox waves became nearly indistinguishable;
only one redox wave at more positive potentials was observed for DBBQ; whereas PAQ exhibited
two reduction waves but almost no oxidation wave were found. PAQ will therefore not be further
studied because of its irreversible nature. Turning eventually to CO2-saturated electrolytes (blue
lines in Figure IV. 19b), one reduction wave was observed for both AQ and DBBQ with its onset
potential remain unchanged, suggesting that CO2 does not participate in the initial electrochemical
step. Interestingly, unlike for DMF electrolytes, a typical passivation behavior, i.e., the current
drops to nearly zero at the end of reduction, is observed in MeCN electrolytes suggesting the
precipitation of a solid product.
Li-CO2 cells were then assembled using in-situ pressure cells to gain more insights into the
reactions among Li+, quinone anions and CO2 in MeCN-based electrolytes. In the absence of
130
DBBQ, Li-CO2 cell exhibited negligible discharge capacity (black line in Figure IV. 21a),
consistent with the negligible current response previously measured with cyclic voltammetry
(Figure A. 4. 8b). In contrary, Li-CO2 cell with DBBQ discharged under the same condition
delivered a notably greater capacity of ~ 6300 mAh g-1 at a voltage plateau of 2.69 V vs Li+/Li (blue
curve in Figure IV. 21a), along with a clear pressure drop (blue curve in Figure IV. 21b),
confirming that the reduction of DBBQ in Li+-containing MeCN electrolyte is associated with the
consumption of CO2. At first glance, our results could naively lead us to conclude that DBBQ
significantly promotes the discharge in Li-CO2 battery. However, a capacity of ~ 7900 mAh g-1 was
obtained (red curve in Figure IV. 21a) when discharging the cell under argon, which is ~ 36 times
higher than the theoretical capacity for the reduction of DBBQ itself ( ~ 216 mAh g-1). This
discrepancy could be explained by the use of Li1-xFePO4 negative electrode (metallic Li was not
used because of its reactivity with MeCN). Indeed, given the high redox potential of Li1-xFePO4 (~
3.45 V vs Li+/Li), the reduced form of DBBQ generated at the positive electrode upon discharge
can diffuse to the negative electrode side to be reoxidized. Therefore, to avoid such “shuttle effect”
of quinones, a two compartment “Ohara” cell that integrates a lithium ion conductive glass ceramic
membrane was thereby implemented. Correspondingly, a capacity of ~ 34 mAh g-1 was obtained in
the cell discharged under argon (red curve in Figure IV. 21c), indicating that the crossover of
DBBQ was effectively avoided. In contrast, a greater capacity of ~ 160 mAh g-1 was achieved when
discharging the cell under CO2 (blue curve in Figure IV. 21c). A slight pressure drop corresponding
to the consumption of CO2 by DBBQ anions was detected (blue curve in Figure IV. 21d), while the
pressure was found constant in argon cell (red curve in Figure IV. 21d). Nevertheless, the small
discharge capacity delivered by this Li-CO2 cell does not allow for a reliable quantification of the
amount of CO2 consumed. Nevertheless, we could demonstrate the formation of Li2CO3 at the end
of discharge by XRD (Figure IV. 22). Accordingly, a high potential (above 4.2 V vs Li+/Li) typical
for the oxidation of Li2CO3 was observed on charge (Figure A. 4. 9). For Li-CO2 cell using AQ, a
lower discharge voltage associated with a small pressure drop were obtained compared to those
measured for DBBQ (Figures IV. 21e & 21f), in agreement with previous voltammetry study
showing smaller reduction current response (Figure IV. 19b), and no crystalline phase could be
assigned as the ultimate discharge product by means of XRD (Figure IV. 22).
131
Figure IV. 21 Galvanostatic discharge curves (a) and their corresponding pressure changes
(b) of Li batteries discharged in 0.1 M LiClO4 /MeCN without DBBQ under CO2 (black curve),
and with 5 mM DBBQ under Ar (red curves) and CO2 (blue curves) using a classical in-situ
pressure cell. Galvanostatic discharge curves (c, e) and their corresponding pressure changes (d, f)
of Li batteries discharged in 0.1 M LiClO4 /MeCN with 5 mM DBBQ (c, d) and 2.5 mM AQ (e , f)
under Ar (red curves) and CO2 (blue curves) using a two-compartment “oraha” cell with a ceramic
membrane to prevent the crossover of quinones.
132
Figure IV. 22 X-ray diffraction of electrodes after discharging in MeCN with 5mM DBBQ
under CO2 (red line) and Ar (blue line); and with 2.5 mM AQ under CO2 (black line) and Ar (green
line). The broad peak at around 26°is referred to carbon Super P.
In light of the fairly small discharge capacity obtained from the Li-CO2 cell using DBBQ as
a mediator, 1H NMR was therefore employed to understand i) if soluble products are formed and ii)
if degradation of electrolyte and/or DBBQ occurs (Figure IV. 23). Two signals at 6.47 ppm and
1.24 ppm were observed for the solution before discharge, corresponding to the aromatic protons
and the protons in the methyl groups of the tert-butyl in DBBQ, respectively (Figure IV. 23a). After
reduction (Figure IV. 23b), two distinct sets of 1H NMR signals were detected in the regions of 6.1 -
6.7 ppm (shaded blue and labeled a) and 2.8 - 4.5 ppm (shaded grey and labeled b). The two peaks
in region a around the signal for the aromatic protons of DBBQ can be assigned to the aromatic
protons of the reduced form of DBBQ. Nevertheless, one cannot rule out that one of these two
signals might arise from the aromatic protons of degradation products arising from the
decomposition of DBBQ. The three new peaks in region b were not observed in the solution before
reduction, indicating the formation of products with very different proton environment. Although
we failed to successfully assign these species using 2D correlation NMR spectroscopy (not shown
here), the use of liquid mass spectrometry suggested a quinone degradation mechanism associated
with a dimerization phenomenon coupled with the loss of oxygen and tert-butyl groups (Figure IV.
24). Moreover, knowledge from Li-O2 battery can be used to tentatively explain the formation of
these peaks. Indeed, redox-active electrolyte additives are susceptible to degrade during battery
133
operation and in some cases can induce degradation of other battery components (e.g., solvents).264
Hence, bearing in mind that reactive radical species, i.e., DBBQ anion radical, was formed after the
initial electron transfer, we can suspect at this stage that these species in region b are likely
originate from either the degradation of DBBQ itself or the nucleophilic attack of MeCN by DBBQ
anion radical. Finally, the singlet at 1.31 ppm was assigned to the protons in the methyl groups of
the tert-butyl group of the reduced DBBQ. Looking in more detail at the peaks at 6.47 ppm and at
1.24 ppm which were still present after reduction, one can deduce that a small amount of DBBQ
remain unreacted. More importantly, the relative intensity measured for these two peaks clearly
increased after bubbling the solution with CO2 (Figure IV. 23c), which can be explained by the
regeneration of DBBQ after CO2 reaction, in a manner similar to the quinone mediated O2
reduction.89 No extra distinct signal and little change for the three peaks in region b were observed
after CO2 reaction, likely suggesting that the parasitic reactions are related to the initial
electrochemical reduction step. To conclude, NMR analysis revealed the formation of side products
during discharge which can possibly explain the limited capacity delivered in DBBQ mediated Li-
CO2 battery. This result further emphasizes the importance of the stability for both organic
electrolyte solvents and their additives in harsh electrochemical environments for potential Li-gas
battery technologies.
134
Figure IV. 23 1H NMR spectra of (a) d3-MeCN + LiClO4 + DBBQ, (b) d3-MeCN + LiClO4
+ DBBQ after reduction and (c) d3-MeCN + LiClO4 + DBBQ after reduction and CO2 reaction.
jmt-tba-dbbq-reduced-plus-co2 28/09/2017 10:26:42
RT: 23.68 - 26.67

25.46 25.48 NL:
100 5.05E8
25.57 TIC MS
90
25.60
Phtalate jmt-tba-
dbbq-
80 reduced-
plus-co2
70 25.70
Relative Abundance
25.76
60
25.80
25.91 26.06
50
26.22
40
25.24 26.29
25.14 26.38
30 24.02 24.98
23.70 24.19 24.38 24.48 24.69
20
10
0
23.8 24.0 24.2 24.4 24.6 24.8 25.0 25.2 25.4 25.6 25.8 26.0 26.2 26.4 26.6
Time (min)
jmt-tba-dbbq-reduced-plus-co2 #6234-6252 RT: 26.20-26.26 AV: 19 SB: 36 26.04-26.16 NL: 2.95E6

T: {0,0} + c EI Full ms [5.00-300.00]
177.14
Figure IV. 24 Possible degradation products detected by liquid mass spectrometry in the
100
90
solution of d3-MeCN
80 with DBBQ after reduction and 161.13
reaction with CO2. The solution was tested
70
after a short exposure to the air, hence the formation of hydroxides.
Relative Abundance
60
164.16
50
40 127.12
30
141.12 284.19
20 121.10
178.15
119.10
10
18.12 26.99 78.10 91.10 195.09 215.08
228.08 269.14
285.14 135
46.10 61.05 251.07
0
20 40 60 80 100 120 140 160 180 200 220 240 260 280
m/z
IV. 3. 3 Conclusion
Firstly, we revealed an intimate chemical interaction between CO2 and quinone dianions
based on a push-pull mechanism. The effects of the chemical structure of quinones, the addition of
Lewis acid cation (i.e., Li+), and the electrolyte solvents on such interaction have also been
considered. More specifically, weaker CO2 interaction is observed for ortho-quinones than para-
quinones because of the steric hindrance and coulombic repulsion induced by the two oxygens in
ortho positions. Li+ might impede the interactions between quinone and CO2 when strong ion
pairings are formed between Li+ and quinone anions. The interaction of quinone and CO2 can be
altered via tuning the solvation strength of quinone anions by the electrolyte solvent, e.g., a stronger
quinone interaction with CO2 is observed in MeCN than in DMF.
Secondly, by adding DBBQ into the MeCN electrolyte, we evaluated the concept of
quinone-mediated CO2 reduction in Li-CO2 batteries. The discharge voltage was found to be
dictated by the reduction potential of DBBQ, which is more positive than that of CO 2 reduction.
Moreover, Li2CO3 was identified as the ultimate discharge product. Nevertheless, the limited
discharge capacity together with NMR results suggested parasitic reactions involving DBBQ itself
and/or other cell components.
To conclude, the use of chemical catalysts is an effective approach to facilitate the

electrochemical reduction of CO2 in Li-CO2 batteries. However, to take advantage of such approach,
the catalyst should be properly selected to provide appropriate redox potential, proper catalytic
activity towards CO2 reduction, along with good electrochemical and chemical stability. Also of
importance is the anode side, which needs to be protected by a solid electrolyte membrane to avoid
the undesirable “shuttle effect”.
136
Chapter V General conclusion and outlook
The ever-growing energy demands of modern society have instigated intensive researches in
i) continuously pushing the energy limits of Li-ion batteries, especially at the positive electrode side,
and ii) exploring new battery chemistries with potentially higher energy storage. Li-rich layered
oxides are among the most promising positive electrode materials for high-energy Li-ion batteries,
provided their superior capacities due to the cumulative cationic and anionic redox processes.
Nevertheless, the activation of anionic redox generally induces structural instabilities associated
with irreversible O2 loss and cation migration, resulting in unfavorable performance decay upon
cycling. In this work, an electrochemical method, i.e., rotating ring disk electrode (RRDE)
voltammetry, was first introduced to follow the O2 evolution phenomenon in Li-rich layered oxides.
The dynamic structural evolutions linked with oxygen redox activities (both irreversible O2
evolution and reversible oxygen redox), along with their influence on the electrochemical properties,
were further investigated for a practically important Li1.2Ni0.13Mn0.54Co0.13O2 phase. At last, the
fundamental aspects of Li-CO2 battery, an alternative to the conventional Li-ion battery, was
explored.
The main contents and conclusions of this work include:
i) RRDE voltammetry was successfully developed as an alternative technique for

monitoring the O2 gas evolution from Li-rich layered oxides during high-potential charge process.
In a typical RRDE measurement, the oxide of interest was deposited onto the disk electrode where
its delithiation was controlled by cyclic voltammetry or galvanostatic cycling, whilst the ring
electrode was fixed at a potential at which the reduction of O2 could effectively takes place. Thanks
to the rotation of the whole RRDE assembly, the evolved O2 gas from the oxide can be transported
to and reduced (detected) at the ring electrode. As a proof of concept, we conducted RRDE
measurements using various Li-rich layered oxides, and the results were systematically compared
with those obtained by two other well-established gas/pressure analysis techniques, i.e., OEMS and
in-situ pressure cell. Through this comparative study, we could demonstrate that the proposed
RRDE method can reliably detect O2 while it has the advantage of being more sensitive, making it
suitable for determining the onset of O2 evolution and examining oxides that release a reduced
amount of O2. Moreover, owing to the simplicity of this approach, and the widespread of RRDE in
137
the fields of battery and electrocatalysis, we believe that it can be used for routine O2 analysis when
designing new materials that are prone to O2 release. Despite of its positive attributes, an obvious
optimization of RRDE method lies in the quantification of the total O2 amount generated from the
oxide. For that purpose, a calibration curve of the increase of reductive ring current as a function of
the O2 concentration at the disk surface may be obtained to deduce the “detection efficiency”.
RRDE voltammetry can also be extended to detect other type of gases such as CO2, provided that
the adequate ring electrode is selected.
ii) In light of previous works, O2 loss from Li-rich layered oxides may facilitate cation
migration and result in structure transformation. Therefore, we carried out a detailed study using a
stellar Li-rich compound with nominal composition of Li1.2Ni0.13Mn0.54Co0.13O2, in order to provide
more in-depth insights into the coupling between structural dynamics and oxygen redox. By
pushing the high-voltage oxidation as well as enlarging the discharge voltage window, we could
propose a more holistic scenario for the structural evolution of Li1.2Ni0.13Mn0.54Co0.13O2 during the
first electrochemical cycle.
Starting on oxidation, the delithiation occurs first through the oxidation of the Ni2+ and Co3+
cations until the removal of 0.4 Li+, followed by the oxidation of oxygen anions that triggers Mn-
migration until 4.6 V (ca. x < 1.0). Afterwards, the delithiation further proceeds concomitantly
with O2 evolution, creating both cationic and anionic vacancies that are filled by migrated Mn ions,
and eventually leading to the fully delithiated phase A’ (∆x = 1.15). Through an arsenal of
techniques including synchrotron X-ray and neutron diffraction, TEM, OEMS and ICP-OES, we
identified phase A’ as a densified O3 bulk structure enlisting Mn migration into the interlayer
octahedral sites, in contrast to previous literatures proposing it as a densified layer48,179 or spinel-
like structure157 which grows only at the surface. Combing in-situ laboratory XRD, OEMS and
RRDE voltammetry, we revealed the conjoint removal of Li+ and oxygen as the primary cause for
the formation of phase A’. In terms of electrochemical properties, pushing the high-voltage
oxidation to obtain the phase A’ induces a drop in both the voltage and capacity on the subsequent
discharge, which thus does not improve the long-term voltage fading. Several factors could
contribute to such performance decay, including Mn-migration, the presence of O1-like defects,
cation reduction (e.g., Co) driven by O2 loss etc. However, further investigations are needed to
138
understand better their correlation, which will also provide useful guidance for designing Li-rich
oxides with improved cycling performance.
Turning to the discharge (until 2.0 V), the anionic and cationic network of the phase A’ are
conjointly reduced along with only part of the migrated Mn ions returning to their initial positions,
forming a discharge phase P’ that differs from the pristine phase (ca. x = 1.03). By applying a
constant voltage step at 2.0 V or discharging to lower voltages (ca. 1.4 V), additional Li+ can be
inserted through a solid-solution process to reach the nearly fully lithiated phase P’ (ca. x = 1.15)
enlisting substantial amount of Mn-migration. This result reveals that a major part of the capacity
loss during the first cycle is actually of kinetic origin and linked to the sluggish Li+ diffusion at high
degree of lithiation. Upon further discharging to 1.2 V, the Li content of phase P’ exceeds that of
the pristine phase (ca. x = 1.43), leading to the growth of a new discharged phase P”. Phase P”
remains in the R3m space group with the same (ccp) oxygen stacking sequence pertaining to the
pristine phase, but with some of the interlayer tetrahedral sites filled by Li+. By gradually increasing
the depth of charge, as well as designing a path to restore the migrated Mn ions in the phase P’, we
demonstrated that the formation of phase P’’ is directly linked to the necessity of triggering the
oxygen redox activity on charge, more specifically, the Mn-migration triggered by the oxygen
redox. The cation mixing in the phase P” likely favors the stabilization of the oxygen sublattice,
preventing the layered to 1T phase transition upon overlithiation. Although we have showed the
positive effects of obtaining the fully lithiated phase P’ in recovering the first-cycle irreversible
capacity loss, application-wise, the exploitation of the low-voltage activity does not provide
substantial improvement regarding the energy density, whilst retains the voltage fade and hysteresis
pertaining to Li1.2Ni0.13Mn0.54Co0.13O2. In light of the recovery of voltage and the original stair-case
charge profile after restoring the cation ordering by the heat treatment, future research directions
could therefore be explored regarding the impact of cation disordering on the electrochemical
characteristics, and to design materials with reversible cation migration.
During the subsequent charge, phase P’ releases Li+ through a solid-solution process, which
differs from the two-phase process (A → A’ phase transition) on the first charge, leading to the
formation of phase A’ at the end of 2nd CC-CV charge. This finding suggests that although O2
evolution is critical to the initial formation of phase A’, once it forms, it can reversibly uptake and
release Li+ without the need for having more anionic vacancies.
139
iii) We investigated the fundamentals of Li-CO2 chemistries which were explored in the
battery community as an alternative end-use of CO2. Firstly, the electrolyte solvents were found to
be very crucial in dictating the reaction mechanisms of Li-O2/CO2 batteries. Two distinct discharge
reaction pathways were identified herein. Following the initial discharge step, which is the
electrochemical reduction of O2 into superoxide, the superoxide intermediate can either react with
Li+ to form Li2O2 that further chemically reacts with CO2 to give Li2CO3 (denoted as the chemical
pathway), or first react with CO2 and then Li+ to electrochemically form Li2CO3 (denoted as the
electrochemical pathway). Through a comparative study using three electrolyte solvents with
various physico-chemical parameters, we revealed that the competition between the two reaction
pathways depends strongly on the Li+ solvation, which is reflected by the donor number (DN) of the
solvent. In high DN solvent like DMSO, due to the strong solvation of Li+, the superoxide tends to
react with CO2; consequently, Li2CO3 is mainly formed through the electrochemical pathway.
Whereas in low DN solvent like DME, the reaction between superoxide and Li+ is more favored
and Li2CO3 is predominantly formed through the chemical pathway. Although the abovementioned
scenario rationalizes the major difference in the discharge processes of Li-O2/CO2 batteries, further
investigation into the electrochemical pathway are necessary to identify the reaction steps following
the reaction between the superoxide and CO2. Concerning the charge, Li-O2/CO2 batteries are not
rechargeable in their present form, as we found that only CO2 is evolved during the oxidation of
Li2CO3, and no O2. To overcome this limitation, interesting future directions might be i) tuning the
oxidation of Li2CO3 into a reversible manner; and ii) altering the discharge reaction so as to avoid
the formation of “troublemaker” Li2CO3.
Secondly, in contrasts with many literatures, our findings reveal that CO2 cannot be directly
reduced in aprotic solvents within their electrochemical window using carbon electrodes (DME,
DMSO and MeCN were investigated herein). Correspondingly, no electrochemical activity was
observed in pure Li-CO2 batteries. Future investigations are therefore necessary to explain such
discrepancy in electrochemical activity.
Finally, bearing in mind the inertness of CO2 in these solvents, we evaluated the use of
quinones as solution catalysts, aiming to facilitate the electrochemical reduction of CO2 in Li-CO2
battery. The concept was validated by the demonstration of a DBBQ-mediated Li-CO2 battery with
Li2CO3 as the ultimate discharge product. Nevertheless, the limited discharge capacity together with
140
NMR analysis highlights the degradation of DBBQ itself and/or other cell components under
operative conditions. These results emphasize that the choice of catalyst is critical. Hence, a suitable
solution catalyst should provide appropriate redox potential, proper catalytic activity towards CO2
reduction, along with good electrochemical and chemical stability. To take advantage of this
mediator approach, it is also essential to protect the anode side by a solid electrolyte membrane to
avoid the undesirable “shuttle effect”.
141
Appendix
Appendix
Appendix
A. 1 Design of experimental equipment
A. 1.1 Gas filling station
Figure A. 1. 1 Schematic of the gas filling station developed in this Ph.D. work which
enables a purging/refilling of the in-situ pressure cell and the electrochemical cell located inside the
glovebox with desired gases.
A. 1. 2 Description of OEMS systems
The closed OEMS system used in this work comprises an electrochemical cell (with a
detailed description in Chapter II. 2. 2), a BioLogic galvanostat /potentiostat and an ExQ gas
analysis system (Hiden Analytical, USA). The ExQ system is composed of a HAL series
quadrupole mass spectrometer, an Ultra-High Vacuum (UHV) mass spectrometer vacuum chamber,
a vacuum pumping system and a QIC series capillary inlet. The whole system is controlled by a
MASsoft 7 professional software.
136
Appendix
Figure A. 1. 2 Schematic and photograph of the closed OEMS system.
Figure A. 1. 3 Schematic and photograph of the open OEMS system developed in this Ph.D.
work.
137
Appendix
Figure A. 1. 4 Schematic and photograph of the electrochemical cell dedicated to the open
OEMS system.
A. 2 Supporting information for Chapter II
A. 2. 1 Methods
Materials. Li1.2Ni0.13Mn0.54Co0.13O2 was prepared by co-precipitation combined with solid-

state synthesis,43 Li2Ru0.75Ti0.25O3 and Li2Ru0.5Ti0.5O3 were prepared by classical solid-state
synthesis, as described in details elsewhere.154 LiNi1/3Mn1/3Co1/3O2 (> 98%) and Carbon Super P
(Csp) were purchased from Sigma-Aldrich and MTI Corporation, respectively. LP30 electrolyte
(99.9 %) made of 1 M LiPF6 dissolved in 1:1 v/v Ethylene carbonate/Dimethyl carbonate was
purchased from Solvionic, France.
Preparation of self-standing Li1-xFePO4 film. Self-standing Li1-xFePO4 film is composed

of 85 wt.% (by weight) LiFePO4, 10 wt.% Csp and 5 wt.% polytetrafluoroethylene (PTFE).151 A
blend of LiFePO4 and Csp was first mix in a SPEX 8000M ball-miller for 15 min before adding
PTFE (60 wt.% dispersion in H2O, Sigma-Aldrich). After several minutes’ hand grinding, a few
drops of isopropanol (Sigma-Aldrich) were gradually added into the mixture and further hand
grinded before obtaining one piece of elastic dough. This dough was tape-casted into a 0.5 mm
thick film and further punched into 11 mm diameter self-standing disks which were washed with a
water/ethanol solution before being dried overnight at 80 °C under vacuum in BÜCHI oven and
138
Appendix
transferred into the glovebox. Every single as-prepared LFP electrode (60 to 65 mg per cm2) was
then pre-charged in a Swagelok-type cell with metallic lithium as counter electrode and LP30 as
electrolyte until it reached a constant plateau with a stable potential of 3.45 V vs Li/Li+. To recover
the pre-charged Li1-xFePO4 electrode, the Swagelok cell was disassembled in the glovebox, washed
with dimethylcarbonate at least three times and dried under vacuum.
RRDE setup. All RRDE experiments were performed in an argon filled glovebox (water
content < 0.5 ppm, oxygen content < 0.1 ppm) with a five-neck glass cell sealed with rubber
stoppers, except a bubbler which was connected to a gas line that allows feeding the cell with dry
argon (5.0 quality, Linde France), and a needle which was inserted into the cell as a gas outlet. The
RRDE tip consists of an inter-changeable glassy carbon disc of 5.0 mm in diameter and a platinum
ring with an internal diameter of 6.5 mm and an external diameter of 7.5 mm (E6 series, Pine
Research Instrumentation, USA). Prior to its use, both platinum ring and glassy carbon disk were
polished sequentially with alumina suspensions of 0.5, 0.3, 0.05 μm (Allied Tech Products Inc.,
USA), followed by sonication with DI water and an electrochemical cleaning procedure in 0.5 M
H2SO4 solution until typical characteristic cyclic voltammograms were obtained.265,266 The
reference electrode (Ag+/Ag, RE-7, ALS Co., Ltd) consists of a silver wire, a glass tube filled with
a MeCN/0.1 M TBAClO4/0.01 M AgNO3 solution (ALS Co., Ltd) and sealed with a Vycor 7930
frit (ALS Co., Ltd). The reference electrode was assembled in the glovebox at least 24 hours before
use and was stored partially immersed in a MeCN/0.1 M TBAClO4 solution. Self-standing Li1-
xFePO4 film was used as the counter electrode. The LP30 electrolyte with water content less than 10
ppm, as deduced by Karl Fischer titration (Coulometric KF titration, Metrohm), was first purged
with argon. The typical volume of electrolyte used was 10 mL for each experiment. The potential
difference between Ag+/Ag reference electrode and Li+/Li redox potential was measured to be 3.19
V in LP30, and all potentials in this work were corrected using this potential difference. All the
experiments were operated on an AFMSRCE rotator (PINE Research Instrumentation, USA) and a
VMP3 Bipotentiostat (BioLogic Science Instruments, France).
Fabrication of layered oxide casted disk electrodes. A blend of active material and Csp at
a 4:1 mass ratio was hand-grinded with N-methyl pyrrolidine (NMP)/poly (vinylidene fluoride)
(PVDF) solution for 10 minutes and then stirred for another 1 hour, yielding a slurry with the final
concentrations of 10 mgoxide mLslurry-1, 2.5 mgCsp mLslurry-1 and 1.25 mgPVDF mLslurry-1. Next, 10 µL of
139
Appendix
as-prepared slurry was drop-casted onto a 0.196 cm²glassy carbon disk. The oxide layer on the
glassy carbon disk was dried under vacuum at room temperature for 1 hour prior to be dried further
for 6 hours at 55 °C under vacuum in BÜCHI oven. The electrode had a final composition of 100
µgoxide, 25 µgCsp, and 12.5 µgpvdf. The uniformity of such as-prepared electrodes was verified by
scanning electron microscopy (SEM) (Figure A. 2. 1). The Csp electrode was prepared in a similar
manner, except that the slurry was obtained with a mixture of Csp and PVDF at a 6:4 mass ratio.
The collection efficiency, N, of the RRDE geometry remained unchanged on a Csp supported disk
at 25.05% 0.15 as evaluated by the redox reaction of ferrocene/ferrocenium (K3Fe(CN)6 was not
used here due to solubility limitation in LP30 electrolyte) (Figure A. 2. 2). Similar collection
efficiencies were obtained on both bare and Csp-supported disk electrodes which 1) justify the
electrode preparation method proposed here and 2) demonstrate that the low collection efficiency
found in the present work is originating from gas solubility issues and not from a modification of
the laminar flow by the particles.
Electrochemical cycling of in-situ pressure cell and OEMS cell. All electrochemical tests
were performed with BioLogic galvanostat/potentiostats (BioLogic Science Instruments, France). A
typical electrochemical cell is made with metallic lithium foil counter electrode (Sigma-Aldrich),
two layers of glass fiber separators (Whatman GF/D, United Kingdom) soaked with LP30
electrolyte and positive electrode in powder form. The positive electrode is composed of active
materials and Csp at a 4:1 mass ratio. Prior to be cycled, the pressure cells were kept under open-
circuit voltage at 25 °C for at least 10 h in a temperature-controlled incubator (IPP260, Memmert)
in order to stabilize the internal cell pressure. After assembly, the OEMS cells were flushed and
pressurized with pure argon to avoid any contamination from the glove box atmosphere. Prior to
perform the galvanostatic cycling, the OEMS cells were typically held at open circuit voltage for
approximately 1.5 hours to reach a gas-liquid equilibrium phase inside the cell, and therefore to
obtain a stable baseline value for all the partial pressure signals. During the OEMS measurements,
the internal cell pressure was continuously measured by the pressure sensor and the produced
gaseous species were continuously sampled from the cell head space to the mass spectrometer via a
thin capillary (1/16″ diameter) at a leak rate of 12.5 µL min-1. The partial pressures are eventually
determined for each gas based on their mass to charge ratio (m/z). After ionization in the ionization
source of the MS, separation in the mass analyzer and further detection in the ion detector, the
partial pressures at m/z = 32, m/z = 44 were used to determine the evolution of O2 and CO2,
140
Appendix
36
respectively. All the partial pressure signals were normalized to the partial pressure of the Ar
isotope to correct for fluctuations of pressure. The specific gas evolution rate is obtained by
processing the first order derivative of the normalized partial pressure vs time profile.
A. 2. 2 Supporting figures
Figure A. 2. 1 Scanning electron microscopy (SEM) images of Li2Ru0.5Ti0.5O3-casted disk

electrode at various resolutions, images of two electrodes were presented to show the
reproducibility: electrode 1 (a ~ c), electrode 2 (d ~ f).
141
Appendix
Figure A. 2. 2 Ring and disk currents for the determination of the collection efficiency, N,
on bare (a) and Csp-casted (c) glassy carbon disk electrodes in LP30 electrolytes with 10mM
Ferrocene, positive scan at 10 mV s-1, Ering = -0.4 V vs Ag+/Ag. Ring currents as a function of disk
currents and the N value of RRDE tips with bare (b) and Csp-casted (d) glassy carbon disk
electrodes, respectively.
142
Appendix
Figure A. 2. 3 Differential capacity (dQ/dV) vs V plots (bottom panels) and their respective
pressure changes (top panels) for Li-rich NMC (a) and NMC 111 (b) derived from the galvanostatic
charging curves in pressure cells charged at a current rate of 0.3 C to 4.8 V vs Li+/Li. RRDE
profiles for Li-rich NMC (c) and NMC 111 (d) measured in LP30 electrolyte. Bottom panels
present the linear scan voltammograms obtained in the disk electrodes at a scan rate of 0.1 mV s-1 to
4.8 V vs Li+/Li. Top panels illustrate their corresponding ring current responses.
143
Appendix
Figure A. 2. 4 Two identical RRDE experiments for (a) Li2Ru0.75Ti0.25O3 and (b)
Li2Ru0.5Ti0.5O3, showing fairly reproducible electrochemical performance in the disk (bottom panel),
and ring current response (top panel, corresponding to O2 detection).
A. 2. 3 Calculation of the quantity of released O2 gas
1, Li2Ru0.5Ti0.5O3 → Li2-xRu0.5Ti0.5O3-2y + x e¯+ x Li+ + y O2
Total charge detected in the ring: 3.04177 e-4 mAh
Collection efficiency: 25 %
Total charge compensated by oxygen redox: 12.16708 e-4 mAh
Total amounts of O atoms loss: 0.0227 µmol
Total O atoms in the material: 2.2 µmol
Percentage of O loss from Li2Ru0.5Ti0.5O3: 1.036 %
2, Li2Ru0.75Ti0.25O3 → Li2-xRu0.75Ti0.25O3-2y + x e¯+ x Li+ + y O2
Total charge detected in the ring: 1.66337 e-4 mAh
Collection efficiency: 25 %
144
Appendix
Total charge compensated by oxygen redox: 6.65348 e-4 mAh
Total amounts of O atoms loss: 0.0123135 µmol
Total O atoms in the material: 2 µmol
Percentage of O loss from Li2Ru0.75Ti0.25O3: 0.62 %
Table A. 2. 1 The final composition for Li-rich compounds after the initial charge calculated
based on RRDE and pressure measurements
oxides RRDE Pressure cell
25% 20% 10% 5%

Li2Ru0.5Ti0.5O3 LiRu0.5Ti0.5O2.97 LiRu0.5Ti0.5O2.96 LiRu0.5Ti0.5O2.94 LiRu0.5Ti0.5O2.88 Li0.65Ru0.5Ti0.5O2.74
Li2Ru0.75Ti0.25O3 Li0.5Ru0.75Ti0.25O2.98 Li0.5Ru0.75Ti0.25O2.97 Li0.5Ru0.75Ti0.25O2.95 Li0.5Ru0.75Ti0.25O2.9 Li0.25Ru0.75Ti0.25O2.69
Li-rich NMC Li0.262(NMC)O1.954 Li0.262(NMC)O1.94 Li0.262(NMC)O1.88 Li0.262(NMC)O1.76 Li0.215(NMC)O1.76
145
Appendix
A. 3 Supporting information for Chapter III
A. 3. 1 Methods
Material synthesis and characterization. Li1.2Ni0.13Mn0.54Co0.13O2 powder was

synthesized by a two-step process involving the co-precipitation of carbonate precursor followed by
two annealing steps.43 Synchrotron X-ray diffraction (XRD) measurements were performed on the
11-BM beamline of the Advanced Photon Source at Argonne National Laboratory, with a
wavelength of 0.4128 Å. Neutron powder diffraction pattern was collected with the high resolution
powder diffractometer SPODI in Forschungs-Neutronenquelle Heinz Maier-Leibnitz (FRM II) at
Technical University of Munich with a wavelength of 1.5482 Å. In-situ XRD measurements were
carried out using an airtight electrochemical cell equipped with a Beryllium window in a BRUKER
D8 Advance diffractometer with Cu K radiation source ( K1 = 1.54056 Å,  K2 = 1.54439 Å).
Electron diffraction (ED) patterns, high angle annular dark field scanning transmission electron
microscopy (HAADF-STEM) images, annular bright field scanning transmission electron
microscopy (ABF-STEM) images were obtained with an aberration-corrected Titan G3 electron
microscope operated at 200 kV. TEM sample was prepared in an Ar-filled glove box by crushing
the crystals in a mortar with dimethyl carbonate and depositing drops of suspension onto holey
carbon grids. A specialized Gatan vacuum transfer holder was also used for the TEM analysis.
Electrochemical tests. All the electrochemical characterization was performed in CR2032-

type coin cells, which were assembled in Ar-filled glovebox with metallic lithium as negative
electrode and two glass-fiber separators (Whatman, United Kingdom) soaked with LP30 electrolyte
(1 M LiPF6 dissolved in 1:1 v/v ethylene carbonate/dimethyl carbonate, Solvionic, France). A blend
of Li1.2Ni0.13Mn0.54Co0.13O2 and carbon super P loose powder at a 4:1 mass ratio was hand-ground
for 15 min before using as the positive electrode. Galvanostatic cycling tests were performed at C/5,
unless specified otherwise. All the ex-situ samples were prepared using Swagelok-type cells with
similar cell stacking described above. The as-recovered samples were washed with dimethyl
carbonate for at least three times before being dried under vacuum. The near - equilibrium voltage
profiles of the first discharge were obtained using galvanostatic intermittent titration technique
(GITT, every ∆x = 0.05) with the relaxation process controlled by either dV/dt 0.01 mV s-1 or
146
Appendix
time = 4 h. The Ohmic resistance was deduced from the voltage jump in a short time (1 s) after the
application of the current pulse.
Gas analyses. A homemade cell was developed for online electrochemical mass
spectrometry (OEMS) experiments as described elsewhere.267 A slurry consisting of 72 wt.% of
cathode active material, 18 wt.% of carbon Super P, and 10 wt.% of Polyvinylidene
fluoride or polyvinylidene difluoride (PVDF) binder was coated on one side of a Celgard 2400
separator (22 mm in diameter) with a wet thickness of 200 μm. Lithium discs with 20 mm in
diameter were used as the counter and reference electrode with 120 μL of LP30 electrolyte.
Rotating ring disk electrode (RRDE) measurements were performed with a four-electrode cell
configuration as described in Chapter A. 2. 1.145 The glassy carbon disk electrode was coated by
slurry with the final concentrations of 10 mgoxide mLslurry−1, 2.5 mgCsp mLslurry−1 and 1.25 mgPVDF
mLslurry−1. The platinum ring electrode was kept at a potential sufficient for oxygen reduction
reaction (i.e., 2.5 V). The reference electrode (Ag+/Ag, RE-7, ALS Co., Ltd) consists of a silver
wire, a glass tube filled with a MeCN/0.1 M TBAClO4/0.01 M AgNO3 solution (ALS Co., Ltd) and
sealed with a Vycor 7930 frit (ALS Co., Ltd). Self-standing Li1-xFePO4 film was used as the counter
electrode following a preparation method as described in Chapter A. 2. 1.151
X-ray absorption spectroscopy (XAS). Self-standing films consisting of 74 wt.% of

Li1.2Ni0.13Mn0.54Co0.13O2, 18 wt.% of carbon Super P, and 8 wt.% of Polytetrafluoroethylene (PTFE)
were used and cycled to the desired state of charge or discharge. After washing with dimethyl
carbonate and drying under vacuum, the cycled electrodes were then placed between two layers of
Kapton tape and sealed under argon atmosphere in an air-tight transparent plastic bag. Ex-situ XAS
measurements at the three transition-metal (Ni, Co, Mn) K-edge were performed in transmission
mode at the ROCK beamline of synchrotron SOLEIL (France).
X-ray photoelectron spectroscopy (XPS). Loose powder consisting of 90 % of

Li1.2Ni0.13Mn0.54Co0.13O2 and 10 % of carbon Super P was used and cycled to the desired state of
charge. XPS measurements were carried out with a Kratos Axis Ultra spectrometer, using focused
monochromatic Al K radiation source (hν = 1486.6 eV). The XPS spectrometer was directly
connected through a transfer chamber to an argon dry box, to avoidmoisture/air exposure of the
samples. The analyzed area of the samples was 300 × 700 µm2. Peaks were recorded with constant
147
Appendix
pass energy of 20 eV. For the Ag 3d5/2 line, the full width at half maximum (FWHM) was 0.58 eV
under the recording conditions.
Mild-temperature annealing of the cycled Li1.2Ni0.13Mn0.54Co0.13O2 electrodes. Loose

powder of Li1.2Ni0.13Mn0.54Co0.13O2 and carbon super P at a 4:1 mass ratio was used as positive
electrode, and charged to 4.6 V prior to be discharged 2.0 V in a Swagelok-type cell. The cycled
samples were recovered in Ar-filled glovebox, washed with dimethyl carbonate for three times and
dried under vacuum before annealing under Ar atmosphere at 250C for 1 h. After the annealing,
the as-obtained powders were assembled in a CR-2032 type coin cell against metallic lithium to
evaluate the impact of annealing on their electrochemical properties.
Figure A. 3. 1 Selective region of the laboratory XRD patterns for ex-situ samples charged
to various states (i.e., the amount of Li removal ∆x = 0.6 (black curve), 0.8 (blue curve), and 1.0
(yellow curve)) and followed with an OCV step for 120 h. The more Li1.2Ni0.13Mn0.54Co0.13O2 is
delithiated, the more growth of phase A’.
148
Appendix
Figure A. 3. 2 RRDE profiles measured for the first charge of Li1.2Ni0.13Mn0.54Co0.13O2 to

4.8 V, following with an OCV step. Top panel presents the galvanostatic charge curves obtained in
the disk electrode. Bottom panel illustrates the corresponding ring current responses. Dash black
line marks the onset of O2 loss that is located at the end of charge (ca. 4.6 V). Filled blue area in the
bottom panel reveals that O2 continues to evolve during the OCV.
149
Appendix
Figure A. 3. 3 Difference-Fourier Maps generated from the neutron (a) and synchrotron (b) XRD
patterns of the A’ phase. The difference was calculated after having removed all the transition
metals from the structure (Ni, Mn and Co), clearly showing density on the octahedral sites, and not
on the tetrahedral sites. Panel (c) presents a view of the structure as deduced from the combined
refinement of the neutron and XRD patterns (oxygen is red, Mn is purple, Li is green, Ni is gray, Co
is dark blue).
150
Appendix
Figure A. 3. 4 ABF-STEM images demonstrating local oxygen environment of the TM

cations in the tetrahedral and octahedral positions.
Figure A. 3. 5 Selective region of the laboratory XRD patterns for ex-situ samples charged
to 4.8 V with an OCV step for 0.5 h (black line, denoted as CC - 0.5 h OCV) and 120 h (blue line,
denoted CC - 120 h OCV), or a CV hold for 8 h (red line, denoted as CC-8 h CV). A mixture of
phase A and phase A’ are obtained after 0.5 h OCV, whereas phase A’ is the dominate phase after
120 h OCV storage or 8 h CV hold. Rietveld refinement of the synchrotron XRD pattern of CC -
120 h OCV charged phase is shown in Figure A. 3. 7d, with corresponding crystallographic
parameters reported in Table A. 3. 1.
151
Appendix
Figure A. 3. 6 Galvanostatic charge-discharge voltage curves of Li1.2Ni0.13Mn0.54Co0.13O2.

The cell has been firstly charged to 4.8 V followed by an OCV step for 120 h, then recharged to 4.8
V. Very small amount of Li+ (~ 0.04 Li+) was extracted after 120 h OCV, suggesting the reinsertion
of Li+ during the OCV period is negligible.
152
Appendix
Figure A. 3. 7 Rietveld refinements of the synchrotron XRD patterns of

Li1.2Ni0.13Mn0.54Co0.13O2 collected at their charged states with the amount of Li+ removal (∆x)
ranging from 0.4 (a), 0.6 (b), 0.8 (c), and 1.0 (d), followed by 120 h OCV. The red line, black line
and bottom green line represent the observed, calculated XRD patterns and the difference,
respectively. Vertical blue tick bars mark the Bragg positions in the R3̅m space group.
153
Appendix
Table A. 3. 1 Crystallographic data and atomic positions of the ∆x = 1.0 (CC - 120 h OCV)
charged phase determined from Rietveld refinement of the synchrotron XRD pattern.
Table A. 3. 2 Crystallographic data and atomic positions of the ∆x = 0.4 charged phase
determined from Rietveld refinement of the synchrotron XRD pattern.
154
Appendix
155
Appendix
Figure A. 3. 8 Rietveld refinement of the synchrotron XRD pattern of

Li1.2Ni0.13Mn0.54Co0.13O2 charged to 4.8 V and then discharged to 2.0 V. The red line, black line and
bottom green line represent the observed, calculated XRD patterns and the difference, respectively.
Vertical blue tick bars mark the Bragg positions in the R3̅m space group.
Table A. 3. 5 Crystallographic data and atomic positions of the sample collected at 2.0 V
discharge state after 4.8 V charge determined from Rietveld refinement of the synchrotron XRD
pattern.
156
Appendix
Figure A. 3. 9 Galvanostatic discharge-charge curves collected during the first discharge

(down to 1.4 V) and 2nd charge of Li1.2Ni0.13Mn0.54Co0.13O2/Li cell after the initial 4.8 V charge and
the corresponding in-situ X-ray diffraction patterns.
Figure A. 3. 10 Galvanostaic discharge curve collected on discharging

Li1.2Ni0.13Mn0.54Co0.13O2 down to 0.5 V after 4.8 V charge (left side) and the corresponding in-situ
X-ray diffraction patterns. Blue and yellow lines show the results before and after the overshooting
point (marked by red arrow), respectively. Grey lines reveal that further pushing the discharge
down to 0.5 V results in a material amorphization.
157
Appendix
5.0
x = 0.8
x = 0.4
4.0
Voltage (V vs. Li+/Li)

3.0 x = 0.0
low-voltage activity associated with Mn
2.0 migration (activation of O redox on charge)
conversion-type reaction
leading to material amorphization
1.0
0.0
0.8 1.6 2.4 3.2
x in LixNi0.13Mn0.54Co0.13O2
Figure A. 3. 11 Voltage-composition profiles of Li1.2Ni0.13Mn0.54Co0.13O2 while varying the

lithium removal during charge (∆x) from 0.0 (pristine, grey line) to 0.4 (black line), and 0.8 (red
line), but fixing the discharge cutoff voltage at 0.5 V.
Table A. 3. 6 Crystallographic data and atomic positions of the 1.4 V discharged phase (P’)
158
Appendix
Table A. 3. 7 Crystallographic data and atomic positions of the 1.2 V discharged phase (P”)
Table A. 3. 8 Crystallographic data and atomic positions of the 4.6 V charged – 2.0 V
discharged phase determined from Rietveld refinement of the synchrotron XRD pattern.
159
Appendix
Table A. 3. 9 Crystallographic data and atomic positions of the annealed 4.6 V charged – 2.0
V discharged phase determined from Rietveld refinement of the synchrotron XRD pattern.
Figure A. 3. 12 Charge-discharge voltage profiles of the pristine Li1.2Ni0.13Mn0.54Co0.13O2

phase (first cycle, marine line), the 2.0 V discharged phase (second cycle, black line) and the
annealed 2.0 V discharged phase (second cycle, red line). All the cells were cycled at a current rate
of C/5.
160
Appendix
Figure A. 3. 13 Voltage-composition profiles of the annealed 2.0 V CV discharged sample

cycled between 4.6 V and 2.0 V.
Figure A. 3. 14 The initial charge-discharge curves of Li1.2Ni0.13Mn0.54Co0.13O2 mixed with

20 wt.% (red line) and 40 wt.% (blue line) carbon super P within a voltage window of 4.8 – 1.2 V at
a current rate of C/5.
161
Appendix
Figure A. 3. 15 Voltage-composition profiles of Li1.2Ni0.13Mn0.54Co0.13O2 (black line) and

Li1.0Ni0.33Mn0.33Co0.33O2 (red line) discharge to 1.2 V at C/5.
Figure A. 3. 16 Voltage-composition profiles of Li1.2Ni0.13Mn0.54Co0.13O2 discharge to 1.2 V

at C/5 (black line) and C/20 (yellow line).
162
Appendix
Figure A. 3. 17 In-situ XRD study of the first cycle and the second charge of
Li1.2Ni0.13Mn0.54Co0.13O2 .
163
Appendix
A. 4 Supporting information for Chapter IV
A. 4. 1 Methods
Chemicals. 1, 2-Dimethoxyethane (DME, 99.9 %, Sigma-Aldrich), dimethyl sulfoxide

(DMSO, ≥ 99.9 %, Sigma-Aldrich), diethylene glycol dimethyl ether (DGME, 99.5 %, Sigma-
Aldrich), tetraethylene glycol dimethyl ether (TGME, ≥ 99 %, Sigma-Aldrich), acetonitrile (MeCN,
anhydrous, 99.8 %, Sigma-Aldrich), deuterated acetonitrile (d3-MeCN, ≥ 99.8 %, Eurisotop) and N,
N-dimethylformamide (DMF, 99.8 %, sigma Aldrich) were dried with activated molecular sieves
(4 Å, Sigma-Aldrich) for at least 2 days before use. Lithium perchlorate (LiClO4, 99.99 %, Sigma-
Aldrich), tetrabutylammonium bis-trifluoromethanesulfonimidate (TBATFSi, ≥ 99.0 %, Sigma-
Aldrich) tetrabutylammonium hexafluorophosphate(TBAPF6, ≥ 99.0 %, Sigma-Aldrich),
tetrabutylammonium perchlorate (TBAClO4, ≥ 98.0 %, Sigma-Aldrich), 9, 10 anthraquinone (AQ,
analytical standard, sigma Aldrich), 9,10-phenanthrenequinone (PAQ, ≥ 99 %, Sigma Aldrich) and
2,5-di-tert-butyl-1,4-benzoquinone (DBBQ, 99 %, sigma Aldrich) were dried overnight at 80 °C
under vacuum in BÜCHI oven, so as to decrease the water content of the as-prepared electrolytes
(solvents and salts) < 20 ppm as deduced by Karl Fischer titration (Coulometric KF titration,
Metrohm). Glass microfiber filters (Whatman) were dried under vacuum at 220 °C overnight.
Cyclic voltammetry measurements. A PTFE embedded glassy carbon disk (5.0 mm diam,
Pine Research Instrumentation) was used as the working electrode unless otherwise specified and a
platinum wire was used as the counter electrode. The reference electrode (Ag/Ag+, RE-7, ALS Co.,
Ltd) consists of a silver wire, a glass tube filled with a MeCN/0.1 M TBAClO4/0.01 M AgNO3
solution (ALS Co., Ltd) and sealed with a Vycor 7930 frit (BioLogic Science Instruments).The
reference electrode was always assembled in the glovebox at least 24 hours before use and was
partially immersed in an MeCN/0.1 M TBAClO4 solution. All measurements were performed in a
glove box with water content < 0.5 ppm. The electrochemical cell was well sealed with rubber
stoppers except a bubbler which was connected to a gas line able to feed with dry Ar (5.0 quality,
Linde France), O2 (5.6 quality, Linde France), CO2 (5.3 quality, Linde France) or 70 % O2 / 30 %
CO2 mixture (Linde France). The water content of electrolytes is found to be < 40 ppm after cyclic
voltammetry measurements.
164
Appendix
Preparation of carbon positive electrodes. 90 wt.% Carbon Super P (Csp, Timcal Ltd) and
10 wt.% polytetrafluorethylene (PTFE, 60 wt.% dispersion in water, Sigma-Aldrich) were firstly
well mixed with an electrical blender in certain amount of isopropanol (≥ 99.9 %, Sigma-Aldrich)
to prepare a slurry. The as-prepared slurry was then drop-casted onto 11 mm diameter stainless steel
meshes (AISI 316, 325 mesh, 0.036 mm diam wire, Alfa Aesar). After being dried in ambient air
and washed with a water/ethanol (1:2 v/v) solution to remove the surfactants in the PTFE dispersion,
these positive electrodes were further dried overnight at 80 °C under vacuum in BÜCHI oven and
stored in glove box. Carbon Super P loading of the resulting electrodes is 0.36 (± 0.02) mg on a
surface of 0.4 (±0.04) cm2.
Preparation of Li1-xFePO4 negative electrodes. Self-standing composite Li1-xFePO4 (LFP)

electrodes (LFP: Super P: PTFE = 85:10:5) were used as lithium source for both electrolytes.
Detailed preparation method can be found in A. 2. 1. Every single as-prepared LFP electrode (60 to
65 mg per cm2) was then pre-charged in a Swagelok cell with metallic lithium as counter electrode
and LP30 as electrolyte until it reached a constant plateau before serving as an excess lithium
source with a stable potential of 3.45 V vs Li/Li+ for Li-O2, Li-CO2, Li-O2/CO2 or Li-Ar cells.
Cell testing. Li-O2/CO2 cells were tested with in-situ pressure cell so as to monitor
uninterruptedly the pressure change during discharge/charge. Li─CO2 cells were tested with a two
compartment “Ohara” in-situ pressure cell that integrates an additional ceramic membrane (Lithium
ion conductive glass ceramics (LICGC), Li1+x+yAlxTi2─xSiyP3─yO12, 200 µm thick, Ohara Inc) in
order to prevent the crossover of quinones. All the cells were assembled in an argon-filled glove
box (water content < 1 ppm, oxygen content < 1 ppm) with carbon positive electrode and Li1-
xFePO4 negative electrode. Two layers of glass fibers wetted with 450 ± 20 μL electrolytes were
used as separators. After assembly, the cells were quickly vacuumed and refilled with O2, CO2,
O2/CO2 mixture or Ar at a constant pressure of 1.55 ± 0.05 bars. Electrochemical tests were
performed with a VMP3 multichannel potentiostat (BioLogic Science Instruments) in a
temperature-controlled oven at 25.0 ± 0.05 °C. The current density and specific capacity are
normalized based on the mass of Carbon Super P.
Online electrochemical mass spectrometry (OEMS). The OEMS cell was first fully
discharged under O2/CO2 mixture before to be quickly vacuumed to remove all residual O2/CO2
165
Appendix
inside and refilled with pure Ar. Detailed description of the routine OEMS measurement can be
found in A. 2. 1.
Nuclear magnetic resonance (NMR). For NMR analyses, deuterated acetonitrile (d3-
MeCN) was used as the supporting solvent for the sake of better sensitivity. In order to generate
exclusively the reduced form of DBBQ, one cell was discharged in pure argon by first decreasing
its potential down to 2 V vs Li+/Li where the reduction takes place, and then hold this potential
until the reduction current gradually decreases close to zero. To differentiate the conditions of
electrolyte solutions after the initial electrochemical reduction of DBBQ and the follow-up CO2
chemical reaction, NMR analyses were performed with the solutions collected right after the
discharge and after further bubbling with CO2, respectively. All NMR spectra were recorded on a
Bruker 300 MHz spectrometer.
166
Appendix
Figure A. 4. 1 Galvanostatic discharge curves and pressure change of Li-CO2 batteries using
0.1 M LiClO4/DME in (a) and (b), 0.1 M LiClO4/DMSO in (c) and (d).Galvanostatic discharge
curves (e) and (f) pressure change of Li-argon batteries using 0.1 M LiClO4/DMSO as electrolyte.
The discharge current density is 100 mA g-1. Blue curve and red curve are representative of two
cells measured under the same condition.
167
Appendix
Figure A. 4. 2 Photos of separators after discharged in (a) Li-O2/CO2 battery down to 2 V vs

Li/Li+, (b) Li-CO2 battery down to 1 V vs Li+/Li and (c) Li-argon battery down to 1 V vs Li+/Li,
using 0.1 M LiClO4/DMSO as electrolytes, at a current density of 100 mA g-1.
Figure A. 4. 3 Galvanostatic discharge curves for Na-CO2 batteries with (a) 1 M

NaClO4/DGME and (b) 1 M NaClO4/TGME (blue curve and yellow curve are two cells measured
under the same condition) as electrolytes, using carbon super p as electrode; and with (b) 0.5 M
NaOTf /DGME as electrolyte, using GDL as electrode (black curve). The discharge current density
is 100 mA g-1.
168
Appendix
Figure A. 4. 4 Cyclic voltammograms at 50 mV s-1 with O2-saturated electrolytes of 0.1 M

TBAPF6/DMSO (dash curve) and 0.1 M LiClO4/DMSO (solid curve).
Figure A. 4. 5 (a) Cyclic voltammograms at 50 mV s-1 with various O2/CO2-saturated

solvents: 0.1 M LiClO4/MeCN (blue curve), 0.1 M LiClO4/DGME (black curve), and 0.1 M
LiClO4/TGME (yellow curve).
169
Appendix
Figure A. 4. 6 SEM images of electrodes fully discharged in pure O2 with an electrolyte of (a)
0.1 M LiClO4/DMSO and (b) 0.1 M LiClO4/DME, at a current density of 50 mA g-1. SEM images
of electrodes fully discharged in O2/CO2 mixture, with 0.1 M LiClO4/DMSO as electrolyte, at
current densities of (c) 100 mA g-1 and (d) 200 mA g-1.
170
Appendix
Figure A. 4. 7 SEM images of electrodes after fully discharged in (a) pure O2 and then
exposed in pure CO2; and (b) O2/CO2 mixture.
Table A. 4. 1 Positive shifts of the second reduction waves (∆E2) due to chemical reactions
of CO2 and Q2¯ in DMF and MeCN, respectively. ∆E2 were obtained by comparing the peak
positions of the second reduction waves in presence of argon and CO2, respectively.
∆E2 (mV)
AQ DMF 167
MeCN 331
PAQ DMF 131
MeCN 147
DBBQ DMF 180
MeCN 408
171
Appendix
Table A. 4. 2 Positive shifts of the reduction potentials for the first and second reduction
waves (∆E1 and ∆E2, respectively) in presence of Li+ in DMF and in MeCN, respectively.
∆E1 (mV) ∆E2 (mV)

AQ DMF 32 498
MeCN 215 NA
PAQ DMF 205 810
MeCN 408 NA
DBBQ DMF 57 659
MeCN 271 NA
0.2 a) DMF 0.2 b) MeCN
0.1 0.1
j (mA/cm )
j (mA/cm )
2
2
0.0 0.0
Ar Ar
-0.1 CO2 -0.1 CO2
-0.2 -0.2
-2.4 -1.6 -0.8 0.0 0.8 -2.4 -1.6 -0.8 0.0 0.8
+ +
E (V vs Ag /Ag) E (V vs Ag /Ag)
Figure A. 4. 8 Cyclic voltammograms at electrolytes with 0.1 M LiClO4 in DMF (a) and
MeCN (b) under argon (green curves) and CO2 (orange curves). Scan rate: 100 mV s-1.
172
Appendix
Table A. 4. 3 Half-wave potentials of the first and second reduction waves (E1 and E2,
respectively) for quinones in DMF and MeCN.
E1 (V) E2 (V)
AQ DMF -1.25 -2.05
MeCN -1.23 -1.88
PAQ DMF -1.01 -1.98
MeCN -0.97 -1.83
DBBQ DMF -1.03 -2.14
MeCN -1.01 -1.93
Figure A. 4. 9 Galvanostatic discharge-charge curve in 0.1 M LiClO4 /MeCN with the

addition of 5 mM DBBQ under CO2 in a two-compartment “oraha” cell with a ceramic membrane
to prevent the crossover of quinones.
173
Bibliography
Bibliography
Bibliography
1 International Energy Agency, O. IEA World Energy Outlook 2017. (2017).
2 Whittingham, M. S. & Jr., F. R. G. The Lithium Intercalates of the Transition Metal

Dichalcogenides. Mater. Res. Bull. 10, 363-371 (1975).
3 Whittingham, M. S. Electrical Energy Storage and Intercalation Chemistry. Science 192,

1126-1127 (1976).
4 Rao, B. M. I., Francis, R. W. & Christopher, H. A. Lithium-Aluminum Electrode. J.

Electrochem. Soc. 124, 1490-1492 (1977).
5 Wang, J., Raistrick, I. D. & Huggins, R. A. Behavior of Some Binary Lithium Alloys as
Negative Electrodes in Organic Solvent‐Based Electrolytes. J. Electrochem. Soc. 123, 457-
460 (1986).
6 Buiel, E. & Dahn, J. R. Reduction of the Irreversible Capacity in Hard‐Carbon Anode

Materials Prepared from Sucrose for Li‐Ion Batteries. J. Electrochem. Soc. 145, 1977-1981
(1998).
7 Han, Y. S., Yu, J. S., Park, G. S. & Lee, J. Y. Effects of Synthesis Temperature on the
Electrochemical Characteristics of Pyrolytic Carbon for Anodes of Lithium‐Ion Secondary
Batteries. J. Electrochem. Soc. 146, 3999-4004 (1999).
8 Magasinski, A. et al. High-Performance Lithium-Ion Anodes Using a Hierarchical Bottom-

Up Approach. Nat. Mater. 9, 353-358 (2010).
9 Poizot, P., Laruelle, S., Grugeon, S., Dupont, L. & Tarascon, J.-M. Nano-Sized Transition-
Metal Oxides as Negative-Electrode Materials for Lithium-Ion Batteries. Nature 407, 496-
499 (2000).
10 Chang, H. J. et al. Investigating Li Microstructure Formation on Li Anodes for Lithium

Batteries by in Situ 6Li/7Li NMR and SEM. J. Phys. Chem. C 119, 16443-16451 (2015).
11 Wang, X. et al. New Insights on the Structure of Electrochemically Deposited Lithium

Metal and Its Solid Electrolyte Interphases via Cryogenic TEM. Nano Lett. 17, 7606-7612
(2017).
12 Ding, F. et al. Dendrite-Free Lithium Deposition via Self-Healing Electrostatic Shield

Mechanism. J. Am. Chem. Soc. 135, 4450-4456 (2013).
13 Monroe, C. & Newman, J. Dendrite Growth in Lithium/Polymer Systems A Propagation

Model for Liquid Electrolytes under Galvanostatic Conditions. J. Electrochem. Soc. 150,
A1377-A1384 (2003).
174
Bibliography
14 Lin, D., Liu, Y. & Cui, Y. Reviving the Lithium Metal Anode for High-Energy Batteries.
Nat. Nanotechnology 12, 194-206 (2017).
15 Liang, X. et al. A Facile Surface Chemistry Route to a Stabilized Lithium Metal Anode. Nat.
Energy 2, 17119-17125 (2017).
16 Zhao, J. et al. Surface Fluorination of Reactive Battery Anode Materials for Enhanced
Stability. J. Am. Chem. Soc. 139, 1550-11558 (2017).
17 Li, L. et al. Self-Heating–Induced Healing of Lithium Dendrites. Science 359, 1513–1516

(2018).
18 Mizushima, K., Jones, P. C., Wiseman, P. J. & Goodenough, J. B. Li xCoO2 (0 < x < 1): A
New Cathode Material for Batteries of High Energy Density. Mater. Res. Bull. 15, 783-789
(1980).
19 Amatucci, G. G., Tarascon, J. M. & Kleinb, L. C. CoO2, The End Member of the LixCoO2
Solid Solution. J. Electrochem. Soc. 143, 1114-1123 (1996).
20 Delmas, C., Fouassier, C. & Hagenmuller, P. Structural Classification and Properties of the
Layered Oxides. Physica B+C 99, 81-85 (1980).
21 Delmas, C. et al. An Overview of the Li(Ni,M)O2 Systems: Syntheses, Structures and

Properties. Electrochim. Acta. 45, 243-253 (1999).
22 Dahn, J. R., Fuller, E. W., Obrovac, M. & Sacken, U. v. Thermal Stability of Li xCoO2,
LixNiO2 and λ-MnO2 and Consequences for the Safety of Li-ion Cells. Solid State Ion. 69,
265-270 (1994).
23 Shao‐Horn, Y. et al. Structural Characterization of Layered LiMnO2 Electrodes by Electron

Diffraction and Lattice Imaging. J. Electrochem. Soc. 146, 2404-2412 (1999).
24 Nitta, N., Wu, F., Lee, J. T. & Yushin, G. Li-Ion Battery Materials: Present and Future.
Mater. Today 18, 252-264 (2015).
25 Andre, D. et al. Future Generations of Cathode Materials: An Automotive Industry

Perspective. J. Mater. Chem. A 3, 6709-6732 (2015).
26 Choi, J. & Manthiram, A. Role of Chemical and Structural Stabilities on the

Electrochemical Properties of Layered LiNi1∕3Mn1∕3Co1∕3O2 Cathodes. J. Electrochem. Soc.
152, A1714-A1718 (2005).
27 Yoon, C. S. et al. High-Energy Ni-Rich Li[NixCoyMn1–x–y]O2 Cathodes via Compositional

Partitioning for Next-Generation Electric Vehicles. Chem. Mater. 29, 10436-10445 (2017).
175
Bibliography
28 Jung, R., Metzger, M., Maglia, F., Stinner, C. & Gasteiger, H. A. Oxygen Release and Its
Effect on the Cycling Stability of LiNixMnyCozO2(NMC) Cathode Materials for Li-Ion
Batteries. J. Electrochem. Soc. 164, A1361-A1377 (2017).
29 Xu, J., Lin, F., Doeff, M. M. & Tong, W. A Review of Ni-Based Layered Oxides for
Rechargeable Li-Ion Batteries. J. Mater. Chem. A 5, 874-901 (2017).
30 Radin, M. D. et al. Narrowing the Gap between Theoretical and Practical Capacities in Li-
Ion Layered Oxide Cathode Materials. Adv. Energy Mater. 7, 1602888-1602920 (2017).
31 Rozier, P. & Tarascon, J. M. Review—Li-Rich Layered Oxide Cathodes for Next-

Generation Li-Ion Batteries: Chances and Challenges. J. Electrochem. Soc. 162, A2490-
A2499 (2015).
32 James, A. C. W. P. & Goodenough, J. B. Structure and Bonding in Li2MoO3 and Li2−xMoO3

(0 ≤ x ≤ 1.7). J. Solid State Chem. 76, 87-96 (1988).
33 Rossouw, M. H. & Thackeray, M. M. Lithium Manganese Oxides from Li2MnO3 for

Rechargeable Lithium Battery Applications. Mater. Res. Bull. 26, 463-473 (1991).
34 Thackeray, M. M. et al. Li2MnO3-Stabilized LiMO2 (M = Mn, Ni, Co) Electrodes for

Lithium-Ion Batteries. J. Mater. Chem. 17, 3112–3125 (2007).
35 Jiang, J., Eberman, K. W., Krause, L. J. & Dahn, J. R. Structure, Electrochemical Properties,
and Thermal Stability Studies of Cathode Materials in the
xLi[Mn1∕2Ni1∕2]O2⋅yLiCoO2⋅zLi[Li1∕3Mn2∕3]O2 Pseudoternary System (x + y + z = 1). J.
Electrochem. Soc. 152, A1879-A1889 (2005).
36 Kalyani, P., Chitra, S., Mohan, T. & Gopukumar, S. Lithium Metal Rechargeable Cells
Using Li2MnO3 as the Positive Electrode. J. Power Sources 80, 103-106 (1999).
37 Park, Y. Structural and Electrochemical Properties of Li[CoxLi1/3-x/3Mn2/3-2x/3]O2 Cathode

Materials. Solid State Ion. 175, 305-309 (2004).
38 Numata, K., Sakaki, C. & Yamanaka, S. Synthesis and Characterization of Layer Structured
Solid Solutions in the System of LiCoO2–Li2MnO3. Solid State Ion. 117, 257–263 (1999).
39 Numata, K., Sakaki, C., Yamanaka, S. & Synthesis of Solid Solutions in a System of
LiCoO2-Li2MnO3 for Cathode Materials of Secondary Lithium Batteries. Chem. Lett. 26,
725-726 (1997).
40 Lu, Z., MacNeil, D. D. & Dahn, J. R. Layered Cathode Materials Li[NixLi1/3−2x/3Mn2/3−x/3]O2

for Lithium-Ion Batteries. Electrochem. Solid State Lett. 4, A191-A194 (2001).
41 Lu, Z. & Dahn, J. R. Understanding the Anomalous Capacity of Li/Li[NixLi1/3-2/3xMn2/3-

x/3]O2 Cells Using In Situ X-Ray Diffraction and Electrochemical Studies. J. Electrochem.
Soc. 149, A815-A822 (2002).
176
Bibliography
42 Kim, J. Synthesis and Electrochemical Behavior of Li[Li0.1Ni0.35−x/2CoxMn0.55−x/2]O2

Cathode Materials. Solid State Ion. 164, 43-49 (2003).
43 Pimenta, V. et al. Synthesis of Li-Rich NMC: A Comprehensive Study. Chem. Mater. 29,
9923-9936 (2017).
44 Armstrong, A. R. et al. Demonstrating Oxygen Loss and Associated Structural

Reorganization in the Lithium Battery Cathode Li[Ni0.2Li0.2Mn0.6]O2. J. Am. Chem. Soc. 128,
8694-8698 (2006).
45 Tran, N. et al. Mechanisms Associated with the "Plateau" Observed at High Voltage for the
Overlithiated Li1.12(Ni0.425Mn0.425Co0.15)0.88O2 System. Chem. Mater. 20, 4815–4825 (2008).
46 Luo, K. et al. Charge-Compensation in 3d-Transition-Metal-Oxide Intercalation Cathodes

Through the Generation of Localized Electron Holes on Oxygen. Nat. Chem. 8, 684-691
(2016).
47 Strehle, B. et al. The Role of Oxygen Release from Li- and Mn-Rich Layered Oxides
During the First Cycles Investigated by On-Line Electrochemical Mass Spectrometry. J.
48 Koga, H. et al. Different Oxygen Redox Participation for Bulk and Surface: A Possible
Global Explanation for the Cycling Mechanism of Li1.20Mn0.54Co0.13Ni0.13O2. J. Power
Sources 236, 250-258 (2013).
49 Boulineau, A., Simonin, L., Colin, J. F., Bourbon, C. & Patoux, S. First Evidence of
Manganese-Nickel Segregation and Densification upon Cycling in Li-rich Layered Oxides
for Lithium Batteries. Nano Lett. 13, 3857-3863 (2013).
50 Teufl, T., Strehle, B., Müller, P., Gasteiger, H. A. & Mendez, M. A. Oxygen Release and
Surface Degradation of Li- and Mn-Rich Layered Oxides in Variation of the Li2MnO3
Content. J. Electrochem. Soc. 165, A2718-A2731 (2018).
51 Sathiya, M. et al. Reversible Anionic Redox Chemistry in High-Capacity Layered-Oxide

Electrodes. Nat. Mater. 12, 827-835 (2013).
52 Xie, Y., Saubanère, M. & Doublet, M. L. Requirements for Reversible Extra-Capacity in Li-
Rich Layered Oxides for Li-Ion Batteries. Energy Environ. Sci. 10, 266-274 (2017).
53 Saubanère, M., McCalla, E., Tarascon, J. M. & Doublet, M. L. The Intriguing Question of
Anionic Redox in High-Energy Density Cathodes for Li-Ion Batteries. Energy Environ. Sci.
9, 984-991 (2016).
54 Seo, D.-H., Urban, A. & Ceder, G. The Structural and Chemical Origin of the Oxygen
Redox Activity in Layered and Cation-Disordered Li-Excess Cathode Materials. Nat. Chem.
8, 692-697 (2016).
177
Bibliography
55 Maitra, U. et al. Oxygen Redox Chemistry without Excess Alkali-Metal Ions in

Na2/3[Mg0.28Mn0.72]O2. Nat. Chem. 10, 288–295 (2018).
56 Mortemard de Boisse, B. et al. Highly Reversible Oxygen-Redox Chemistry at 4.1 V in

Na4/7−x[1/7Mn6/7]O2 (: Mn Vacancy). Adv. Energy Mater. 8, 1800409-1800415 (2018).
57 Perez, A. J. et al. Approaching the Limits of Cationic and Anionic Electrochemical Activity
with the Li-Rich Layered Rocksalt Li3IrO4. Nat. Energy 2, 954-962 (2017).
58 Yabuuchi, N. et al. High-Capacity Electrode Materials for Rechargeable Lithium Batteries:

Li3NbO4-Based System with Cation-Disordered Rocksalt Structure. Proc. Natl. Acad. Sci.
112, 7650-7655 (2015).
59 McCalla, E. et al. Visualization of O-O Peroxo-Like Dimers in High-Capacity Layered

Oxides for Li-ion Batteries. Science 350, 1516-1521 (2015).
60 Pearce, P. E. et al. Evidence for Anionic Redox Activity in a Tridimensional-Ordered Li-

Rich Positive Electrode β-Li2IrO3. Nat. Mater. 16, 580–586 (2017).
61 Yahia, M. B., Vergnet, J., Saubanère, M. & Doublet, M.-L. Unified Picture of Anionic
Redox in Li/Na-Ion Batteries. Nat. Mater. 18, 496–502 (2019).
62 Manthiram, A. & Goodenough, J. B. Lithium Insertion into Fe2(SO4)3 Frameworks. J.

Power Sources 26, 403-408 (1989).
63 Goodenough, J. B. & Kim, Y. Challenges for Rechargeable Li Batteries. Chem. Mater. 22,
587-603 (2010).
64 De Dompablo, M. E. A. y., Amador, U. & Tarascon, J. M. A Computational Investigation

on Fluorinated-Polyanionic Compounds as Positive Electrode for Lithium Batteries. J.
Power Sources 174, 1251-1257 (2007).
65 Nagahama, M., Hasegawa, N. & Okada, S. High Voltage Performances of Li2NiPO4F

Cathode with Dinitrile-Based Electrolytes. J. Electrochem. Soc. 157, A748-A752 (2010).
66 Goodenough, J. B. Evolution of Strategies for Modern Rechargeable Batteries. Acc. Chem.

Res. 46, 1053-1061 (2013).
67 Gao, J., Shi, S.-Q. & Li, H. Brief Overview of Electrochemical Potential in Lithium Ion
Batteries. Chin. Phys. B 25, 018210-018224 (2016).
68 Wang, H., Xia, H., Lai, M. O. & Lu, L. Enhancements of Rate Capability and Cyclic
Performance of Spinel LiNi0.5Mn1.5O4 by Trace Ru-Doping. Electrochem. Commun. 11,
1539-1542 (2009).
178
Bibliography
69 Amatucci, G., Pasquier, A. D., A. Blyra, T. Z. & Tarascon, J.-M. The Elevated Temperature
Performance of the LiMn2O4/C System: Failure and Solutions. Electrochim. Acta 45, 255-
271 (1999).
70 Yang, L., Ravdel, B. & Lucht, B. L. Electrolyte Reactions with the Surface of High Voltage
LiNi0.5Mn1.5O4 Cathodes for Lithium-Ion Batteries. Electrochem. Solid State Lett. 13, A95-
A97 (2010).
71 Pieczonka, N. P. W. et al. Understanding Transition-Metal Dissolution Behavior in

LiNi0.5Mn1.5O4 High-Voltage Spinel for Lithium Ion Batteries. J. Phys. Chem. C 117,
15947-15957 (2013).
72 Dedryvère, R. et al. Electrode/Electrolyte Interface Reactivity in High-Voltage Spinel

LiMn1.6Ni0.4O4/Li4Ti5O12 Lithium-Ion Battery. J. Phys. Chem. C 114, 10999-11008 (2010).
73 Amatucci, G. G., Blyrc, A., Sigala, C., Alfonse, P. & Tarascon, J.-M. Surface Treatments of
Li1+xMn2-xO4 Spinels for Improved Elevated Temperature Performance. Solid State Ion. 104,
13-25 (1997).
74 Zhang, Z. et al. Fluorinated Electrolytes for 5 V Lithium-Ion Battery Chemistry. Energy

Environ. Sci. 6, 1806-1811 (2013).
75 Pieczonka, N. P. W. et al. Impact of Lithium Bis(oxalate)borate Electrolyte Additive on the

Performance of High-Voltage Spinel/Graphite Li-Ion Batteries. J. Phys. Chem. C 117,
22603-22612 (2013).
76 Kang, H.-B., Myung, S.-T., Amine, K., Lee, S.-M. & Sun, Y.-K. Improved Electrochemical
Properties of BiOF-Coated 5V Spinel Li[Ni0.5Mn1.5]O4 for Rechargeable Lithium Batteries.
J. Power Sources 195, 2023-2028 (2010).
77 Xu, K., Zhang, S. S., Lee, U., Allen, J. L. & Jow, T. R. LiBOB: Is It an Alternative Salt for
Lithium Ion Chemistry? J. Power Sources 146, 79-85 (2005).
78 Wang, J. et al. Superconcentrated Electrolytes for a High-Voltage Lithium-Ion Battery. Nat.

Commun. 7, 12032-12040 (2016).
79 Jung, S.-K. et al. Lithium-Free Transition Metal Monoxides for Positive Electrodes in
Lithium-Ion Batteries. Nat. Energy 2, 16208, doi:10.1038/nenergy.2016.208 (2017).
80 Zhang, L., Chen, G., Berg, E. J. & Tarascon, J.-M. Triggering the In Situ Electrochemical
Formation of High Capacity Cathode Material from MnO. Adv. Energy Mater. 7, 1602200-
1602205 (2017).
81 Liu, T. et al. Cycling Li-O2 Batteries via LiOH Formation and Decomposition. Science 350,
530-533 (2015).
179
Bibliography
82 Johnson, L. et al. The Role of LiO2 Solubility in O2 Reduction in Aprotic Solvents and Its
Consequences for Li–O2 Batteries. Nat. Chem., 1091–1099 (2014).
83 Knudsen, K. B., Vegge, T., McCloskey, B. D. & Hjelm, J. An Electrochemical Impedance

Spectroscopy Study on the Effects of the Surface- and Solution-Based Mechanisms in Li-O2
Cells. J. Electrochem. Soc. 163, A2065-A2071 (2016).
84 Kwabi, D. G. et al. Experimental and Computational Analysis of the Solvent-Dependent O2

/Li+-O2¯ Redox Couple: Standard Potentials, Coupling Strength, and Implications for
Lithium-Oxygen Batteries. Angew. Chem. Int. Ed. Engl. 55, 3129-3134 (2016).
85 Burke, C. M., Pande, V., Khetan, A., Viswanathan, V. & McCloskey, B. D. Enhancing
Electrochemical Intermediate Solvation Through Electrolyte Anion Selection to Increase
Nonaqueous Li–O2 Battery Capacity. Proc. Natl. Acad. Sci. 112, 9293-9298 (2015).
86 Aetukuri, N. B. et al. Solvating Additives Drive Solution-Mediated Electrochemistry and

Enhance Toroid Growth in Non-aqueous Li-O2 Batteries. Nat. Chem. 7, 50-56 (2015).
87 Yao, W. et al. Tuning Li2O2 Formation Routes by Facet-Engineering of MnO2 Cathode

Catalysts. J. Am. Chem. Soc. 141, 12832-12838 (2019).
88 Khetan, A., Luntz, A. & Viswanathan, V. Trade-Offs in Capacity and Rechargeability in

Nonaqueous Li-O2 Batteries: Solution-Driven Growth versus Nucleophilic Stability. J. Phys.
Chem. Lett. 6, 1254-1259 (2015).
89 Gao, X., Chen, Y., Johnson, L. & Bruce, P. G. Promoting Solution Phase Discharge in Li-O2
Batteries Containing Weakly Solvating Electrolyte Solutions. Nat. Mater. 15, 882-888
(2016).
90 Dai, W. et al. Defect Chemistry in Discharge Products of Li–O2 Batteries. Small Methods 3,
1800358-1800375 (2018).
91 Kang, S., Mo, Y., Ong, S. P. & Ceder, G. A Facile Mechanism for Recharging Li 2O2 in Li–
O2 Batteries. Chem. Mater. 25, 3328-3336 (2013).
92 Geng, W. T., He, B. L. & Ohno, T. Grain Boundary Induced Conductivity in Li2O2. J. Phys.
Chem. C 117, 25222-25228 (2013).
93 Yang, Y. et al. Tuning the Morphology and Crystal Structure of Li2O2: A Graphene Model
Electrode Study for Li-O2 Battery. ACS Appl. Mater. Interfaces 8, 21350-21357 (2016).
94 Zhang, Y. et al. Amorphous Li2O2 Chemical Synthesis and Electrochemical Properties.

Angew. Chem. Int. Ed. 55, 1-6 (2016).
95 Gallagher, K. G. et al. Quantifying the Promise of Lithium–Air Batteries for Electric

Vehicles. Energy Environ. Sci. 7, 1555-1563 (2014).
180
Bibliography
96 Kwabi, D. G. et al. The Effect of Water on Discharge Product Growth and Chemistry in Li-
O2 Batteries. Phys. Chem. Chem. Phys. 18, 24944-24953 (2016).
97 Tułodziecki, M. et al. The Role of Iodide on the Formation of Lithium Hydroxide in

Lithium-Oxygen Batteries. Energy Environ. Sci. 10, 1828-1842 (2017).
98 Qiao, Y. et al. From O2¯ to HO2¯ : Reducing By-Products and Overpotential in Li-O2
Batteries by Water Addition. Angew. Chem. Int. Ed. Engl. 56, 4960-4964 (2017).
99 Liu, T. et al. The Effect of Water on Quinone Redox Mediators in Nonaqueous Li-O2
Batteries. J. Am. Chem. Soc. 140, 1428-1437 (2018).
100 Mekonnen, Y. S. et al. The Influence of CO2 Poisoning on Overvoltages and Discharge
Capacity in Non-Aqueous Li-Air Batteries. J. Chem. Phys. 140, 121101-121106 (2014).
101 Yang, S., He, P. & Zhou, H. Exploring the Electrochemical Reaction Mechanism of
Carbonate Oxidation in Li-Air/CO2 Battery Through Tracing Missing Oxygen. Energy
Environ. Sci. 9, 1650-1654 (2016).
102 Takechi, K., Shiga, T. & Asaoka, T. A Li-O2/CO2 Battery. Chem. Commun. 47, 3463-3465
(2011).
103 Okuoka, S.-i. et al. A New Sealed Lithium-Peroxide Battery with a Co-Doped Li2O Cathode
in a Superconcentrated Lithium Bis(fluorosulfonyl)amide Electrolyte. Sci. Rep. 4, 5684-
5689 (2014).
104 Zhu, Z. et al. Anion-Redox Nanolithia Cathodes for Li-Ion Batteries. Nat. Energy 1, 16111-
16127 (2016).
105 Shimada, Y. et al. Fluorine and Copper Codoping for High Performance Li2O-Based
Cathode Utilizing Solid-State Oxygen Redox. ACS Appl. Energy Mater. 2, 4389-4394
(2019).
106 McCalla, E. et al. Understanding the Roles of Anionic Redox and Oxygen Release During
Electrochemical Cycling of Lithium-Rich Layered Li4FeSbO6. J. Am. Chem. Soc. 137,
4804-4814 (2015).
107 La Mantia, F., Rosciano, F., Tran, N. & Novák, P. Quantification of Oxygen Loss from
Li1+x(Ni1/3Mn1/3Co1/3)1−xO2 at High Potentials by Differential Electrochemical Mass
Spectrometry. J. Electrochem. Soc. 156, A823-A827 (2009).
108 La Mantia, F., Rosciano, F., Tran, N. & Novák, P. Direct Evidence of Oxygen Evolution
from Li1+x(Ni1/3Mn1/3Co1/3)1−xO2 at High Potentials. J. Appl. Electrochem. 38, 893-896
(2008).
109 Lanz, P., Sommer, H., Schulz-Dobrick, M. & Novák, P. Oxygen Release from High-energy
xLi2MnO3·(1−x)LiMO2 (M=Mn, Ni, Co): Electrochemical, Differential Electrochemical
181
Bibliography
Mass Spectrometric, In Situ Pressure, and In Situ Temperature Characterization.

Electrochim. Acta 93, 114-119 (2013).
110 Michalak, B. et al. Gas Evolution in LiNi0.5Mn1.5O4/Graphite Cells Studied In Operando by

a Combination of Differential Electrochemical Mass Spectrometry, Neutron Imaging, and
Pressure Measurements. Anal. Chem. 88, 2877-2883 (2016).
111 Castel, E., Berg, E. J., El Kazzi, M., Novák, P. & Villevieille, C. Differential
Electrochemical Mass Spectrometry Study of the Interface of xLi2MnO3·(1–x)LiMO2 (M =
Ni, Co, and Mn) Material as a Positive Electrode in Li-Ion Batteries. Chem. Mater. 26,
5051-5057 (2014).
112 Singer, A. et al. Nucleation of Dislocations and Their Dynamics in Layered Oxide Cathode
Materials during Battery Charging. Nat. Energy 3, 641-647 (2018).
113 Xiong, D. J. et al. Measuring Oxygen Release from Delithiated LiNixMnyCo1-x-yO2 and Its
Effects on the Performance of High Voltage Li-Ion Cells. J. Electrochem. Soc. 164, A3025-
A3037 (2017).
114 Shin, Y. & Persson, K. A. Surface Morphology and Surface Stability against Oxygen Loss
of the Lithium-Excess Li2MnO3 Cathode Material as a Function of Lithium Concentration.
ACS Appl. Mater. Interfaces 8, 25595-25602 (2016).
115 Lee, J. et al. Mitigating Oxygen Loss to Improve the Cycling Performance of High Capacity
Cation-Disordered Cathode Materials. Nat. Commun. 8, 981-990 (2017).
116 Peng, Z., Freunberger, S. A., Chen, Y. & Bruce, P. G. A Reversible and Higher-Rate Li-O2
Battery. Science 337, 563-566 (2012).
117 Barile, C. J. & Gewirth, A. A. Investigating the Li-O2 Battery in an Ether-Based Electrolyte
Using Differential Electrochemical Mass Spectrometry. J. Electrochem. Soc. 160, A549-
A552 (2013).
118 Tsiouvaras, N., Meini, S., Buchberger, I. & Gasteiger, H. A. A Novel On-Line Mass
Spectrometer Design for the Study of Multiple Charging Cycles of a Li-O2 Battery. J.
119 Eggert, G. & Heitbaum, J. Electrochemical Reactions of Propylenecarbonate and

Electrolytes Solved Therein—A DEMS Study. Electrochim. Acta 31, 1443-1448 (1986).
120 Imhof, R. & Novák, P. In Situ Investigation of the Electrochemical Reduction of Carbonate
Electrolyte Solutions at Graphite Electrodes. J. Electrochem. Soc. 145, 1081-1087 (1998).
121 Vetter, J. et al. In Situ Study on CO2 Evolution at Lithium-Ion Battery Cathodes. J. Power
Sources 159, 277-281 (2006).
182
Bibliography
122 Holzapfel, M., Würsig, A., Scheifele, W., Vetter, J. & Novák, P. Oxygen, Hydrogen,
Ethylene and CO2 Development in Lithium-Ion Batteries. J. Power Sources 174, 1156-1160
(2007).
123 He, M. et al. In Situ Gas Analysis of Li4Ti5O12 Based Electrodes at Elevated Temperatures.
J. Electrochem. Soc. 162, A870-A876 (2015).
124 Guéguen, A. et al. Decomposition of LiPF6 in High Energy Lithium-Ion Batteries Studied
with Online Electrochemical Mass Spectrometry. J. Electrochem. Soc. 163, A1095-A1100
(2016).
125 Uddin, J. et al. Lithium Nitrate As Regenerable SEI Stabilizing Agent for Rechargeable
Li/O2 Batteries. J. Phys. Chem. Lett. 4, 3760-3765 (2013).
126 Giordani, V. et al. A Molten Salt Lithium-Oxygen Battery. J. Am. Chem. Soc. 138, 2656-
2663 (2016).
127 Lepoivre, F., Grimaud, A., Larcher, D. & Tarascon, J. M. Long-Time and Reliable Gas
Monitoring in Li-O2 Batteries via a Swagelok Derived Electrochemical Cell. J. Electrochem.
Soc. 163, A923-A929 (2016).
128 Jia, Z., Yin, G. & Zhang, J. Rotating Ring-Disk Electrode Method. 199-229 (2014).
129 Albery, W. J. & Drur, J. S. Ring-Disc Electrodes Part 1.-A New Approach to the Theory.
Trans. Faraday Soc. 62, 1915-1919 (1966).
130 Dalton, F. Historical Origins of the Rotating Ring-Disk Electrode. Electrochem. Soc.
Interface 25, 50-59 (2016).
131 Zoski, C. G. Review—Advances in Scanning Electrochemical Microscopy (SECM). J.

Electrochem. Soc. 163, H3088-H3100 (2015).
132 Wang, Y. et al. A Solvent-Controlled Oxidation Mechanism of Li2O2 in Lithium-Oxygen

Batteries. Joule 2, 2364-2380 (2018).
133 Wang, W., Lai, N. C., Liang, Z., Wang, Y. & Lu, Y. C. Superoxide Stabilization and a
Universal KO2 Growth Mechanism in Potassium-Oxygen Batteries. Angew. Chem. Int. Ed.
57, 5042-5046 (2018).
134 Behm, R. J., Paulus, U. A., Schmidt, T. J. & Gasteiger, H. A. Oxygen Reduction on a High-
Surface Area Pt/Vulcan Carbon Catalyst: a Thin-Film Rotating Ring-Disk Electrode Study.
J. Electroanal. Chem. 495, 134–145 (2001).
135 Jaouen, F. d. r. & Dodelet, J.-P. O2 Reduction Mechanism on Non-Noble Metal Catalysts for
PEM Fuel Cells. Part II: A Porous-Electrode Model To Predict the Quantity of H2O2
Detected by Rotating Ring-Disk Electrode. J. Phys. Chem. C 113, 15422-15432 (2009).
183
Bibliography
136 Herranz, J., Garsuch, A. & Gasteiger, H. A. Using Rotating Ring Disc Electrode
Voltammetry to Quantify the Superoxide Radical Stability of Aprotic Li–Air Battery
Electrolytes. J. Phys. Chem. C 116, 19084-19094 (2012).
137 Xia, C., Black, R., Fernandes, R., Adams, B. & Nazar, L. F. The Critical Role of Phase-
Transfer Catalysis in Aprotic Sodium Oxygen Batteries. Nat. Chem. 7, 496-501 (2015).
138 Sankarasubramanian, S., Seo, J., Mizuno, F., Singh, N. & Prakash, J. Elucidating the
Oxygen Reduction Reaction Kinetics and the Origins of the Anomalous Tafel Behavior at
the Lithium–Oxygen Cell Cathode. J. Phys. Chem. C 121, 4789-4798 (2017).
139 Lu, Y.-C., He, Q. & Gasteiger, H. A. Probing the Lithium–Sulfur Redox Reactions: A
Rotating-Ring Disk Electrode Study. J. Phys. Chem. C 118, 5733-5741 (2014).
140 Wang, L.-F., Ou, C.-C., Striebel, K. A. & Chen, J.-S. Study of Mn Dissolution from
LiMn2O4 Spinel Electrodes Using Rotating Ring-Disk Collection Experiments. J.
141 Buchheit, R. G., Martinez, M. A. & Montes, L. P. Evidence for Cu Ion Formation by
Dissolution and Dealloying the Al2CuMg Intermetallic Compound in Rotating Ring-Disk
Collection Experiments. J. Electrochem. Soc. 147, 119-124 (2000).
142 Li, J., Sun, W., Hurley, B., Luo, A. A. & Buchheit, R. G. Cu Redistribution Study during the
Corrosion of AZ91 using a Rotating Ring-Disk Collection Experiment. Corros. Sci. 112,
760-764 (2016).
143 McCrory, C. C., Jung, S., Peters, J. C. & Jaramillo, T. F. Benchmarking Heterogeneous
Electrocatalysts for the Oxygen Evolution Reaction. J. Am. Chem. Soc. 135, 16977-16987
(2013).
144 Scholz, J. et al. Rotating Ring–Disk Electrode Study of Oxygen Evolution at a Perovskite
Surface: Correlating Activity to Manganese Concentration. J. Phys. Chem. C 120, 27746-
27756 (2016).
145 Yin, W., Mariyappan, S., Grimaud, A. & Tarascon, J. M. Rotating Ring Disk Electrode for
Monitoring the Oxygen Release at High Potentials in Li-Rich Layered Oxides. J.
146 Lepoivre, F. Study and Improvement of Non-Aqueous Lithium-Air Batteries via the
Development of a Silicon-Based Anode. Ph.D Thesis, Université Pierre et Marie Curie -
Paris VI (2016).
147 Ashton, S. J. Design, Construction and Research Application of a Differential

Electrochemical Mass Spectrometer (DEMS). Springer 8, 9-27 (2012).
184
Bibliography
148 Lu, Y. C., Gasteiger, H. A. & Shao-Horn, Y. Catalytic Activity Trends of Oxygen
Reduction Reaction for Nonaqueous Li-Air Batteries. J. Am. Chem. Soc. 133, 19048-19051
(2011).
149 Strmcnik, D. et al. The Role of Non-Covalent Interactions in Electrocatalytic Fuel-Cell

Reactions on Platinum. Nat. Chem. 1, 466-472 (2009).
150 Bard, A. J. & Faulkner, L. R. Electrochemical Methods,Fundamentals and Applications.

Wiley New York (2001).
151 Yin, W., Grimaud, A., Lepoivre, F., Yang, C. & Tarascon, J.-M. Chemical vs
Electrochemical Formation of Li2CO3 as a Discharge Product in Li-O2/CO2 Batteries by
Controlling the Superoxide Intermediate. J. Phys. Chem. Lett. 8, 214−222 (2017).
152 Chen, H. & Islam, M. S. Lithium Extraction Mechanism in Li-Rich Li2MnO3 Involving
Oxygen Hole Formation and Dimerization. Chem. Mater. 28, 6656-6663 (2016).
153 Li, B. et al. Understanding the Stability for Li-Rich Layered Oxide Li2RuO3 Cathode. Adv.
Funct. Mater. 26, 1330-1337 (2016).
154 Sathiya, M. et al. Origin of Voltage Decay in High-Capacity Layered Oxide Electrodes. Nat.
Mater. 14, 230-238 (2014).
155 Gao, Q., Ranjan, C., Pavlovic, Z., Blume, R. & Schlögl, R. Enhancement of Stability and
Activity of MnOx/Au Electrocatalysts for Oxygen Evolution Through Adequate Electrolyte
Composition. ACS Catal. 5, 7265-7275 (2015).
156 Zhang, X. et al. Structural and Electrochemical Study of Al2O3 and TiO2 Coated
Li1.2Ni0.13Mn0.54Co0.13O2 Cathode Material Using ALD. Adv. Energy Mater. 3, 1299-1307
(2013).
157 Yu, H. et al. Temperature-Sensitive Structure Evolution of Lithium-Manganese-Rich

Layered Oxides for Lithium-Ion Batteries. J. Am. Chem. Soc. 140, 15279-15289 (2018).
158 Xu, B., Fell, C. R., Chi, M. & Meng, Y. S. Identifying Surface Structural Changes in
Layered Li-Excess Nickel Manganese Oxides in High Voltage Lithium Ion Batteries: A
Joint Experimental and Theoretical Study. Energy Environ. Sci. 4, 2223-2233 (2011).
159 Fell, C. R. et al. Correlation Between Oxygen Vacancy, Microstrain, and Cation
Distribution in Lithium-Excess Layered Oxides During the First Electrochemical Cycle.
Chem. Mater. 25, 1621-1629 (2013).
160 Jiang, M., Key, B., Meng, Y. S. & Grey, C. P. Electrochemical and Structural Study of the
Layered, “Li-Excess” Lithium-Ion Battery Electrode Material Li[Li1/9Ni1/3Mn5/9]O2. Chem.
Mater. 21, 2733-2745 (2009).
185
Bibliography
161 Van Bommel, A., Krause, L. J. & Dahn, J. R. Investigation of the Irreversible Capacity Loss
in the Lithium-Rich Oxide Li[Li1/5Ni1/5Mn3/5]O2. J. Electrochem. Soc. 158, A731-A735
(2011).
162 Shunmugasundaram, R., Senthil Arumugam, R. & Dahn, J. R. High Capacity Li-Rich
Positive Electrode Materials with Reduced First-Cycle Irreversible Capacity Loss. Chem.
Mater. 27, 757-767 (2015).
163 Kasnatscheew, J. et al. The Truth about the 1st Cycle Coulombic Efficiency of
LiNi1/3Co1/3Mn1/3O2 (NCM) Cathodes. Phys. Chem. Chem. Phys. 18, 3956-3965 (2016).
164 Zhou, H., Xin, F., Pei, B. & Whittingham, M. S. What Limits the Capacity of Layered
Oxide Cathodes in Lithium Batteries? ACS Energy Lett. 4, 1902-1906 (2019).
165 Song, B. et al. Mitigated Phase Transition during First Cycle of a Li-Rich Layered Cathode
Studied by In Operando Synchrotron X-Ray Powder Diffraction. Phys. Chem. Chem. Phys.
18, 4745-4752 (2016).
166 Yabuuchi, N., Yoshii, K., Myung, S. T., Nakai, I. & Komaba, S. Detailed Studies of a High-
Capacity Electrode Material for Rechargeable Batteries, Li2MnO3-LiCo1/3Ni1/3Mn1/3O2. J.
Am. Chem. Soc. 133, 4404-4419 (2011).
167 Mohanty, D. et al. Structural Transformation of a Lithium-Rich Li1.2Co0.1Mn0.55Ni0.15O2

Cathode During High Voltage Cycling Resolved by In Situ X-Ray Diffraction. J. Power
Sources 229, 239-248 (2013).
168 Hatsukade, T., Schiele, A., Hartmann, P., Brezesinski, T. & Janek, J. The Origin of Carbon
Dioxide Evolved during Cycling of Nickel-Rich Layered NCM Cathodes. ACS Appl. Mater.
Interfaces 10, 38892-38899 (2018).
169 Renfrew, S. E. & McCloskey, B. D. Residual Lithium Carbonate Predominantly Accounts

for First Cycle CO2 and CO Outgassing of Li-Stoichiometric and Li-Rich Layered
Transition Metal Oxides. J. Am. Chem. Soc. 139, 17853-17960 (2017).
170 Jung, R., Metzger, M., Maglia, F., Stinner, C. & Gasteiger, H. A. Chemical versus
Electrochemical Electrolyte Oxidation on NMC111, NMC622, NMC811, LNMO, and
Conductive Carbon. J. Phys. Chem. Lett. 8, 4820-4825 (2017).
171 Mohanty, D. et al. Unraveling the Voltage-Fade Mechanism in High-Energy-Density

Lithium-Ion Batteries: Origin of the Tetrahedral Cations for Spinel Conversion. Chem.
Mater. 26, 6272-6280 (2014).
172 Dogan, F. et al. Reentrant Lithium Local Environments and Defect Driven Electrochemistry
of Li- and Mn-rich Li-ion Battery Cathodes. J. Am. Chem. Soc. 137, 2328-2335 (2015).
173 Croy, J. R. et al. First-charge Instabilities of Layered-Layered Lithium-Ion-Battery

Materials. Phys. Chem. Chem. Phys. 17, 24382-24391 (2015).
186
Bibliography
174 Benedek, R. & Iddir, H. Simulation of First-Charge Oxygen-Dimerization and Mn-

Migration in Li-Rich Layered Oxides xLi2MnO3·(1 – x)LiMO2 and Implications for Voltage
Fade. J. Phy. Chem. C 121, 6492-6499 (2017).
175 Robert, R. & Novák, P. Switch of the Charge Storage Mechanism of LixNi0.80Co0.15Al0.05O2
at Overdischarge Conditions. Chem. Mater. 30, 1907-1911 (2018).
176 Davidson, I., Greedan, J. E., Sacken, U. v., Michal, C. A. & Dahn, J. R. Structure of 1T-
Li2NiO2 from Powder Neutron Diffraction. Solid State Ion. 46 243-247 (1991).
177 Lyu, Y. et al. Correlations between Transition-Metal Chemistry, Local Structure, and
Global Structure in Li2Ru0.5Mn0.5O3 Investigated in a Wide Voltage Window. Chem. Mater.
29, 9053-9065 (2017).
178 Lyu, Y. et al. Probing Reversible Multi-electron Transfer and Structure Evolution of
Li1.2Cr0.4Mn0.4O2 Cathode Material for Li-ion Batteries in a Voltage Range of 1.0 - 4.8 V.
Chem. Mater. 27, 5238-5252 (2015).
179 Genevois, C. et al. Insight into the Atomic Structure of Cycled Lithium-Rich Layered Oxide
Li1.20Mn0.54Co0.13Ni0.13O2 Using HAADF STEM and Electron Nanodiffraction. J. Phys.
Chem. C 119, 75-83 (2014).
180 Bridges, N. J., Gutowski, K. E. & Rogers, R. D. Investigation of Aqueous Biphasic Systems
Formed from Solutions of Chaotropic Salts with Kosmotropic Salts (salt–salt ABS). Green
Chem. 9, 177-183 (2007).
181 Yin, S.-C., Rho, Y.-H., Swainson, I. & Nazar, L. F. X-ray/Neutron Diffraction and
Electrochemical Studies of Lithium De/Re-Intercalation in Li1-xCo1/3Ni1/3Mn1/3O2 ( x = 0 →
1). Chem. Mater. 18, 1901-1910 (2006).
182 Mueller-Neuhaus, J. R., Dunlap, R. A. & Dahn, J. R. Understanding Irreversible Capacity in

LixNi1-yFeyO2 Cathode Materials. J. Electrochem. Soc. 147, 3598-3605 (2000).
183 Kang, S.-H., Yoon, W.-S., Nam, K.-W., Yang, X.-Q. & Abraham, D. P. Investigating the
First-Cycle Irreversibility of Lithium Metal Oxide Cathodes for Li Batteries. Journal of
Materials Science 43, 4701-4706 (2008).
184 Mondal, M., Khanra, S., Tiwari, O. N., Gayen, K. & Halder, G. N. Role of Carbonic
Anhydrase on the Way to Biological Carbon Capture Through Microalgae-A Mini Review.
Enrivon. Prog. Sustain. 35, 1605-1615 (2016).
185 Nugent, P. et al. Porous Materials with Optimal Adsorption Thermodynamics and Kinetics
for CO2 Separation. Nature 495, 80-84 (2013).
186 Mistry, H. et al. Highly Selective Plasma-Activated Copper Catalysts for Carbon Dioxide
Reduction to Ethylene. Nat. Commun. 7, 12123-12131 (2016).
187
Bibliography
187 Studt, F. et al. Discovery of a Ni-Ga Catalyst for Carbon Dioxide Reduction to Methanol.
Nat. Chem. 6, 320–324 (2014).
188 Chen, Y., Li, C. W. & Kanan, M. W. Aqueous CO2 Reduction at Very Low Overpotential
on Oxide-Derived Au Nanoparticles. J. Am. Chem. Soc. 134, 19969-19972 (2012).
189 Wei, J. et al. New Insights Into the Effect of Sodium on Fe3O4- Based Nanocatalysts for
CO2 Hydrogenation to Light Olefins. Catal. Sci. Technol. 6, 4786-4793 (2016).
190 Wei, J. et al. Directly Converting CO2 into a Gasoline Fuel. Nat. Commun. 8, 15174-15181
(2017).
191 Zhu, Y. et al. Catalytic Conversion of Carbon Dioxide to Methane on Ruthenium–Cobalt

Bimetallic Nanocatalysts and Correlation between Surface Chemistry of Catalysts under
Reaction Conditions and Catalytic Performances. ACS Catal. 2, 2403-2408 (2012).
192 Moret, S., Dyson, P. J. & Laurenczy, G. Direct Synthesis of Formic Acid from Carbon
Dioxide by Hydrogenation in Acidic Media. Nat. Commun. 5, 4017-4023 (2014).
193 Angamuthu, R., Byers, P., Lutz, M., Spek, A. L. & Bouwman, E. Electrocatalytic CO 2
Conversion to Oxalate by a Copper Complex. Science 327, 313-315 (2010).
194 Qiao, Y. et al. Li-CO2 Electrochemistry: A New Strategy for CO2 Fixation and Energy
Storage. Joule 1, 1-12 (2017).
195 Abraham, K. M. & Jiang, Z. A Polymer Electrolyte-Based Rechargeable lithium/Oxygen

Battery. J. Electrochem. Soc. 143, 1-5 (1996).
196 Bruce, P. G., Freunberger, S. A., Hardwick, L. J. & Tarascon, J. M. Li-O2 and Li-S Batteries
with High Energy Storage. Nat. Mater. 11, 19-29 (2012).
197 Cho, M. H. et al. The Effects of Moisture Contamination in the Li-O2 Battery. J. Power
Sources 268, 565-574 (2014).
198 Gowda, S. R., Brunet, A., Wallraff, G. M. & McCloskey, B. D. Implications of CO 2

Contamination in Rechargeable Nonaqueous Li-O2 Batteries. J. Phys. Chem. Lett. 4, 276-
279 (2013).
199 Freunberger, S. A. et al. The Lithium-Oxygen Battery with Ether-Based Electrolytes. Angew.
Chem. Int. Ed. Engl. 50, 8609-8613 (2011).
200 Chen, Y., Freunberger, S. A., Peng, Z., Barde, F. & Bruce, P. G. Li-O2 Battery with a
Dimethylformamide Electrolyte. J. Am. Chem. Soc. 134, 7952-7957 (2012).
201 Freunberger, S. A. et al. Reactions in the Rechargeable Lithium-O2 Battery with Alkyl
Carbonate Electrolytes. J. Am. Chem. Soc. 133, 8040-8047 (2011).
188
Bibliography
202 McCloskey, B. D., Bethune, D. S., Shelby, R. M., Girishkumar, G. & Luntz, A. C. Solvents'
Critical Role in Nonaqueous Lithium-Oxygen Battery Electrochemistry. J. Phys. Chem. Lett.
2, 1161-1166 (2011).
203 McCloskey, B. D. et al. Limitations in Rechargeability of Li-O2 Batteries and Possible

Origins. J. Phys. Chem. Lett. 3, 3043-3047 (2012).
204 Ottakam Thotiyl, M. M., Freunberger, S. A., Peng, Z. & Bruce, P. G. The Carbon Electrode
in Nonaqueous Li-O2 Cells. J. Am. Chem. Soc. 135, 494-500 (2013).
205 McCloskey, B. D. et al. Twin Problems of Interfacial Carbonate Formation in Nonaqueous

Li-O2 Batteries. J. Phys. Chem. Lett. 3, 997-1001 (2012).
206 Gallant, B. M. et al. Chemical and Morphological Changes of Li–O2 Battery Electrodes
upon Cycling. J. Phys. Chem. C 116, 20800-20805 (2012).
207 Lu, Y.-C. et al. Influence of Hydrocarbon and CO2 on the Reversibility of Li–O2 Chemistry
Using In Situ Ambient Pressure X-ray Photoelectron Spectroscopy. J. Phys. Chem. C 117,
25948-25954 (2013).
208 Liu, J. et al. Pathways for Practical High-Energy Long-Cycling Lithium Metal Batteries.
Nat. Energy 4, 180–186 (2019).
209 Ling, C., Zhang, R., Takechi, K. & Mizuno, F. Intrinsic Barrier to Electrochemically
Decompose Li2CO3 and LiOH. J. Phys. Chem. C 118, 26591-26598 (2014).
210 Meini, S. et al. Rechargeability of Li-Air Cathodes Pre-Filled with Discharge Products
Using an Ether-Based Electrolyte Solution: Implications for Cycle-Life of Li-Air Cells.
Phys. Chem. Chem. Phys. 15, 11478-11493 (2013).
211 Julian L. Roberts, J., Calderwood, T. S. & Sawyer, D. T. Nucleophilic Oxygenation of

Carbon Dioxide by Superoxide Ion in Aprotic Media To Form the C2O62- Species. J. Am.
Chem. Soc. 106, 4667-4670 (1984).
212 Buzzeo, M. C. et al. Kinetic Analysis of the Reaction between Electrogenerated Superoxide
and Carbon Dioxide in the Room Temperature Ionic Liquids 1-Ethyl-3-methylimidazolium
Bis(trifluoromethylsulfonyl)imide and Hexyltriethylammonium
Bis(trifluoromethylsulfonyl)imide. J. Phys. Chem. B 108, 3947-3954 (2004).
213 Liu, Y., Wang, R., Lyu, Y., Li, H. & Chen, L. Rechargeable Li/CO2–O2 (2 : 1) Battery and
Li/CO2 Battery. Energy Environ. Sci. 7, 677-681 (2014).
214 Lim, H. K. et al. Toward a Lithium-"Air" Battery: the Effect of CO2 on the Chemistry of a
Lithium-Oxygen Cell. J. Am. Chem. Soc. 135, 9733-9742 (2013).
189
Bibliography
215 Mekonnen, Y. S., Garcia-Lastra, J. M., Hummelshøj, J. S., Jin, C. & Vegge, T. Role of
Li2O2@Li2CO3 Interfaces on Charge Transport in Nonaqueous Li–Air Batteries. J. Phys.
Chem. C 119, 18066-18073 (2015).
216 Laoire, C. O., Mukerjee, S., Abraham, K. M., Plichta, E. J. & Hendrickson, M. A. Influence
of Nonaqueous Solvents on the Electrochemistry of Oxygen in the Rechargeable
Lithium−Air Battery. J. Phys. Chem. C 114, 9178-9186 (2010).
217 Wadhawan, J. D., Welford, P. J., McPeak, H. B., Hahn, C. E. W. & Compton, R. G. The
Simultaneous Voltammetric Determination and Detection of Oxygen and Carbon Dioxide A
Study of the Kinetics of the Reaction Between Superoxide and Carbon Dioxide in Non-
Aqueous Media using Membrane-Free Gold Disc Microelectrodes. Sens. Actuator B 88, 40-
52 (2003).
218 Hartmann, P. et al. Pressure Dynamics in Metal–Oxygen (Metal–Air) Batteries: A Case

Study on Sodium Superoxide Cells. J. Phys. Chem. C 118, 1461-1471 (2014).
219 Wadhawan, J. D. et al. Microelectrode Studies of the Reaction of Superoxide with Carbon
Dioxide in DimethylSulfoxide. J. Phys. Chem. B 105, 10659-10668 (2001).
220 Casadei, M. A., Cesa, S. & Moracci, F. M. Activation of Carbon Dioxide by

Electrogenerated Superoxide Ion: A New Carboxylating Reagent. J. Org. Chem. 61, 380-
383 (1996).
221 Vassiliev, Y. B., Bagotzky, V. S., Khazova, O. A. & Mayorova, N. A. Electroreduction of

Carbon Dioxide: Part II. The Mechanism of Reduction in Aprotic Solvents. J. Electroanal.
Chem. Interfacial Electrochem. 189, 295-309 (1985).
222 Hu, X. et al. Rechargeable Room-Temperature Na–CO2 Batteries. Angew. Chem. Int. Ed. 55,
6482-6486 (2016).
223 Zhang, Z. et al. The First Introduction of Graphene to Rechargeable Li-CO2 Batteries.
Angew. Chem. Int. Ed. 54, 6550-6553 (2015).
224 Zhang, X. et al. Rechargeable Li–CO2 Batteries with Carbon Nanotubes as Air Cathodes.
Chem. Commun. 51, 14636-14639 (2015).
225 Lutz, L. et al. High Capacity Na–O2 Batteries: Key Parameters for Solution-Mediated
Discharge. J. Phys. Chem. C 120, 20068-20076 (2016).
226 Sharon, D. et al. Oxidation of Dimethyl Sulfoxide Solutions by Electrochemical Reduction

of Oxygen. J. Phys. Chem. Lett. 4, 3115-3119 (2013).
227 Kwabi, D. G. et al. Chemical Instability of Dimethyl Sulfoxide in Lithium-Air Batteries. J.

Phys. Chem. Lett. 5, 2850-2856 (2014).
190
Bibliography
228 Abraham, K. M. Electrolyte-Directed Reactions of the Oxygen Electrode in Lithium-Air

Batteries. J. Electrochem. Soc. 162, A3021-A3031 (2014).
229 Adams, B. D. et al. Current Density Dependence of Peroxide Formation in the Li–O2
Battery and Its Effect on Charge. Energy Environ. Sci. 6, 1772-1778 (2013).
230 Gallant, B. M. et al. Influence of Li2O2 Morphology on Oxygen Reduction and Evolution
Kinetics in Li–O2 Batteries. Energy Environ. Sci. 6, 2518-2528 (2013).
231 Zhai, D. et al. Raman Evidence for Late Stage Disproportionation in a Li-O2 Battery. J.
Phys. Chem. Lett. 5, 2705-2710 (2014).
232 Mitchell, R. R., Gallant, B. M., Thompson, C. V. & Shao-Horn, Y. All-Carbon-Nanofiber

Electrodes for High-Energy Rechargeable Li–O2 Batteries. Energy Environ. Sci. 4, 2952-
2958 (2011).
233 Lau, S. & Archer, L. A. Nucleation and Growth of Lithium Peroxide in the Li–O2 Battery.
Nano Lett. 15, 5995-6002 (2015).
234 Horstmann, B. et al. Rate-Dependent Morphology of Li2O2 Growth in Li-O2 Batteries. J.

Phys. Chem. Lett. 4, 4217-4122 (2013).
235 Wang, L. et al. Monodispersed Ru Nanoparticles Functionalized Graphene Nanosheets as

Efficient Cathode Catalysts for O2-Assisted Li–CO2 Battery. ACS Omega 2, 9280-9286
(2017).
236 Zhao, Z., Su, Y. & Peng, Z. Probing Lithium Carbonate Formation in Trace O2-Assisted
Aprotic Li-CO2 Batteries Using In-Situ Surface Enhanced Raman Spectroscopy. J. Phys.
Chem. Lett. 10, 322-328 (2019).
237 Qiao, Y. et al. A Li2CO3-Free Li-O2/CO2 Battery with Peroxide Discharge Product. Energy
Environ. Sci. 11, 1211-1217 (2018).
238 Jain, P. S. & Lal, S. Electrolytic Reduction of Oxygen at Solid Electrodes in Aprotic
Solvents- the Superoxide Ion. Electrochim. Acta 27, 759-763 (1982).
239 Sullivan, B. P. Electrochemical and Electrocatalytic Reactions of Carbon Dioxide. Elsevier,

Laramie, WY, USZ (1993).
240 Lamy, E., Nadjo, L. & Saveant, J. M. Standard Potential and Kinetic Parameters of the
Electrochemical Reduciton of Carbon Dioxide in Dimethyformamide. J. Electroanal. Chem.
78, 403-407 (1977).
241 Xu, S., Das, S. K. & Archer, L. A. The Li–CO2 Battery: A Novel Method for CO2 Capture
and Utilization. RSC Adv. 3, 6656-6660 (2013).
191
Bibliography
242 Qie, L., Lin, Y., Connell, J. W., Xu, J. & Liming, D. Highly Rechargeable Lithium-CO2
Batteries with a Boron- and Nitrogen-Codoped Holey-Graphene Cathode. Angew. Chem. Int.
Ed. 56, 6970–6974 (2017).
243 Yang, S. et al. A Reversible Lithium–CO2 Battery with Ru Nanoparticles as a Cathode

Catalyst. Energy Environ. Sci. 10, 972-978 (2017).
244 Costentin, C., Robert, M. & Saveant, J. M. Catalysis of the Electrochemical Reduction of
Carbon Dioxide. Chem. Soc. Rev. 42, 2423-2436 (2013).
245 Savéant, J. M. Molecular Catalysis of Electrochemical Reactions. Mechanistic Aspects.

Chem. Rev. 208, 2348–2378 (2008).
246 Andrieux, C. P., Dumas-Bouchiat, J. M. & Saveant, J. M. Homogeneous Redox Catalysis of

Electrochemical Reactions: Part I. Introduction. J. Electroanal. Chem. Interfacial
Electrochem. 87, 39-53 (1978).
247 Bhugun, I., Lexa, D. & Saveant, J.-M. Ultraefficient Selective Homogeneous Catalysis of
the Electrochemical Reduction of Carbon Dioxide by an Iron(0) Porphyrin Associated with
a Weak Bronsted Acid Cocatalyst. J. Am. Chem. Soc. 116, 5015-5016 (1994).
248 Costentin, C., Drouet, S., Robert, M. & Saveant, J. M. A Local Proton Source Enhances
CO2 Electroreduction to CO by a Molecular Fe Catalyst. Science 338, 90-94 (2012).
249 Rheinhardt, J. H., Singh, P., Tarakeshwar, P. & Buttry, D. A. Electrochemical Capture and
Release of Carbon Dioxide. ACS Energy Lett. 2, 454-461 (2017).
250 Mizen, M. B. & Wrighton, M. S. Reductive Addition of CO2 to 9, 10-Phenanthrenequinone.

J. Electrochem. Soc. 136, 941-946 (1989).
251 Gurkan, B., Simeon, F. & Hatton, T. A. Quinone Reduction in Ionic Liquids for
Electrochemical CO2 Separation. ACS Sustain. Chem. Eng. 3, 1394-1405 (2015).
252 Bhugun, I., Lexa, D. & Saveant, J. M. Catalysis of the Electrochemical Reduction of Carbon
Dioxide by Iron(0) Porphyrins. Synergistic Effect of Lewis Acid Cations. J. Phys. Chem.
100, 19981-19985 (1996).
253 Ding, Y., Li, Y. & Yu, G. Exploring Bio-inspired Quinone-Based Organic Redox Flow
Batteries: A Combined Experimental and Computational Study. Chem 1, 790-801 (2016).
254 Wilfrod, J. H. & Archer, M. D. Solvent Effects on the Redox Potentials of Benzoquinone. J.
Electroanal. Chem. 190, 271-277 (1985).
255 Neeraj, G. & Henry, L. Hydrogen-Bonding and Protonation Effects in Electrochemistry of

Quinones in Aprotic Solvents. J. Am. Chem. Soc. 119, 6384-6391 (1997).
192
Bibliography
256 Guin, P. S., Das, S. & Mandal, P. C. Electrochemical Reduction of Quinones in Different
Media: A Review. Int. J. Electrochem. 2011, 1-22 (2011).
257 Tsutomu, N., Nobuyuki, N., Koji, F. & Kotaro, O. Mechanisms of Reductive Addition of
CO2 to Quinones in Acetonitrile. J. Electroanal. Chem. 322, 383-389 (1992).
258 Saveant, J. M. Elements of Molecular and Biomolecular Electrochemistry. Wiley, Paris,

France (2006).
259 Fujinaga, T., Izutsu, K. & Nomura, T. Effect of Metal Ions on the Polarographic Reduction
of Organic Compounds in Dipolar Aprotic Solvents. J. Electroanal. Chem. 29, 203-209
(1971).
260 Kim, K. C., Liu, T., Lee, S. W. & Jang, S. S. First-Principles Density Functional Theory
Modeling of Li Binding: Thermodynamics and Redox Properties of Quinone Derivatives for
Lithium-Ion Batteries. J. Am. Chem. Soc. 138, 2374-2382 (2016).
261 Marek, K. K. & Barbara, T. G. Ion-pair Effects in the Electrochemistry of Aromatic

Compounds Effects of Metal Ions and Solvents on the Polarographic Reduction of Quinones.
J. Electroana. Chem. Interfacial Electrochem. 55, 277-286 (1974).
262 Jaworski, J. S., Lesniewska, E. & Kalinowski, M. K. Solvent Effect on the Redox Potential
of Quinone-Semiquinone Systems. J. Electroanal. Chem. 105, 329-334 (1979).
263 Staley, P. A. The Electrochemistry of Quinones in Aprotic Solvents. Ph.D Thesis,

University of California, San Diego, San Diego State University (2016).
264 McCloskey, B. D. & Addison, D. A Viewpoint on Heterogeneous Electrocatalysis and

Redox Mediation in Nonaqueous Li-O2 Batteries. ACS Catal. 7, 772-778 (2017).
265 Markovic, N. M., Gasteiger, H. A. & Philip N. Ross, J. Oxygen Reduction on Platinum
Low-Index Single-Crystal Surfaces in Sulfuric Acid Solution: Rotating Ring-Pt(hkl) Disk
Studies. J. Phys. Chem. 99, 3411-3415 (1995).
266 Solla-Gullo´n, J., Montiel, V., Aldaz, A. & Clavilier, J. Electrochemical Characterisation of
Platinum Nanoparticles Prepared by Microemulsion: How to Clean Them Without Loss of
Crystalline Surface Structure. J. Electroanal. Chem. 491, 69-77 (2000).
267 Berg, E. J. & P. Novák. Recent Progress on Li-O2 Batteries at PSI. ECL Annual Report,
Paul Scherrer Institut, Villigen, Switzerland (2012).
193
Abstract/Résumé
Abstract
Fundamental understanding of high-energy-density Li-ion batteries and

beyond: structure-property relationships and reaction chemistries
Abstract:
The increasing energy demands placed on energy storage devices for both transportation and
mobile applications have stimulated the development of high-capacity positive materials for Li-ion
batteries. Li-rich layered oxides are among the leading candidates provided their staggering
capacities, owing to the participation of anionic redox in the charge compensation mechanism aside
from the conventional transition metal redox. Nevertheless, Li-rich layered oxides are yet to reach
commercial success since the extra capacities offered by oxygen anions generally come with
structural instability, leading to significant first-cycle irreversibility and performance deterioration
upon cycling. Understanding the structural evolution in these materials and its effect on their
electrochemical properties are therefore of vital importance. This work addressed the major
degradation phenomenon in Li-rich layered oxides, i.e., irreversible oxygen release, by introducing
a simple while sensitive oxygen gas detection technique. The strong coupling between the structural
evolution and the electrochemical behaviors was further investigated using a practically important
Li1.2Ni0.13Mn0.54Co0.13O2 compound. Other than Li-ion batteries based on intercalation chemistry,
numerous alternative battery chemistries are undergoing intensive research in the battery
community, seeking to move beyond the limits of Li-ion technology. Within this context, the
fundamental aspects of Li-CO2 batteries, which include the reaction mechanisms and the catalytic
CO2 reduction, were explored in the second part of this work.
Keywords: Li-ion batteries; Anionic redox; Li-rich layered oxides; Li-CO2 batteries;
Reaction mechanisms; Homogeneous catalysis
194
Résumé
Compréhension fondamentale des batteries Li-ion àhaute densité

d'énergie et au-delà: relations structure-propriétéet chimies de réaction
Résumé:
La recherche de dispositifs de stockage d'énergie plus denses en énergie pour les

applications de transport et d'électronique portable a stimuléle développement de matériaux de
cathode de grande capacité pour les batteries Li-ion. Les oxydes lamellaires riches en lithium
figurent ainsi parmi les principaux candidats. Leur capacitéimportante résulte de la participation,
en plus de la rédox cationique conventionnel, de la rédox anionique lors de
l'insertion/désinsertion du lithium. Néanmoins, ils n'ont pas encore rencontré de succès
commercial car la capacité supplémentaire offerte par la rédox anionique s'accompagne
généralement d'une instabilité structurelle, ce qui entraîne une irréversibilité significative au
premier cycle et une dégradation des performances lors du cyclage. Comprendre l'évolution
structurelle de ces matériaux et son effet sur leurs propriétés électrochimiques revêt donc une
importance vitale. Les travaux réalisés lors de cette thèse ont porté sur le phénomène de
dégradation prédominant dans les oxydes lamellaires riches en lithium, à savoir la libération
irréversible d'oxygène, en introduisant une technique de détection simple mais sensible à
l'oxygène gazeux. Le couplage fort entre l'évolution de la structure et les comportements
électrochimiques a étéétudiéplus en détail pour Li1.2Ni0.13Mn0.54Co0.13O2 qui présente un grand
intérêt industriel. Afin de dépasser les limites de la technologie Li-ion, de nombreuses chimies
alternatives font actuellement l'objet de recherches approfondies. Dans ce contexte, les aspects
fondamentaux des batteries Li-CO2, qui incluent les mécanismes de réaction et la réduction
catalytique de CO2, ont aussi étéexplorés.
Mot clés: Batteries Li-ion; Redox anionique; Oxydes lamellaires riches en lithium;
Batteries Li-CO2; Mécanismes de réaction; Catalyseurs homogènes
195
Résumé
Résumé
RésuméCompréhension fondamentale des batteries Li-ion à

haute densitéd'énergie et au-delà: relations structure-
propriétéet chimies de réaction
Les demandes énergétiques croissantes de la sociétémoderne ont entraînédes recherches

intensives afin de i) repousser sans cesse les limites de densité d’énergie des batteries Li-ion, en
particulier du côtédes électrodes positives, et ii) d’explorer de nouvelles chimies permettant de
stocker plus d’énergie. Les oxydes lamellaires riches en Li sont parmi les matériaux d'électrodes
positives les plus prometteurs pour les batteries Li-ion à haute densité d’énergie. En effet, les
capacités de stockage de ces matériaux sont très élevées et ce grâce àla participation du réseau
d’oxygène dans les mécanismes de compensation de charge lors du cyclage. Néanmoins,
l’activation du rédox anionique induit généralement des instabilités structurelles associées àune
perte irréversible d'oxygène et à une migration des cations, entraînant une dégradation des
performances lors du cycle. Dans ce travail, une méthode électrochimique, i.e. la
voltampérométrie cyclique utilisant une électrode àdisque et àanneau tournante (RRDE), a été
introduite pour la première fois pour suivre le phénomène de dégagement d’oxygène dans les
oxydes lamellaires riches en Li. Les évolutions structurelles dynamiques liées aux activités redox
de l'oxygène (l'évolution irréversible de l'oxygène et redox de l'oxygène réversible), ainsi que leur
influence sur les propriétés électrochimiques ont été étudiées plus en détail en utilisant un
matériau d’intérêt pratique de formule chimique Li1.2Ni0.13Mn0.54Co0.13O2. Enfin, les aspects
fondamentaux de la batterie Li-CO2, une alternative àla batterie conventionnelle Li-ion, ont été
explorés.
Le contenu principal et les conclusions de ce travail incluent:
i) La voltampérométrie RRDE a été développée avec succès en tant que technique

alternative pour surveiller le dégagement d'oxygène gazeux àpartir d'oxydes lamellaires riches en
Li au cours d'un processus de charge à hauts potentiels. La Figure 1 présente un schéma du
système RRDE proposé. Dans une mesure RRDE typique, l’oxyde d’intérêt a été déposé sur
l’électrode à disque où sa délithiation était contrôlée par voltampérométrie cyclique ou par
196
Résumé
cyclage galvanostatique, tandis que l’électrode anneau était fixée à un potentiel permettant la
réduction de l’oxygène. Grâce à la rotation de l'ensemble de la RRDE, l’oxygène dégagé de
l'oxyde peut être transportéet réduit (détecté) au niveau de l'électrode anneau. Comme preuve de
concept, nous avons effectuédes mesures RRDE en utilisant divers oxydes lamellaires riches en
Li, et les résultats ont étésystématiquement comparés àceux obtenus par deux autres techniques
bien établies d’analyse de gaz/pression, i.e. la spectrométrie de masse électrochimique en ligne
(OEMS) et le suivi operando de la pression grâce à un capteur monté sur une cellule
électrochimique. Grâce àcette étude comparative, nous avons démontréque la méthode RRDE
proposée peut détecter de manière fiable l’oxygène avec une précision supérieure aux autres
techniques. Cette technique est donc appropriée pour déterminer le début de l’évolution de
l’oxygène et examiner les oxydes libérant une quantité faible d’oxygène. De plus, grâce à la
simplicitéde cette approche et de la généralisation de la RRDE dans les domaines de la batterie et
de l'électrocatalyse, nous pensons qu'elle peut être utilisée pour les analyses de routine de
l'oxygène lors de la conception de nouveaux matériaux susceptibles de dégager de l'oxygène.
Malgré ses attributs positifs, une optimisation évidente de la méthode RRDE réside dans la
quantification de la quantitétotale d'oxygène générée àpartir de l'oxyde. Dans ce but, une courbe
d’étalonnage de l’augmentation des courants de réduction annulaires en fonction des
concentrations en oxygène àla surface du disque peut être obtenue pour déduire “l’efficacité de
détection”. La voltampérométrie RRDE peut également être étendue pour détecter d’autres types
de gaz tels que le CO2, àcondition qu’une électrode anneau adéquate soit sélectionnée.
197
Résumé
Figure 1 (a) Schéma du système RRDE et de la configuration de cellule à quatre

électrodes, WE1, WE2, CE et RE représentent respectivement l'électrode de travail 1 (électrode à
disque), l'électrode de travail 2 (électrode àanneau), la contre-électrode et l'électrode de référence.
(b) Une vue agrandie de la pointe RRDE et de sa région proche de l'électrolyte décrit les
processus de transport et redox se produisant dans le système RRDE conçu pour évaluer le
phénomène de libération d'oxygène dans les oxydes lamellaires riches en Li.
ii) À la lumière des travaux précédents, une perte irréversible en oxygène des oxydes
lamellaires riches en Li pourrait faciliter la migration des cations et entraîner une transformation
de la structure. Par conséquent, nous avons effectuéune étude détaillée en utilisant un composé
riche en Li de composition nominale en Li1.2Ni0.13Mn0.54Co0.13O2, afin de fournir des informations
plus approfondies sur le couplage fort entre les changements structuraux et le redox de l’oxygène.
En élargissant la fenêtre de potentiel lors du cyclage nous avons proposé un scénario plus
holistique pour l'évolution structurale de Li1.2Ni0.13Mn0.54Co0.13O2 au cours du premier cycle
électrochimique.
Premièrement, en appliquant une tension constante à4,8 V, nous avons obtenu pour la
première fois une seule phase complètement chargée A’, permettant ainsi sa caractérisation
structurale détaillée. En combinant la diffraction des rayons X synchrotron, la diffraction des
neutrons sur poudre, la microscopie électronique à transmission (TEM), l’OEMS et le
Spectrométrie optique àplasma àcouplage inductif (ICP-OES), nous avons identifiéla phase A'
198
Résumé
comme ayant une structure de type O3 densifiée impliquant la migration de Mn dans les sites
octaédriques inter-lamellaires, contrairement aux rapports précédents qui déclaraient qu’il
s’agissait d’une couche densifiée1,2 ou d’une structure en forme de spinelle3 présente qu’à la
surface des particules. De plus, nous avons révéléque sa formation est causée par l'élimination
conjointe de Li+ et d'oxygène, ce qui crée des lacunes cationiques et anioniques. Les lacunes
cationiques sont partiellement comblées par des métaux de transition par migration dans les sites
octaédriques inter-lamellaires, ainsi supprimant arrangement cationique Li-M “en nid d'abeille”
(M étant un métal de transition). Les lacunes anioniques migrent à la surface et sont ainsi
comblées. Les lacunes anioniques migrent àla surface et sont ainsi comblées. En conséquence, le
ratio M/O augmente et de fait la structure se densifie.
Deuxièmement, nous avons démontréqu'une grande partie de la perte de capacitéau cours

du premier cycle est due àune barrière cinétique et peut être récupérées en appliquant une étape
de potentiel constant ou en abaissant la limite basse de potentiel. Notez que cette dernière
méthode n’est pas spécifique à Li1.2Ni0.13Mn0.54Co0.13O2. En effet, en étudiant l'efficacité
coulombique de LiNi1/3Mn1/3Co1/3O2 au premier cycle, Kasnatscheew et al. ont montré qu'une
partie importante de la perte de capacitéest d'origine cinétique et est liée àune diffusion lente de
Li+; par conséquent, ils ont recommandéd'appliquer une étape de potentiel constant àla fin de la
décharge pour combler les lacunes de Li.4 Une conclusion similaire a ététirée par Zhou et al. lors
de l'étude de l'électrode LiNi0,8Mn0,1Co0,1O2.5 En effet, une baisse du coefficient de diffusion du
Li+ àun état de décharge élevéa étéobservée pour de nombreux oxydes lamellaires,5,6 et peut
s'expliquer par i) la contraction du paramètre maille c se produisant généralement àune teneur
élevée en Li (x > 0,6, x étant la quantitéde Li);7,8 et ii) la lithiation inhomogène conduisant àune
très forte concentration de Li (manque de lacunes) àla surface des particules.9
Troisièmement, en utilisant des conditions très réductrices, nous avons révéléla formation
d’une nouvelle phase déchargée P”, responsable d’une capacité supplémentaire à bas potentiel.
En faisant varier progressivement la profondeur de charge de x = 0.0 (x étant la quantitéde Li
extraite pendant la charge) àx = 0.4 (région impliquant les métaux de transition uniquement,
redox cationique), et finalement àx = 0,6, 0,8 et 1,0 (région impliquant le réseau d’oxygène,
rédox anionique), nous avons en outre démontréque l'apparition de la phase P” est directement
199
Résumé
liée à l'activité redox de l'oxygène pendant la charge. L’insertion de Li+ supplémentaire à des
potentiels inférieurs à2,0 V a étérapportée pour d'autres oxydes lamellaires riches en Li10,11 et
Li-stœchiométrique.12,13 La sur-lithiation conduit généralement àla formation d'une phase de type
Li2MO2 avec une structure P3m1, confirmée par Robert et al.13 Dans le cadre de cette étude, nous
avons montréexpérimentalement que du Li+ supplémentaire pouvait également être insérédans
Li1.0Ni0. 33Mn0.33Co0.33O2, lors de la décharge du matériau non cycléà1,2 V. Par ailleurs, nous
avons montréque le composéLi1.2Ni0.13Mn0.54Co0.13O2 non cycléne peut pas être sur-lithiée et ce
même lorsque la tension limite de décharge est abaissée à1,2 V. Cependant, une fois le processus
redox de l’oxygène activé à haut potentiel, il est possible d’insérer du Li supplémentaire à bas
potentiel mais cela ne conduit pas à une transition de phase O3 → 1T comme observé pour
d’autres matériaux lamellaires. Ce comportement pourrait être associé à la présence d’ions Li+
dans les couches de métaux de transition dans Li1.2Ni0.13Mn0.54Co0.13O2, ce qui modifie la stabilité
respective de la phase P’ vs P’’. Toutefois, il a étémontréque d'autres oxydes lamellaires riches
en Li étaient capables d’insérer de manière réversible des ions Li+ supplémentaires (Li2+xIrO3,
Li3+xIrO4)14,15 entrainant une transition structurale de R3m àP3m1 doublant ainsi le nombre de
sites disponibles dans les couches de Li. Une explication plus raisonnable réside dans la
migration de Mn observée dans Li1.2Ni0.13Mn0.54Co0.13O2, déclenchée par l'activité redox
anionique, déséquilibrant les interactions électrostatiques entre les couches de MO2 ainsi que la
répartition intra-inter des couches de Li. De plus, la présence d'ions Mn dans l’inter-feuillet
pourrait servir de “pilier” pour empêcher le glissement des plans, ce qui est nécessaire pour
obtenir la structure 1T décrite précédemment pour Li2NiO2.16
Enfin, nous avons conçu un protocole inspirédes travaux de Singer et al,17 permettant de
restaurer les migrations de Mn dans la phase P' déchargée par recuit àtempérature modérée (>
250 °C). En effet, le processus redox observé à bas potentiel ainsi que le Mn dans l’inter-feuillet
disparaissent après traitement thermique, confirmant la corrélation entre la migration de Mn et
l'activitéredox à bas potentiel. De plus, la récupération de la superstructure et des plateaux de
potentiel en charge après le traitement indiquent que le désordre cationique est responsable du
changement de comportement électrochimique après la première charge. L’assemblage de
batteries tout solide pouvant fonctionner dans une large fenêtre de température permettrait de
200
Résumé
mieux évaluer l’impact des migrations cationiques sur les propriétés électrochimiques de ce
matériau.
Bien que ce nouveau protocole permet une récupération partielle de la capacité perdu
après la première charge, la densité énergétique n’a pas pu être augmentée ni par l’exploitation
des processus redox a bas potentiel ni par l’insertion de Li+ supplémentaire. De plus, la chute
progressive de potentiel au cours du cyclage persiste.
L'évolution structurale de Li1.2Ni0.13Mn0.54Co0.13O2 au cours du premier cycle a été

élucidée et est décrite dans la Figure 2. Premièrement, l'oxydation des cations Ni2+ et Co3+ a lieu
jusqu'à∆x = 0,4, suivie par l'oxydation de O2- qui déclenche la migration du Mn jusqu'à4,6 V (∆x
< 1,0). Ensuite, une partie de Li+ peut encore être extraite de manière concomitante avec le
dégagement d’oxygène pour atteindre la phase presque complétement délithiée, laissant des
lacunes d’oxygène dans la structure qui sont compensées par la migration de Mn dans l’inter-
feuillet. La phase ainsi obtenue après charge jusqu’à 4.8 V correspond à la phase “densifiée”
nommée A’ (∆x = 1.15). La densification implique que les lacunes d’oxygène formées au cours
de l’étape du dégagement d'oxygène migrent du cœur vers la surface, avec l'augmente de ratio
atomique M/O. Une phase pure A’ ne peut être obtenue que si une quantitésuffisante de Li+ est
éliminée et qu’on dispose d’un délai suffisant pour la migration des métaux de transition, ce qui
pourrait expliquer pourquoi la phase A’ n’a jamais été reportée pure. Lors de la décharge, les
réseaux anioniques et cationiques de la phase A' sont conjointement réduits et seule une partie
des cations migrés recouvre sa position initiale, ce qui explique que la phase déchargéP' diffère
de la phase initiale. Ces mécanismes sont réversibles au cours du cyclage (dans la présente étude
jusqu’à 26 cycles) mais le dégagement d’oxygène n’est observé qu’au cours de la première
charge.
201
Résumé
Figure 2 Schéma de la transition de phase A → A’ → P’.
iii) Nous avons étudié les principes fondamentaux liés à la chimie Li-CO2 qui a été
explorée par la communautédes batteries comme une valorisation du dioxyde de carbone.
Premièrement, nous avons démontré que différents mécanismes réactionnels sont mis
effectivement en jeu lors de la décharge dans les batteries Li-O2/CO2. Ces mécanismes dépendent
de la force de solvatation des solvants de l’électrolyte sur les ions Li+. Comme décrit dans le
Schéma 1, l'étape initiale lors de la décharge est la réduction électrochimique de l’oxygène en
superoxyde qu’elle est lieue en diméthylsulfoxyde (DMSO) ou en 1,2-diméthoxyéthane (DME).
Ensuite, le superoxyde réagit favorablement avec le CO2 dans le DMSO, solvant ànombre élevé
de donneurs, en raison de la forte solvatation du Li+ par le DMSO alors qu'il réagit
préférentiellement avec le Li+ dans le DME, solvant àfaible nombre de donneurs, pour former du
superoxyde de lithium. En conséquence, Li2CO3 est forméélectrochimiquement dans du DMSO
(désigné par voie électrochimique). Alors que dans le DME, Li2O2 est formé en premier de la
même manière que dans les batteries Li-O2 avant de réagir chimiquement avec le CO2 pour
former Li2CO3 (désignécomme voie chimique). Malgréla réactivitédifférente des superoxydes
mise en évidence dans ce travail, le carbonate de lithium est le seul produit final pour les batteries
Li-O2/CO2, indépendamment des solvants. Il convient de noter que, àla suite de nos travaux, un
scénario similaire a également étéétabli pour une batterie Li-CO2 assistée par des traces d’O2 (ce
qui peut être vue intuitivement comme une batterie Li-O2/CO2 avec un ratio CO2/O2 très élevé)
basée sur la spectroscopie Raman in-situ.18 Bien que le scénario susmentionné rationalise la
principale différence entre les mécanismes de décharges des batteries Li-O2/CO2, les étapes
réactionnelles suivant la réaction du superoxyde avec le CO2 dans le DMSO, restent àidentifier
(les voies de réaction du superoxyde donnant lieu àla formation de Li2O2 dans le DME ont déjà
étéétudiées en détail dans le contexte des batteries Li-O2). Dans le même sens, Qiao et al. ont
202
Résumé
mis l’accent, plus tard, sur l’importance du ratio CO2/O2 dans la détermination du mécansime
réactionnel du superoxyde avec le CO2 sur la base de mesures Raman.19 Plus précisément, le
peroxodicarbonate (C2O62¯) tend à se former dans des conditions avec un ratio CO2/O2 élevé,
tandis que le peroxomonocarbonate (CO42¯) est l’intermédiaire de réaction dominant à faible
ratio CO2/O2.
Deuxièmement, ces multiples voies de réaction entraînent des morphologies du Li2CO3

très différentes les unes des autres (forme d'oursin dans le DMSO contrairement àdes particule
dans le DME). De plus, la différence de capacitéentre les batteries Li-O2/CO2 et les batteries Li-
O2 dépend de la densité de courant de décharge, pouvant résulter de la compétition entre
nucléation et croissance, ainsi que de la différence de transfert de charge entre Li2CO3 et Li2O2.
Schéma 1 Illustration schématique de la formation de Li2CO3 dans des batteries Li-

O2/CO2 avec des électrolytes àbase de DMSO et de DME, suivant la voie électrochimique ou la
voie chimique.
Troisièmement, nous avons démontréque les batteries Li-O2/CO2 ne peuvent pas être des
batteries rechargeables àcause de ces deux raisonss: i) seul le CO2 est dégagémais pas d'O2 lors
de l'oxydation du Li2CO3 en charge et ii) les très hauts potentiels requis pour l'oxydation de
Li2CO3 conduisent àdes réactions parasitiques dramatiques. Pour surmonter ces limitations, une
203
Résumé
future direction intéressante consisterait àmodifier la réaction de décharge de manière àéviter le

“fauteur de troubles” qu’est le Li2CO3.
Enfin, nos résultats ont révélé que le CO2 ne peut pas être directement réduit dans les
solvants aprotiques dans leur fenêtre électrochimique (DME, DMSO et MeCN ont été étudiés
dans ce travail) en utilisant du carbone Super P comme électrode positive. De même, aucune
activité électrochimique dans les batteries Li-CO2 n'a été observée, contrairement à de
nombreuses publications. Il convient de mentionner une étude récente qui a démontréque, bien
que la batterie Li-CO2 présente une capacitéde décharge négligeable (en utlisant des électrodes
positives de graphène nues ou avec nanoparticules de Ru monodispersées à leur surface),
l'introduction de 2 % d'O2 (ratio volumique) peut conduire àune capacitéélevée de 4742 mAh g-1
(appelée batterie Li-CO2 assistée par O2).20 En général, la batterie Li-CO2 en est encore à ses
balbutiements et de nouvelles connaissances sont nécessaires pour expliquer cette différence
d'activitéélectrochimique.
iv) Compte tenu de la non réactivité du CO2 dans les solvants non aqueux utilisé dans
cette étude, nous avons évaluél'utilisation de quinones en tant que catalyseurs en solution, dans
le but de faciliter la réduction électrochimique du CO2 dans les batteries Li-CO2.
Tout d’abord, nous avons mis en évidence une complexation entre le CO2 et les dianions
de quinone. Aussi, nous avons étudié l’influence de la nature de la quinone, de la présence d’un
cation acide de Lewis (i.e. Li+) ou du solvant sur cette interaction. Plus spécifiquement, une
interaction plus faible avec le CO2 est observée pour les ortho-quinones que pour les para-
quinones en raison de l’empêchement stérique et de la répulsion coulombique induite par les
deux oxygènes en position ortho. Le cation lithium pourrait entraver les interactions entre la
quinone et le CO2 lorsque de forts appariements d’ions se forment entre le Li+ et les anions de
quinone. L'interaction de la quinone et du CO2 peut être modifiée en ajustant la force de
solvatation des anions de quinone par le solvant, par exemple, une interaction de quinone plus
forte avec le CO2 est observée dans le MeCN que dans le DMF.
Deuxièmement, en ajoutant du 2,5-di-tert-butyl-1,4-benzoquinone (DBBQ) àl'électrolyte

dans l’acétonitrile (MeCN), nous avons évalué le concept de réduction du CO2 induite par la
204
Résumé
quinone dans une batterie Li-CO2. Le potentiel de décharge est dictépar le potentiel de réduction
du DBBQ, qui est plus positif que celui de la réduction de CO2. De plus, le Li2CO3 a étéidentifié
comme le produit de décharge finale. Néanmoins, la capacitéde décharge limitée ainsi que les
résultats de résonance magnétique nucléaire suggéraient des réactions parasites impliquant le
DBBQ lui-même et/ou d’autres composants de la cellule.
Dans l’ensemble, l'utilisation de catalyseurs chimiques est une approche efficace pour
faciliter la réduction électrochimique du CO2 dans les batteries Li-CO2. Cependant, pour tirer
parti de cette approche, le catalyseur doit être correctement choisi pour fournir le potentiel rédox
approprié, une activité catalytique appropriée pour la réduction du CO2, ainsi qu'une bonne
stabilité électrochimique et chimique. Le côté de l’anode est également important et doit être
protégé par une membrane en électrolyte solide afin d'éviter l'effet “navette” (shuttle effect)
indésirable.
Références:
1 Koga, H. et al. Different Oxygen Redox Participation for Bulk and Surface: A Possible
Global Explanation for the Cycling Mechanism of Li1.20Mn0.54Co0.13Ni0.13O2. J. Power
Sources 236, 250-258 (2013).
2 Genevois, C. et al. Insight into the Atomic Structure of Cycled Lithium-Rich Layered
Oxide Li1.20Mn0.54Co0.13Ni0.13O2 Using HAADF STEM and Electron Nanodiffraction. J.
Phys. Chem. C 119, 75-83 (2014).
3 Yu, H. et al. Temperature-Sensitive Structure Evolution of Lithium-Manganese-Rich

Layered Oxides for Lithium-Ion Batteries. J. Am. Chem. Soc. 140, 15279-15289 (2018).
4 Kasnatscheew, J. et al. The Truth about the 1st Cycle Coulombic Efficiency of
LiNi1/3Co1/3Mn1/3O2 (NCM) Cathodes. Phys. Chem. Chem. Phys. 18, 3956-3965 (2016).
5 Zhou, H., Xin, F., Pei, B. & Whittingham, M. S. What Limits the Capacity of Layered
Oxide Cathodes in Lithium Batteries? ACS Energy Lett. 4, 1902-1906 (2019).
6 Radin, M. D. et al. Narrowing the Gap between Theoretical and Practical Capacities in
Li-Ion Layered Oxide Cathode Materials. Adv. Energy Mater. 7, 1602888-1602920
(2017).
205
Résumé
7 Song, B. et al. Mitigated Phase Transition during First Cycle of a Li-Rich Layered
Cathode Studied by In Operando Synchrotron X-Ray Powder Diffraction. Phys. Chem.
Chem. Phys. 18, 4745-4752 (2016).
8 Yin, S.-C., Rho, Y.-H., Swainson, I. & Nazar, L. F. X-ray/Neutron Diffraction and
Electrochemical Studies of Lithium De/Re-Intercalation in Li1-xCo1/3Ni1/3Mn1/3O2 ( x = 0
→ 1). Chem. Mater. 18, 1901-1910 (2006).
9 Mueller-Neuhaus, J. R., Dunlap, R. A. & Dahn, J. R. Understanding Irreversible Capacity

in LixNi1-yFeyO2 Cathode Materials. J. Electrochem. Soc. 147, 3598-3605 (2000).
10 Lyu, Y. et al. Correlations between Transition-Metal Chemistry, Local Structure, and

Global Structure in Li2Ru0.5Mn0.5O3 Investigated in a Wide Voltage Window. Chem.
Mater. 29, 9053-9065 (2017).
11 Lyu, Y. et al. Probing Reversible Multi-electron Transfer and Structure Evolution of

Li1.2Cr0.4Mn0.4O2 Cathode Material for Li-ion Batteries in a Voltage Range of 1.0 - 4.8 V.
Chem. Mater. 27, 5238-5252 (2015).
12 Kang, S.-H., Yoon, W.-S., Nam, K.-W., Yang, X.-Q. & Abraham, D. P. Investigating the
first-cycle irreversibility of lithium metal oxide cathodes for Li batteries. Journal of
Materials Science 43, 4701-4706 (2008).
13 Robert, R. & Novák, P. Switch of the Charge Storage Mechanism of

LixNi0.80Co0.15Al0.05O2 at Overdischarge Conditions. Chem. Mater. 30, 1907-1911 (2018).
14 Perez, A. J. et al. Approaching the Limits of Cationic and Anionic Electrochemical

Activity with the Li-Rich Layered Rocksalt Li3IrO4. Nat. Energy 2, 954-962 (2017).
15 Pearce, P. E. et al. Evidence for Anionic Redox Activity in a Tridimensional-Ordered Li-

Rich Positive Electrode β-Li2IrO3. Nat. Mater. 16, 580–586 (2017).
16 Davidson, I., Greedan, J. E., Sacken, U. v., Michal, C. A. & Dahn, J. R. Structure of 1T-
Li2NiO2 from Powder Neutron Diffraction. Solid State Ion. 46 243-247 (1991).
17 Singer, A. et al. Nucleation of Dislocations and Their Dynamics in Layered Oxide

Cathode Materials during Battery Charging. Nat. Energy 3, 641-647 (2018).
18 Zhao, Z., Su, Y. & Peng, Z. Probing Lithium Carbonate Formation in Trace O2-Assisted
Aprotic Li-CO2 Batteries Using In-Situ Surface Enhanced Raman Spectroscopy. J. Phys.
Chem. Lett. 10, 322-328 (2019).
19 Qiao, Y. et al. A Li2CO3-Free Li-O2/CO2 Battery with Peroxide Discharge Product.

Energy Environ. Sci. 11, 1211-1217 (2018).
206
Résumé
20 Wang, L. et al. Monodispersed Ru Nanoparticles Functionalized Graphene Nanosheets as

Efficient Cathode Catalysts for O2-Assisted Li–CO2 Battery. ACS Omega 2, 9280-9286
(2017).
207

WEI Yin 2019

Uploaded by

Copyright:

Available Formats

WEI Yin 2019

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

WEI Yin 2019

Uploaded by

Copyright:

Available Formats

Sorbonne Université

ED 397 – Physique et Chimie des Matériaux

Fundamental understanding of high-energy-density Li-ion batteries

Supervised by Jean-Marie TARASCON and Alexis GRIMAUD

To be presented and defended in public on 18 Octorber 2019 in front of the jury:

Dr. Fanny Bardé Program manager, IMEC, Belgium Referee

Prof. Mathieu Salanne Professor, Sorbonne Université, Paris Examiner

Dr. Alexis Grimaud CNRS researcher, Collège de France, Paris Co-supervisor

Table of contents ...................................................................................................... 1

Broader context and thesis outline ......................................................................... 4

Outline of thesis ..................................................................................................... 6

I. 1 Basic concepts of Li-based batteries and insertion-type electrodes .......... 8

I. 1. 2 From Li battery to Li-ion battery.................................................................................... 9

I. 2 Increase the energy density of positive electrode materials for Li-ion

I. 2. 2 Li substitution and the participation of oxygen redox .................................................. 14

I. 2. 3 Positive electrode materials with elevated (de)intercalation voltage ........................... 17

I. 3 Searching for new chemistries with potentially higher energy storage .. 19

I. 3. 2 Li-O2 battery ................................................................................................................. 22

II. 1 Background and motivation ...................................................................... 30

II. 2. 2 Online electrochemical mass spectrometry ................................................................. 33

II. 2. 3 Rotating ring disk electrode voltammetry ................................................................... 34

III. 1 Background and motivation..................................................................... 51

III. 2. 2 Detailed structural characterization of phase A’ ........................................................ 60

III. 2. 3 The impact of phase transition on the electrochemical properties ............................. 70

III. 3 Discussion and chapter conclusion .......................................................... 89

IV. 1 Background and motivation ..................................................................... 94

IV. 2. 2 Results and discussion ............................................................................................... 99

IV. 2. 3 Discussion and conclusion....................................................................................... 118

IV. 3 Li−CO2 batteries: CO2 conversion mediated by quinone derivatives 121

IV. 3. 2 Results and discussion ............................................................................................. 122

IV. 3. 3 Conclusion ............................................................................................................... 136

Chapter V General conclusion and outlook ...................................................... 137

Appendix ............................................................................................................... 136

A. 1. 2 Description of OEMS systems .................................................................................. 136

A. 2 Supporting information for Chapter II .................................................. 138

A. 2. 2 Supporting figures ..................................................................................................... 141

A. 2. 3 Calculation of the quantity of released O2 gas .......................................................... 144

A. 3 Supporting information for Chapter III................................................. 146

A. 3. 2 Supporting figures ..................................................................................................... 148

A. 4 Supporting information for Chapter IV ................................................. 164

A. 4. 2 Supporting figures ..................................................................................................... 167

Bibliography ......................................................................................................... 174

Broader context and thesis outline

This work is thus divided into five chapters:

Chapter II addresses the degradation phenomenon of Li-rich positive electrode materials

Going beyond the horizon of Li-ion batteries, Chapter IV is specifically dedicated to a

Chapter I General introduction

I. 1 Basic concepts of Li-based batteries and insertion-type electrodes

I. 1. 1 Intercalation chemistry and the advent of Li battery

I. 1. 2 From Li battery to Li-ion battery

I. 1. 3 Positive electrode materials for Li-ion batteries and their energy

I. 2 Increase the energy density of positive electrode materials for Li-

I. 2. 1 M/M’ chemical substitution

Figure I. 3 Typical discharge voltage profiles of different intercalation-type positive

Similarly, NMC series can be prepared by partially substituting Co with Ni and Mn in

I. 2. 2 Li substitution and the participation of oxygen redox

In an attempt to achieve higher capacities than LiMO2-type layered compounds, Li-rich

I. 2. 3 Positive electrode materials with elevated (de)intercalation voltage

Nevertheless, the practical application of these LiNi0.5Mn1.5O4-based spinel electrodes poses

ɛ DN AN O2 CO2 105Do2 105Dco2 Conductivity Viscosity