fulltext

Digital Comprehensive Summaries of Uppsala Dissertations
from the Faculty of Science and Technology 1362
Si negative electrodes for Li-ion
batteries
Aging mechanism studies by electrochemistry and
photoelectron spectroscopy
FREDRIK LINDGREN
ACTA
UNIVERSITATIS
UPSALIENSIS
UPPSALA
2016
ISSN 1651-6214
ISBN 978-91-554-9533-6
urn:nbn:se:uu:diva-281694
Dissertation presented at Uppsala University to be publicly examined in Polhemsalen,
Ångströmlaboratoriet, Lägerhyddsvägen 1, Uppsala, Thursday, 2 June 2016 at 13:15 for
the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty
examiner: Ph.D. Miguel Ángel Muñoz.
Abstract
Lindgren, F. 2016. Si negative electrodes for Li-ion batteries. Aging mechanism studies
by electrochemistry and photoelectron spectroscopy. Digital Comprehensive Summaries of
Uppsala Dissertations from the Faculty of Science and Technology 1362. 67 pp. Uppsala:
Acta Universitatis Upsaliensis. ISBN 978-91-554-9533-6.
This thesis is focusing on the challenges when using Si as a possible new negative electrode
material in Li-ion batteries. The overall aim is to contribute to a general understanding of the
processes in the Si electrode, to identify aging mechanisms, and to evaluate how they influence
the cycling performance. Another objective is to investigate how photoelectron spectroscopy
(PES) can be used to analyze these mechanisms.
LiPF6 based electrolytes are aggressive towards the oxide layer present at the surface of
the Si particles. With the use of fluoroethylene carbonate (FEC) as an electrolyte additive the
cycling performance is improved, but the oxide layer is still affected. A recently developed
salt, lithium 4,5-dicyano-2-(trifluoromethyl)imidazolide (LiTDI), is shown not to have any
detrimental effects on the oxide. The SEI with FEC and vinylene carbonate (VC) as contains
a high concentration of LiF and polymeric carbonate species and this composition seems
to be beneficial for the cycling performance, but the results indicate that additional aging
mechanisms occur. Therefore, electrochemical analysis is performed and confirms a continuous
SEI formation. However, it also reveals a self-discharge mechanism and that a considerable
amount of Li is remaining in the Si material after standard cycling.
PES is used in this work to analyze the SEI-layers as well as the surface and the bulk of the
Si material. With this technique it is hence possible to distinguish changes in the Si material as
a function of lithiation. To improve the data interpretation of PES spectra, a range of battery
electrode model systems are investigated. These results show shifts of the SEI peaks relative to
the electrode specific peaks as a result of the SEI thickness and the presence of a dipole layer.
Also other electronically insulating composite electrode components show relative peak shifts
as a function of the electrochemical potential.
To summarize, these studies investigate a number of well recognized aging mechanisms in
detail and also establish additional processes contributing to aging in Si electrodes. Furthermore,
this work highlights phenomena that influence data interpretation of PES measurements from
battery materials.
Fredrik Lindgren, Department of Chemistry - Ångström, Structural Chemistry, Box 538,
Uppsala University, SE-751 21 Uppsala, Sweden.
© Fredrik Lindgren 2016
ISSN 1651-6214
ISBN 978-91-554-9533-6
urn:nbn:se:uu:diva-281694 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-281694)
List of Papers
This thesis is based on the following papers, which are referred to in the text
by their Roman numerals.
I
II
III
IV
V
VI
Xu, C., Lindgren, F., Philippe, B., Gorgoi, M., Björefors, F.,
Edström, K., Gustafsson, T. (2015) Improved Performance of
the Silicon Anode for Li-ion Batteries: Understanding the Surface Modification Mechanism of Fluoroethylene Carbonate as
an Effective Electrolyte Additive. Chemistry of Materials, 27
:2591–2599
Lindgren, F., Xu, C., Maibach, J., Andersson A.M., Marcinek,
M., Niedzicki, L., Gustafsson, T., Björefors, F., Edström, K.
(2016) A hard X-ray photoelectron spectroscopy study on the
solid electrolyte interphase of a lithium 4,5-dicyano-2(trifluoromethyl)imidazolide based electrolyte for Si-electrodes.
Journal of Power Sources, 301:105-112
Lindgren, F., Xu, C., Marcinek, M., Niedzicki, L., Gustafsson,
T., Björefors, F., Edström, K., Younesi, R. (2016) SEI Formation and Interfacial Stability of a Si electrode in a LiTDI
Based Electrolyte with FEC and VC additives. Submitted manuscript
Lindgren, F., Rehnlund, D., Younesi, R., Xu, C., Edström, K.,
Nyholm, L. (2016) Aging mechanisms of composite nano-Si
electrodes in a Li-ion battery half-cell. In manuscript
Maibach, J., Lindgren, F., Eriksson, H., Edström, K., Hahlin,
M. (2016) The electric potential gradient at the buried interface
between lithium-ion battery electrodes and the SEI observed using photoelectron spectroscopy. Submitted manuscript
Lindgren, F., Hahlin, M., Edström, K., Maibach, J. (2016) The
Influence of Electrochemical Cycling and Side Reactions on
Battery Electrode Materials Analyzed by Photoelectron Spectroscopy. In manuscript
Reprints were made with kind permission from the respective publishers.
Contributions to the papers
Paper I
Contributed to the experimental work and took part in all discussions
Paper II
Planned and performed most of the experimental work, took
part in all discussions and wrote the manuscript.
Paper III
Planned and performed most of the experimental work, took
part in all discussions and wrote the manuscript.
Paper IV
Planned and performed most of the experimental work, took
part in all discussions and wrote the manuscript.
Paper V
Contributed to the discussions on PES data interpretation and
the manuscript.
Paper VI
Planned and performed the experimental work, took part in all
discussions and wrote the manuscript.
Other publications to which the author has contributed that are not included
in the present thesis:
Philippe, B., Derdryvere, R., Allouche, J., Lindgren, F., Gorgoi,
M., Rensmo, H., Gonbeau, D., Edström, K. (2012) Nanosilicon
Electrodes for Lithium-Ion Batteries: Interfacial Mechanisms
Studied by Hard and Soft X-ray Photoelectron Spectroscopy.
Chemistry of Materials, 24:1107-1115
Valvo, M., Lindgren, F., Lafont, H., Björefors, F., Edström, K.
(2014) Towards more sustainable negative electrodes in Na-ion
batteries via nanostructured iron oxide. Journal of Power
Sources, 245:967-978
Philippe, B., Valvo, M., Lindgren, F., Rensmo, H., Edström, K.
(2014) Investigation of the Electrode/Electrolyte Interphase of
Fe2O3 Composite Electrodes. Chemistry of Materials, 26:50285041
Valvo, M., Philippe, B., Lindgren, F., Tai, C., Edström, K.
(2016) Insight into the processes controlling the electrochemical
reactions of nanostructured iron oxide electrodes in Li- and Nahalf cells. Electrochimica Acta, 194:74-83
Contents
Summary in Swedish ...................................................................................... 9 Introduction ................................................................................................... 12 Li-ion batteries .............................................................................................. 14 Li-ion battery construction and operating principles................................ 17 Materials in the Li-ion battery .................................................................. 19 Reactions and aging mechanisms in the Li-ion battery ............................ 25 Silicon as negative electrode material ...................................................... 27 Scope of the thesis ........................................................................................ 32 Methods ........................................................................................................ 33 Half-cell assembly .................................................................................... 33 Electrochemical characterization ............................................................. 34 Scanning electron microscopy.................................................................. 35 Photoelectron spectroscopy ...................................................................... 35 Results and discussion .................................................................................. 41 The SEI ................................................................................................ 42 Cycling performance and aging of the Si material .............................. 46 Electrode morphology ......................................................................... 50 The Si material analyzed by PES......................................................... 51 Energy calibration in PES.................................................................... 56 Conclusions ................................................................................................... 60 Outlook ......................................................................................................... 61 Acknowledgements ....................................................................................... 62 References ..................................................................................................... 63 Summary in Swedish
Grundämnet kisel har en möjlighet att ersätta och förbättra dagens negativa
elektrodmaterial i litium-jonbatterier. Detta skulle innebära att batteriet kan
lagra mer energi per vikt och volym och att batteriernas prestanda ökar. Därigenom kan lagring av förnybar energi från sol och vindkraft bli mer lönsam
samtidigt som fossila bränslen i exempelvis fordon kan ersättas med bättre
batterier.
Kisel är ett av de vanligaste grundämnena på jorden och ingår i bland annat sand och många mineral. Tillgången är alltså god och kisel har inga negativa effekter på miljö eller människors hälsa. Tillverkningen av rent kisel
sker redan på stor industriell skala inom halvledarinstrustrin där det till exempel används i olika elektronikkomponenter och solceller.
När kisel används i ett Li-jonbatteri reagerar varje kiselatom med nästan
fyra litiumjoner. Detta medför att volymen ökar med upp till 280% vilket
kan medföra att materialet spricker sönder. Denna volymsökning är ett stort
hinder för att kunna tillverka effektiva elektroder baserade på kisel. De negativa effekterna av denna volymsändring kan minskas genom att kiselpartiklarna minskas i storlek. Med nanometerstora partiklar motstår materialet
volymsförändringar på ett bättre sätt och spricker därmed i mindre utsträckning. Att använda nanopartiklar har både positiva och negativa effekter. Det
är positivt eftersom kemiska reaktioner går fortare om partiklarna är mindre
och upp- och urladdningshastigheter borde därmed kunna ökas. Dock innebär små partiklar att elektrodens interna ytarea blir större. Med större ytarea
ökar också de oönskade sidoreaktionerna som ofta sker just på ytor och detta
leder till att batteriet får en förkortad livslängd. Ett sätt att mäta livslängd för
batterier är genom upprepade laddningscykler där kapaciteten mäts i varje
upp- och urladdning. Ett exempel på en sådan mätning kan ses i Figur 1 där
det framgår hur den tillgängliga kapaciteten miskar med antal cykler.
Det finns många sidoreaktioner som bidrar till att ett batteri tappar kapacitet eller ”åldras”. För just Li-jonbatterier är ett vanligt exempel på en sådan
sidoreaktion bildandet av ett ytskikt som i batterisammanhang kallas för
”solid electrolyte interphase” eller kort SEI. Detta SEI-skikt bildas på negativa elektrodmaterial på grund av att de lösningsmedel som används i elektrolyten inte är stabila vid de låga spänningar som dessa elektroder arbetar
vid. När SEI-skiktet bildas genom nedbrytning av lösningsmedel förbrukas
också batteriets laddning och batteriets prestanda försämras. För att inte batteriet ska tappa kapacitet är det viktigt att SEI-bildandet avstannar. SEI9
150
100
-1
Kapacitet [mAh g av aktivt material]
skiktets uppgift blir alltså helt enkelt att skydda lösningsmedlen i elektrolyten från vidare nedbrytning och konsumtion av laddning. För grafitelektroder
har acceptabla nivåer för nedbrytning av elektrolyt uppnåtts med tillsatser av
olika ämnen som bildar ett fördelaktigt SEI-skikt. I det här arbetet har liknande tillvägagångssätt använts för kiselbaserade elektroder och det har inneburit en väsentlig förbättring av livslängden. Resultaten visar att tillsatserna bidrar till att bilda SEI-produkter som är mer stabila och att detta även
bidrar till minskad SEI-bildning. Dock är SEI-lagret på kiselelektroden inte
tillräckligt stabilt för kommersiell användning. Detta beror på att för kiselelektroder måste SEI-lagret också kunna hantera stora volymsändringar från
den aktiva kiselpartikeln.
50
0
0
5
10
15
20
25
Antal urladdningar
Figur 1. Urladdningskapacitet som en funktion av antal laddningscykler för ett batteri där den positiva elektroden består av litiumjärnfosfat och den negativa elektroden är baserad på nanometerstora kiselpartiklar.
Också saltet i elektrolyten påverkar kiselpartiklarnas yta och det traditionellt sett mest användna saltet för Li-jonbatterier, LiPF6, kan vara aggressivt
mot den naturliga oxiden på partiklarnas yta. I den här avhandlingen har ett
nytt salt, LiTDI, prövats tillsammans med kiselelektroder. Ytanalyser har då
visat att dessa sidoreaktioner försvinner. Även om detta är en mycket positiv
effekt är det svårt att visa hur detta påverkar en kiselelektrod i det långa loppet. Detta beror på att det finns många andra sidoreaktioner som har större
inverkan på elektrodens prestanda än bara partiklarna ytstabilitet.
I detta arbete har även en annan process som bidrar till att kapaciteten
minskar i kiselelektroden identifierats. Detta sker genom att Li hålls fast i
kiselmaterialet och blir svårt att extrahera. Med långa urladdningsperioder
går en stor del av detta Li att få tillbaka men det är troligt att en ännu större
del ändå finns kvar i kiselpartiklarna.
10
Många av de reaktioner som pågår i en kiselektrod sker på partiklarnas
ytor. Därför är ytanalys av dessa ett av de bästa verktygen för att spåra de
reaktioner som skett och följa, bland annat, hur SEI-lagret byggs upp.
Ett verktyg för ytanalys som använts mycket i den här avhandlingen är fotoelektronspektroskopi. Detta är en mycket ytkänslig metod där tunna skikt
kan analyseras och metoden visar vilka kemiska föreningar en yta består av.
Med högenergetisk röntgenstrålning, som erbjuds vid till exempel synkrotronanläggningen BESSY II i Berlin, kan fotoelektronspektroskopi dessutom
användas för att få relativt sett mer signal djupare ner i provet. Detta gör att
materialens uppbyggnad kan identifieras vilket ökar förståelsen för SEIskiktets egenskaper. När både ytor och underliggande material kan analyseras samtidigt uppstår nya utmaningar i kalibreringen av resultat från mätningar med fotoelektronspektroskopi. Det är viktigt att denna kalibrering
görs korrekt för att slutsatserna inte skall bli missvisande. I denna avhandling exemplifieras svårigheterna när batterimaterial analyseras med fotoelektronspektroskopi och viktiga parametrar som påverkar kalibreringen lyfts
fram. Resultaten kan appliceras för att förstå resultaten från fotoelektronspektroskopimätningar på kisel men har även ett universellt syfte och påvisar
effekter som är viktiga att känna till för många material i olika slags elektroder.
Avslutningsvis har detta arbete har bidragit med att påvisa effekter från
traditionella elektrolyter. Dessutom har ett nytt elektrolytalternativ baserat på
saltet LiTDI utarbetats och dess egenskaper för kiselbaserade elektroder
påvisats. Vidare har många processer och sidoreaktioner i en kiselelektrod
identifierats och deras påverkan på elektrodens prestanda har utvärderats.
Högenergetisk fotoelektronspektroskopi är en av få metoder som kan användas för att analysera kisel och dess reaktionsprodukter när de används i litiumjonbatterier. En ökad insikt i hur resultat från sådana mätningar påverkas
av olika elektrodmaterial och processer som pågår i batterier kommer också
bidra till en förbättrad tolkning av framtida resultat.
Förhoppningen är att denna avhandling leder till vidare undersökningar av
de mekanismer som påverkar kiselelektrodens funktion. Högenergetisk fotoelektronspektroskopi är ett verktyg som kan påvisa hur kiselektroden påverkas under upp- och urladdningar och ytterligare mätningar kan bidra till en
ökad förståelse för processer som bidrar till försämrad kapacitet. Inte heller
arbetet med att kartlägga de mekanismer som påverkar kalibreringen av fotoelektronspektroskopidata slutar här utan är en process som komemr att
behöva uppdateras, i synnerhet när nya elektrodmaterial analyseras.
11
Introduction
Our society relies on a continuous energy supply to function properly. At
present, a large fraction of this energy comes from fossil resources and green
house gas emissions from these threaten the stability of our climate. To mitigate these effects, more renewable energy need to be implemented in the
energy system [1]. Examples of renewable energy sources are: biopower,
hydropower, as well as wind and solar power [2]. A challenge for all these
systems is the energy storage. Biomass degrades over time and requires well
developed infrastructure [3]. Hydropower requires large water reservoirs
which affect nature and wildlife [4]. Solar and wind power produce energy
only when the sun is shining or when the wind is blowing. If the energy is
not used right away, some kind of storage is needed.
A wide variety of energy storage systems are currently in use and these
stretch over a broad range of technical solutions. Hydropower and pumped
hydro rely on the potential energy of water at elevated altitudes. Electromechanical storage utilizes both the potential energy in e.g. compressed air
but also the kinetic energy in flywheels etc. Thermal energy storage is used
both domestically but also on larger scales. In batteries, electrical energy is
transformed to chemical energy by electrochemical reactions [5].
In Figure 2 the worldwide energy storage for cycle durations longer than
four hours is considered. The installed capacity from pumped hydro accounts
for a quantity on the GWh scale while the capacities of the other techniques
are better presented in MWh. However, the expansion of pumped hydro
storage technology is limited to a few geographical locations and requires
large infrastructure investments. To enable a more delocalized and versatile
energy storage from renewable sources, electrochemical storage in batteries
is a possible approach which fairly easily could be implemented in the power
grid [6]. Batteries can even offer a mobile energy storage system that already
today powers e.g. electric vehicles.
12
Electrochemical storage
50 MWh
Electromechanical storage
6 MWh
Pumped hydro storage
37 GWh
Thermal energy storage
120 MWh
Hydrogen storage
8 MWh
Figure 2. Installed energy storage worldwide designed for energy delivery durations
longer than 4 h [7].
Although batteries present a promising way of storing energy, it is so far an
expensive technology. Also the environmental impact from battery production should be considered for the whole life cycle of the battery [8]. By improving the performance and lowering the cost of Li-ion batteries, more renewable energy can replace fossil fuels and thus decrease the environmental
impact from human activities.
13
Li-ion batteries
A battery is a device that uses chemical reactions to produce electric energy.
If it is possible to reverse the chemical reactions by supplying electrical energy, the device can be recharged and is then referred to as a secondary battery. The very first battery was constructed by Volta in 1799 and was built
up by zinc and silver discs using cardboard soaked in acid as separator and
electrolyte, respectively [9]. Many different battery systems have been constructed since then. Some examples of the more familiar ones in the category
of secondary batteries are the so called nickel cadmium batteries, lead acid
batteries, nickel metal hydride batteries and the family of Li-ion batteries
[10]. Each battery chemistry has different properties, but the Li-ion batteries
generally offer higher capacities and longer lifetimes compared to the others
and yet the performance of these kinds of batteries is still expected to improve [11]. The benefit of using Li in batteries stems from the high eagerness of Li to share its electron and its low atomic weight and size [12].
Li has been used in battery research since the 1970s [13] and the first
commercial Li-ion battery was developed by Sony and released in 1991 [14].
This battery used lithium cobalt oxide as positive electrode and a carbonaceous material on the negative side [15]. More or less the same materials are
still in use but nowadays also other alternative materials have reached the
commercial stage [12]. Examples of commercialized positive electrode materials are lithium iron phosphate and various types of lithium metal oxides
based on nickel, manganese and cobalt. The negative side is still dominated
by carbon-based compounds but lithium titanate and also small additions of
silicon (3-5%) is being used for commercial cells [16]. The electrolyte is
mainly based on organic carbonates and a lithium salt but commercial cells
usually have many different additives to achieve certain properties [17].
The Li-ion battery can be produced in various sizes and shapes depending
on the application. Some different examples of Li based batteries are presented in Figure 3. The variety of electrode materials and electrolyte compositions also allow for many different possible combinations to make up a
complete Li-ion battery. Each different material has its unique properties and
by careful choice of materials, certain properties can be enhanced to render a
battery suitable for specific conditions.
14
4
1
3
5
2
6
Figure 3. Different kinds, formats and shapes of Li based batteries. 1 - Lab scale
half-cell (Li-metal vs. Si), 2 - Cell phone battery, 3 – Laptop cell (18650-type) 4 –
Primary Li-metal button cell, 5 – Laptop cell, 6 – Electric vehicle cell.
For all batteries, cost, safety and environmental impact are crucial parameters. Anyhow, batteries for electric vehicles require high safety even more
and also fast charge and discharge possibilities. Batteries for large scale storage, on the other hand, may have other systems to control safety, thus cost
and environmental impact are perhaps the most important parameters. Batteries for consumer electronics require high volumetric capacities but are
less cost sensitive.
The safety concern is a righteous one due to incidents reported for consumer electronics, electric vehicles as well as aircrafts relying on these batteries [18]. Indeed, a battery as an energy provider is in general a risky device as the two reacting components (negative and positive electrode) are
kept at micrometer distances in a container. In this aspect, batteries are very
similar to explosives where both fuel and oxidizers are kept together and the
energy can be released by relatively low activation energy. When a battery is
working according to plan, the separator makes sure that both electrodes are
not in contact and the outer circuit restricts the rate of the chemical reactions
to avoid excess heat and detrimental side products. In the case of unforeseen
actions such as overcharging, overheating or short circuit, the reactions may
15
proceed spontaneously and the risk of thermal runaway resulting in fire or
explosion of the flammable organic solvents is impending. However, a lot of
work has been conducted in order to improve Li-ion battery safety and with
proper means and precautions the safety seems to be within control [19-20].
If energy storage in Li-ion batteries should become more implemented in
the energy system, it must be assured that the production, use and recycling
of these actually have a positive impact on the climate. In this sense, life
cycle assessment of battery systems should be performed from the harvest of
new materials to the recycling process. To reveal the true impact of the use
of batteries, any emissions from production and recycling should be divided
by the total capacity delivered by the battery over its lifetime. Indeed, Li-ion
batteries have a good lifetime relative to other battery technologies and for
this reason, they also have a relatively low environmental impact [21]. However, it should be emphasized that nothing is gained, in terms of environmental impact, unless it is renewable energy that is stored in the batteries.
Also, when considering the cost, the most important parameter is the lifetime of the battery. If the initial cost is divided by the amount of energy that
the battery can store during its lifetime, the true cost of the battery is revealed. With this in mind, a battery that has an infinite lifetime could be
allowed to have a very high initial cost. Unfortunately, the lifetime of Li-ion
batteries is not infinite. Perhaps even worse, the lifetime is very complicated
to predict. Thus, it is challenging to make environmental and economical
predictions. However, it is clear that the performance and the lifetime of Liion batteries are parameters that need to be further improved.
16
Li-ion battery construction and operating principles
The Li-ion battery is comprised of a positive and a negative electrode that
both can store Li. It also requires an electrolyte for Li-ion transport between
the electrodes and an external circuit for electron conduction. In the charged
state Li is stored in the negative electrode. During discharge the external
circuit is closed and a Li-ion and an electron will leave the negative electrode and start to move towards the positive electrode material. However, the
ions can only be transported through the electrolyte and electrons are only
able to flow in the external circuit. When the electrons are moving through
the circuit they can be used to power an electrical device. A schematic illustration of this is presented in Figure 4.
In a secondary Li-ion battery, the Li-ion and the electron are possible to
force in the reversed direction by applying an external potential and by doing
this, the battery is charged. These processes can be repeated over and over
again and a Li-ion battery can maintain most of its original capacity for
thousands of cycles [22].
In Figure 4, the so called half-cell reaction that takes place at the negative
electrode during cycling is
LiC6  C6 + Li+ + e-
(1)
and the half-cell reaction proceeding at the positive electrode is the following
2 Li1/2CoO2 + Li+ + e-  2 LiCoO2
(2)
Together these two half-cell reactions results in the overall cell reaction
LiC6 + 2 Li1/2CoO2  C6 + 2 LiCoO2
(3)
When discharging the cell, these reactions proceed from left to right until the
graphite material is basically empty or the LiCoO2 is full. At that point the
cell potential decreases abruptly and the discharge operation is terminated.
An example of how this appears in a typical cycling curve is presented in
Figure 5 for a cell using a LiFePO4 positive electrode and a Si based negative electrode.
When charging the battery, reaction schemes 1, 2 and 3 proceed from
right to left. During the charge, the cell potential increases slowly until the
point where the negative electrode material is fully lithiated or when the
positive material is completely empty. As this happens, the cell potential will
increase rapidly and the charge is stopped at a certain cut off value. Also this
is depicted in Figure 5.
17
The overall cell reaction proceeds at different rates depending on battery
specific factors such as electrode material, electrical conductivity and electrolyte properties but also on external factors such as temperature etc. If these reactions can occur at high rates it allows for large charge and discharge
currents, which means that the battery can be charged and discharged within
short time frames. However, if these reactions proceed at too high rates,
excess heat will be produced with the risk of battery degradation and failure.
A common metric for batteries is the C-rate describing the current required
for a certain charge or discharge duration. The discharge rate C/10 e.g.
means that the full capacity of the battery is delivered over 10 hours. At 1C
the capacity is extracted in one hour and at 10C the discharge lasts for only
one tenth of an hour.
However, even with moderate C-rates, also unwanted reactions occur in
the battery that are not completely reversible and over time these may worsen the battery functionality. All these reactions and mechanisms that degrade
the battery performance are commonly called aging. To make the Li-ion
battery work better, it is important to identify the aging mechanisms and to
find materials that can minimize side reactions.
Load
_
External circuit
Electron
+
Current collector
Current collector
Electrolyte
Li ion
Graphite
Negative electrode
Separator
LiCoO2
Positive electrode
Figure 4. Schematic picture of a Li-ion battery with a LiCoO2 positive electrode and
a graphite negative electrode during discharge.
18
Materials in the Li-ion battery
To achieve a battery with a high energy density, both the cell potential and
the electrode capacities should be high. The cell potential (Ecell) is the difference between the half-cell reaction potentials for the two electrodes and can
be expressed as:
Ecell = Epostive electrode – Enegative electrode
(4)
The resulting cell potential for a full cell Li-ion battery can thereby be deduced from Figure 6 for some selected electrode materials. An example of a
cell comprised of a LiFePO4 (LFP) positive electrode and a Si negative electrode is presented in Figure 5 where the cell average working potential is
about 3 V vs. Li+/Li.
In Figure 6 also the electrode gravimetric capacities are presented. This
capacity is determined by the amount of Li that can be cycled reversibly (in
theory or in practice) divided by the mass of electrode material. Note that, in
Figure 6, the scale of the positive electrode materials is one tenth of that for
the negative electrode materials. The gravimetric capacity should normally
be considered relative to the weight of the entire battery unit. However, in
research on electrode materials, gravimetric capacities are usually given for
the electrode materials that are the focus of the research.
Battery voltage [V]
4
End of charge
3
2
End of discharge
1
0
Time [arb. unit]
Figure 5. Cycling curve of a lab scale battery comprising a LFP positive electrode
vs. a nano-Si negative electrode. The capacity as a function of cycle number for this
cell is presented in Figur 1.
19
-1
Positive electrode capacity [mAh g ]
0
100
LiMn2O4
300
400
500
LiCoO2
+
4
200
Positive
electrodes
LiFePO4
3
Li2FeSiO4
Cell potential
2
LTO
1
Graphite
Tin
0
0
1000
Lithium metal
2000
3000
Silicon
4000
Negative
electrodes
Potential [V vs. Li /Li]
5
5000
-1
Negative electrode capacity [mAh g ]
Figure 6. Potential and gravimetric capacities for different positive and negative
electrode materials.
Positive electrodes
The positive electrode materials are dominated by transition metal oxides
and lithium polyanionic compounds. The very first commercial cathode material used was LiCoO2 and this material is still employed today [12]. Cobalt,
however, has some drawbacks in terms of structural stability [23] and mining of cobalt is many times a questionable business [24]. To circumvent
these issues other metal oxides such as LiMnO2, LiNiO2 and LiFeO2 have
been investigated but they all suffer from poor stability and insufficient cycling performance [25]. By synthesizing oxides with mixed metals, the structural stability can be increased and nowadays there are examples of commercial cathode materials such as LiNi0.80Co0.15Al0.05O2 and LiNi1/3Mn1/3Co1/3O2.
Also other types of oxides have been investigated, and spinel LiMn2O4
(LMO) is one example that has reached the commercial market [16].
Instead of relying on the O22- anion, larger polyanions such as SiO44-,
PO43- and SO42- may provide increased structural stability [26]. A well
known member of the polyanionic class of positive electrode materials is the
olivine type LiFePO4 (LFP). The successful commercialization of this material, despite its relatively poor electronic and ionic conductivity, is possible
due to nano-sized particles and conductive coatings [27]. The achievements
regarding LFP have triggered research on similar polyanionic materials such
as Li2FeSiO4 (LFS) [28], LiFeSO4F [29] etc. and it is possible that future
positive electrodes will stem from this group of materials.
20
Negative electrodes
The ideal negative electrode for the Li-ion battery would be metallic Li. This
would give both high volumetric (2047 mAh cm-3) and gravimetric capacity
(3862 mAh g-1) [30], a high cell potential and no need of an additional current collector. Unfortunately, the use of metallic Li is a risky choice because
during charge, uneven deposition of Li may form so called dendrites which
could lead short circuit and thermal runaway [31]. There have been attempts
to avoid dendrite formation by employing different additives in the electrolyte and by replacing the liquid electrolyte with a solid electrolyte or a protective membrane [32-34]. A commercial example of the solid polymer electrolyte used with lithium metal electrodes is the batteries provided by
Bolloré [35]. Even if this is a step forward for the Li metal negative electrode, the solid polymer electrolytes do not allow for the high charge and
discharge rates required in all applications.
To circumvent dendrite problems, various types of carbon materials are
used as negative electrodes which can significantly decrease the dendrite
formation problem. The first commercial Li-ion battery comprised a petroleum coke amorphous carbon negative electrode [15] although nowadays,
natural graphite power the majority of commercial Li-ion batteries. Compared to lithium metal, carbon materials have a slightly higher lithiation
voltage and the theoretical capacity for the fully lithiated graphite is 372
mAh g-1. Although this capacity is more than twice as large as those for most
cathode materials, it is about ten times lower compared to the value for metallic lithium. In the graphite structure, Li-ions are intercalated in between
graphene layers and up to one Li for each six C atoms can be stored, forming
the LiC6 phase [36]. Also synthetic graphite has been investigated and the
properties of this material can be tuned to suit certain electrochemical applications, but the production requires high temperatures and it is not a cost
competitive alternative. Hard amorphous carbons have higher reversible
gravimetric capacities but the volumetric capacities are worse than for
graphite [37-38].
Another intercalation material that is successfully commercialized as a
negative electrode is Li4Ti5O12 (LTO) [16]. The electrochemical performance of LTO offers a flat lithiation plateau at 1.55 V vs. Li+/Li and a theoretical capacity of 175 mAh g-1 [39]. The relatively high lithiation potential
and the low gravimetric capacity result in an electrode with a fairly low energy density compared to graphite. On the other hand, the high potential
results in almost no side reactions with the electrolyte [40] and also enables
the use of aluminum current collectors. This is both a cheaper and lighter
alternative compared to the copper current collectors used for most other
negative electrodes [41]. Although LTO is a poor electron conductor and the
lithiation kinetics is fairly slow, modifications such as particle size decrease
and surface coatings now present LTO as a material with high charge and
21
discharge rate capabilities [16]. Cycling of LTO negative electrodes has been
successful for more than ten thousand cycles [42].
A second class of materials relies on the formation of intermetallic or alloy compounds with lithium. This could potentially increase the performance
of the Li-ion battery since the theoretical volumetric and gravimetric capacities are considerably higher than for the intercalation compounds [30]. Materials that are included in this group are i.e. Si, Sn, Sb, Al and Mg and mixtures of them. Among these Si has the highest energy densities and has
therefore been the subject for intense research efforts. Common for all these
materials is the large volume expansions that make reversible cycling problematic and commercially only small additions of Si in negative electrodes
have been reported [43]. Since Si is the main focus in this thesis, more in
depth information will be shared in a coming section.
A third class of possible future negative electrode materials also exists
and includes so called conversion type materials. A wide variety of compounds such as transition metal oxides, hydrides, fluorides, carbonates and
many more have the possibility to take up and release Li-ions reversibly.
These materials are generally simple and cheap to produce and offer higher
capacities than graphite. On the downside, the voltage hysteresis is relatively
large and the charge and discharge processes thus feature a poor energy efficiency [44].
Electrolytes
In commercial batteries the most widely used electrolyte system is based on
a Li salt dissolved in an organic carbonate liquid phase. This is a fairly simple electrolyte with excellent Li transport properties. However, the organic
carbonates operate outside their stability window on the negative electrode
side and are depending on the formation of an efficient solid electrolyte interphase (SEI). Also the flammable nature of these carbonates poses a safety
concern [15]. Polymer electrolytes have been used commercially as mentioned earlier and are a safer alternative but require slower charge and discharge rates [45]. A gel polymer electrolyte is a mixture of the solvent based
electrolyte and the solid polymer electrolyte and thereby combines the properties of the two [46] and has also been utilized in commercial cells [47].
The development of the existing electrolyte alternatives is an ongoing research field but investigations of new classes of possible future electrolyte
systems are also in progress. Among these, ceramic [45] and ionic liquid
electrolytes [48] have unique properties that may be suitable for certain battery applications. In this thesis organic carbonate electrolytes are used with
both well known and newly introduced salts as well as additives aimed to
improve the electrode performance.
Several salts have been investigated for use in Li-ion batteries with varying success. These stretch from purely inorganic salts such as LiAsF6, LiPF6
and LiClO4 etc. to more organic salts like lithium bis(trifluoromethyl)
22
sulfonimide) (LiTFSI) and lithium 4,5-dicyano-2-(trifluoromethyl)
imidazolide (LiTDI). The structural formula of LiPF6 and LiTDI are presented in Figure 7. A number of the salts are considered toxic or explosive while
other yield inefficient SEI layers while some have negative effects on aluminum current collectors [49]. The most widely used salt in commercial Li-ion
batteries is undoubtedly LiPF6 although the chemical and thermal stabilities
of this compound are questioned. Also, in case of battery malfunction and
accidents the LiPF6 salt may form toxic substances [50]. It is worth mentioning that different electrode material may have special requirements in terms
of salt properties and it is important that possible decomposition products are
not harmful to any components in the battery.
Solvents for electrolytes must have a polarity high enough to allow Li salt
dissolution and at the same time withstand the potentials present in the Liion battery without decomposing. These requirements are not easily fulfilled
since a range between 0 and 5 V vs. Li+/Li is reality for most Li-ion batteries. In commercial cells mixtures of e.g. ethylene carbonate (EC), diethyl
carbonate (DEC), dimethyl carbonate (DMC) etc. are usually used which
yield a good solubility of Li salts [51]. The structural formula of DMC and
EC is presented in Figure 7.
Electrolyte additives are used to provide different properties related to
SEI formation, safety and overcharge protection. SEI forming additives such
as FEC and VC (also in Figure 7) can form polymeric species improving the
stability and the safety of the negative electrode. Also flame retardant additives as well as overcharging redox shuttles have been investigated in order
to enhance safety. Commercial cells may have a complicated electrolyte
composition with many additives to ensure specific properties of the battery
[17].
F
F
Li+
F
N
F
N
N
F
F
P
F
N
F
Li+
Lithium 4,5-dicyano-2-(trifluoromethyl)imidazolide
(LiTDI)
F
Lithium hexafluorophosphate - LiPF6
F
O
H3CO
O
OCH3
Dimethyl carbonate
(DMC)
O
O
Ethylene carbonate
(EC)
O
O
O
O
O
O
Vinylene carbonate Fluoroethylene carbonate
(FEC)
(VC)
Figure 7. Structural formula for LiPF6, LiTDI, DMC, EC, VC and FEC.
23
Inactive battery components
The previously mentioned materials are all active in the chemical reactions.
To support these materials a number of inactive battery components are required which also are crucial to the battery performance. These are e.g. binders, conductive additives, separators, current collectors and the outer casing.
Modern Li-ion battery electrodes are so called composite electrodes. These usually consist of micro- or nanometer sized particles, a polymeric binder
and a conductive additive. The purpose of the binder is to keep the active
particles together and attached to the current collector. Binders are traditionally based on polyvinylidene difluoride (PVdF) but the benefits of using
water soluble binders are becoming more and more apparent, especially in
the research on new negative electrode materials. Replacing solvent based
chemistries with water based alternatives is good from both a health and an
environmental perspective. To enhance the electrical wiring a so called conductive additive is usually included in the composite. These comprise various types of soft carbons, carbon nanotubes or fibers that are dispersed in the
composite coating.
To keep the positive and the negative electrodes apart a so called separator is required. This is usually an electronically insulating and porous sheet
that allows for Li-ion diffusion. Many times these comprise a polymeric
membrane based on polyethylene or polypropylene. The separator is subject
to a considerable amount of research since it is an expensive component.
Also, the possibility to stop dendrite formation by the separator is considered
a viable way for metallic Li to be used in commercial cells.
Electrode composites are deposited on the current collector (a metallic
foil) that governs the electron transport out of the battery. The positive electrode usually uses an aluminum current collector while the negative electrodes are coated on copper. This part of the battery has a major impact on
the overall environmental and cost balance since aluminum production is an
energy demanding process and copper is expensive due to limited resources.
Finally some kind of packaging is required to protect the components
from air and moisture which otherwise would destroy the battery. Materials
used for packaging usually include air and moisture tight metallic and plastic
containers.
24
Reactions and aging mechanisms in the Li-ion battery
An example of the full cell reaction in a Li-ion battery is given in Scheme
(3). The redox reactions of the active materials are the primary reactions in
the battery and ideally these would be the only ones. However, as observed
in the example of LiCoO2, only half of the Li-ions in the material are possible to cycle reversibly. If more than half of the Li-ions are extracted, the
atomic structure of the host material may collapse and the battery will lose
capacity. The structural instability of the materials is only one of many
sources for battery degradation or aging. In Figur 1, a typical plot of the capacity vs. cycle number shows decreasing capacity due to one or several
aging mechanisms. Another rather familiar observation of aging is the inflation of Li-ion batteries due to internal gas evolution as a result of overcharging and decomposition of electrolyte solvents [52]. One such example of an
inflated Li-ion battery is shown in Figure 8.
Figure 8. Gas evolution inside a Li-ion battery due to overcharging (left) and
an identical cell before overcharging (right)
Aging mechanisms for positive electrode materials are often connected to
changes in the crystallographic structure as in the example of the LiCoO2
above. The intercalation mechanism of these materials also depends on a
stable structure to maintain a good Li-ion transportation. Another mechanism that degrades the performance is the cation dissolution into the electro25
lyte. In this process, the redox active metal is lost from the active material
and the amount of redox active sites decreases and in turn, the capacity fades
[53]. Another problem associated with the cation dissolution is that cations
may end up at the negative electrode and cause an acceleration of the SEI
growth on the negative electrode, also resulting in a fading capacity [54].
SEI formation is indeed a major source for aging in negative electrodes
[55]. It is necessary that the SEI offers good Li-ion conductivity and electrical insulation. The SEI should also be an impermeable layer that prevents
solvent diffusion to the electrode. The SEI layer usually consist of compounds such as LiF, Li alkyl carbonates, Li2O etc. In the formation of these
compounds, charge is irreversibly consumed and the battery capacity will
decrease. Therefore, the SEI properties are crucial and the SEI formation
must be kept at a minimum level to avoid capacity losses in full cell batteries. Also the structural integrity is threatened in, e.g. graphite negative electrodes as not only Li may be intercalated into its structure. An exfoliation of
the graphene sheets may occur as result of intercalation of larger molecules
such as solvents, also resulting in decreased performance [56].
Even the integrity of the entire composite electrode and inactive components may be subject to aging mechanisms. This could be due to, i.e. mechanical stress as a function of volume changes during cycling [55] or due to
binder decomposition resulting in deterioration of the composite structure
[57].
All aging mechanisms mentioned herein may affect the electron and Liion transport in the electrolyte and in the active materials. If the kinetics of
these processes is hindered this would result in an increasing overall cell
resistance. In turn the half-cell reaction potential would be shifted and the
energy efficiency would decrease. Also any cutoff limits would be reached
earlier meaning that the battery effectively will lose capacity.
To summarize, the Li-ion battery may be subject to many different side
reactions that may impair its performance and each component of a battery
has its individual aging mechanisms. Efforts on many fronts are therefore
required to mitigate battery aging and to improve the lifetime. By increasing
the lifetimes, the cost and environmental impact of batteries may be decreased.
26
Si as negative electrode material
Si is one of the most abundant elements on earth, it is non toxic, cheap, and
is produced in large quantities for the semiconductor industry. The very first
electrochemical lithiation of Si was performed at elevated temperatures in
the 1970s and proved that reversible cycling was possible [58]. During
lithiation several crystalline phases corresponding to the Li-Si phase diagram
have been observed in the galvanostatic cycling curve. The most lithiated
phase, Li4.4Si, has a theoretical gravimetric capacity of about 4200 mAh g-1
which even exceeds that for lithium metal itself [59-60].
At room temperature the lithiation of crystalline Si appears differently and
none of the phases observed at elevated temperatures are present. Instead,
the lithiation of crystalline Si has a flat plateau around 0.1 V vs. Li+/Li, see
Figure 9, indicating that a two phase addition reaction is taking place. It is
shown by XRD that during this two-phase reaction, the crystalline Si is
transformed into an amorphous Li-Si compound [61]. By the end of the
lithiation, when approaching 0 V vs. Li+/Li the crystalline Li3.75Si phase has
sometimes been observed [62-64] and sometimes not [65-66]. The formation
of Li3.75Si proceeds through the following reaction scheme:
3.75 Li + Si  Li3.75Si
(5)
and the resulting theoretical capacity for this phase is 3579 mAh g-1. Even
though this phase is sometimes detected and sometimes not, its presence is
revealed during de-lithiation where the curve experiences an extra plateau
around 0.45 V vs. Li+/Li. This short plateau is also observed during the first
de-lithiation (Figure 9).
1.0
Cell voltage [V]
2.4
+
Cell voltage [V vs. Li /Li]
0.8
1.6
1.2
0.8
0.4
0.0
0.6
SEI formation and
lithiation of CB and SiO2
2.0
0
40
80
120
160
200
-1
Capacity [mAh g ]
amorp
ho
0.4
us Si
amor
phous
Li Si
x
0.2
crystalline Si
0.0
0
1000
amorphous Li Si
x
2000
3000
4000
-1
Capacity [mAh g of Si]
Figure 9. The first lithiation and de-lithiation of a crystalline nano Si electrode. The
inset presents the details during the SEI formation at an early stage in the lithiation
step.
27
The presence of this phase seems to depend on the cycling rate as it is
formed during lithiation at low C-rates [43]. If this phase is avoided, either
by a cutoff value or by the use of high cycling rates, the de-lithiation curve
exhibits a more sloping shape with two quasi plateaus [64, 67]. The lower
potential quasi plateau has been suggested to stem from de-lithiation of an
environment where the majority of the closest neighbors are Li atoms while
the higher plateau would stem from de-lithiation of an environment where
the Si atoms make up the closest neighbors [68-69]. Likewise, during
lithiation the higher quasi plateau would be insertion of Li with mainly Si
neighbors and the lower plateau, Li insertion with mainly Li in the closest
vicinity. The overall sloping behavior of the cycling curve resembles the
lithiation and de-lithiation curves seen for a solid solution [43]. The detailed
local order in Li-Si materials formed at room temperature in Si electrodes is
complex and numerous investigations have presented a wide variety of local
Li-Si arrangements [70-74].
When inserting as much as 3.75 Li for each Si atom, a volume increase in
the range of 280% is expected [62]. This is schematically depicted in Figure
10 where a gradual transformation of Si to Li3.75Si is presented from left to
right. This causes internal stress and may result in fractures of the Si material
which in turn may lead to loss of electrical contacts and poor cycling performance [75]. By decreasing the Si particle size to the nanometer scale major improvements have been achieved [76-78]. For Si particles, the upper
size limit to avoid fractures has been reported to be in the range of 150 nm
[79]. Even with particles of approximately this size reports have indicated
formation of smaller particles [80]. A similar example from this work is
presented in Figure 11 where SEM micrographs before and after 100 cycles
reveal a particle size decrease as a function of cycling.
Figure 10. Simulation of the lithiation reaction of crystalline Si (left) gradually transformed to Li3.75Si (right). Reprinted with permission from [81]. Copyright (2012)
American Chemical Society.
28
Rather poor cycling performance of Si based electrodes is achieved when
using PVdF based binders [77, 82], which otherwise are the standard binding
agents for other electrode materials. By substituting the PVdF binder with
water based polymeric binders such as Na-CMC [82-84], poly acrylic acid
[85-86] and alginate [87] major improvements have been achieved concerning the cycling performance. One common denominator for these binders is
the carboxylate group which has been attributed to have good binding properties towards the active Si material [84, 87], a binding mechanism that can
be improved further by the use of a pH 3 buffered solution during the composite preparation step [88-89].
The electrolytes used for the vast majority of research on Si based electrodes are undoubtedly based on the LiPF6 salt in different carbonate mixtures [90-97]. Only a few studies of alternative electrolyte salts have been
reported with some improvements in cycling performance [92, 98-99]. However, the major improvement in the cycling performance of electrolytes for
Si-based electrodes came with the introduction of SEI forming additives
such as FEC and VC into the electrolyte [100-104].
The SEI formed by LiPF6 based electrolyte on Si consists of e.g. Li alkyl
carbonates, LiF, Li2CO3, Li2O, etc. from the breakdown of the electrolyte
[90-97]. In general the products that make up the SEI on Si electrodes are the
same as those found in the SEI on graphite and Li-metal negative electrodes
[105]. This is perhaps not surprising as the electrolytes investigated so far
have similar compositions. However, the volume changes for Si electrodes
during cycling are much more significant than for graphitic electrodes. With
volume expansion new surfaces may be exposed which require further SEI
formation and this may become a continuous process. Another important
aspect is that graphite electrodes generally consist of micrometer sized particles while Si electrodes often are comprised of nanometer sized particles.
This means that the overall surface area of a Si electrode is many times larger and since SEI formation occurs on surfaces, the amount of SEI formation
will increase with smaller particles.
With FEC additions to the electrolyte, the amount of LiF in these SEI layers increases significantly. Also, both FEC and VC have the possibility to
form polymeric carbonates [101-104]. These two additives have attracted the
largest interest and with the use of these, the cycling performance is enhanced. The benefit of these additives is generally considered to be due to
the thinner and more resistant SEI. This suggests that an inorganic and polymeric rich SEI layer is preferred.
Although the above just mentioned concept may be the general consensus
regarding the Si SEI, other studies report somewhat different results. Also
FEC free electrolytes have shown SEI layers with high LiF contents (from
LiPF6 decomposition), good passivating properties and relatively low levels
of other Li carbonate-like species. However, in this study an electrolyte with
the FEC additive was suggested to further enhance passivation due to the
29
polymeric decomposition product rather than LiF [99]. Similar results were
reported where the actual benefit of FEC addition were suggested to include
the formation of a polymeric layer closest to the particles surface [106]. The
SEI thickness has also been investigated using in situ ellipsometry and it was
actually reported to be thicker with FEC and VC additives [107]. In addition,
in this study the SEI was found to be dynamic in the sense that it was increasing during lithiation and decreasing during de-lithiation. Similar results
are obtained by ex situ PES during the first cycle of Si electrodes [90] as
well as on graphite electrodes [108]. An explanation for this behavior could
be the dissolution of SEI components into the electrolyte [109-110].
It is important to emphasize that almost all investigations on Si electrodes
have been performed in half-cells with a Li metal counter electrode, meaning
that a large surplus of Li (and charge) is available. Thus, any consumption of
Li in SEI formation will not lead to a capacity decrease of the Si material as
new Li can be retrieved from the Li metal whenever required. However, if a
thicker SEI results in slower Li transport, an increased polarization of the
cell would be seen. With increased polarization the cutoff voltage is reached
earlier and it will appear as a capacity loss.
Nonetheless, in a full cell the Li resources are limited and the continuous
SEI formation is a feature that prevents Si electrodes from being commercialized. However, continuous SEI formation is by far not the only aging
mechanism that exists for Si electrodes. As mentioned earlier, HF generated
from the LiPF6 and water is able to decompose and etch the Si oxide surface
[91, 97, 111-112]. It seems reasonable that such an effect would be detrimental to the electrode but exactly how the oxide layer and its composition
influence the cycling performance of Si based electrodes remains unclear
[113-114].
Other mechanisms responsible for fading capacity are often connected to
the volume variations during cycling resulting in decreased electrical contacts within the composite matrix or by complete loss of active material
[115]. However, careful cycling data analysis has identified incomplete delithiation as the largest source for irreversible capacity in Si electrodes [116].
Li remaining within the amorphous Si material after de-lithiation has also
been suggested [117] and recently detected with neutron reflectometry [118].
In connection with this, the slow reaction kinetics of Li-Si phases [67] and
also the capture of Li by unsaturated Si atoms [119] could be a reasonable
explanation. A similar discussion also exists for other intermetallic negative
electrode materials [120-122].
Although a tremendous amount of work has been spent on finding the
failure mechanisms of Si based electrodes and the SEI formation, a clear
picture has not yet been realized. This indicates the importance to continue
the research in new ways with different electrolytes and alternative approaches.
30
Pristine electrode
a
100 µm
c
1µm
After 100 cycles
b
100 µm
d
1µm
e
f
200 nm
200 nm
Figure 11. SEM micrographs showing the porous structure and the Si nanoparticles
of a composite electrode in pristine condition at three different magnifications (a, c
and e) and an identical electrode after 100 cycles in LiTDI based electrolyte with
FEC and VC additives (b, d and f). From the top and down the magnification increases. From Paper III.
31
Scope of the thesis
This thesis is focusing on Si and some of its challenges associated with its
use as a possible new negative electrode material in Li-ion batteries. The
overall aim is to contribute to the general understanding of processes taking
place in the Si electrode during lithiation and de-lithiation. One of the main
objectives is to identify and pinpoint different aging mechanisms and how
they influence the cycling performance. The purpose with this is to map the
most important negative effects in order to establish where the effort for
further improvements should be concentrated. One specific aging mechanism that should be addressed is the detrimental side effects on the Si material when LiPF6 based electrolytes are used.
This thesis also targets to assess the desired properties of a more efficient
SEI layer in order to provide a better insight of the functionality of this layer.
A wider understanding of the SEI layer may be useful also for other battery
materials.
Another main purpose with this work is to contribute to the knowledge
base when using different kinds of PES as tools to investigate Si and its SEI
formation. More specific, fundamental insights are required in the energy
calibration of PES spectra not only for Si based electrode materials but also
for other battery materials in general.
32
Methods
Half-cell assembly
To be able to evaluate the Si material during cycling, composite electrodes
were produced according to the following procedures. The basis for these
electrodes was a Si material, a binder and a conductive additive. The binder
was a Na-CMC salt dissolved in a water:ethanol mixture or in a pH 3 buffered solution of KOH and citric acid. An amorphous carbon was also used to
increase the electronic conductivity of the composite. These components
were mixed thoroughly by ball milling to ensure a homogeneous mixing.
The resulting paste was bar coated on Cu foil and dried. Also other types of
electrodes were produced in this way to obtain electrodes with different active materials. Composite electrodes with positive electrode materials were
fabricated in the same way with the same carbon black conductive additive
but the binder in these cases was PVdF-HFP dissolved in N-methyl-2pyrrolidone (NMP). These composite slurries were bar coated on Al foil.
The active material made up roughly 80 wt% of the dry components, the
binder about 10% and the carbon black (CB) conductive additive also about
10%. With the use of the buffered solution, the Si active material comprised
67% of the dried composite mass. In some experimental setups the working
electrode was a pure Au or Cu disc.
All electrodes were dried under vacuum inside an Ar containing glovebox
for at least 5 h at 120°C to remove water residues. Half-cells were then assembled where a lithium foil and the electrode were separated by a porous
polyethylene sheet. A schematic drawing of this is presented in Figure 12.
The two electrodes and the separator were placed in an aluminum reinforced
plastic bag with metal tabs connected to the backside of the electrodes. The
separator was soaked in electrolyte before sealing the pouch under vacuum.
Since one of the main focuses of this work was the interfacial stability of
the active Si-material and the electrolyte, some different electrolytes have
been used. LiPF6 based electrolytes were used for Si electrodes as well as the
positive electrode materials and the electrolyte based on LiTDI was investigated for Si-electrodes. Additives and their effect on the electrochemical
performance on Si material were investigated for both salts.
33
Li foil counter electrode
Porous separator
soaked in electrolyte
Si composite electrode
Ni connectors
Cu foil current collector
Figure 12. Schematic drawing of a composite Si electrode half-cell.
Electrochemical characterization
Electrochemical methods are naturally the main analysis techniques for battery materials. By monitoring the potentials during the lithiation and delithiation steps, information regarding the development of the active material
can be retrieved. A wide range of electrochemical tools for battery analysis
exist and the ones used in this thesis are briefly described below.
Galvanostatic cycling or cyclic chronopotentiometry
In galvanostatic cycling a constant current is applied or extracted from the
battery [123] and the cell voltage as a function of time is recorded. This
presentation is a so called cycling curve and examples of cycling curves are
presented in Figure 5 and Figure 9. The information obtained from cycling
curves includes overall reaction potentials, kinetics and also the amount of
charge or capacity that can be inserted or extracted during repeated cycling
procedures. By comparing the charge obtained during the lithiation and delithiation processes, the amount of irreversible reactions can be monitored.
The ratio of charge extracted divided by the charge inserted into the working
electrode during a cycle is usually referred to as the cycling efficiency or
coulombic efficiency.
In this work, galvanostatic cycling is the major technique used for battery
evaluation and sample preparation for PES measurements. Depending on the
material and the desired test, different currents, cutoff voltages or limited
cycling capacities were used. For Si electrodes, both cutoff voltages as well
as charge limited cycling were evaluated. The first cycle of Si based materials was generally performed with a considerably lower current than that used
for extended cycling to allow for the initial amorphisation of Si.
Cyclic voltammetry
In cyclic voltammetry (CV) the voltage of the working electrode is increased
or decreased at a constant rate between defined voltage values and the cur34
rent response is recorded [123]. The resulting voltammogram reveals the
potentials required to achieve oxidation and reduction. CV was used as a
complementary technique to form SEI-layers on metallic substrates.
Chronoamperometry
Chronoamperometry is a potentiostatic technique where a constant voltage is
applied and the current response as a function of time is recorded. In this
work, voltage steps were applied to extract residual amount of Li from the Si
electrode and also to mimic continuous SEI formation.
Electrochemical impedance spectroscopy
This technique records the current response as function of a wide range of
sinusoidal alternating voltage frequencies. Results from this technique are
generally challenging to interpret but could provide details of the kinetics of
processes within the active material and the SEI. However, in this work impedance spectroscopy was merely used to get a general overview of the total
impedance of half-cells.
Scanning electron microscopy
In the scanning electron microscope (SEM), an electron beam is directed
onto the sample and a detector collects the scattered electrons which are
converted to a micrograph over the surface. SEM was used to get an overview of the surface morphology as a function of cycling for electrodes. Sample transfer to the SEM analysis chamber could not be made completely inert
but the atmospheric exposure was minimized to about 1 minute.
Photoelectron spectroscopy
PES utilizes the photoelectric effect where high energy photons radiate a
sample which responds by emitting photoelectrons. The energy of the emitted photoelectrons can be detected and yields a spectrum over the photoemissions as a function of binding energy. This process is schematically presented in Figure 13 where different energy levels result in a photoemission
of a certain binding energy. The resulting binding energies can be used to
identify the elemental composition of the sample and all elements except
hydrogen can be subject to PES analysis.
35
PES spectrum of Cu
Intensity [arb. unit]
2p
3/2
2p
1/2
Auger electrons
2s
3p
3s 3d
1400
1200
2s
1s
1000
2p1/2
800
600
400
Binding energy [eV]
200
0
2p3/2
3s
3d
3p
Figure 13. Schematic representation of the photoelectric effect and the resulting
PES spectrum of Cu.
Depending on the closest neighbors of a certain atom, the measured binding
energy of a specific core level may vary to some extent. This variation is
enough to get a hint on the character of the surrounding atoms. Together
with core level spectra of other atoms in the sample, a picture of the composition can be obtained. An illustrative example showing the possibilities to
distinguish between different chemical surroundings is presented in Figure
14 from the C1s photoemission of ethyl trifluoroacetate. Different neighboring atoms give rise to a peak shift - this is what usually is referred to as the
chemical shift. The chemical shift arises due to the variation in core levels
that are altered with the number of neighboring atoms and their electronic
properties. In ethyl trifluoroacetate in Figure 14, the more electronegative
surrounding of the carbon atoms results in a more positive chemical shift. As
a result of this, the measured binding energies of these carbons increase with
a more electronegative surrounding.
36
Figure 14. C1s PES spectrum of ethyl trifluoroacetate. Reprinted from [124] with
permission from the publisher.
The spectrum is generated by analyzing the kinetic energy of the emitted
photoelectrons from the sample. Binding energies are correlated to the kinetic energy by the following equation
EK = hν – EB – ϕS – qϕ
(6)
where EK is the measured photoelectron kinetic energy, hν is the incoming
photon energy causing the photoemission (excitation energy), EB is the binding energy of the emitted photoelectron, ϕS is the spectrometer workfunction,
q is the electron charge and ϕ is the surface potential of the studied photoemission. Out of these parameters, the kinetic energy is measured but the
excitation energy as well as the spectrometer workfunction are known. For
the sake of simplicity, the surface potential is many times set to zero allowing the binding energy to be calculated. For highly conducting samples this
is an appropriate simplification but for insulating materials this surface potential should be considered and a further discussion on this topic is presented in a later section. A schematic energy diagram of the relationship between
these parameters is presented in Figure 15. In this figure the Fermi level (EF)
of the sample and the spectrometer are aligned since they are in electrical
contact.
37
Figure 15. Energy diagrams of sample and spectrometer for PES measurements.
Reprinted from [125] with permission from the publisher.
Surface sensitivity
Photoelectrons that travel through a sample are likely to interfere with other
atoms and may then lose some of their kinetic energy. These are the so
called elastically scattered photoelectrons which will produce a background
in the spectrum. Photoelectrons that manage to reach outside the sample
without losing its energy are the inelastically scattered photoelectrons and
they give rise to the emission lines or peaks observed in the spectra as in,
e.g. Figure 13 and Figure 14. The intensity (I) of photoelectrons from the
distance (d) into the sample can be expressed by the following equation
^
/ sinΘ
(7)
where I0 is the intensity due to photoelectrons at the very surface, λ is the
inelastic mean free path (IMFP) for a photoelectron moving through the
sample and Θ is the angle between the sample surface and the detector. In
Figure 16a, the probability of a photoelectron to reach the surface is depicted
as a function of distance into the sample expressed in λ. The major part of
the photoelectrons that manage to escape the sample without being elastically scattered thus originate from the atoms close to the surface. According to
Equation 7, the probability for ineastically scattered electrons decays exponentially and therefore 95% comes from a distance of 3λ within the sample.
38
a)
1.0
I(d)
0.8
0.6
0.4
0.2
0.0
63%
0
1
24%
2
8%
3
d [ ]
5%
4
5
Inelastic mean free path [nm]
b)
Elements
Inorganic compounds
Organic compounds
10000
1000
100
10
1
0.1
1
10
100
1000 10000
Electron kinetic energy [eV]
Figure 16. a) The probability of inelastically scattered photoelectrons as a function
of d expressed as IMFP or λ. b) How the IMFP varies as a function of electron kinetic energy for pure elements, inorganic and organic compounds as described by
Seah and Dench [126].
The value of the IMFP depends on the kinetic energy of the emitted photoelectron an in Figure 16b, the relationship between IMFP and photoelectron kinetic energy is presented. In PES photoelectrons of kinetic energies
from 100 eV up to 10 000 eV are analyzed.
The kinetic energy of the photoelectron can be altered by changing the
energy of the incoming photon. With soft x-rays (hundreds of eVs) this implies that λ is less than 1 nm. With Al Kα radiation (1.49 keV) the same
number is 1 - 2 nm and with hard x-rays (2 keV to 10 keV) it is possible to
obtain λ values in the range of 3 – 10 nm. By varying the incoming excitation energy a depth resolved picture can be obtained. This is schematically
depicted in Figure 17 where an idealized layered structure is presented. It
should be noted that these interpretations are valid for a specific photoemission, a constant IMFP and where each layer has a unique chemical shift.
Nonetheless, this figure illustrates the concept of depth resolved PES analysis using different excitation energies.
If the sample is built up by a random distribution of components with different chemical shift, the appearance of the spectra will be independent of
the excitation energy. This is further illustrated in the random structure sample presented in Figure 17 where the different components have different
chemical shift and thus appear at different binding energies.
39
Layered structure
Random structure
Bulk material D
Short inelastic mean free path
C
I(d)
0.8
0.6
A
87%
8
0.4
0.2
0.0
13%
87%
0
1
2 3
d [
6
4
2
0
Binding energy [eV]
C
B
C
D
A
B
25%
25% 25% 25%
B
B
8
0.4
6
2
0C
C4
C Binding energy
C [eV]
0.2
0.0
<1%
4
0.6
B
0
5
1
2 3
d []
4
D
D
D - 5%
C - 8%
6
4
2
0
Binding energy [eV]
0.4
0.2
0.0
63%
0
8%
24%
1
2
4
D
0.8
5
C
A
D
CD
B
D
25% 25%
C 25% 25%B
BA
8
4
2D 0
Binding energy [eV]
D6
C
C
0.4
0.0
C
C
0.6
D
A
0.2
5%
3
d []
Intensity [arb. unit]
8
B - 24%
Intensity [arb. unit]
0.6
I(d)
I(d)
0.8
B
B
A mean free path
Long inelastic
A
1.0
D
C
A
Long inelastic mean free path
C
A
5 B
A
B
A
63%
B
D
B
1.0
B
A
D
D
0.8
C & D <1%
I(d)
1.0
A
B
A
1.0
B - 13%
Intensity [arb. unit]
Short inelastic mean free path
Intensity [arb. unit]
Layer A Layer B
Layer C
surface
D
B
0
1
2
3
4
5
d [
Figure 17. Schematic representation of a layered and a randomly structured material and the corresponding PES spectra when analyzed with different excitation energies.
In this work PES is used to analyze mainly composite electrodes with a
three dimensional structure of particles in the nano- and micrometer sized
regime. Since the size of the analyzed surface is many times larger than that
of a single particle, this means that the escape angle of the photoelectron is
effectively independent on the sample angle vs. the analyzer. Another effect
that should be considered for PES on a nanomaterial is the relative increase
in signal from surface layers compared to flat surfaces [127]. This effect as
well as the changing surface morphology makes surface layer thickness estimations complex.
PES in this work is used with different excitation energies to achieve a
depth profile of the surface layers. The HIKE end station [128] at the KMC1 beamline at the synchrotron facility BESSY II operated by the HelmholtzZentrum Berlin [129], was used for achieving high kinetic energy photoelectrons. Low kinetic energy PES is performed at the MaxIV synchrotron facility in Lund, Sweden, beamline I411 [130]. Also in-house PES (or XPS) was
used with a AlKα radiation source of 1486.7 eV.
40
Results and discussion
The basis for this thesis is work presented in six articles. A short summary of
the basis for this thesis and the general findings follows here.
The etching of the Si surface by the use of LiPF6 based salts is a recognized aging mechanism for Si electrodes. To address this issue, the FEC
electrolyte additive and the resulting SEI layer is investigated for Si based
electrodes. Although FEC enhances the cycling performance, the surfaces of
the particles are still affected by the electrolyte (Paper I). A new electrolyte
salt (LiTDI) is therefore investigated and the Si is shown to be stable towards this salt and its decomposition products. However, the cycling performance with this electrolyte is poor (Paper II). By using FEC and VC additives together with the LiTDI salt, a cycling performance comparable to that
for LiPF6 based electrolyte with FEC is achieved and the Si surface is not
affected by etching (Paper III).
Although the LiTDI based electrolyte with additives presents a more stable alternative, the cycling performance is not significantly increased. Thus,
other aging mechanisms must be more important. In order to evaluate other
possible aging mechanisms, a thorough electrochemical analysis is performed. This study reveals a self-discharge mechanism during storage in the
lithiated state and also that a significant amount of Li is lost in the Si material during standard cycling (Paper IV).
To get further insights into the mechanisms behind aging of the Si material, more analysis tools are required. One such tool that is able to probe both
the formation of SEI layers, the surface of the active material and also a part
of the bulk material is HAXPES. Although general PES is a well established
technique, battery materials are a relatively new class of samples that require
further insights to provide accurate interpretations. Therefore, a model to
explain the effects observed in PES spectra due to SEI formation is presented (Paper V). To further highlight the effect of SEI formation and other parameters influencing the measured spectra, a series of model systems is investigated (Paper VI) and the results indicates the influence of local potential
variations within the electrode.
The following sections will more deeply explain and discuss the results of
this work.
41
The SEI
The SEI formation is investigated with different kinds of PES at synchrotron
facilities as well as with in-house PES. With different incoming photon energies a depth profiling of the material is achieved which indicates the
buildup of the SEI, similar to what is described in Figure 17. Although these
investigations have included different salts, solvents and additives, the formation of the SEI has several features in common.
Initial SEI formation and continuous formation during cycling
In the cycling curves, the first SEI formation is indicated already at 1.6 V vs.
Li+/Li, see e.g. the inset in Figure 9. With the most surface sensitive measurements electrolyte decomposition is observed at 0.9 V (Paper I) but the
major SEI formation occurs at potentials around 0.5 V and is clearly observed in cycling curves, as exemplified by Figure 9.
The initial SEI formation is completed when the lithiation plateau of Si is
reached at about 0.1 – 0.2 V. At this stage the Si material is starting to increase in volume and if the SEI is not persistent, new surfaces will be exposed to electrolyte and form more SEI. This is what is considered continuous SEI formation due to volume changes. Another kind of continuous SEI
formation is suggested to originate from solvent diffusion through the SEI
and where it is reduced on the Si surface in a self-discharge mechanism.
The continuous SEI formation is proceeding throughout the first 50 cycles
as indicated by the changes in the C1s spectra in Figure 18. Between 50 and
100 cycles the C1s spectra is almost identical and the SEI composition is
stable. It is challenging to conclude which of the two continuous SEI formation mechanisms that is dominating and both of them may occur simultaneously. The decrease in particle size and the rearrangement of the composite electrode would suggest that new surfaces are formed while in coming
sections both self-discharge and continuous SEI formation is observed.
One effect that is observed with more cycling is that the relative amount
of CB signals decreases. This is observed in Figure 18 (black peak) and Figure 20 (yellow peak). This could simply be explained by an increase in the
SEI layer thickness. However, another possible explanation is that with cycling, more surfaces are formed that also need to be passivated by SEI formation. Thus, the relative amount of SEI vs. CB increases and thereby the
relative CB signal should decrease, although the SEI layer may be of the
same thickness. Again, both mechanisms are probable.
When observing the SEM micrographs in Figure 24, significant morphological differences are observed between electrodes cycled with and without
the FEC additive. Despite this difference, the corresponding spectra reveal
rather similar CB signals (Figure 20, yellow peaks). This suggests that the
CB signal is perhaps not the most reliable indication of SEI layer thicknesses.
42
2005 eV excitation energy 6015 eV excitation energy
CB
CB
CF3
CF 3
OCV
Intensity [arb.unit]
C-H/C-C
C-H/C-C
SEI formation
CO3
CO3
1 cycle
Poly CO3
Poly CO3
25 cycles
C-O/C-N
C-O/C-N
50 cycles
O-C-O
O-C-O
100 cycles
295
290
285
Binding energy [eV]
295
290
285
Binding energy [eV]
Figure 18. C1s spectra recorded at excitation energies of 2005 and 6015 eV for
electrodes cycled in 0.6 M LiTDI electrolyte with FEC and VC additives. From
Paper III.
The chemical composition of the SEI
Since the SEI is composed of the decomposition products from the electrolyte, it is not surprising that different electrolytes result in different SEI layers. Without the FEC and VC additives, the composition of carbon and oxygen containing compounds is essentially the same both for LiPF6 and for
LiTDI-based electrolytes. Mainly Li alkyl carbonates, LiF, and other solvent
decomposition products are observed. With these electrolytes the LiF content is approximately 4%. If FEC is added to either of the electrolytes, the
LiF content increases significantly to 15-20%.
It is interesting to compare the relative Li and F content. As the major part
of F in the SEI is present as LiF, the amount of Li available for other compounds can be roughly estimated. Without additives, about 10% Li is available for other compounds such as Li alkyl carbonates while with additives,
about 5% and even less Li is found in something other than LiF. The SEI
layers with additives also comprise more of a compound considered to originate from the FEC and VC decomposition into polycarbonate-like species.
43
This confirms the decomposition mechanisms suggested for these additives
[102, 138]. The peak due to these compounds is located at around 291 eV
and is present in both Figure 18 (green peak) and Figure 20 (red peak).
The SEI structure
It is interesting to know how the SEI is built up as it reflects the SEI formation process. The depth profiling is achieved using different excitation
energies during PES analysis. It is also valuable to analyze samples frequently during the SEI formation to be able to follow the development of new
features.
Without the use of additives, the buildup of the SEI is essentially homogeneous. This is observed as different excitation energies result in similar
spectra, see Figure 19 and Figure 20. The buildup of an SEI layer can be
followed by the spectra in C1s, Figure 18, from the LiTDI based electrolyte
with FEC and VC additives. A layered structure is observed already after the
first cycle as there is a difference concerning the relative amounts of the
polycarbonate (green in Figure 18) and Li alkyl carbonates (yellow) between
the two excitation energies. With more cycling, the polycarbonate species
are increasing in relative amounts closer to the surface. This can be observed
since the green peak is more intense relative to the yellow peak at 2005 eV
than it is at 6015 eV. The same observation is made also for LiPF6 based
electrolytes when additives are used, see Figure 20.
With the most surface sensitive measurements in Paper I, the formation of
this polycarbonate compound is found to occur at higher potentials, or before, the formation of other carbonate-like species in the first cycle. With
this observation in mind, it is interesting that Li alkyl carbonate species at a
later stage are formed beneath this layer, or closer to the surface of the Si
material. This is interpreted as a result of two different mechanisms. The
first is the possibility of dissolution of Li alkyl carbonate species from the
outermost SEI layer into the electrolyte, thus enriching the more stable/less
soluble polymeric carbonates. In the other mechanism solvent molecules are
diffusing through the SEI layer and are reduced on the electrode surface.
This continuous SEI formation would then consume charge from the electrode in a self-discharge process.
44
x
-CO3
-CO3
lithium
silicate
-CF3
-CF3
Si1s - 6015 eV
Li Si
295 290 285 295 290 285
Intensity [arb. unit]
C1s - 6015 eV
Pure LiTDI
electrolyte
LiTDI with
additives
C1s - 2005 eV
1843 1837
2005 eV
6015 eV
N1s - 2005 eV O1s
LiTDI with
additives
LiTDI
Poly C=O
_
2005 eV
6015 eV
F1s
LiF
C=O
_
-C-O_
C-F
Li O
_
Pure LiTDI
electrolyte
2
403
398
533
528
688
Binding energy [eV]
Intensity [arb. unit]
Binding energy [eV]
683
Figure 19. Comparison of PES spectra from electrodes cycled 50 times with LiTDI
based electrolyte either with FEC and VC additives or without additives (pure LiTDI
electrolyte). From Paper III.
Stability of the electrolyte salt
Both LiTDI and LiPF6 salt decomposition is observed already after assembling the half-cells and storing them at the open circuit potential. This is
revealed by the presence of LiF in the electrodes stored in the pure electrolytes as the only source of F in these cases is the LiTDI or LiPF6 molecule.
Other decomposition products are observed by monitoring the salt specific photoemissions. For LiPF6, P is unique for the salt in these electrolytes
and decomposition products such as phosphates are observed. With the use
of FEC, an additional peak is observed in the P1s spectra in Figure 20. This
indicates that a different decomposition product is present with the use of the
FEC additive. The HF formation in this electrolyte is indicated due to the
silicon-oxygen/fluoride species observed in the Si spectra in Figure 28.
Thus, the FEC does not provide protection of the silicon oxide surface.
Also the LiTDI salt is shown to decompose but little is known about the
degradation reactions of this salt. One decomposition product of this salt is
LiF as indicated by the F1s spectra with pure LiTDI electrolyte in Figure 19.
45
Also an additional peak (grey in Figure 19) is indicated in the N1s spectra
both with and without the use of additives, revealing that a new N surrounding exists.
To summarize these findings on SEI characterization: the SEI composition, structure and electrode morphology are clearly affected by the choice of
electrolyte. With the use of additives a layered SEI structure is observed but
the salt decomposition continues. Also the SEI formation is a continuous
process that may proceed due to different mechanisms. In the following sections it will be demonstrated that the additives provide an enhanced cycling
performance. Therefore it is at hand to suggest that the layered SEI, rich in
polycarbonates and LiF is preferable over mainly a homogeneous SEI
formed by Li alkyl carbonate-like compounds.
Figure 20. PES results from SEI formed on Si electrodes by LiPF6 based electrolytes
with additives (black spectra) and without additives (blue spectra). From Paper I.
Cycling performance and aging of the Si material
Cycling of the composite Si electrodes in this thesis is performed
galvanostatically with either a charge limited lithiation or a lithiation to a fix
cutoff voltage. By restricting the lithiation with these limits only a part of the
available Li storing capacity is utilized and volume variations are therefore
limited. De-lithiation is in all cases performed with a cutoff value of 0.9 or
1.0 V vs. Li+/Li.
With the use of electrolytes without additives, the capacity losses in each
cycle are fairly high. This is especially the case for pure LiTDI electrolyte
which has a coulombic efficiency of only 96% and also a poor overall cycling performance. The LiPF6 based electrolyte yields a higher coulombic
efficiency but the major improvement is achieved with the use of FEC and
VC additives. With these, coulombic efficiencies of about 99% can be
achieved and it even increases slightly with cycling. With capacity limited
46
cycling to 1200 mAh g-1 of Si, it is possible to achieve more than 200 cycles
with the LiTDI based electrolyte with additives. It is a considerable improvement compared to electrolytes without additives but still, 99%
coulombic efficiency indicates far too much side reactions and lost capacity.
By monitoring the voltage profiles from extended cycling, presented in
Figure 21, it is observed that the lithiation plateaus are shifted to lower voltages with more cycles. The voltage profile during de-lithiation is essentially
unaffected but the quasi-plateau is extended and the voltage increase towards
the end of these curves becomes sharper with more cycles. These observations are in line with the conclusion that Li is remaining in the Si material
even after the de-lithiation [116].
Cycle 50
Cycle 100
Cycle 200
0.8
+
Cell voltage [V vs. Li /Li]
1.0
0.6
0.4
Decreasing
0.2
0.0
0
200
400
lithiation vo
ltage
600
800
1000
1200
-1
Capacity [mAh g ]
Figure 21. Cycling curves for a composite Si electrode lithiated to 1200 mAh g-1
and de-lithiated to 1.0 V vs. Li+/Li. The electrolyte is 0.6 M LiTDI with FEC and VC
additives. From Paper IV.
Incomplete de-lithiation of the Si material
To investigate the possibility that Li is remaining in the Si material after
cycling, a constant voltage step is added to the standard cycling procedure.
After every 10th or 20th de-lithiation the voltage is maintained at 1.0 V until
the current reaches below 1 µA. During such a step, additional time is allowed for slow processes to occur and release Li. The extracted capacity is
presented in Figure 22a) as the sum of the standard de-lithiation capacity and
the capacity extracted by the voltage step.
It is obvious that a considerable amount of capacity is possible to extract
with these voltage steps. With more cycles, the capacity that is possible to
extract increases and after 200 cycles an additional 330 mAh g-1 of Si is re47
1600
1200
800
De-lithiation capacity
including a voltage step
De-lithiation capacity in the
cycle after the voltage step
400
0
0
50
100
150
Cycle number
200
c)
150
300
100
200
50
100
0
Voltage step duration
0
1.0
Cycle 50
Cycle 100
Cycle 200
0.8
+
+
Extracted capacity
in voltage step
400
50
100
150
Cycle number
200
0
d)
1.0
Cell voltage [V vs. Li /Li]
Extracted capacity in constant
-1
voltage step [mAh g of Si]
b)
Lithiation capacity
De-lithiation capacity
0.6
Beginning of second
quasi-plateau
0.4
0.2
0.0
First quasi-plateau
0
400
Cell voltage [V vs. Li /Li]
-1
Capacity [mAh g of Si]
a)
Constant voltage step duration [h]
trieved. Interestingly, the time and the capacity in each voltage step are correlated and increase similarly, see Figure 22b). This suggests that with even
longer step durations, more capacity would be possible to extract. Therefore
the true amount of “lost” Li could be higher and this experiment merely reveals that a considerable amount of Li remains in the Si after cycling.
In the cycling curves before the voltage step (Figure 22c) it is observed
that the major part of the cycling curve is in the first quasi-plateau and that
the second quasi-plateau is reached at the end of the cycle. In Figure 22d),
the cycling curves after a voltage step is presented. In these cycling curves it
is apparent that essentially the first quasi-plateau is extended and the second
quasi-plateau is not possible to observe. The capacity lost in the following
cycles is also obvious as the de-lithiation plateau is shortened.
800
-1
Capacity [mAh g of Si]
1200
Cycle 51
Cycle 101
Cycle 201
0.8
0.6
0.4
0.2
0.0
First quasi-plateau
0
400
800
1200
-1
Capacity [mAh g of Si]
Figure 22. Standard cycling with voltage step. In a) the lithiation and de-lithiation
capacities as a function of cycle number are presented. Also the additional capacity
extracted in the potential step is shown as well as the capacity loss in the cycle following a potential step. In b) the capacity extracted by the potential step and the
potential step duration is presented as a function of cycle number. In c) and d) the
cycling curves before and after selected potential steps are presented, respectively.
From Paper IV.
Self-discharge in the lithiated state
As mentioned earlier, a self-discharge process by a continuous SEI formation is suggested by PES results on the SEI. In order to estimate the magnitude of this process and the amount of capacity it consumes, a selfdischarge experiment is evaluated. This is done for the LiTDI electrolyte
48
with additives used in paper IV. An electrode is paused for a number of
hours in the lithiated state during cycling. After the pause, the standard delithiation step is performed and the capacity extracted is plotted against the
pause time, see Figure 23a). This procedure is repeated but with increasing
pause time durations.
A significant capacity loss is observed with longer pauses. However, the
rate at which capacity is lost decreases with the pause time according to a
non-linear relationship. If the self-discharge is governed by a linear diffusion, the de-lithiation capacity vs. the square root of the pause time should
result in a linear relationship. This presentation is given in Figure 23b) and a
decent approximation is achieved. If the de-lithiation capacity is instead
plotted vs. time0.7, an almost perfect linear relationship is observed (see Paper IV). This suggests that the self-discharge process very well may be diffusion controlled, but perhaps not in a strictly linear fashion.
The origin of this self-discharge mechanism is not established but it is
possible that continuous SEI formation and other mechanisms are responsible. Nonetheless, the capacity losses due to shorter pauses are significant and
if these processes are active also during cycling, this mechanism should be
responsible for a major part of the capacity losses in each cycle.
b)
1200
1160
1120
1080
0
200
400
600
800
Pause in lithiated state [h]
De-lithiation capacity
-1
after pause [mAh g of Si]
De-lithiation capacity
-1
after pause [mAh g of Si]
a)
1200
1160
1120
Linear fit
y = 1193 - 3.3x
2
R = 0.984
1080
0
10
20
30
1/2
Pause in lithiated state [h ]
Figure 23. a) De-lithiation capacity as a function of pause time in the lithiated state.
b) De-lithiation capacity as a function of the pause time in the lithiated state1/2.
From Paper IV.
In summary, the cycling performance of the Si electrode is depending on the
electrolyte where FEC and VC additives enhance the cycling performance a
great deal. Still, with the use of additives and non aggressive salts, a large
capacity loss is observed during cycling. The electrochemical methods with
potential step de-lithiation and the self-discharge test indicate the presence of
effects that can explain large fractions of the capacity losses in each cycle.
Thus it is clear that more than the electrolyte needs to be considered to obtain efficient Si electrodes.
49
Electrode morphology
The pristine composite electrode has a porous structure without cracks and
individual particles are easily distinguished. Examples of this are presented
in Figure 11a), c) and d) and also in Figure 24a) and b). After cycling the
morphology of the electrodes are heavily altered. With LiPF6 based electrolyte without additives, the porous structure is significantly changed (Figure
24d) and cracks are observed with low magnification (Figure 24c) after 85
cycles. When the FEC additive is used, the electrode still has the porous
structure after 85 cycles and the particles appear less smooth, presumably
due to SEI formation and surface reconstructions. There is no indication of
crack formation or loss of active material with the use of the FEC additive.
Similar observations are also made for LiTDI based electrolytes with and
without additives. In Figure 11, a larger Si particle size is used and it is observed that the overall particle size is decreasing upon cycling. Similar observations are also reported in the literature [80].
Thus, with the use of additives, the porous structure is effectively maintained and there is no sign of cracks or loss of active material that would
explain the capacity losses. The decrease in particle size is easily observed
with the larger particles but it is possible that also the smaller nanoparticles
may decrease in particle size due to extended cycling, although it is more
challenging to observe by SEM.
Figure 24. SEM micrographs of a pristine electrode (a and b), an electrode cycled
in a LiPF6 based electrolyte without FEC additive (c and d) and with FEC additive
(e and f) after 85 cycles. From Paper I.
50
The Si material analyzed by PES
Intensity [arb. unit]
The active Si material in this thesis is based on Si particles with a size of
either about 50 nm (paper I and II) or 100-200 nm (paper III and IV). The
spectra for the Si1s signal of a pristine electrode are presented in Figure 25.
With low electron kinetic energy, the SiO2 signal is strong and also nonstoichiometric oxides (SiOx) are observed. With 6015 eV excitation energy
the bulk Si peak is now more intense and the SiO2 peak is only a small fraction of the total signal. This confirms that a thin native oxide layer is covering the Si particles. From this data it is challenging to estimate the oxide
thickness [127] but native oxides for Si materials are thin and generally in
the range of only a few atoms thick [131].
Si1s measured at 2005 eV
Bulk Si
Bulk Si
SiO 2
SiO x
1848
Si1s measured at 6015 eV
1844
SiO 2
1840
Binding energy [eV]
1848
1844
1840
Binding energy [eV]
Figure 25. The Si1s photoemissions measured at 2005 and 6015 eV of a pristine
composite Si electrode. The spectra are uncalibrated. Previously unpublished data.
With thick SEI layers the Si signal is heavily attenuated. However, the high
kinetic photoelectron energies obtained in HAXPES makes it possible to
probe Si materials covered by relatively thick SEI layers. Although it is possible to measure the photoemission spectra from the Si particles, the interpretation of the spectral changes due to a lithiation reaction is complex. This
is due to fundamental aspects of Si being a semiconductor, the nature of LiSi compounds and the lack of stable energy reference points.
The lithiation mechanism studied by PES
In the first lithiation the particles undergo a transformation from crystalline
Si to amorphous LixSi: this is evident due to the lithiation plateau in e.g.
Figure 9 [62]. In order to investigate how this lithiation mechanism proceeds
and what the corresponding spectral response is, several electrodes at different state of lithiation are analyzed. The spectrum from electrochemically
lithiated electrodes and reference Si wafers with different treatments are
presented in Figure 26.
The reference Si wafer, where the non analyzed side had been allowed to
react with Li metal (Figure 26b), revealed a shift to higher binding energies
and a similar shift is also observed for the Si electrode after SEI formation
(Figure 26e). This could be interpreted as a doping effect where the Fermi
51
level of the Si is increased. With increasing lithiation of the Si electrode, a
new, presumably more lithiated phase emerges, indicated with a yellow to
brown peak, denoted LizSi, in Figure 26f to i. With increasing lithiation, the
highly lithiated peak increases in relative intensity. At the same time, both
the grey (LixSi) peak and the more lithiated phase (LizSi) are shifted similarly to lower binding energies.
Si1s
C1s
Si
a)
0.0 eV
SiO
b)
0.0 eV
SiO
C-H
Pristine Si wafer
2
Li x Si
"Li doped" Si wafer
2
Lithiated
c)
oxide
0.0 eV
Li z Si
Lithiated Si wafer
CB
Intensity [arb. unit]
Si
d)
-0.5 eV
SiO
C-H
Pristine nano-Si
electrode
2
Li x Si
Lithiated
e)
oxide
-0.3 eV
After SEI
formation
-1
Lithiated to 700 mAh g
f)
0.0 eV
Li Si
z
g)
0.8 eV
Lithiated to
-1
1200 mAh g
h)
-0.3 eV
Lithiated to
-1
2000 mAh g
Lithiated to
-1
4000 mAh g
i)
-0.1 eV
1848
1844
1840
Binding energy [eV]
1836
297
293
289
285
Binding energy [eV]
Figure 26. Si1s and C1s spectra for a pristine Si wafer a), Li doped Si wafer b),
lithiated Si wafer c) and composite electrodes at different degrees of lithiation (d to
i). The spectra from wafers are uncalibrated, the pristine electrode d) is calibrated
with respect to the SiO2 peak in the pristine wafer a) and the other electrodes (e to i)
are calibrated vs. the lithiated oxide signal in the lithiated Si wafer c). The calibration required to place each spectra at the current position is indicated, for d) the
calibration shift was -0.5 eV, for e) the shift was -0.3 eV etc. From Paper VI.
52
The lithiated and de-lithiated states of Si electrodes are also presented in
Figure 27 (from paper IV) after the first, 25th and 50th cycle (at 1200 mAh g-1
of Si). It is seen that more and more of the highly lithiated phase is required
to achieve the desired lithiation capacity. Also in this case both bulk peaks
are shifted to lower binding energies with more cycling, a sign that the overall Li content in the Si increases. This observation is in good agreement with
the hypothesis regarding remaining Li within the Si material as discussed in
a previous section. However, in the de-lithiated state, practically no similar
peak shifts are observed although presumably more and more Li should be
present in the electrode with more cycles. This could be explained if the
remaining Li is located in the interior of the particles or deeper in the electrode and does not allow for detection. Alternatively, the Li remaining in the
structure is evenly distributed so that all bulk Si atoms has identical surroundings and thus only one chemical shift.
Intensity
[arb. unit]
Pristine electrode
Bulk Si
Si 1s
Surface oxide
1845
1841
1837
Binding energy [eV]
Lithiated state
Si 1s
De-lithiated state
Si 1s
Intensity [arb. unit]
Li x Si
Surface oxide
Li x Si
First cycle
Li z Si
Surface oxide
25 cycles
*
50 cycles
1845
1841
1837 1845
1841
1837
Binding energy [eV]
Binding energy [eV]
Figure 27. Si electrode in the pristine state as well as lithiated and de-lithiated electrodes after the first, 25th and 50th cycle with capacity limited cycling. The electrolyte is 0.6 M LiTDI with FEC and VC additives. After 50 cycles, in the lithiated
state, a new peak is observed attributed to a more lithiated oxide, indicated with a *
in the figure. From Paper IV.
53
1852
Li x Si
Si-O
Si-O / Si-F
Si-F
Intensity [arb. unit]
Si surface reactions
The active Si materials have also been analyzed by PES to trace the surface
stability and the formation of new phases. It is shown in other studies that
the Si surface is affected by the LiPF6 based electrolytes where presumably
water residues results in HF formation which then gives rise to etching of the
oxide surface and the formation of new species [91, 111-112]. This is also
seen in Figure 28 (unpublished data) where oxygen and fluorine containing
Si compounds are formed after 70 cycles in a LiPF6 based electrolyte.
With the use of LiTDI based electrolytes there is no sign of the formation
of fluorinated Si species or other unwanted reaction products at any stage.
Thus, this salt and its decomposition products seem to be harmless towards
the Si material.
The relative amount of silicon oxide is increasing with cycling. This can
be observed in Figure 27 but also in another study with a different electrolyte
[92]. One reason for this increase may be due to the particle size decrease, as
revealed by SEM. This would result in relatively more surface area and since
PES is surface sensitive technique, the surface signal would increase. A
thicker oxide layer could also explain the increase in the oxide signal.
The reversible cycling of different silicon oxides is demonstrated in a
number of studies e.g. [132-134]. It is therefore a reasonable assumption that
also the thin oxide layer in the nanoparticles electrode may be able to reversibly cycle Li. By comparing the oxide signal for lithiated and de-lithiated
electrodes in Figure 27, only a small peak (indicated *) positioned in between surface oxide and bulk peaks is observed after 50 cycles. This peak
only tells us that minor or insignificant amount of a more lithiated oxide
exists and this phase should give only a minor contribution to the overall
cycling capacity. However, also the major oxide peak could potentially store
some Li. To be able to differentiate between a lithiated and a de-lithiated
oxide, a different spectrum calibration is required.
1848
1844
Binding energy [eV]
1840
Figure 28. Si 1s spectrum measured at 2600 eV excitation energy of a nano-Si electrode after 70 capacity limited cycles. The electrolyte is 1 M LiPF6 with FEC and
VC additives. Previously unpublished data.
54
One challenging aspect for the PES analysis of composite Si electrodes is
the choice of energy reference point. When the surface hydrocarbon peak is
used as the energy reference, the formation of the SEI layer results in relative peak shifts of the underlying material which is why this calibration is
unsuitable. The CB peak offers a good option as it should be in good electrical contact with the active material and the relative peak position of CB is
also well established. However, the CB material may likewise undergo a
lithiation reaction and how this affects the measured binding energy is not
fully understood. The native oxide on the Si particle is indeed also in good
contact with the bulk Si but the oxide may be subject to a binding energy
change due to a reaction with Li. This makes the comparison between spectra in the lithiated and de-lithiated state virtually impossible and calibrations
should be specific for the lithiated or the de-lithiated state, respectively. Still,
with the oxide as the internal reference, relative binding energy changes are
probed and provide valuable information regarding changes in the bulk material with respect to the surface oxide.
HAXPES is a useful tool to probe the changes in a Si material as a function of cycling and both bulk and surface changes can be observed. However, the changes in the Si occurring during lithiation are not fully understood.
A more fundamental insight into the lithiation mechanism would provide
meaningful details to the aid data interpretation.
55
Energy calibration in PES
The basis of all data interpretation in PES measurements is the energy calibration of the spectra. These interpretations enable the assignment of peaks
to certain compounds. If the energy calibration is not carefully performed,
the interpretations may be incorrect. In this section some phenomena occurring in PES analyses of battery electrodes affecting the binding energy and
energy calibration are exemplified.
Peak shifts and changes due to the degree of lithiation
When analyzing active materials in the Li-ion battery, a change in chemical
shift is expected due to lithiation or de-lithiation. These shifts depend largely
on the involved mechanisms. As an example, LMO electrodes follow a fairly
well defined redox chemistry with increasing/decreasing peaks as a response
to the lithium content (Paper VI). In this sense, lithiation of Si is more complex with both new peaks and shifting peak positions. The difference can be
explained by the different lithiation mechanisms involved. LMO is an intercalation material where one Li is inserted for each formula unit while
lithiation of Si results in an intermetallic compound where also the semiconducting properties of the material may change when up to 3.75 Li for each Si
is inserted.
For the LMO example, only the +IV oxidation state is expected in the delithiated state. However also +III and +II oxidation states of Mn exist. With
increasing Li content the +III and +II oxidation states increase as expected.
A detailed peak de-convolution is necessary to provide a well resolved picture of the changes in the active material. With LMO, the spectral changes
are due to the insertion of Li+ which increase or decrease the amount of Mn
atoms with a certain oxidation state. In these spectra, the peak position from
the CB conductive additive is stable and effectively no internal energy referencing is needed for these peaks.
Peak shifts due to SEI formation
It is well known from other areas of PES that insulating layers on conductive
substrates are subject to relative peak shifts as a function of layer thicknesses
[135-138]. The formation of a SEI layer on a more conducting electrode is a
similar situation and peak shifts relative to underlying materials due to SEI
formation are seen for Si electrodes in Paper II and for graphite and
Ni0.5TiOPO4 electrodes in Paper V.
To further establish the relationship between the SEI formation and the
peak shifts, the build-up of an SEI layer on a Au electrode is investigated.
The uncalibrated spectra from this system are presented in Figure 29a) and
c). These investigations allow for a combined measurement of both the bulk
Au4d5/2 peak and the SEI components in the C1s spectra. The decrease in
Au4d5/2 peak is proportional to the increase in SEI thickness. Indeed, the
56
peaks for the SEI components are shifted to higher binding energies for
thicker SEI layers. However, the peak shift is most pronounced (1.4 eV) in
the first cycle. This is observed by comparing peak positions for the pristine
Au-disc sample with those for the “1 cycle” sample in Figure 29c. The
Au4d5/2 peak decreases to about one third for these samples. Comparing the
Au4d5/2 peak after 1 and 150 cycles, the intensity decreases to less than one
tenth and yet the peak shift is “only” 0.5 eV. Thus, the main part of the observed peak shift is due to the formation of a surface dipole layer that is established during the first cycle.
a)
Spectra as measured
Au 4d 5/2
Surface hydrocarbon
calibration
b)
Au 4d 5/2
C1s
C1s
285 eV
Intensity [arb. unit]
285 eV
Pristine Au-disc
Pristine Au-disc
Soaked in electrolyte
Soaked in electrolyte
Start of SEI formation
Start of SEI formation
1 cycle
1 cycle
x10
50 cycles
x10
150 cycles
x 10
x 10 150 cycles
335 325 315 305 295 285
335 325 315 305 295 285
Binding energy [eV]
c)
50 cycles
Spectra as measured
Binding energy [eV]
d)
Surface hydrocarbon
calibration
C1s
C1s
285 eV
Intensity [arb. unit]
285 eV
Pristine Au-disc
Pristine Au-disc
Soaked in electrolyte
Soaked in electrolyte
Start of SEI formation
Start of SEI formation
1 cycle
1 cycle
50 cycles
50 cycles
150 cycles
150 cycles
295
290
285
Binding energy [eV]
295
290
285
Binding energy [eV]
Figure 29. a) and b), combined Au4d5/2 and C1s spectra recorded for SEI deposited
on gold for increasing amount of CV cycles. c) and d) show the C1s region for the
spectra in a) and b), respectively. Spectra in a) and c) are uncalibrated and b) and
d) are calibrated vs. surface hydrocarbons. From Paper VI.
57
With this dipole layer, the local potential of the SEI layer will be shifted
with respect to the spectrometer and therefore shifted in binding energy. In
Figure 15 and Equation 6, this surface potential would be represented by ϕ.
A more detailed representation of how the energy levels and the results are
affected by such a dipole is presented in Paper V.
In Figure 29b) and d) the spectra are calibrated vs. surface hydrocarbons
set to 285 eV. This kind of calibration is meaningful for further analysis of
the SEI components where characteristic SEI compounds such as Li alkyl
carbonates now align at ~290 eV. With this calibration the Au4d5/2 peak is
shifted to lower binding energies. This illustrates how inappropriate surface
hydrocarbon calibration is for the bulk materials as the shift would suggest
that the Au material is subject to e.g. a chemical shift due to a chemical reaction, which is not the case.
The influence of the electrode potential
As discussed earlier, changes in the spectra are expected for the active materials as a function of lithiation. However, also peak positions shifts for electrochemically inactive components are observed between the lithiated and
de-lithiated states in e.g. the Ni0.5TiOPO4 electrodes in Paper V. This effect
is perhaps even more pronounced when the material has a large potential
difference between lithiated and de-lithiated states. To further investigate
this effect, a composite electrode with four different materials was assembled. This allows the electrode to be stopped during cycling at four well defined potentials at which the electrode is analyzed with PES.
The results, presented in Figure 30, reveal that inactive components such
as the PVdF binder (green peaks in Figure 30a) and b) are shifted as a function of electrode potential. However, a similar shift is also observed for the
LTO active material (Figure 30c) while peak positions from the CB conductive additive and the LMO active material (Figure 30d) remain essentially
unaffected. The common denominator for materials that exhibit peak shifts
(LTO and PVdF) as a function of electrode potential is that they are semiconductors or insulating materials while LMO and CB, on the other hand,
are good electron conductors. Good electronic conductivity ensures that a
material is properly grounded to the spectrometer during the measurement.
This means that the material and the spectrometer are on the same potential
(or Fermi level (EF)). With insulating components, a different surface potential is achieved when compared to the more conductive parts, similarly to the
SEI formation on Au in the previous section.
Thus, the observed binding energy shifts for active and inactive components in composite electrodes should be considered as a function of electrode
potential and the electronic conductivity of the specific compound. With
these insights, it is again clear that different types of energy calibration are
preferable depending on the point of interest and that in some cases; no internal energy calibration is required at all.
58
a) C1s
CB
b)
F1s
Pristine
electrode
LiF
A - 1.58 V
(LTO)
Intensity [arb.unit]
2
Intensity [arb.unit]
2
-CF 2
-CF - CH -
B - 2.95 V
(LFS)
C - 3.42 V
(LFP)
D - 4.13 V
(LMO)
296 292 288 284
Binding energy [eV]
c) Ti2p
692
688
684
Binding energy [eV]
d) Mn2p
Intensity [arb.unit]
Intensity [arb.unit]
Pristine
electrode
A - 1.58 V
(LTO)
B - 2.95 V
(LFS)
C - 3.42 V
(LFP)
D - 4.13 V
(LMO)
468 464 460 456
Binding energy [eV]
656 652 648 644 640
Binding energy [eV]
Figure 30. a) C1s, b) F1s, c) Ti2p and d) Mn2p spectra for a mixed materials electrode measured at different electrode potentials. From Paper VI.
59
Conclusions
The SEI that results in better cycling performance is rich in LiF and polycarbonate species from the FEC and VC decomposition. The polycarbonate
compounds are located in the outer parts of the SEI while the inner SEI is
rich in Li alkyl carbonates. Without additives the SEI is built up by essentially LiF and Li alkyl carbonate compounds in a homogeneous mixture.
The LiTDI salt or its decomposition products does not affect the Si particle surface. It is, however, not suitable to use this salt without electrolyte
additives since the SEI formed by the pure electrolyte results in poor cycling
performance. A considerable increase in surface oxide is observed with increasing cycling and the increase is more pronounced without electrolyte
additives.
Remaining Li in the Si material is a major source of capacity losses during cycling. A self-discharge mechanism is observed that to at least some
part is diffusion limited. This mechanism also contributes to a large fraction
of the capacity loss in each cycle. The electrochemical methods that are used
to identify these processes could easily be applied to different Si materials
and also other electrode materials.
HAXPES is a suitable technique to analyze Si electrode materials. It is
shown in this work that new peaks emerge from a more lithiated phase and
that the overall peak positions from bulk materials are altered as a function
of the lithium content. Also, the surface modifications as a function of cycling is possible to probe and a relative increase in surface oxides are observed with increased cycling. HAXPES is thereby one of few techniques
that can be used for the analysis of both the bulk Si reactions, the surface
oxide and the SEI simultaneously. In addition to this, HAXPES with different excitation energies also allows for a depth profiling of the SEI layer.
This work has encountered and pinpointed several important processes
that affect the measured binding energies in PES analysis of battery materials. Most importantly, it is shown that the binding energy of poorly conducting components are affected by the electrode potential and that SEI formation causes a peak position shift with respect to the underlying electrode.
The energy scale calibration of PES spectra should be carefully considered,
especially when new materials and structures are analyzed. The mapping of
these effects should provide a good starting point for future PES analysis on
battery materials.
60
Outlook
The commercialization of the Si electrode is challenged by several mechanisms that lower the performance and prevents the use of Si electrodes in
full-cell batteries. By efficient additives the cycling performance is enhanced
but still the continuous SEI formation must be decreased further to obtain a
commercially viable alternative.
Remaining Li in the Si electrode is a seldom considered source for capacity losses but as indicated by the results in the present work, this effect should
be seriously considered. Further investigations are needed to determine the
mechanism behind the Li lost in the Si material and to, if possible, suggest
ways to circumvent this problem.
More analysis tools would be helpful to analyze the mechanisms and reactions in the Si during cycling. HAXPES is a technique that can probe
changes in the Si material during lithiation and de-lithiation processes, but a
complete understanding of the lithiation mechanisms of Si is needed to link
PES the spectra to certain phases or transformations of the material.
Finally, PES analysis of battery materials is still a young research area
where efforts should be focused on understanding mechanisms and processes
affecting the measured binding energies. This could make PES analysis an
even more powerful tool in studying aging mechanisms of battery materials.
61
Acknowledgements
Dear all!
I want to thank everyone hanging out in the battery lab, making it a nice,
friendly and fun place to work in. And thanks to all you others that haven’t
had the luck to work in the battery lab, but still makes Ångström house 2 and
3 a great place.
And then I want thank my supervisors, Fredrik, Kristina and Anna, for giving me this opportunity to work with something really interesting. Thank
you for all feedback and encouragement.
I also want to thank all of you who have been working with me in these projects. Chao, thank you for always giving superfast feedback and for always
listening to my ideas, if there is anything I can help out with, just ask me.
Thank you Julia, Maria, Reza, David, Torbjörn and Leif for your support in
writing. Thank you Adam and Mattis for being great friends. Thank you
Henrik for all support in the lab! Thank you Fernanda and Kasia for being
super officemates. Thank you all.
Karin, Einar och Lotta, ni är mina favoriter, tack för att ni är så bäst! I övrigt
så är det många andra som också är bäst. Det är förstås min mamma och
pappa, Matte och Janna, Mormor och Farfar, och det är Elisabet, Mats, Matilda, Johanna, Adrian och August. Tack för att ni tar hand om mig och min
familj när jag bara jobbar. Nu har jag skrivit klart!
62
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
IPCC, Climate Change 2014: Synthesis report, IPCC, Geneva, Switzerland
(2014).
REN21, Renewables 2015 Global Status Report, Paris: REN 21 Secretariat
(2015).
P. Lamers, M. Junginger, C. Hamelinck, A. Faaij, Renew. Sustainable Energy
Rev. 16 (2012) (5) 3176-3199.
WCD, Dams and Development - A new Framework for Decision-making,
Earthscan Publications Ltd, London (2000).
R.A. Huggins, Energy Storage, Springer, New York (2010).
F.P. Sioshansi, Smart Grid - Integrating Renewable, Distributed and Efficient
Energy, Elsevier - Academic Press, Amsterdam (2012).
The DOE Global Energy Storage Database, Sandia National Laboratories,
http://www.energystorageexchange.org/ Accessed: 2016-03-10.
V. Julch, T. Telsnig, M. Schulz, N. Hartmann, J. Thomsen, L. Eltrop, T.
Schlegl, Energy Procedia 73 (2015) 18-28.
S. Trasatti, J. Electroanal. Chem. 460 (1999) (1-2) 1-4.
R.M. Dell, Solid State Ion. 134 (2000) (1-2) 139-158.
R. Van Noorden, Nature 507 (2014) (7490) 26-28.
E.C. Evarts, Nature 526 (2015) (7575) S93-95.
M.S. Whittingham, Science 192 (1976) (4244) 1126-1127.
Sony
Energy
Devices
Corporation,
http://www.sonyenergydevices.co.jp/en/keyword/, Accessed: 2015-12-24.
K. Xu, Chem. Rev. 104 (2004) (10) 4303-4417.
N. Nitta, F.X. Wu, J.T. Lee, G. Yushin, Mater. Today 18 (2015) (5) 252-264.
J. Kalhoff, G.G. Eshetu, D. Bresser, S. Passerini, Chemsuschem 8 (2015) (17)
2765-2765.
Q.S. Wang, P. Ping, X.J. Zhao, G.Q. Chu, J.H. Sun, C.H. Chen, J. Power
Sources 208 (2012) 210-224.
D. Doughty, E.P. Roth, Electrochem. Soc. Interface Vol. 21, No. 2 (2012)
p.37 - p.44.
D. Lisbona, T. Snee, Process Saf. Environ. 89 (2011) (6) 434-442.
M. Hiremath, K. Derendorf, T. Vogt, Environ. Sci. Technol. 49 (2015) (8)
4825-4833.
M. Fuhs, KIT-test: Some battery cells last five times longer than others,
http://www.pv-magazine.com/news/details/beitrag/kit-test--some-batterycells-last-five-times-longer-than-others_100018517/#axzz3U72AlRtU,
Accessed: 2015-12-28.
H.F. Wang, Y.I. Jang, B.Y. Huang, D.R. Sadoway, Y.T. Chiang, J.
Electrochem. Soc. 146 (1999) (2) 473-480.
S. Nordbrand, P. Bolme, Powering the Mobile World, SwedWatch (2007).
M.S. Whittingham, Chem. Rev. 104 (2004) (10) 4271-4301.
63
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
[50]
[51]
[52]
[53]
[54]
[55]
[56]
[57]
64
C. Masquelier, L. Croguennec, Chem. Rev. 113 (2013) (8) 6552-6591.
M. Armand, M. Gauthier, J.F. Magnan, N. Ravet, Method for synthesis of
carbon-coated redox materials for with controlled size, World Patent
WO02/27823 A1, (2001).
A. Nyten, A. Abouimrane, M. Armand, T. Gustafsson, J.O. Thomas,
Electrochem. Commun. 7 (2005) (2) 156-160.
N. Recham, J.N. Chotard, L. Dupont, C. Delacourt, W. Walker, M. Armand,
J.M. Tarascon, Nature Materials 9 (2010) (1) 68-74.
W.-J. Zhang, J. Power Sources 196 (2011) (1) 13-24.
W. Xu, J.L. Wang, F. Ding, X.L. Chen, E. Nasybutin, Y.H. Zhang, J.G.
Zhang, Energy Environ. Sci. 7 (2014) (2) 513-537.
R.D. Rauh, S.B. Brummer, Electrochim. Acta 22 (1977) (1) 75-83.
D. Aurbach, A. Zaban, Y. Gofer, Y.E. Ely, I. Weissman, O. Chusid, O.
Abramson, J. Power Sources 54 (1995) (1) 76-84.
M.A. Thomas, K. Amit, V.K. Sergei, J. Stephen, Nanotechnology 23 (2012)
(32) 325402.
J. Motavalli, Nature 526 (2015) (7575) S96-S97.
R.A. Huggins, Negative Electrodes in Lithium Cells, Springer US (2009).
Z. Ogumi, H. Wang, In: M. Yoshio, R. Brodd, A. Kozawa, Editors, LithiumIon Batteries, Springer New York (2009), p.49-73.
N.A. Kaskhedikar, J. Maier, Advanced materials (Weinheim) 21 (2009) (2526) 2664-2680.
K.M. Colbow, J.R. Dahn, R.R. Haering, J. Power Sources 26 (1989) (3) 397402.
Z. Chen, I. Belharouak, Y.K. Sun, K. Amine, Adv. Funct. Mater. 23 (2013)
(8) 959-969.
A.H. Whitehead, M. Schreiber, J. Electrochem. Soc. 152 (2005) (11) A2105A2113.
K. Zaghib, M. Dontigny, A. Guerfi, P. Charest, I. Rodrigues, A. Mauger,
C.M. Julien, J. Power Sources 196 (2011) (8) 3949-3954.
M.N. Obrovac, V.L. Chevrier, Chem. Rev. 114 (2014) (23) 11444-11502.
J. Cabana, L. Monconduit, D. Larcher, M.R. Palacín, Adv. Mater. 22 (2010)
(35) E170-E192.
J.W. Fergus, J. Power Sources 195 (2010) (15) 4554-4569.
J.Y. Song, Y.Y. Wang, C.C. Wan, J. Power Sources 77 (1999) (2) 183-197.
G. Pistoia, Batteries for Portable Devices, Elsevier, Amsterdam (2005).
A. Lewandowski, A. Świderska-Mocek, J. Power Sources 194 (2009) (2)
601-609.
V. Aravindan, J. Gnanaraj, S. Madhavi, H.K. Liu, Chem. Eur. J. 17 (2011)
(51) 14326-14346.
A. Hammami, N. Raymond, M. Armand, Nature 424 (2003) (6949) 635-636.
K. Xu, Chem. Rev. 114 (2014) (23) 11503-11618.
B. Michalak, H. Sommer, D. Mannes, A. Kaestner, T. Brezesinski, J. Janek,
Sci. Rep. 5 (2015) 15627.
G.G. Amatucci, J.M. Tarascon, L.C. Klein, Solid State Ion. 83 (1996) (1–2)
167-173.
H. Shin, J. Park, A.M. Sastry, W. Lu, J. Power Sources 284 (2015) 416-427.
J. Vetter, P. Novák, M.R. Wagner, C. Veit, K.C. Möller, J.O. Besenhard, M.
Winter, M. Wohlfahrt-Mehrens, C. Vogler, A. Hammouche, J. Power
Sources 147 (2005) (1–2) 269-281.
M. Winter, Z. Phys. Chem. 223 (2009) 1395-1406.
R. Spotnitz, J. Franklin, J. Power Sources 113 (2003) (1) 81-100.
[58]
[59]
[60]
[61]
[62]
[63]
[64]
[65]
[66]
[67]
[68]
[69]
[70]
[71]
[72]
[73]
[74]
[75]
[76]
[77]
[78]
[79]
[80]
[81]
[82]
[83]
[84]
[85]
R.A. Sharma, R.N. Seefurth, J. Electrochem. Soc. 123 (1976) (12) 17631768.
C.J. Wen, R.A. Huggins, J. Solid State Chem. 37 (1981) (3) 271-278.
B.A. Boukamp, G.C. Lesh, R.A. Huggins, J. Electrochem. Soc. 128 (1981)
(4) 725-729.
P. Limthongkul, Y.-I. Jang, N.J. Dudney, Y.-M. Chiang, J. Power Sources
119–121 (2003) 604-609.
M.N. Obrovac, L. Christensen, Electrochem. Solid-State Lett. 7 (2004) (5)
A93-A96.
T.D. Hatchard, J.R. Dahn, J. Electrochem. Soc. 151 (2004) (6) A838-A842.
J. Li, J.R. Dahn, J. Electrochem. Soc. 154 (2007) (3) A156-A161.
C.K. Chan, H. Peng, G. Liu, K. McIlwrath, X.F. Zhang, R.A. Huggins, Y.
Cui, Nat. Nano 3 (2008) (1) 31-35.
S. Misra, N. Liu, J. Nelson, S.S. Hong, Y. Cui, M.F. Toney, ACS Nano 6
(2012) (6) 5465-5473.
F. Wang, L.J. Wu, B. Key, X.Q. Yang, C.P. Grey, Y.M. Zhu, J. Graetz, Adv.
Energy Mater. 3 (2013) (10) 1324-1331.
J. Li, A. Smith, R.J. Sanderson, T.D. Hatchard, R.A. Dunlap, J.R. Dahn, J.
Electrochem. Soc. 156 (2009) (4) A283-A288.
V.L. Chevrier, J.R. Dahn, J. Electrochem. Soc. 157 (2010) (4) A392-A398.
T.K.J. Köster, E. Salager, A.J. Morris, B. Key, V. Seznec, M. Morcrette, C.J.
Pickard, C.P. Grey, Angew. Chem. Int. Ed. 50 (2011) (52) 12591-12594.
B. Key, M. Morcrette, J.-M. Tarascon, C.P. Grey, J. Am. Chem. Soc. 133
(2010) (3) 503-512.
B. Key, R. Bhattacharyya, M. Morcrette, V. Seznec, J.M. Tarascon, C.P.
Grey, J. Am. Chem. Soc. 131 (2009) (26) 9239-9249.
Y. Kubota, M.C.S. Escaño, H. Nakanishi, H. Kasai, J. Appl. Phys. 102 (2007)
(5) 053704.
E.D. Cubuk, E. Kaxiras, Nano Lett. 14 (2014) (7) 4065-4070.
J.H. Ryu, J.W. Kim, Y.-E. Sung, S.M. Oh, Electrochem. Solid-State Lett. 7
(2004) (10) A306-A309.
J. Graetz, C.C. Ahn, R. Yazami, B. Fultz, Electrochem. Solid-State Lett. 6
(2003) (9) A194-A197.
H. Li, X. Huang, L. Chen, Z. Wu, Y. Liang, Electrochem. Solid-State Lett. 2
(1999) (11) 547-549.
N. Liu, Z. Lu, J. Zhao, M.T. McDowell, H.-W. Lee, W. Zhao, Y. Cui, Nat.
Nano 9 (2014) (3) 187-192.
X.H. Liu, L. Zhong, S. Huang, S.X. Mao, T. Zhu, J.Y. Huang, Acs Nano 6
(2012) (2) 1522-1531.
E. Radvanyi, W. Porcher, E. De Vito, A. Montani, S. Franger, S.J.S. Larbi,
Phys. Chem. Chem. Phys. 16 (2014) (32) 17142-17153.
M.K.Y. Chan, C. Wolverton, J.P. Greeley, J. Am. Chem. Soc. 134 (2012) (35)
14362-14374.
N.S. Hochgatterer, M.R. Schweiger, S. Koller, P.R. Raimann, T. Wohrle, C.
Wurm, M. Winter, Electrochem. Solid-State Lett. 11 (2008) (5) A76-A80.
J. Li, R.B. Lewis, J.R. Dahn, Electrochem. Solid-State Lett. 10 (2007) (2)
A17-A20.
J.S. Bridel, T. Azais, M. Morcrette, J.M. Tarascon, D. Larcher, Chem. Mater.
22 (2010) (3) 1229-1241.
A. Magasinski, B. Zdyrko, I. Kovalenko, B. Hertzberg, R. Burtovyy, C.F.
Huebner, T.F. Fuller, I. Luzinov, G. Yushin, ACS Appl. Mater. Interfaces 2
(2010) (11) 3004-3010.
65
[86]
[87]
[88]
[89]
[90]
[91]
[92]
[93]
[94]
[95]
[96]
[97]
[98]
[99]
[100]
[101]
[102]
[103]
[104]
[105]
[106]
[107]
[108]
[109]
[110]
[111]
66
S. Komaba, K. Shimomura, N. Yabuuchi, T. Ozeki, H. Yui, K. Konno, J.
Phys. Chem. C 115 (2011) (27) 13487-13495.
I. Kovalenko, B. Zdyrko, A. Magasinski, B. Hertzberg, Z. Milicev, R.
Burtovyy, I. Luzinov, G. Yushin, Science 334 (2011) (6052) 75-79.
D. Mazouzi, B. Lestriez, L. Roue, D. Guyomard, Electrochem. Solid-State
Lett. 12 (2009) (11) A215-A218.
D. Mazouzi, Z. Karkar, C.R. Hernandez, P.J. Manero, D. Guyomard, L.
Roue, B. Lestriez, J. Power Sources 280 (2015) 533-549.
B. Philippe, R. Dedryvère, J. Allouche, F. Lindgren, M. Gorgoi, H. Rensmo,
D. Gonbeau, K. Edström, Chem. Mater. 24 (2012) (6) 1107-1115.
B. Philippe, R. Dedryvère, M. Gorgoi, H. Rensmo, D. Gonbeau, K. Edström,
Chem. Mater. 25 (2013) (3) 394-404.
B. Philippe, R. Dedryvère, M. Gorgoi, H. Rensmo, D. Gonbeau, K. Edström,
J. Am. Chem. Soc. 135 (2013) (26) 9829-9842.
W.L. Xu, S.S.S. Vegunta, J.C. Flake, J. Power Sources 196 (2011) (20)
8583-8589.
C.K. Chan, R. Ruffo, S.S. Hong, Y. Cui, J. Power Sources 189 (2009) (2)
1132-1140.
Y.M. Lee, J.Y. Lee, H.-T. Shim, J.K. Lee, J.-K. Park, J. Electrochem. Soc.
154 (2007) (6) A515-A519.
K.W. Schroder, H. Celio, L.J. Webb, K.J. Stevenson, J. Phys. Chem. C 116
(2012) (37) 19737-19747.
K.W. Schroder, A.G. Dylla, S.J. Harris, L.J. Webb, K.J. Stevenson, ACS
Appl. Mater. Interfaces 6 (2014) (23) 21510-21524.
N.S. Choi, K.H. Yew, H. Kim, S.S. Kim, W.U. Choi, J. Power Sources 172
(2007) (1) 404-409.
S. Dalavi, P. Guduru, B.L. Lucht, J. Electrochem. Soc. 159 (2012) (5) A642A646.
V. Etacheri, O. Haik, Y. Goffer, G.A. Roberts, I.C. Stefan, R. Fasching, D.
Aurbach, Langmuir 28 (2012) (1) 965-976.
H. Nakai, T. Kubota, A. Kita, A. Kawashima, J. Electrochem. Soc. 158
(2011) (7) A798-A801.
M.Y. Nie, D.P. Abraham, Y.J. Chen, A. Bose, B.L. Lucht, J. Phys. Chem. C
117 (2013) (26) 13403-13412.
K. Schroder, J. Avarado, T.A. Yersak, J.C. Li, N. Dudney, L.J. Webb, Y.S.
Meng, K.J. Stevenson, Chem. Mater. 27 (2015) (16) 5531-5542.
N.S. Choi, K.H. Yew, K.Y. Lee, M. Sung, H. Kim, S.S. Kim, J. Power
Sources 161 (2006) (2) 1254-1259.
M. Gauthier, T.J. Carney, A. Grimaud, L. Giordano, N. Pour, H.-H. Chang,
D.P. Fenning, S.F. Lux, O. Paschos, C. Bauer, F. Maglia, S. Lupart, P. Lamp,
Y. Shao-Horn, J. Phys. Chem. Lett. 6 (2015) (22) 4653-4672.
T. Jaumann, J. Balach, M. Klose, S. Oswald, U. Langklotz, A. Michaelis, J.
Eckert, L. Giebeler, Phys. Chem. Chem. Phys. 17 (2015) (38) 24956-24967.
M.A. McArthur, S. Trussler, J.R. Dahn, J. Electrochem. Soc. 159 (2012) (3)
A198-A207.
K. Ciosek Högström, S. Malmgren, M. Hahlin, M. Gorgoi, L. Nyholm, H.
Rensmo, K. Edström, Electrochim. Acta 138 (2014) (0) 430-436.
K. Tasaki, A. Goldberg, J.-J. Lian, M. Walker, A. Timmons, S.J. Harris, J.
Electrochem. Soc. 156 (2009) (12) A1019-A1027.
K. Tasaki, S.J. Harris, J. Phys. Chem. C 114 (2010) (17) 8076-8083.
S.F. Lux, I.T. Lucas, E. Pollak, S. Passerini, M. Winter, R. Kostecki,
Electrochem. Commun. 14 (2012) (1) 47-50.
[112] S.F. Lux, J. Chevalier, I.T. Lucas, R. Kostecki, ECS Electrochem. Lett. 2
(2013) (12) A121-A123.
[113] N. Delpuech, D. Mazouzi, N. Dupré, P. Moreau, M. Cerbelaud, J.S. Bridel,
J.C. Badot, E. De Vito, D. Guyomard, B. Lestriez, B. Humbert, J. Phys.
Chem. C 118 (2014) (31) 17318-17331.
[114] B.-C. Yu, Y. Hwa, C.-M. Park, J.-H. Kim, H.-J. Sohn, RSC Adv. 3 (2013)
(24) 9408-9413.
[115] J.H. Ryu, J.W. Kim, Y.E. Sung, S.M. Oh, Electrochem. Solid-State Lett. 7
(2004) (10) A306-A309.
[116] T. Yoon, C.C. Nguyen, D.M. Seo, B.L. Lucht, J. Electrochem. Soc. 162
(2015) (12) A2325-A2330.
[117] M.N. Obrovac, L.J. Krause, J. Electrochem. Soc. 154 (2007) (2) A103-A108.
[118] B. Jerliu, L. Dorrer, E. Huger, G. Borchardt, R. Steitz, U. Geckle, V. Oberst,
M. Bruns, O. Schneider, H. Schmidt, Phys. Chem. Chem. Phys. 15 (2013)
(20) 7777-7784.
[119] H. Li, X. Huang, L. Chen, G. Zhou, Z. Zhang, D. Yu, Y. Jun Mo, N. Pei,
Solid State Ion. 135 (2000) (1–4) 181-191.
[120] J. Yang, M. Wachtler, M. Winter, J.O. Besenhard, Electrochem. Solid-State
Lett. 2 (1999) (4) 161-163.
[121] L. Liu, M.N. Obrovac, ECS Electrochem. Lett. 1 (2012) (1) A10-A12.
[122] P.P. Ferguson, R.A. Dunlap, J.R. Dahn, J. Electrochem. Soc. 157 (2010) (3)
A326-A332.
[123] A.J. Bard, L.R. Faulkner, Electrochemical Methods Fundamental and
Applications, John Wiley & Sons, Inc, New York (2001).
[124] U. Gelius, E. Basilier, S. Svensson, T. Bergmark, K. Siegbahn, J. Electron.
Spectrosc. Relat. Phenom. 2 (1973) (5) 405-434.
[125] S. Iwata, A. Ishizaka, J. Appl. Phys. 79 (1996) (9) 6653-6713.
[126] M.P. Seah, W.A. Dench, Surf. Interface Anal. 1 (1979) (1) 2-11.
[127] D.R. Baer, M.H. Engelhard, J. Electron. Spectrosc. Relat. Phenom. 178–179
(2010) 415-432.
[128] M. Gorgoi, S. Svensson, F. Schafers, W. Braun, W. Eberhardt, Eur. Phys. JSpec. Top. 169 (2009) 221-225.
[129] F. Schaefers, M. Mertin, M. Gorgoi, Rev. Sci. Instrum. 78 (2007) (12)
123102.
[130] M. Bässler, A. Ausmees, M. Jurvansuu, R. Feifel, J.O. Forsell, P. de Tarso
Fonseca, A. Kivimäki, S. Sundin, S.L. Sorensen, R. Nyholm, O. Björneholm,
S. Aksela, S. Svensson, Nucl. Instrum. Meth. A 469 (2001) (3) 382-393.
[131] N.N. Greenwood, A. Earnshaw, Chemistry of the Elements (2nd Edition),
Butterworth-Heinemann, Oxford (1997).
[132] M. Yamada, A. Inaba, A. Ueda, K. Matsumoto, T. Iwasaki, T. Ohzuku, J.
Electrochem. Soc. 159 (2012) (10) A1630-A1635.
[133] J. Yang, Y. Takeda, N. Imanishi, C. Capiglia, J.Y. Xie, O. Yamamoto, Solid
State Ion. 152 (2002) 125-129.
[134] J.-W. Song, C.C. Nguyen, S.-W. Song, RSC Adv. 2 (2012) (5) 2003-2009.
[135] D.T. Clark, H.R. Thomas, J. Polym. Sci., Part A: Polym. Chem. 16 (1978) (7)
1461-1474.
[136] D.T. Clark, A. Dilks, H.R. Thomas, D. Shuttleworth, J. Polym. Sci., Part A:
Polym. Chem. 17 (1979) (3) 627-638.
[137] A. Cros, J. Electron. Spectrosc. Relat. Phenom. 59 (1992) (1) 1-14.
[138] H. Sezen, G. Ertas, S. Suzer, J. Electron. Spectrosc. Relat. Phenom. 178–179
(2010) 373-379.
67
Acta Universitatis Upsaliensis
Digital Comprehensive Summaries of Uppsala Dissertations
from the Faculty of Science and Technology 1362
Editor: The Dean of the Faculty of Science and Technology
A doctoral dissertation from the Faculty of Science and
Technology, Uppsala University, is usually a summary of a
number of papers. A few copies of the complete dissertation
are kept at major Swedish research libraries, while the
summary alone is distributed internationally through
the series Digital Comprehensive Summaries of Uppsala
Dissertations from the Faculty of Science and Technology.
(Prior to January, 2005, the series was published under the
title “Comprehensive Summaries of Uppsala Dissertations
from the Faculty of Science and Technology”.)
Distribution: publications.uu.se
urn:nbn:se:uu:diva-281694
ACTA
UNIVERSITATIS
UPSALIENSIS
UPPSALA
2016