A study of poly(vinyl alcohol) as a solid polymer electrolyte for lithium-ion batteries

UPTEC K 16013

Examensarbete 30 hp Juli 2016

A study of poly(vinyl alcohol) as a solid polymer electrolyte for lithium ion batteries

Gustav Ek

Teknisk- naturvetenskaplig fakultet UTH-enheten

Besöksadress:

Ångströmlaboratoriet Lägerhyddsvägen 1 Hus 4, Plan 0

Postadress:

Box 536 751 21 Uppsala

Telefon:

018 – 471 30 03

Telefax:

018 – 471 30 00

Hemsida:

http://www.teknat.uu.se/student

Abstract

A study of poly(vinyl alcohol) as a solid polymer electrolyte for lithium ion batteries

Gustav Ek

The use of solid polymer electrolytes in lithium-ion batteries has the advantage in terms of safety and processability, however they often lack in terms of performance.

This is of major concern in applications where high current densities or rapidly changing currents are important. Such applications include electrical vehicles and energy storage of the electrical grid to accommodate fluctuations when using renewable energy sources such as wind and solar.

In this study, the use of commercial poly(vinyl alcohol) (PVA) as a solid polymer electrolyte for use in lithium-ion batteries has been evaluated. Films were prepared using various lithium salts such as lithium bis(trifluoromethane)sulfonimide (LiTFSI) and casting techniques. Solvent free films were produced by substituting the solvent Dimethyl sulfoxide (DMSO) with water and rigouros drying or by employing a hot-pressing technique. The best performing system studied was PVA-LiTFSI-DMSO, which reached ionic conductivities of 4.5E-5 S/cm at room temperature and 0.45 mS/cm at 60 °C. The solvent free films showed a drop of ionic conductivity by roughly one order of magnitude compared to films with residual DMSO present. High ionic conductivities in PVA-LiTFSI-DMSO electrolytes are thus ascribed to fast lithium ion transport through the liquid domain of DMSO, or by plasticizing effects of salt and solvent on the polymer.

Thermal analysis of the films showed a clear plasticizing effect of DMSO by a decrease in the glass transition temperature. FTIR analysis showed complexation of all the lithium salts investigated with the OH-groups of the polymer by a shift in the characteristic frequencies of both salts and polymer.

For the first time, prototype battery cells containing PVA electrolytes were manufactured and evaluated by galvanostatic cycling. PVA-LiTFSI-DMSO showed stable cycling performance for 15 cycles. Solvent free electrolytes were also investigated but did not result in any stable cycling performance.

Ämnesgranskare: Tim Bowden

Handledare: Daniel Brandell

Popul¨ arvetenskaplig sammanfattning

Dagens samh¨ alle st˚ ar inf¨ or stora utmaningar i och med klimatf¨ or¨ andringar p˚ a jorden till f¨ oljd av anv¨ andandet av fossila br¨ anslen. F¨ or att ¨ overkomma detta beh¨ over vi ist¨ allet anv¨ anda f¨ ornybara energik¨ allor s˚ asom vind- och solkraft, fast dessa har problem med att till exempel vinden inte alltid bl˚ aser och att solen inte skiner p˚ a natten. Detta leder till att vi inte alltid kan f˚ a ut den energi vi beh¨ over f¨ or att samh¨ allet skall fungera, och vi beh¨ over d¨ arf¨ or spara

¨

overskottet fr˚ an n¨ ar det produceras som mest.

En typ av energilagringsteknik som skulle kunna vara en l¨ osning p˚ a detta ¨ ar litium-jon batterier eftersom de kan lagra v¨ aldigt mycket energi i j¨ amf¨ orelse med andra batterier, och har ocks˚ a v¨ aldigt sm˚ a f¨ orluster vid upp- och urladdning.

Men eftersom de kan lagra s˚ apass mycket energi, s˚ a uppst˚ ar det s¨ akerhet- srisker vid storskalig anv¨ andning. Detta h¨ or oftast samman med anv¨ and- ningen av brandfarliga flytande organiska f¨ oreningar i den s˚ a kallade elek- trolyten, vars funktion ¨ ar att f¨ orflytta litium-joner mellan elektroderna i batteriet. F¨ or att ¨ oka s¨ akerheten kan man ist¨ allet anv¨ anda sig av en fast elektrolyt best˚ aende av en syntetisk polymer. Problemet med dessa fasta varianter ¨ ar att de har s¨ amre prestanda ¨ an de flytande med avseende p˚ a bland annat jonledningsf¨ orm˚ aga.

I detta examensarbete har polymeren poly(vinyl alkohol) studerats f¨ or anv¨ and- ning som just fast elektrolyt i litium-jon batterier. Polymeren har blandats med olika litiumsalter och l¨ osningsmedel f¨ or att kunna se skillnader i pre- standa. Egenskaper s˚ asom jonledningsf¨ orm˚ aga, v¨ armeegenskaper och inter- aktioner mellan salt och polymer har studerats. Prototyp-batteri celler in- neh˚ allande poly(vinyl alkohol) som elektrolyt har ocks˚ a tillverkats och utsatts f¨ or repeterad upp- och urladdning.

Resultatet av studien ¨ ar att poly(vinyl alkohol) kan fungera i dessa till¨ amp-

ningar. Men f¨ or att uppn˚ a tillr¨ ackligt h¨ og jonledningsf¨ orm˚ aga s˚ a kr¨ avs att

en viss m¨ angd av l¨ osningsmedlet dimetylsulfoxid finns kvar efter tillverkning

av elektrolytfilmerna. Detta pekar p˚ a att polymeren inte leder joner i sin

kristallina fas eftersom detta l¨ osningsmedel g¨ or polymeren mjukare.

List of abbreviations

DMC Dimethyl carbonate ((CH

)

CO) DMSO Dimethyl sulfoxide ((CH

)

OS) DSC Differential scanning calorimetry EC Ethylene carbonate ((CH

O)

CO) EDLC Electric double layer capacitor

EIS Electrochemical impedance spectroscopy FTIR Fourier transform infrared spectroscopy LFP Lithium iron phosphate (LiFePO

)

TFSI Bis(trifluoromethane)sulfonimide (N(CF

SO

)

) Tf Trifluoromethanesulfonate (CF

SO

) Ni-MH Nickel-metal hydride

NMR Nuclear magnetic resonance PVA Poly(vinyl alcohol)

PEO Poly(ethylene oxide)

PTFE Poly(tetrafluoroethylene)

T

Glass transition temperature

TGA Thermograviometric analysis

SPE Solid polymer electrolyte

VTF Vogel-Tammann-Fulcher

Contents

1 Introduction 1

1.1 Background . . . . 1

1.1.1 Li-ion batteries . . . . 2

1.1.2 Solid polymer electrolytes (SPE) . . . . 3

1.1.3 Poly(vinyl alcohol) . . . . 5

1.2 PVA-based polymer electrolytes . . . . 6

2 Scope of this thesis 9 3 Theory 10 3.1 Ionic conductivity . . . . 10

3.2 Ion-transport mechanisms in solid polymers . . . . 11

4 Methodology 14 4.1 Electrode and electrolyte preparation . . . . 14

4.1.1 Polymer electrolyte films . . . . 14

4.1.2 Electrodes . . . . 14

4.1.3 Battery cell assembly . . . . 15

4.2 Analysis . . . . 15

4.2.1 Thermograviometric Analysis (TGA) . . . . 15

4.2.2 Differential Scanning Calorimetry (DSC) . . . . 16

4.2.3 Electrochemical Impedance Spectroscopy (EIS) . . . . 17

4.2.4 Galvanostatic cell cycling . . . . 20

4.2.5 Fourier Transform Infrared Spectroscopy (FTIR) . . . 20

5 Experimental 21 5.1 Electrode and electrolyte preparation . . . . 21

5.1.1 Polymer electrolyte films . . . . 21

5.1.2 Electrodes . . . . 23

5.1.3 Battery cell assembly . . . . 23

5.2 Analysis . . . . 24

5.2.1 Thermograviometric Analysis (TGA) . . . . 24

5.2.2 Differential Scanning Caliometry (DSC) . . . . 24

5.2.3 Electrochemical Impedance Spectroscopy (EIS) . . . . 24

5.2.4 Galvanostatic cell cycling . . . . 25

5.2.5 Fourier Transform Infrared Spectroscopy (FTIR) . . . 25

6 Results and discussion 26 6.1 Ionic conductivity . . . . 26

6.1.1 Variation of salt content in the PVA-LiTFSI-DMSO system . . . . 26

6.1.2 Effects of different salts and solvents . . . . 27

6.2 Thermal properties . . . . 30

6.2.1 TGA . . . . 30

6.2.2 DSC . . . . 32

6.3 Fourier Transform Infrared Spectroscopy (FTIR) . . . . 35

6.3.1 Polymer-salt interactions . . . . 35

6.3.2 Interactions between lithium salt and functional groups 39 6.4 Cycling . . . . 41

7 Conclusions 44

8 Acknowledgements 46

1 Introduction

1.1 Background

Today’s society faces a number of major challenges in the process of switching from fossil to renewable energy resources. One of these challenges is the nature of renewable energy sources (e.g wind and solar). Energy storage systems need to be put in place in order to accomendate fluctuations in the electrical grid, for instance such as at peak hours [1, 2]. Today, pumped- storage hydroelectric is most used for this purpose, although the technique lacks in term of efficiency seeing at least 15 % of the energy lost [3]. Another challenge is converting the vehicle fleet from gasoline into electric powered or fuel cell based/hydrogen powered. This also calls for not only high energy storage, but also for light weight solutions.

Lithium-ion batteries are very promising solutions to these problems because of their superior energy and power density in comparison to other battery systems such as lead-acid and Ni-MH (figure 1.1). The system is also superior in terms of energy conversion efficiency compared to pumped-storage hydro- electricity; the charge-discharge efficiency of commercial Li-ion batteries is around 90 % [4].

As mentioned above, the storage most used today is pumped-hydro which

has an energy efficiency of 70-80 % compared to the 90 % of the lithium-

ion battery. Hydrogen storage and fuel cells are also alternative solutions to

grid stabilisation. These systems however presents problems such as the risk

of explosion with compressed hydrogen gas and considerably lower energy

efficiencies of around 40 % [5].

NiCd, NiMH

Fossil fuels

Lithium-ion

Lead Acid

E n e rg y d e n si ty [ W h /k g]

Power density [W/kg]

Figure 1.1: Ragone chart of different energy storage technologies (redrawn from ref [6])

1.1.1 Li-ion batteries

The state-of-the-art Li-ion battery is made up of two electrodes (a negative

anode and a positive cathode), an electrolyte and a separator. It is often

referred to as the ”rocking chair battery”, and functions by shuttling Li

between the electrodes through the electrolyte (figure 1.2), thus generating

a current during discharge. The electrodes are made out of compounds that

have a layered or channeled structure, so that they can store Li

in be-

tween the layers/channels. This process is called intercalation, and common

materials used today are LiCoO

as the positive electrode (”cathode”), and

graphite as the negative electrode (”anode”). The electrolyte is usually a

mixture of organic solvents such as ethylene carbonate (EC) and a variety

of linear alkyl carbonates, such as dimethyl carbonate (DMC) and a lithium

salt, typically LiPF

for sufficient ionic transport at high concentrations. To

keep the cell from short-circuiting a porous separator is soaked in the liquid electrolyte and placed between the electrodes.

e

Li

A n o d e C at ho d e

Electrolyte

Figure 1.2: schematic image of a Li-ion electrochemical cell during discharge

A big problem with modern Li-ion batteries is the safety concerns regarding the organic liquid electrolyte due to the low flash points and high vapour pressures at elevated temperatures of the organic solvents, and the reactivity of lithiated graphite which can ignite spontantanously in air [7]. To com- pensate for these shortcomings, flame retardants and other compounds are added into the electrolyte which will influence the operation of the battery.

Another concern is that the separator between the electrodes has a low ther- mal and mechanical stability which could lead to short-circuit and thermal runaway. This limits the use of these systems in large scale applications such as electrical grid stabilisation and in electric vehicles.

1.1.2 Solid polymer electrolytes (SPE)

Solid polymer electrolytes have the potential to replace conventional liquid

organic variants due to that they show improved properties in terms of in-

creased electrochemical stability, reduced risk of leakage and fire hazard as

well as preventing the growth of lithium dendrites when metallic Li is used as the anode material [7, 8]. They also act as separators, replacing the tra- ditional porous materials with low thermal and mechanical stability, and reducing overall weight. The most explored polymer in literature for Li-ion battery applications is polyethylene oxide (PEO), although recent research has shown advantages of alternative polymer materials in terms of thermal and electrochemical stability [8].

In order for polymer electrolytes to compete with liquid organic counterparts in commercial applications, they should posses the following properties [9] :

• Ionic conductivity σ ≥ 10

S cm

at room temperature

In order to supply enough power for commercial applications, the ionic conductivity has to be somewhat in the proximity of liquid electrolytes.

• Ionic transference number of the conducting ion close to unity; t

∼ 1 The polymer electrolyte should act as an effective conducting medium for the ion contributing to the reaction generating the electrical current.

To achieve this it is desired that the mobility of the cation Li

is much greater than the anion to avoid polarisation effects.

• High chemical, thermal and electrochemical stabilities

For safety reasons, the electrolyte should be stable in the environment of the cell. This is to avoid unwanted chemical reactions, and to in- crease the span of temperature and potential window at which it can be operated.

• Mechanically stable

Large scale processing demands the materials to be mechanically stable.

This also prevents short-circuits associated with cell abuse.

• Be compatible with commercial electrode materials

In order for polymer electrolyte to be a commercial success, it needs to

work (i.e. be stable against oxidation and reduction) with well estab-

lished electrode materials to reduce costs of further development.

1.1.3 Poly(vinyl alcohol)

Poly(vinyl alcohol), often abbreviated PVA, PVAL or PVOH, is a synthetic polymer that was first prepared by Hermann and Haehnel in 1924 by hy- drolizing polyvinyl acetate [10]. Because of its water solubility, it is used in a wide range of commercial applications such as lamination of safety glass, pro- tective coatings, binding of pigments and the production of other synthetic polymers [10].

OH

Figure 1.3: The reapeating unit of PVA

The polymer consists of a carbon backbone with OH side-chains (figure 1.3) and has the chemical formula [CH

CH(OH)]

. PVA is synthezised by poly- merisation of vinyl acetate, and then the resulting poly(vinyl acetate) is converted into PVA by a base-catalyzed addition of methanol (figure 1.4) [11]. This makes PVA different from other vinyl polymers that are made directly from their corresponding monomers.

Table 1.1: General properties of PVA [10]

Density 1.19-1.31g/cm

Melting point 200

C

Decomposition temperature 230

C

Glass transition temperature 85

C

OH O

O

+ CH3OH NaOH + CH3OAC

O

O O

O

Free radical polymerisation

Figure 1.4: General reaction formula for PVA synthesis

The content of acetate functionality left after synthesis (degree of hydrolysis) has a direct impact on the properties of the material. This is because a larger amount of OH-groups instead of acetates, leads to a more crystalline material due to crosslinking hydrogen bonds forming crystalline domains[10].

PVA materials are often categorized into two different segments of hydrolysis;

partial (<98%) and full (>98%), where properties such as solubility differs.

1.2 PVA-based polymer electrolytes

PVA was found to form salt complexes with large amounts of lithium salt

by Kanbara et al. in 1989 [12, 13]. It was reported that PVA forms trans-

parent and homogeneous composite films together with various lithium salts

(LiClO

, LiBF

and LiCF

SO

) up to about 60 wt-% salt, that showed con-

ductivites as high as 2 × 10

S cm

. The films however were prepared by

solution casting in dimethyl sulfoxide (DMSO) which left residual solvent in

the films. This was believed to be caused by complexation of DMSO and

PVA, since drying of DMSO-solutions with lithium salts produced DMSO- free salts.

Further studies on PVA-salt mixtures as solid polymer electrolytes were done by Yamamoto et al. in 1994 [14]. Here PVA was compared with poly(parabanic acid) in terms of its ability to dissolve alkali metal salts and ionic conductivity. It was found that both polymers can dissolve high amounts of salt and show at first glance Arrhenius-like behaviour in conduc- tivity, although PVA has a higher value at the same salt:polymer ratio. IR studies of PVA-LiCF

SO

indicated that the interaction between the Li salt and the OH-group of PVA is not strong.

Lithium ion mobility in PVA:LiCF

SO

based electrolytes was studied by Every et al. with means of

Li NMR spectroscopy [15]. It was found that PVA differs from traditional polyether-based systems with regards on cation transport being decoupled from the polymer relaxation. It was suggested that ion conductivity is more likely to occur by other mechanisms.

Rajendran et al. investigated ionic conductivity in PVA mixed with PMMA (poly(methyl methacrylate)) and ethylene carbonate (EC) plasticizer [16, 17].

These studies showed that the PVA-PMMA-EC-LiBF

complexes have non linear trends in Arrhenius plots, and that the ionic conductivity is therefore coupled to the segmental motion of the polymer backbone.

Since residual DMSO in the PVA films acts as a plasticizer, lowering the T

and thus improving segmental motion of the polymer chains, it should have an impact on the ionic conductivity. The effect of DMSO on the conductivity in PVA-salt mixtures was investigated by MacFarlane et al. in 1998 [18]. This was done by hot pressing a powder-mixture of polymer and salt to avoid solvents. It was found that solvent-free films of PVA and lithium salts show lower conductivities, in the range of 10

S cm

at room temperature, which is believed to be due to the absence of the plasticizing effect of DMSO.

More recent research on the PVA-LiCF

SO

system has been done by Malathi

et al. in 2010 [19]. Electrolyte films prepared by solution casting and char-

acterized with FTIR, XRD, DSC and AC impedance. FTIR confirmed a

polymer-salt complex (in contradiction to Yamamoto et al.) and the XRD

revealed its amorphous nature. Thermal analysis showed that increasing

salt concentration leads to a lowering of T

due to a plasticizing effect. The

highest conductivity reported was 7 × 10

S cm

at 25 mol% LiCF

SO

.

PVA has been tested as a solid polymer electrolyte in electrical double-layer capacitor (EDLC) by both Kanbara et al. [13] and C.-S. Lim et al. [20].

Both papers indicate that PVA is well suited as a solid electrolyte in EDLC applications with specific capacitance about 2.5 Fg

. It is also shown that the addition of TiO

particles can enhance the performance in these applications, reaching an electrochemical stability window between -2.3 V and 2.3 V in a cell with carbon electrodes up to 1000 cycles.

Kim et al. [21] prepared electrolyte films based on a blend of PVA and various concentrations of LiCF

SO

. A dramatic increase in conductivity was observed between 40 and 50 wt-%. AFM studies showed aggregation of salt ions, and it was concluded that Li-ions move by a hopping mecha- nism between aggregates. It was also reported that the PVA-salt mixture is electrochemically stable up to 4.5 V vs Li/Li

.

Chatterjee et al. [22] reported the highest ionic conductivity of PVA (> 10

S cm

at room temperature) in a polymer gel electrolyte using NMP and LiCF

SO

.

To summarize, the characteristic features of PVA-based polymer electrolytes

are their ability to solvate large amounts of alkali-metal salt, and their high

ionic conductivities at temperatures below the glass transition. Solution

casted films using DMSO however, have the problem of residual solvent which

is believed to influence the behaviour of these systems. The mechanism of

ionic motion in PVA electrolytes is still not understood, but generally linear-

behaviour in ionic conductivity with temperature is observed, and it should

therefore not be governed by the motion of the polymeric backbone. No

evaluation of PVA has been found regarding its stability in prototype battery

cells, or the compatibility with more modern salts such as LiTFSI.

2 Scope of this thesis

PVA is rather poorly investigated in scientific literature, although earlier re- search has shown ionic conductivity in these materials which means potential uses as electrolytes [12, 15, 19, 23] [14, 16, 17, 21]. In those studies it has been shown that the temperature dependence on the ionic conductivity for PVA does not follow that of conventional polymer electrolytes (so called Vogel- Tammann-Fulcher dependence). How ions are conducted in PVA-based elec- trolytes, which are hydrogen bonded systems, are therefore of fundamental scientific interest.

In this work, electrolyte membranes consisting of PVA-salt mixtures will be evaluated with regards to their use in battery applications. Commercial PVA polymers will be mixed with various Li-salts and various solvents, since residual DMSO has previously been reported to have a major impact on the performance. Ionic conductivity will be measured using Electrochemical Impedance Spectroscopy (EIS) at temperatures between 20 and 100

C, and polymer-salt interactions will be studied using spectroscopic techniques such as FTIR. Residual solvent and plasticizing effects from both salt and solvent will be monitored with gravimetric techniques (TGA, DSC).

Also, lithium metal — lithium iron phosphate battery cells utilizing PVA-

based electrolytes will be assembled and tested using repeated charge and

discharged cycling with a constant current.

3 Theory

3.1 Ionic conductivity

The most important feature of the electrolyte is its ability to be electroni- cally insulating, while still conducting ions. The general expression for ionic conductivity, σ, is given in eq. 3.1

σ = X

i

n

q

µ

(3.1)

where n is the concentration, q the charge and µ the mobility of ion i. σ is measured in Scm

, where S = Siemens = Ω

.

In order for a battery to have a high power density, a sufficiently high ionic conductivity is needed in the electrolyte. This is often no problem for liquid systems, where the mobility of the ions is high. In solid polymers however, this is not the case. Besides this, a high mobility of Li

is wanted since it is the most useful in Li-battery applications. The ionic conductivity contains contributions from both anion and cation, so conduction mostly by the means of the cation Li

is desired. This contribution to the total ionic conductivity can be described using the transfer number of the positive ion (eq. 3.2).

t

= µ

µ

+ µ

(3.2)

In order to get an energy efficient electrolyte, a transfer number close to

unity (t

∼ 1) is desired. This is often achieved by choosing anions that

easily dissociate from Li

. Common ones used in polymeric systems are

triflate (CF

SO

) and TFSI (N(CF

SO

)

) which are shown in figure 3.1.

Transferance numbers of Li

in organic solvents are often rather low ∼0.3.

Figure 3.1: Chemical structure of the anions triflate (left) and TFSI (right)

3.2 Ion-transport mechanisms in solid poly- mers

Several models have been proposed for the mechanisms of ion- transport in polymeric matrices. In the case of polymer electrolytes, the polymer back- bone is often found to form complexes with the salt. The ion is believed to then be transferred by either the segmental motion of the backbone or of a hopping mechanism where lithium ions ”jump” from one coordination site into an empty neighbouring one illustrated by fig 3.2.

Figure 3.2: The mechanism for Li

transport assisted by segmental motion of the polymer backbone [24]

The motion of the polymer backbone is strongly governed by the T

of the

system, and the free volume available, which can be influenced with different

salt concentrations, fillers and plasticizers which will affect the free volume of the polymer. Systems following this type of mechanism often follow the Vogel-Tamann-Fulcher (VTF) behaviour with temperature, that is character- ized by a curved line in the Arrhenius plot (illustrated in figure 3.3) following equation 3.3 and is typically found in polymer electrolytes based on PEO.

Vogel Tamann Fulcher (VTF)-type:

σ = σ

e

(−

B

T − T

(3.3) Traditional Arrhenius behaviour on the other hand is often seen in systems where the conductivity is believed to be due to a hopping mechanisms of the ions between adjacent vacant sites. This behaviour is represented by a straight line in the Arrhenius plot (figure 3.3) following equation 3.4. This type of behaviour often presents itself in solid state ionic conductors such as fluorides and iodides.

Arrhenius-type:

σ = σ

e

E

RT

(3.4)

Log(Conductivity)

1/T

Figure 3.3: Typical trends of conductivity with temperature for VTF (red)

and Arrhenius (blue) type behaviours

4 Methodology

4.1 Electrode and electrolyte preparation

4.1.1 Polymer electrolyte films

Solution casting

In order to produce self standing polymer electrolyte films for characteriza- tion, a simple technique called solution casting can be employed. The poly- mer and electrolyte salt are dissolved in an appropriate solvent and casted into moulds for gentle drying under vacuum at elevated temperatures, yield- ing a free standing, thin polymer electrolyte film.

Hot-pressing

Hot-pressing is the procedure of applying simultaneous heat and pressure on a powder in order to induce sintering and creeping processes. This will in turn form the powder into a compact body.

4.1.2 Electrodes

Electrode films were also prepared by solution casting. A slurry consisting of

intercalation material, binder and conducting carbon was prepared. It is then

coated onto a current collector foil by doctor blading and left to evaporate.

4.1.3 Battery cell assembly

To test how polymer electrolytes behave in device applications, they have to be tested in electrochemical cells. In this thesis, the cells used are in the format of soft pouches (so called pouch-cells) constructed out of poly- mer coated aluminium foil. This format has many advantages compared to other in terms of ease of fabrication and disassembly. The polymer films are sandwiched between a commercial cathode material (LiFePO

) and Li-metal, and connected to current collectors. This is then vacuum sealed inside of a glovebox to avoid moisture contamination.

4.2 Analysis

4.2.1 Thermograviometric Analysis (TGA)

Thermogravimetric analysis (TGA) is a technique for measuring the change

of mass in a sample with temperature. This is done by placing a sample

onto a microbalance that is then lowered into a furnace with a controlled

atmosphere. The main goal of TGA is to analyze the thermal stability and

decomposition of materials, and the data is plotted as mass-change vs. tem-

perature, which is illustrated in figure 4.1. Mass losses are related to the loss

of solvent or due to decomposition of the sample or evaporation of compo-

nents, while mass gain can be due to oxidation.

Reaction interval Temperature Weight

percentage

Figure 4.1: Typical mass loss due to decomposition in a sample using TGA

4.2.2 Differential Scanning Calorimetry (DSC)

A common technique to study phase transitions such as glass transition,

crystallization and melting is differential scanning calorimetry (DSC). It is

a thermal analysis technique that heats/cools both a sample, and an empty

reference pan in a furnace. The difference in heat-flow applied to maintain

the same temperature in sample and reference is then measured. The en-

ergy difference observed will indicate processes that either consume energy

(endothermic) or release it (exothermic). The data is plotted as heat-flow vs

temperature, and different process temperatures such as the temperature of

melting (T

), crystallisation (T

) and glass-transition (T

) can be determined

using the characteristic shape of the peaks (see figure 4.2).

Glass transition

Crystallization

Melting Temperature

Heat flux

En d ot h er m ic Ex ot her m ic

Figure 4.2: Schematic DSC-curve for a polymer

4.2.3 Electrochemical Impedance Spectroscopy (EIS)

Electrochemical impedance spectroscopy is a technique commonly used to ob- tain information about ionic conductivity and transference numbers for both the bulk and at different interfaces. The method is rapid and non-destructive which is convenient. EIS is performed by applying a small sinusoidal volt- age pertubation of around 10 mV through the sample, and measuring the current response. The current response is somewhat shifted from the input sinusoidal pattern and is also dependant on the frequency. From this the complex impedance can be calculated.

Impedance (Z) is the equivalent of resistance in direct current (DC), but for

alternating current (AC). In other words, it is the ability of the circuit to

oppose current when a voltage is applied. Unlike resistance, impedance is a

complex number (that can also be expressed as a sum of sinus and cosinus

functions) that has both a magnitude and a phase angle. The phase angle

is frequency dependant and corresponds to the phase shift between current

and voltage. Impedance is often written in its Cartesian form (eq 4.1) and

displayed using a Nyquist plot (figure 4.3).

-Z

Z

|Z|

Figure 4.3: A vector illustration of impedance in a Nyquist plot

Z = Z

+ jZ

(4.1)

Electrochemical cells can be represented by equivalent circuits using standard electrical components such as resistors and capacitors. These circuits can be made to contain all of the cell’s electrochemical characteristics such as bulk resistance (R

) and double layer capacitance (C

). These standard electrical components all have varying impedance responses with frequency, which enables separation of the individual contributions.

For solid polymer electrolytes, an equivalent circuit called a Randle circuit

(RC) is often used to describe the electrical behaviour (figure 4.4). The

Nyquist plot show two characteristic regions: a semicircle at high frequency’s

caused by the bulk properties of the electrolyte, and a tail at lower frequency’s

attributed to ion diffusion (figure 4.5).

R

C

C

R

Figure 4.4: Modified Randle circuit where R

is the internal resistance of the equipment, C

is the dielectric capacitance of a relatively resistive solid, R

is the bulk resistance of the polymer electrolyte and C

is the double layer capacitance

R

R

-Z

Z

Figure 4.5: Typical shape of the Nyquist plot for a solid polymer electrolyte

After data fitting to the modified Randle curcuit, the ionic conductivity can be calculated using equation 4.2, where R

is the bulk resistance which is taken as the width of the semicircle in fig 4.5, A is the area of the electrode, and d the thickness of the electrolyte.

σ = d

R

A (4.2)

4.2.4 Galvanostatic cell cycling

In order to monitor how materials behave in actual battery devices, repeated charge and discharge cycles are performed. There are many ways to cycle an electrochemical cell. One is by keeping a constant controlled current and is called galvanostatic cycling. It measures the response in voltage according to Ohm’s law as the current is applied. The cycle is conducted by applying a small current until a certain voltage or time is reached, and the current is then reversed. This charges and discharges the battery repeatedly. The rate of the charge or discharge respectively is called the C-rate, and is specified as the time for a full charge or discharge cycle of the theoretical capacity, i.e.

1 C = 1 h, C

20 = 20 h, 2 C = 30 min.

4.2.5 Fourier Transform Infrared Spectroscopy (FTIR)

Molecular vibrations typically have energies corresponding to the infrared region of the electromagnetic spectrum. Therefore, they absorb light corre- sponding to the characteristic energy of the vibration. FTIR is a scpectro- scopic technique where a sample is subjected to a range of wavelengths in the IR-region, and analyses the reflected light by fourier transform to process the data.

In the present work, FTIR is used to study polymer-salt interactions. This

is done by observing the shift in characteristic bond frequencies, such as the

broadening of the OH-peak in the presence of hydrogen bonding.

5 Experimental

5.1 Electrode and electrolyte preparation

5.1.1 Polymer electrolyte films

For the casting of the polymer electrolyte films, commercial PVA (Mowiol 20-98) with a molecular weight of ∼ 125,000 and 98 % degree of hydrolysis was used as recieved.

DMSO-based

Films were prepared inside of a glovebox by weighing 1 g of polymer that

was then dissolved in 10 ml of DMSO by heating to 60

C under stirring for

several hours. The heat was then turned off and the solution were left to stir

overnight. Appropriate amounts of salt (LiTFSI, LiTf, LiClO

) were added,

and once again left to stir for several hours. The standard amount of salt was

chosen to be 1 Li

per 4 repeating polymer units (henceforth denoted as the

OH:Li

ratio), since this concentration had the highest ionic conductivity

found in literature [19]. 1.5 ml of the solutions were then poured into PTFE

moulds for solvent evaporation. They were treated with a drying procedure

in a vacuum furnace by carefully pumping down from 200 to 0.9 mbar over

20 h. This pressure was then kept for an additional 3 days before heating to

60

C for 20 h at the same pressure. A summary of DMSO samples can be

found in table 5.1.

Table 5.1: Summary of DMSO-based samples (OH:Li

denotes Li per re- peating unit)

Salt: OH:Li: wt% salt:

LiTf 4:1 47

LiClO

4:1 37

LiTFSI 4:1 62

LiTFSI 10:1 40

LiTFSI 50:1 12

LiTFSI 100:1 6

Water-based

Water-based PVA-salt solutions were made in similar fashion as the DMSO equivalents, with the exceptions of being outside of the glovebox and using 10 ml of distilled H

O and heating to 100

C. 1.5 ml of each solution was poured into PTFE moulds and the solvent was evaporated in a furnace using a temperature program set to 40

C, 60

C and 80

C for 4 h each. The program was chosen in order to get a smooth solvent evaporation and to avoid bubbles forming. The films were then transferred into a glovebox, and some were subjected to a drying procedure in a vacuum furnace at 120

C with a pressure of 0.9 mbar to remove any possible residual water. The contents of all the H

O casted samples can be seen in table 5.2.

Table 5.2: Summary of water-based samples Salt: OH:Li: wt% salt: Dried:

LiTf 4:1 47 No

LiClO

4:1 37 No

LiTFSI 4:1 62 No

LiTf 4:1 47 Yes

LiClO

4:1 37 Yes

LiTFSI 4:1 62 Yes

Hot-pressing

Polymer and salt powders with compositions presented in table 5.3 were prepared by ball-milling for 1.5 h at 300 rpm. Samples were then hot pressed with 3 tons of pressure at 200

C for 1 h in an inert atmosphere. The powders were pressed into circular moulds made out of 100 µm thick brass foil. Hot-pressed films with LiClO

underwent thermal decomposition and will therefore be excluded from the results.

Table 5.3: Summary Hotpressed samples Salt: OH:Li: wt% salt:

LiTf 4:1 47

LiClO

4:1 37

LiTFSI 4:1 62

5.1.2 Electrodes

Electrode slurrys were made by mixing a total of 2 g consisting of 80 wt%

of LiFePO

(LFP), 10 wt % conducting carbon and 10 wt% of PVA-binder dissolved in distilled H

O. PVA was chosen as binder to achieve good ad- hesion between the electrodes and polymer electrolytes. The mixture was then ball-milled for 1 h at 300 rpm using a Retsch PM 100 planetary ball mill. The slurry was then coated onto carbon-coated Al-foil and left to dry at ambient conditions. Circular electrodes were then punched from the foil and transferred into a glovebox where they were dried in a vacuum furnace at 120

C for 12 h.

5.1.3 Battery cell assembly

Electrochemical cells for battery testing were assembled in a pouch-cell for-

mat by first assembling current collectors and the polymer coated aluminium

foil pieces by heat-sealing. The pouch was then brought into a glovebox where

the LFP electrodes, Li-metal foil and polymer electrolyte were carefully as-

sembled to avoid short circuiting the cell. The cell was then vacuum sealed

and taken out of the glovebox for cycling experiments.

5.2 Analysis

5.2.1 Thermograviometric Analysis (TGA)

TGA was performed using a TA-instruments TGA Q500. Polymer electrolyte samples were taken out of the glovebox shortly before the experiments started and put in alumina pans to minimize moisture contamination. Data was collected from room temperature to 550

C with a heating rate of 5

C/min under N

.

5.2.2 Differential Scanning Caliometry (DSC)

Samples were sealed in air-tight aluminium pans inside of a glovebox and loaded onto a TA-instruments Q200. The pans were put on a heat-cool-heat cycle from room temperature to 150

C, cooled to -20

C and then heated to 150

C, all at 5

C/min

5.2.3 Electrochemical Impedance Spectroscopy (EIS)

Polymer electrolyte films were put between two stainless steel blocking elec-

trodes in a spring-loaded Swagelok-type cell seen in figure 5.1. The cell was

thoroughly sealed and then taken out of the glovebox and put into a fur-

nace, where it was heated to 115

C and kept for at least 30 min in order

to establish good surface contact. Impedance measurements where taken

on the cooling run with intervals of 20

C using a Schlumberger SI 1260

Impedance/Gain-phase analyzer. Spectra where recorded at 81 points in the

interval 1 Hz - 10 MHz with an amplitude of 10 mV. The data was fitted in

ZView v 3.2b using the equivalent curcuit presented in figure 4.4.

Figure 5.1: Swagelok type cell used for impedance measurements

5.2.4 Galvanostatic cell cycling

Cell cycling was performed using an Arbin battery tester and a furnace where the cells were kept at 60

C. A resting period of 24h was used before cycling began to establish a better electrode-electrolyte interface. Cycles were per- formed between 3 to 4 V at rates C/50, C/25 and C/10.

5.2.5 Fourier Transform Infrared Spectroscopy (FTIR)

Samples were taken out of the glovebox and pressed into an Attenuated Total

Reflectance (ATR) sample holder on a Perkin Elmer Instruments Spectrum

One. IR spectra were recorded between 4000 and 650 cm

with a resolution

of 4 cm

for 16 scans.

6 Results and discussion

6.1 Ionic conductivity

6.1.1 Variation of salt content in the PVA-LiTFSI-DMSO system

Presented below in figure 6.1 are the results of ionic conductivities in PVA- LiTFSI polymer electrolyte films casted with DMSO. It can be observed from the left graph that large amounts of salt are necessary to obtain sufficiently high ionic conductivity values that are required for battery applications.

0 1 0 2 0 3 0 4 0 5 0 6 0 7 0

- 2 0 - 1 8 - 1 6 - 1 4 - 1 2 - 1 0 - 8 - 6 - 4 - 2

log(σ [S/cm])

w t % s a l t R o o m t e m p e r a t u r e

2 . 6 2 . 8 3 . 0 3 . 2 3 . 4

- 2 0 - 1 8 - 1 6 - 1 4 - 1 2 - 1 0 - 8 - 6 - 4 - 2

log(σ [S/cm])

1 0 0 0 / T [ K ^{- 1}] 4 : 1

1 0 : 1 5 0 : 1 1 0 0 : 1

a) b)

Figure 6.1: Variation of ionic conductivity with a) increased content of

LiTFSI salt at room temperature b) temperature dependence (salt concen-

trations given by the OH:Li

ratio)

It has previously been reported that a dramatic increase (”kink”) in conduc- tivity of the PVA-LiTf-DMSO system between 40 and 50 wt%. However, the LiTFSI does not show this behaviour in figure 6.1 a) where the difference is only one order of magnitude at high salt concentrations [21] [18]. At lower concentrations, a large increase from 10

to 10

S cm

can be seen due to the increase in the amount of charge carriers.

In figure 6.1 b), the temperature dependence of ionic conductivity is shown for the different salt concentrations. At the lowest concentration, a clear thermal activation can be observed between 40 and 60

C (3.2-3.0 1000/T) where the ionic conductivity increases from 10

to 10

S/cm (this is also present in the 50:1 concentration, although not as distinct). The activation could be due to that these concentrations reaches their glass transition at these temperature, and therefore enabling more motion of the polymer chain, whereas the higher concentrations are plastiziced enough to be conducting even at room temperature. In the case the two highly concentrated samples (black and red), the conductivity only changes about one order of magnitude in the whole temperature range. The black and red curves in figure 6.1 b) also show slightly curved trends, indicating VTF-type behaviour with ion conduction by segmental motion of the polymer. This is believed to be due to both the increase in the number of charge carriers at higher salt concentrations, and that more salt leads to more plasticization from the anions.

6.1.2 Effects of different salts and solvents

Salt

Presented in figure 6.2 is a comparison of the ionic conductivities between

PVA electrolyte films complexed with various lithium salts. LiTFSI has the

highest recorded conductivity of 0.4 mS/cm at 60

C (3.0 1000/T) and 45

µS/cm at room temperature (3.4 1000/T), followed by LiClO

and LiTf

making the trend LiTFSI > LiClO

> LiTf (for a full list of maximum

conductivities, see table 6.1). The fact that TFSI anion gives the highest

conductivities is probably due to it being the biggest of the three, and thus

having the highest dissociation from Li

and the largest plasticizing effect

on the polymer.

Table 6.1: Ionic conductivities of PVA electrolyte films (25 mol% salt) Salt: Solvent: σ(60

C) [S/cm] σ(20

C) [S/cm]

LiTFSI DMSO 4.2 × 10

4.5 × 10

LiTf DMSO 4.2 × 10

3.9 × 10

LiClO

DMSO 1.9 × 10

1.5 × 10

The general trend of the LiTFSI samples presented in all the subfigures in figure 6.2 is more or less linear with the exception of the lower temperature.

In this temperature region the conductivities tend to show a convergence towards a single value in the water-based and hotpressed samples. PVA- LiTFSI-DMSO (fig 6.2 a) black) however stands out from this this behaviour due to remarkably higher room temperature conductivity. To be noted here is that the TFSI anion in itself does not seems to be plasticizing enough to make PVA sufficiently ionically conducting at lower temperatures, since solvent free samples containing these salts do not show high enough ionic conductivities.

The PVA-LiTf-DMSO (fig 6.2 a) red) sample shows a smooth curved line in resemblance of VTF-behaviour, but does not maintain its high conductivity at lower temperatures as LiTFSI (black). Otherwise, the same trends can be seen irrespective of casting with the exception of the dried water sample in figure 6.2 d). The same drop in conductivity with temperature can be observed for the LiClO

and will be discussed further.

Solvent

When looking at figure 6.2 with regards to solvent, several interesting obser-

vations can be made; firstly, the system PVA-LiTFSI-DMSO in figure 6.2 a)

works particularly well compared to the other systems. In figure 6.2 c), LiTf

shows slightly higher conductivities than LiClO

in the water-casted films,

which is not the case in the DMSO-casted in 6.2 a). Secondly, hot-pressed in

fig 6.2 b) and non dried water-casted samples (fig 6.2 c)) show similar values

and trends in ionic conductivity, which would be expected if they are both

solvent free (shown below in the TGA results). One would expect that the

dried water samples in figure 6.2 d) would also show similar conductivities,

but the drying procedure influences the ionic conductivity about one order of

magnitude in comparison with the non-”dried”. This difference is believed to

be due to the degree of crystallinity of the polymer since the samples contain the same amount of residual solvent (shown in section 6.2.1). Lastly, the general conductivity trend in terms of solvent used is; DMSO > Water >

Hotpressed > Dried water. The fact that DMSO outperforms all the other could be both to plasticizing effects, or to ionic conduction of the liquid phase of residual solvent.

2 . 6 2 . 8 3 . 0 3 . 2 3 . 4

- 1 0 - 8 - 6 - 4 - 2

log(σ [S/cm])

1 0 0 0 / T [ K ^{- 1}] L i T F S I

L i T f L i C l O 4 D M S O

2 . 5 2 . 6 2 . 7 2 . 8 2 . 9 3 . 0 3 . 1 3 . 2 3 . 3 3 . 4

- 1 0 - 8 - 6 - 4 - 2

log(σ [S/cm])

1 0 0 0 / T [ K ^{- 1}] L i T F S I

L i T f H o t p r e s s e d

a) b)

2 . 5 2 . 6 2 . 7 2 . 8 2 . 9 3 . 0 3 . 1 3 . 2 3 . 3 3 . 4

- 1 0 - 8 - 6 - 4 - 2

log(σ [S/cm])

1 0 0 0 / T [ K ^{- 1}] L i T F S I

L i T f L i C l O 4 W a t e r

2 . 5 2 . 6 2 . 7 2 . 8 2 . 9 3 . 0 3 . 1 3 . 2 3 . 3 3 . 4

- 1 0 - 8 - 6 - 4 - 2

log(σ [S/cm])

1 0 0 0 / T [ K ^{- 1}] L i T F S I

L i T f L i C l O 4 D r i e d W a t e r

c) d)

Figure 6.2: Comparison of the temperature dependence of ionic conductivity with various solvents and salts; a) DMSO b) Hotpressed, solvent free, LiClO

missing due to decomposition c) Water d) ”Dried” water

The fact that LiClO

outperforms LiTf in the DMSO-casted films (fig 6.2 a)) could be due to some favourable coordination with the residual solvent.

This could have been confirmed with the hot-pressing results if the LiClO

sample did not decompose (further studies needed). The water-casted, hot- pressed and DMSO samples all show very consistent trends with temperature, displaying linear regions at elevated temperatures and curved near lower (figures 6.2 a), b) and c)). DMSO-films (fig 6.2 a)) in particular tend to have very typical VTF-like shapes indicating that the ion-conduction is performed by the segmental motion of the polymer backbone due to the plasticizing effect of DMSO.

Hot-pressed and water-casted films in figure 6.2 b) and c) show similar conductivities, although somewhat higher at elevated temperatures for the water-casted. This indicates that there are small amounts of residual solvent left in the samples, even before the drying step. One would expect lower conductivities in the dried water samples (6.2 d)) due to the fact that water also acts as a plasticizer in PVA. Also, the additional heat treatment could promote the growth of crystalline regions, meaning less traditional ion con- duction that takes place in the amorphous phase. This is also supported by the fact that the dried water films were more brittle and visually cloudier than regular water-casted ones where the extra heat treatment was skipped.

6.2 Thermal properties

6.2.1 TGA

Thermal stability

The thermal decomposition of the as-received PVA (figure 6.3 a), black)

starts with the evaporation of moisture at ∼100

C, followed by the elim-

ination of OH-sidegroups between 200 and 300

C and breakdown of the

backbone at 350

C and beyond [25]. LiTFSI (figure 6.3 a), red) shows just

a single decomposition temperature at 320

C.

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0

0

2 0 4 0 6 0 8 0 1 0 0

Weight %

T e m p e r a t u r e [ ° C ] P V A

L i T F S I

P V A - L i T F S I - D M S O

0 1 0 0 2 0 0 3 0 0 4 0 0 5 0 0 6 0 0

0

2 0 4 0 6 0 8 0 1 0 0

Weight %

T e m p e r a t u r e [ ° C ] P V A - L i T F S I - H 2 O ( u n d r i e d ) P V A - L i T F S I - H 2 O ( d r i e d )

a) b)

Figure 6.3: Thermal decomposition of a) PVA, LiTFSI and DMSO-casted PVA-LiTFSI film b) Water casted PVA-LiTFSI films

All prepared PVA-based electrolyte films show lower decomposition tempera- tures than the pure polymer, although they should all be sufficient for normal battery applications. DMSO-casted film in figure 6.3 a) blue line, start off with evaporation of residual solvent at round 100

C (boiling point of DMSO is 189

C), but does not start its polymer decomposition until 240

C. The water-casted electrolytes in figure 6.3 b) show no clear signs of residual sol- vent and no clear difference between the dried and undried samples, which is interesting because there is a clear difference in conductivity between the two. The difference in conductivity presented in figure 6.2 c) and d) should therefore not be due to plasticization of residual water solvent present in the films. They also show almost identical decomposition temperatures as the DMSO-casted (6.3 a)), which means that these films are thermally stable for ambient battery applications, and could be considered for high temperature applications.

Residual solvent

Figure 6.4 presents a closer look at the regions of solvent evaporation of DMSO and water-casted PVA electrolyte films. The DMSO content is taken as the weight loss before the start of the main polymer decomposition at 240

◦

C in figure 6.4 a) and is found to be 10 wt%, which is comparable to what

was reported previously by MacFarlane et.al. [18].

2 0 0 3 0 0 4 0 0 6 0

7 0 8 0 9 0 1 0 0 1 1 0

Weight %

T e m p e r a t u r e [ ° C ] P V A

P V A - L i T F S I - D M S O

1 0 0 2 0 0 3 0 0

9 0 9 5 1 0 0 1 0 5

Weight %

T e m p e r a t u r e [ ° C ] P V A - L i T F S I - H 2 O ( u n d r i e d ) P V A - L i T F S I - H 2 O ( d r i e d )

a) b)

Figure 6.4: Magnification of thermal decomposition curves for a)DMSO- casted films b) water-casted films

The solvent content in the water-casted films turned out to be much lower than expected. Looking at figure 6.4 b), the difference between the samples are only 0.43 wt% at 240

C, which indicates that the second drying step inside of the glovebox was not necessary in terms of moisture contamination.

This extra ”drying” is although to be considered as a thermal annealing that should be taken into consideration in terms of growth of crystalline regions that do not contribute to the ionic conductivity of the electrolyte films as could be seen in the ionic conductivities above in section 6.1.2. The small difference in water content should not be significant enough to contribute a plasticizing effect.

6.2.2 DSC

DSC data collected on pure PVA follows a steady decline during the lower part of the curves which represents the heating cycle in figure 6.5 (black lines) which is probably the cause of absorbed moisture. In order to get a more distinct value, the polymer has to be dried beforehand. However there is a slight change in the slope between 70 and 120

C in figure 6.5 that gives a T

of 80

C (compared to 85

C found in the literature [10]). Variations from

literature is to be expected since it is influenced by properties such as degree

of polymerization and hydrolysis.

- 4 0 - 2 0 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0 - 2 . 0

- 1 . 5 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0 1 . 5

Heat Flow [W/g]

T e m p e r a t u r e [ ° C ] P u r e P V A

L i T F S I D M S O L i T F S I H o t p r e s s L i T F S I H 2 O L i T F S I H 2 O d r i e d

- 4 0 - 2 0 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0

- 2 . 0 - 1 . 5 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0 1 . 5

Heat Flow [W/g]

T e m p e r a t u r e [ ° C ] P u r e P V A

L i T f D M S O L i T f H o t p r e s s L i T f H 2 O L i T f H 2 O d r i e d

a) b)

- 4 0 - 2 0 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0

- 2 . 5 - 2 . 0 - 1 . 5 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0 1 . 5

Heat Flow [W/g]

T e m p e r a t u r e [ ° C ] P u r e P V A

L i C l O 4 D M S O L i C l O 4 H o t p r e s s L i C l O 4 H 2 O L i C l O 4 H 2 O d r i e d

c)

Figure 6.5: DSC traces of a) PVA-LiTFSI b) PVA-LiTf c) PVA-LiClO

Water-casted samples show clearly defined T

regions in figure 6.5 along with the hot-pressed films. It was expected that the ”dried” water samples were to have higher T

values attributed to the growth of crystalline domains during thermal annealing. This is confirmed for LiTf and LiClO

containing samples in figure 6.5 b) and c), but stays roughly the same in the LiTFSI films 6.5 a)) which is belived to be due to the plasticizing effect of the TFSI anion.

The increase in T

further supports the theory discussed in the conductivity

results that ions do not migrate in the crystalline regions of the polymer, but

are conducted by the motion of the polymer chains in the amorphous phase.

Another feature of the DSC results in figure 6.5 is that the DMSO-casted films have no distinct glass transition in the studied temperature range. In the case of LiTf and LiClO

, there are similar features of glass transiotion between -10 and 20

C (figure 6.5 b) and c)), but this is attributed to the phase transition of DMSO from liquid to solid, since it has a melting point of 19

C. The decline of DMSO films in figure 6.5 at temperatures beyond the T

of pure PVA is attributed to evaporation of residual solvent with reference to the TGA results above in section 6.2.1. In order to see the glass transi- tion of these films, one would have to expand the temperature range. This was however not possible in this study due to time restrictions. It should be noted that the DMSO-casted films are heavily plasticized with T

values below room temperature, which could be the cause of their superior conduc- tivities. Nevertheless, the PVA-LiTFSI-DMSO show a somewhat stable glass transition of around 60

C in figure 6.5 a) (this sample also has a higher ionic conductivity), and the other have been estimated at start of the heating cycle from figure 6.5 b) and c).

DMSO H2O H2O dried Hotpress DMSO H2O H2O dried Hotpress DMSO H2O H2O dried Hotpress

C l O 4 T f T F S I

0

2 0 4 0 6 0 8 0

Temperature (C)

P u r e P V A

Figure 6.6: T

values of PVA-based polymer electrolyte films

Presented in figure 6.6 is a summary of glass transition temperatures mea-

sured from figure 6.5, where the hot-pressed samples show T

often compa-

rable to that of the water-casted. This further indicates the fact that water-

casted are solvent free, as could be seen with the TGA. The variation from

samples were subjected to increased temperature for a shorter time period than the water-casted, inhibiting the growth of larger crystalline domains.

6.3 Fourier Transform Infrared Spectroscopy (FTIR)

6.3.1 Polymer-salt interactions

PVA-LiTFSI

Presented in figure 6.7 are the collected IR spectra for the PVA-LiTFSI system with various solvents, where the bottom line (black) is that of pure LiTFSI and its characteristic peaks are summarized in table 6.2. First glance of the spectra in figure 6.7 observes that the dominant intensity comes from the lithium salt, and not the polymer. This is to be expected given the high salt concentration of 62 wt%, and it has been previously suggested that the IR-spectra of PVA complexed with lithium salts are just superimposed [14].

The top spectrum (green) in figure 6.7 is that of ball-milled polymer and

salt that is later used for the hot-pressing, meaning no complexation should

have occured and thus giving a simple addition of the intensities of pure PVA

and LiTFSI. This is accurate for the peaks in the region of low wavenumbers

(>1500 cm

), but the polymer peak of C – – O stretching at 1720 cm

is

shifted towards 1640 cm

and the OH-stretch around 3300 cm

towards

3500 cm

.

4 0 0 0 3 5 0 0 3 0 0 0 2 5 0 0 2 0 0 0 1 5 0 0 1 0 0 0 5 0 0

A b s o rp ti o n ( a .u ) W a v e n u m b e r ( c m ^ - 1 )

P V A + L i T F S I ( p o w d e r ) P V A + L i T F S I ( D M S O c a s t e d ) P V A + L i T F S I ( w a t e r c a s t e d ) P V A

L i T F S I

Figure 6.7: FTIR spectrum of PVA-LiTFSI samples

Table 6.2: Characteristic IR-peaks for LiTFSI [26]

Wavenumber: Group: Mode:

1327 cm

CF

Anti-symmetric stretch 1201 cm

SO

Stretch

811 cm

S

N Symmetric stretch 746 cm

S

N Symmetric stretch

The IR-spectra of the casted samples (blue and pink in fig 6.7) all have

slightly shifted salt peaks that are near identical, the CF

anti-symmetric

stretch at 1327 cm

is shifted towards 1325 cm

, SO

stretch at 1201 cm

to 1183 cm

and the S

N peaks at 811 cm

and 746 cm

to 791 cm

and

739 cm

. The salt peaks are thereby all shifted towards lower wavenumbers

along with a strong deformation and shift of the OH-stretching around 3500

to 3300 cm

. This is strong indication of that the spectra are not simply superimposed, but that there are interactions between the lithium salt and the OH-sidegroups of the polymer.

In the DMSO-casted sample (pink), an extra peak at 946 cm

is observed attributed to the stretching mode of sulfoxide from residual solvent.

PVA-LiTf

In figure 6.8, one can see similarities with that of the LiTFSI in figure 6.7.

Salt peaks are dominant in intensity, and the casted samples (blue and pink) all look very similar with the exception of OH-stretching and the sulfox- ide stretching. It is also observed that the salt peaks (presented in table 6.3) of SO

at 1291 cm

and CF

at 1250 cm

, are shifted towards lower wavenumber and start to overlap.

Table 6.3: Characteristic IR-peaks for LiTf [27]

Wavenumber: Group: Mode:

1291 cm

Characteristic LiTf -

1250 cm

SO

Anti-symmetric stretch

1183 cm

CF

Anti-symmetric stretch

1042 cm

SO

Symmetric stretch

4 0 0 0 3 5 0 0 3 0 0 0 2 5 0 0 2 0 0 0 1 5 0 0 1 0 0 0 5 0 0

A b s o rp ti o n ( a .u ) W a v e n u m b e r ( c m ^ - 1 )

P V A + L i T f ( p o w d e r ) P V A + L i T f ( D M S O c a s t e d ) P V A + L i T f ( w a t e r c a s t e d ) P V A

L i T f

Figure 6.8: FTIR spectrum of PVA-LiTf samples

PVA-LiClO

The two characteristic peaks of LiClO

presented in table 6.4, which can

clearly be seen in the ball milled powder sample in figure 6.9 (green). The

two peaks present at 3570 cm

and 3523 cm

in the pristine salt (black)

is attributed to crystal water. In this system, one major difference from the

other two salts in figures 6.7 and 6.8 is made; the intensities from the polymer

to salt is much larger in the casted films (blue and pink) than the ball-milled

powders (green). This could be due to complexation between polymer and

salt suppresses the stretching mode of the ClO

ion at 1063 cm

.

4 0 0 0 3 5 0 0 3 0 0 0 2 5 0 0 2 0 0 0 1 5 0 0 1 0 0 0 5 0 0

A b s o rp ti o n ( a .u ) W a v e n u m b e r ( c m ^ - 1 )

P V A + L i C l O 4 ( p o w d e r ) P V A + L i C l O 4 ( D M S O c a s t e d ) P V A + L i C l O 4 ( w a t e r c a s t e d ) P V A

L i C l O 4

Figure 6.9: FTIR-spectra of PVA-LiClO

with various solvents

Table 6.4: Characteristic IR-peaks for LiClO

[28, 29]

Wavenumber: Group: Mode:

1624 cm

LiClO

1063 cm

ClO

Stretch

6.3.2 Interactions between lithium salt and functional groups

In order to get a better understanding of the complexation of lithium salts

with PVA, one should take a closer look at the OH-stretching peak around

3300 cm

in the pristine PVA (figure 6.10). What is present for all three salts

(6.10 a), b) and c)), is that the peak is shifted towards higher wavenumbers,