High Energy Density Lithium-Sulfur Bat- teries obtained using Functional Binders

Master Thesis

High Energy Density Lithium-Sulfur Bat- teries obtained using Functional Binders

Author:

Viking Österlund

Supervisor:

Matthew Lacey

UPPSALA UNIVERSITY

High Energy Density Lithium-Sulfur Batteries obtained using Functional

Binders

by

Viking Österlund

Supervisor: Matthew Lacey

A thesis submitted in partial fulfillment for the degree of Master of Science

in the

Faculty of Science and Technology Department of Chemistry

July 2015

“Reality consists of infinite possibilities. The reality we perceive is the mere manifestation of possibilities.”

Unknown author

UPPSALA UNIVERSITY

Abstract

Faculty of Science and Technology Department of Chemistry

Master of Chemistry

by Viking Österlund Supervisor: Matthew Lacey

The optimization of the positive electrode in lithium-sulfur batteries was investigated in depth in this thesis. The work was mainly focused on using polyethylene oxide(PEO):

polyvinylpyrrolidone(PVP) as a functional binder mixture together with positive elec-

trode compositions fully consisting of commercially available components, and the elec-

trode preparation was purely water-based. Using this functional binder mixture, 20 %

higher discharge capacities and significantly improved rate-capability could be demon-

strated relative to using a standard water-based carboxymethylcellulose(CMC):styrene-

butadiene rubber(SBR) binder mixture as reference. Discharge capacities of up to 600

mAh g ⁻¹ _electrode and up to 4 mAh cm ⁻² areal capacities were achieved, which represent

positive electrode energy densities twice as high as the lithium-ion layered oxide positive

electrode equivalents. The results of this project show that high energy density posi-

tive electrodes can be achieved without exotic nanomaterials and complicated electrode

preparation.

Popular scientific summary

The world is facing a number of tremendous challenges in the following decades including climate changes, depleting fossile fuels and growing energy consumption resulting from a drastically increasing population. To overcome these challenges, scientists and experts agree that major changes in the energy production from fossile-fuel based technologies to renewable energy based technologies need to be applied on a global scale. Major prob- lems with renewable energy technologies such as solar photovoltaics and wind turbines is that the production is heavily dependent on the weather and that the produced electric- ity cannot be utilized directly in fossile-fuel based motor vehicles. Good energy storage techniques need to be developed that can store the excess electricity produced during good weather conditions to be utilized when the energy consumption exceeds the energy production. Furthermore, motors using renewable energy need to be employed. Bat- teries can be used for both these purposes, efficiently storing electricity and delivering it when needed. The state-of-the-art commercially used lithium-ion (Li-Ion) batteries which nowadays dominate the portable electronics market have inherently insufficient energy densities to allow electric vehicles with more than 20 km driving range. The U.S.

Department of Energy and the car industries have set up a target to increase the driving range to 500 km. To reach this target, battery research needs to look at alternative battery systems with higher energy densities. A promising candidate is lithium-sulfur (Li-S) batteries which are predicted to reach energy densities 2-5 times larger than for the Li-Ion batteries and this would easily suffice to reach the target of 500 km driving range. However, major challenges with the Li-S batteries need to be overcome;

i) The electrically insulating nature of sulfur requires conductive carbon host structures with high surface areas

ii) The shuttle of polysulfide intermediates between the electrodes, both during battery usage and storage, which decreases the lifetime of the batteries by parasitic side-reactions at the lithium electrode

iii) The drastic volume change of ca. 80% during discharge which requires carbon hosts with high porosity and/or good mechanical integrity of the positive electrode structure Major efforts are done within the Li-S research field to overcome these challenges and the main solutions involve;

i) Synthesis of sophisticated carbon structures with high surface area and porosity ii) Including active/passive agents into the positive electrode that retains the polysulfides at the positive electrode

iii) Developing electrolytes that restricts the polysulfide solution.

Researches within the Ångström Advanced Battery Center (ÅABC), Uppsala University,

have previously discovered that the polymer mixture polyethylene oxide (PEO):polyvinylpyrrolidone (PVP) can be used in Li-S batteries to enhance the performance. They used simple

sulfur-carbon compositions together with the functional binder mixture PEO:PVP in water-based positive electrodes and observed improved energy densities and stability when using these in batteries. However, they worked with commercially unrealistic com- positions with low loadings of active material. Furthermore, the effect of using PEO:PVP is not yet well understood. The aim in this thesis work is to achieve high energy density positive electrodes based on the PEO:PVP functional binder concept by;

• Optimizing the positive electrode formulation and preparation -Increase the active material loadings

-Use ketjen black carbon with higher surface area and porosity

• Trying out alternative functional binder mixtures

• Functionalizing ketjen black carbon with PVP functionality to study the effect of using PEO:PVP as a functional binder mixture

The main result is achieving improved positive electrodes for Li-S batteries, through ex-

tensive optimization, with twice the energy density compared to the conventional com-

mercial Li-Ion layered oxide positive electrodes and comparable high areal capacities. The

electrode preparation only consisted of commercially, abundant, environmental friendly,

cheap components and was purely water-based. Combining these aspects with the good

performance shows that the developed positive electrodes have excellent possibilities of

being commercialized in the future. However, more work need to be done on the negative

electrode and electrolyte to increase the battery lifetime.

Acknowledgements

My sincere thanks for all the help I have got achieving my thesis by my brilliant super- visor Dr.Phil Matthew Lacey and talented graduate student Fabian Jeschull. Thanks to Prof. Kristina Edström and Assoc. Daniel Brandell for financing and administrating the project. I would also like to thank Carl Tengstedt, Julia Maibach, Steven Renault and Mario Valvo for their generous help with performing various characterization measure- ments. Thanks to Henrik Eriksson for the practical help with handling the lab equipment.

Thanks also to all the other nice people at Structural Chemistry and Ångström Advanced

Battery Center, Uppsala University, for providing a good atmosphere to do successful

research.

Contents

Abstract ii

Abstract Page ii

Popular scientific summary iii

Acknowledgements v

List of Figures viii

List of Tables xii

Abbreviations xiii

1 Introduction 1

1.0.1 Optimizing the cathode formulation and preparation to increase

the energy density . . . . 2

1.0.2 Comparing different functional binders to enhance the electrochem- ical battery performance . . . . 3

1.0.3 Functionalizing ketjen black carbon . . . . 3

2 Theory 4 2.1 The Lithium-Sulfur Battery . . . . 4

2.2 Electrochemistry . . . . 6

2.2.1 Internal resistance and iR drop . . . . 7

2.2.2 Polarization and overpotential . . . . 7

2.2.3 Capacity . . . . 8

2.2.4 Coulombic efficiency . . . . 9

2.2.5 Energy efficiency . . . . 9

2.2.6 Galvanostatic cycling . . . 10

2.2.7 C-rate and battery rates . . . 10

2.3 Scanning electron microscopy and Energy dispersive X-ray mapping . . . 11

2.4 X-ray photoelectron spectroscopy . . . 11

2.5 Thermogravimetric analysis . . . 12

2.6 Nuclear magnetic resonance spectroscopy . . . 13

Contents vii

3 Experimental information 14

3.1 Materials . . . 14

3.1.1 Materials used for cathode and electrolyte preparation . . . 14

3.1.2 Materials used for esterification of ketjen black carbon . . . 14

3.2 Functionalization of ketjen black carbon . . . 15

3.2.1 Oxidation of ketjen black carbon . . . 15

3.2.2 Steglich esterification of oxidized ketjen black carbon . . . 15

3.3 Preparation of Batteries . . . 15

3.3.1 Cathode preparation . . . 16

3.3.1.1 Cathode preparation based on Ketjen Black carbon . . . 16

3.3.1.2 Cathode preparation based on functionalized Ketjen Black carbon . . . 17

3.3.2 Battery assembly . . . 17

3.4 Experimental characterization . . . 17

3.4.1 Galvanostatic cycling . . . 17

3.4.2 Cycle-wait test . . . 18

3.4.3 Scanning electron microscopy and Energy dispersive x-ray mapping 18 3.4.4 X-ray photoelectron spectroscopy . . . 18

3.4.5 Thermogravimetric analysis . . . 19

3.4.6 Nuclear magnetic resonance . . . 19

4 Results and Discussion 20 4.1 Optimizing the cathode formulation and preparation to increase the en- ergy density . . . 20

4.1.1 Morphologies of the cathodes . . . 24

4.1.2 The effect of super C65 carbon black additive . . . 25

4.1.3 Effects of sulfur loading . . . 26

4.1.4 Varying the sulfur:ketjen black carbon ratio . . . 34

4.1.5 Effects of calendering . . . 36

4.1.6 Effects of carbon-coated aluminium . . . 38

4.2 Comparing different functional binders to enhance the electrochemical bat- tery performance . . . 40

4.2.1 Rate capability test . . . 44

4.2.2 Cycle-wait test . . . 46

4.3 Functionalizing ketjen black carbon . . . 47

4.4 General discussion . . . 53

5 Conclusions 54

A Appendix 56

Bibliography 57

List of Figures

2.1 a) Schematic drawing of a Lithium-Sulfur battery adapted from reference [3] b) a typical voltage profile of a Lithium-Sulfur battery adapted from reference [6]. . . . 4 2.2 Schematic drawing of XPS principles adapted from reference [32]. a) The

photoelectrical effect, b) an example of a XPS spectrum and c) a simplified scheme of the experimental setup. . . 12 4.1 Visual comparison of optimized cathode films with (Left) Z-1-ref and

(Right) Z-1 compositions. The 3.5 wt-% CNF is introduced in Z-1 but not in Z-1-ref. Both films are coated with doctor-blading using 50 µm slit-thickness. . . 22 4.2 Average measured cathode film thicknesses versus (upper) areal sulfur

loadings and (lower) densities of the cathodes versus sulfur loadings. The densities were calculated from the areal sulfur loadings, the sulfur wt-%, the average measured cathode thicknesses and the cathode diameters. . . 23 4.3 SEM pictures on cathodes comparing the (a,c) Z-1 composition with (b,d)

cathodes without CNF with a 58:28:7:7 wt-% composition of S:KB:C65:PEO- PVP (4:1 PEO:PVP). The latter is cathodes from an early stage in the optimization process. The scale in (a) and (b) is 200 µm while in (c) and (d) it is 1 µm. . . 24 4.4 Merged SEM and sulfur EDX-mapping pictures comparing morphologies

of cathodes containing different binders. The binders compared are (a) CMC:SBR, (b) PEO:PVP, (c) Pluronic:PVP and (d) PEO:PEOx. The sulfur is marked in yellow. . . . 26 4.5 Battery performance of cathodes with and without super C65 carbon black

additive evaluated by galvanostatic cycling at C/10 (167.2 mA g ⁻¹ ). The sulfur loadings were 0.9-1.1 mg S cm ⁻² and the discharge capacities are presented in mAh g ⁻¹ sulfur. The cathodes had the composition 58:28:7:7 wt-% of S:KB:X:PEO-PVP where X is super C65 carbon black or KB additive. . . 27 4.6 Apparent internal resistances of cathodes with and without super C65

carbon black additive. The sulfur loadings were 0.9-1.1 mg S cm ⁻² and

the cathodes had the composition 58:28:7:7 wt-% of S:KB:X:PEO-PVP

where X is super C65 carbon black or KB. . . 27

List of Figures ix 4.7 Battery performance of Z-1-C cathodes with varying sulfur loading eval-

uated by galvanostatic cycling at C/10 (167.2 mA g ⁻¹ ). (a) Discharge capacities, (b) voltage profiles during the 10th cycle, (c) coulombic ef- ficiencies and (d) areal capacities calculated from the average discharge capacities between the 10th and 50th cycles. For the cathodes that broke before 50 cycles, the average discharge capacities were calculated between the 10th and last cycle. All discharge capacities are presented in mAh g ⁻¹ sulfur and all cathodes contained 65 wt-% S, 21 wt-% KB, 3.5 wt-%

CNF additive, 3.5 wt-% super C65 carbon black additive and 7.0 wt-%

PEO:PVP as binder (4:1). . . 28 4.8 The apparent internal resistances of the batteries based on Z-1-C cathodes

with different sulfur loadings. All calculations where based on the voltage drops in the beginning and end of a discharge cycle as the current is turned on and off, see section 2.2.1. . . 32 4.9 Battery performance of cathodes with different S:KB ratios evaluated by

galvanostatic cycling at C/10 (167.2 mA g ⁻¹ ). (a) Discharge capacities, (b) voltage profiles during the 10th cycle, (c) coulombic efficiencies and (d) voltage profiles during the 40th cycle. The sulfur loadings were 1.1-1.2 mg S cm ⁻² and the discharge capacities are presented in mAh g ⁻¹ sulfur.

All the cathodes contained 90 wt-% S:KB with varying ratios, 3.0 wt-%

CNF additive and 7.0 wt-% PEO:PVP as binder (4:1). . . 35 4.10 Comparison of battery performance with and without calendering Z-1 and

the calendered Z-2, evaluated by galvanostatic cycling at C/10 (167.2 mA g ⁻¹ ). (a) Discharge capacities, (b) voltage profiles during the 10th cycle, (c) coulombic efficiencies and (d) voltage profiles during the 80th cycle.

The sulfur loadings were 1.9-2.0 mg S cm ⁻² and the discharge capacities are presented in mAh g ⁻¹ sulfur. All cathodes contained 65 wt-% S, 21 wt-

% KB, 3.5 wt-% CNF additive, 3.5 wt-% super C65 carbon black additive and 7.0 wt-% PEO:PVP as binder (4:1). . . 37 4.11 Comparison of battery performance with and without using carbon-coated

aluminium as substrate in the Z-1 composition cathode, evaluated by gal- vanostatic cycling at C/10 (167.2 mA g ⁻¹ ). (a) Discharge capacities, (b) voltage profiles during the 10th cycle, (c) coulombic efficiencies and (d) voltage profiles during the 97th cycle. The sulfur loadings were 1.87-1.93 mg S cm ⁻² and the discharge capacities are presented in mAh g ⁻¹ sulfur.

Both cathodes contained 65 wt-% S, 21 wt-% KB, 3.5 wt-% CNF additive, 3.5 wt-% super C65 carbon black additive and 7.0 wt-% binder. . . 39 4.12 Battery performance using cathodes with different binders evaluated by

galvanostatic cycling at C/10 (167.2 mA g ⁻¹ ). (a) Discharge capacities,

(b) voltage profiles during the 10th cycle, (c) coulombic efficiencies and (d)

voltage profiles during the 97th cycle. The sulfur loadings were 1.9-2.0 mg

S cm ⁻² and the discharge capacities are presented in mAh g ⁻¹ sulfur. All

the cathodes contained 65 wt-% S, 21 wt-% KB, 3.5 wt-% CNF additive,

3.5 wt-% super C65 carbon black additive and 7.0 wt-% binder. . . 41

4.13 Chemical structures of PVP and PEOx. . . 41

List of Figures x 4.14 Rate capability test performed on batteries made with CMC:SBR, PEO:PVP

and PEO:PEOx cathodes. The discharge rate was varied between C/10 and 1C (167.2 to 1672 mA g ⁻¹ ) whereas the charge rate was kept constant at C/10. The sulfur loadings were 2.0-2.1 mg S cm ⁻² and the discharge capacities are presented in mAh g ⁻¹ sulfur. All the cathodes contained 65 wt-% S, 21 wt-% KB, 3.5 wt-% CNF additive, 3.5 wt-% super C65 carbon black additive and 7.0 wt-% binder. . . 45 4.15 Cycle-wait test performed on batteries made with CMC:SBR, PEO:PVP

and Pluronic:PVP cathodes respectively. The charge and discharge rate was kept constant at C/10 (167.2 mA g ⁻¹ ). The upper figure shows how the discharge capacities varies with cycle number including two loops of cycle-wait repetitions with the resting periods being 0.5, 1, 3, 7 and 14 days respectively, see section 3.4.2. The sulfur loadings were 2.0-2.1 mg S cm ⁻² and the discharge capacities are presented in mAh g ⁻¹ sulfur. All the cathodes contained 65 wt-% S, 21 wt-% KB, 3.5 wt-% CNF additive, 3.5 wt-% super C65 carbon black additive and 7.0 wt-% binder. . . 47 4.16 The reaction scheme used for functionalization of ketjen black carbon

starting with an oxidation of ketjen black carbon powder followed by a Steglich esterification using the pyrrolidone unit 1-(2-Hydroxyethyl)- 2-pyrrolidone (HEP) as a linking agent. . . 48 4.17 NMR spectras in DCCl 3 solvent on (a) functionalized ketjen black carbon

(S-7) after water-ethanol washing, (b) functionalized ketjen black carbon after acetonitrile washing, (c) HEP reference, (d) DIC reference and (e) DMAP reference. The reference spectra were all obtained from Sigma- Aldrich. The samples measured were dissolved in DCCl 3 for 3 days prior to the measurement. . . 49 4.18 TGA on (black) unmodified KB carbon, (red) oxidized KB carbon and

(green, dashed) functionalized ketjen black carbon. The samples were all in dried powder-form when measured and oxygen atmosphere was used with a ramp-heating procedure of 5 ^◦ C/minute up to 500 ^◦ C. . . 49 4.19 SEM picture and EDX mapping of (yellow) sulfur and (blue) nitrogen on

a functionalized KB cathode. . . 50 4.20 XPS spectra of C1s, N1s and O1s binding energies. The samples measured

were KB powder (green), oxidized KB powder (gray), functionalized KB powder (red), a casted functionalized KB cathode (black) and a HEP film (blue) casted upon aluminium foil. . . 51 4.21 Battery performance of the functionalized ketjen black carbon cathode

compared with a non-functionalized reference and a PEO:PVP reference,

evaluated by galvanostatic cycling at C/10 (167.2 mA g ⁻¹ ). (a) Discharge

capacities, (b) voltage profiles during the 10th cycle, (c) coulombic effi-

ciencies and (d) voltage profiles during the 40th cycle. The sulfur loadings

were 1.1-1.2 mg S cm ⁻² and the discharge capacities are presented in mAh

g ⁻¹ sulfur. The PEO binder cathodes had the composition 65:21:3.5:3.5:7

wt-% of S:X:CNF:C65:PEO where X is functionalized ketjen black or

pure ketjen black. The PEO:PVP binder cathode had the composition

65:21:3.5:3.5:7 wt-% of S:KB:CNF:C65:PEO-PVP. . . 52

List of Figures xi A.1 Photographs of cathodes with and without ethanol added to the solvent.

The cathodes had the composition 58:28:5:2:7 wt-% of S:KB:CNF:C65:PEO- PVP. The solvent when not using ethanol was 8 ml water and the solvent when using ethanol was 8 ml water mixed with 1 ml ethanol. . . 56 Frontpage figure: Schematic of a lithium-sulfur battery and a smart grid adapted from reference [4] and the website http://www.hitachi.com/environment/showcase/solution/energy/

smartgrid.html, respectively.

List of Tables

3.1 Compositions and sulfur loadings of the different cathodes prepared in- cluding the cathodes optimized with KB (Z-1, Z-2, Z-3 and S-7-1), cath- odes with varying S:KB ratio (V-0, V-1, V-2 and V-3) and cathodes uti- lizing functionalized KB with the Steglich esterification approach (S-7-C and S-7-A). The largely varying sulfur loadings were the result of using doctor-blading with varying slit-thickness when coating the cathode films, the slit-thicknesses presented in the column indexed "slit-thickness". The average cathode thicknesses were measured using an ABSOLUTE microm- eter (Mitutoyo, Mexico). Some of the cathodes were calendered and the average thicknesses of these are also included and indexed "Calendered". . 16 4.1 Data on the effects of sulfur loading extracted from the galvanostatic cy-

cling data presented in Figure 4.7. The calculated data consist of initial discharge capacity q d,i [mAh g ⁻¹ ], average discharge capacity q d,a [mAh g ⁻¹ ], average capacity lost per cycle q l,a (%) , average coulombic efficiency C _E,a (%) , average energy efficiency η ef f,a (%) and the average areal ca- pacity q a [mAh cm ⁻² ]. All averages were taken between the 10th and 50th cycle except for the batteries which broke before 50 cycles where the average is taken between the 10th and last working cycle instead. The discharge capacities are presented as mAh g ⁻¹ of sulfur. . . 31 4.2 Data on the effects in battery performance of using cathodes with different

binders extracted from the galvanostatic cycling data presented in Figure

4.12. The calculated data consists of initial discharge capacity q d,i , aver-

age discharge capacity q d,a , average capacity lost per cycle q l,a , average

coulombic efficiency C E,a , average energy efficiency η ef f,a and the average

areal capacity q a [mAh cm ⁻² ] . All averages are calculated between the

10th and 50th cycle and the discharge capacities are presented as mAh

g ⁻¹ of sulfur. . . 43

Abbreviations

S Sulfur

PS Polysulfides Li-S Lithium-sulfur Li-Ion Lithium-ion

PEO Polyethylene(oxide) PVP Polyvinylpyrrolidone PEOx Poly(2-ethyl-2-oxazoline)

KB Ketjen Black Carbon

CNF Carbon nanofibers

C65 Super C65 conductive carbon black HEP 1-(2-hydroxyethyl)-2-pyrrolidone SEM Scanning electron microscopy

EDX Energy dispersive X-ray spectroscopy XPS X-ray photoelectron spectroscopy TGA Thermogravimetric analysis XAS X-ray absorption spectroscopy RDE Rotating disc electrode

DOL 1,3-dioxolane

DME 1,2-dimethoxyethane

LiTFSI lithium bis(trifluoromethyl)sulfonimide SEI Solid electrolyte interphase

OCV Open-circuit voltage

Abbreviations xiv

CMC Carboxymethylcellulose

SBR Styrene-butadiene rubber

DMAP 4-(Dimethylamino)pyridine

DMF N,N-dimethylformamide

DIC N,N’-Diisopropylcarbodiimide

Dedicated to my lovely lifepartner Johanna, my supporting mother, my inspiring father and the rest of my wonderful family and friends

who believes in me . . .

1 | Introduction

Lithium-ion batteries represent the state-of-the-art commercially available battery tech-

nologies with maximum theoretical specific capacities of 800 Wh kg ^-1 and maximum

practical energy densities of ∼ 200-240 Wh kg ⁻¹ [1, 2]. However, this is not enough to

meet the targeted driving range of ≥ 500 km for electrical vehicles [3, 4], which pushes

researchers towards systems with higher energy density. The Li-sulfur redox couple has

a high theoretical specific energy of 2600 Wh kg ⁻¹ [5, 6] calculated from the interconver-

sion redox reaction S 8 + 16Li ⁺ + 16e ⁻ 8Li 2 S with the polysulfide (PS) intermediates

(Li ₂ S _X , X = 4−8) [7, 8] and sulfur (S) itself has a specific theoretical capacity of 1672 Ah

kg ⁻¹ . From the high theoretical specific energy of the Li-S redox couple, one can predict

that Li-S batteries can be made with practical energy densities of ∼ 400-600 Wh kg ⁻¹

[3, 4, 9] which is much higher than the maximum practical energy densities of the Li-ion

batteries and meets the target of driving ranges exceeding ∼ 500 km. This together with

the low price and abundance of sulfur has made them the target of a rapidly increas-

ing surge of research developing lithium-sulfur batteries [5, 6]. Big challenges remain

before commercializing these batteries, including the insulating nature of lithium sulfur

discharge and charge products, intermediates leaking out to the electrolyte causing capac-

ity fade during cycling, and inherent volume expansion during discharge causing cracking

of the cathode structure leading to battery failure [3, 5, 6]. Current research is mainly fo-

cused on developing sophisticated conductive host structures to deal with these problems

including spherical carbon[10], graphene [11], metal oxides [12, 13], carbon nanofibers

[13] and nitrogen-doped carbon [14], [15]. Some successfully demonstrated strategies to

tackle the soluble intermediates dissolution is to coat the host structure with Lewis acids

interacting with the polysulfide anions[11–13], physical containment through cathode

porosity design [10], ion-selective separators [16] and the use of bifunctional binders such

as Polyvinylpyrrolidone (PVP) which interacts strongly with the discharge products and

displays major capacity retention [17, 18]. Polyethylene oxide (PEO):PVP composites as

binder have previously been proven successful, with an optimized ratio of 4:1, achieving

high initial capacities ∼ 1200 mAh g ⁻¹ and excellent capacity retention of ∼ 800 mAh

g ⁻¹ after 200 cycles at 1C (1672 mA g ⁻¹ ) combining the improved kinetic effects of PEO

and the stabilizing effects of PVP [18]. The performance of this system, corresponding

to approximately 1 mAh cm ⁻² areal capacity [18] does not reach the target of 2-4 mAh

cm ⁻² to be competitive with the best Li-ion batteries [15]. However, the water-solubility

Introduction 2 of these binders, the low-cost, high commercial production and enviromental friendliness make this lithium-sulfur system an excellent candidate for future commercialization. Fur- thermore, this approach avoids exotic nanomaterials that are expensive to produce, hard to characterize and, in some cases, complicate the cathode production. Zheng et al.

[19], have recently expressed the importance of reproducible cathodes with reproducible battery performance and the fact that nanomaterials in many cases may yield cathode films of poor quality due to the large volume shrinkage when drying the cathodes using nano-dimension particles with high surface area and porosity. The ultimate aim of this work was to utilize the promising concept of functional binders based on PEO:PVP to produce reproducible high energy density batteries with long lifetime, competitive for commercialization. To achieve this, the project work is roughly divided into three partly correlated areas of research which are presented in sections 1.0.1, 1.0.2 and 1.0.3.

1.0.1 Optimizing the cathode formulation and preparation to increase the energy density

To obtain commercially competitive energy densities, the areal capacity needs to be im- proved from the current 1 mAh cm ⁻² to the range 2-4 mAh cm ⁻² and preferably beyond.

The areal capacity of the system can be increased by increasing the sulfur ratio of the to-

tal cathode composition and thickness of the cathode film which requires good adhesion

of the composite to the cathode substrate and a proper solvent composition to prevent

cracking effects and detachment, posing a challenging trade-off to maintain good battery

performance. High surface area and large pore volume have previously been shown to

be important for good interfacial contact between reactants and enabling deposition of

solid discharge products while maintaining the cathode structural integrity [20], both

properties enhancing the battery performance. Higher porosity may also improve the

rate capability of the battery due to enhanced diffusion of intermediates as pores may be

blocked by solid discharge products during discharge. Furthermore, higher sulfur ratio

in the cathode composition also requires higher surface area and porosity of the carbon

host since the percentage of carbon needs to be decreased. Therefore, XPB Carbon black

(BET surface area ∼ 1100 m ² g ⁻¹ , pore volume ∼ 2.5 cm ³ g ⁻¹ ) used in earlier work is

exchanged for Ketjen Black EC-600 (KB) carbon (BET surface area ∼ 1400 m ² g ⁻¹ ,

pore volume ∼ 3.0 cm ³ g ⁻¹ ) to improve the battery performance and allow higher sulfur

ratio cathodes. However, increasing the surface area increases the surface tension lead-

ing to more agglomeration and the higher porosity may reduce the mechanical integrity

of the cathode films. Additionally, increasing the cathode thickness gives rise to larger

volume shrinkage when the cathode slurries are drying which requires very good mechan-

ical integrity to prevent the films from cracking [19]. These phenomena pose formidable

Introduction 3 challenges for successful cathode film preparation which are usually solved by using good dispersing agents and binders that provide higher mechanical integrity. Working with the highly porous KB carbon with very high surface area while achieving thick cathodes to unlock higher energy densities requires extra efforts which are investigated at depth in this thesis.

1.0.2 Comparing different functional binders to enhance the electro- chemical battery performance

Different types of binders based on PEO and PVP are used in batteries to further study the synergistic effect of PEO:PVP and effects of alternating the chemical structure.

Furthermore, studies on the morphology and sulfur distribution of these binders are included to characterize the possible connections between morpology and electrochem- istry. Spatial effects on the electrochemical performance are investigated by introducing the surfactant Pluronic F-127 which is similar to PEO. Poly(2-ethyl-2-oxazoline), with similar structure as PVP, is tested to study the effects of different chemical structures on the interaction with polysulfide intermediates. These binders are compared with the binder 2:3 carboxylmethylcellulose (CMC):Styrene-butadiene rubber (SBR) as reference to separate effects of the binder from the rest of the cathode components.

1.0.3 Functionalizing ketjen black carbon

In earlier work with the PEO:PVP binder, the battery performance was observed to

decrease after a few hundred galvanostatic cycles. The hypothesis is that due to the

different polarites of PEO and PVP, they tend to phase-separate gradually as the bat-

teries are cycled [18]. The possible phase-separation may be decreased by attaching a

pyrrolidone unit with the same functionality as PVP, 1-(2-Hydroxyethyl)-2-pyrrolidone

(HEP) onto the KB carbon while still maintaining the stabilizing effect on the capacities

when using it together with PEO binder. Furthermore, it is interesting to investigate

the electrochemical effects of having PVP on the surface and the PEO binder in the

electrolyte when cycling them in batteries since it may affect the reaction kinetics and

could provide more information on the synergistic effect of PEO:PVP.

2 | Theory

This chapter provides the theoretical background for the present thesis.

2.1 The Lithium-Sulfur Battery

The lithium-sulfur battery operates by a redox reaction which can be expressed by the overall reaction; [7, 8]

S ₈ + 16 Li ⁺ + 16 e ⁻ −−* )−− 8 Li ₂ S (2.1) The mechanism operates via several PS intermediates (Li 2 S _X , X = 4 − 8) which are soluble in organic electrolytes and therefore can diffuse to the anode to be reduced and then may be re-oxidized at the anode, diffusing back to the cathode in a redox shuttle mechanism (see Figure 2.1a) [3, 5–8]. Some of the reduction products at the anode are irreversibly formed and results in capacity loss which may be diminished if the PS diffusion flux to the anode can be decreased [3, 5, 6, 21].

Figure 2.1: a) Schematic drawing of a Lithium-Sulfur battery adapted from reference [3] b) a typical voltage profile of a Lithium-Sulfur battery adapted from reference [6].

As shown in figure 2.1a and 2.1b the reaction scheme is very complex due to all the

different intermediates involved and the dissolution-precipitation mechanism of the sulfur

species;

Theory 5

(Charged state) solid S 8 soluble PS (Li 2 S X , X = 4 − 8) insoluble Li 2 S (Discharged state) (2.2) This reaction pathway involves the reduction of S 8 to form soluble PS intermediates (Li 2 S _X , X = 4 − 8) which are ultimately reduced to the insoluble Li 2 S discharge product.

The reaction kinetics is therefore complex since it involves diffusion of multiple species while simultaneously, the diffusion of the soluble intermediates to the reaction surfaces competes with diffusion out to the anode. From in-situ probe [21], in-operando X-ray absorption spectroscopy (XAS) [22] and rotating disc electrode (RDE) [23] studies a more detailed reaction scheme is suggested for the discharge reaction pathway;

(1) Reduction of solid S 8 at 2.3 V

S ₈ + 2 e ⁻ −−* )−− S ₈ ²⁻ (2.3)

(2) S ²⁻ ₈ is dissolved into the organic electrolyte and disproportionates due to thermody- namic reasons. The formation of S ²⁻ ₆ leads to an increased viscosity which increases the overpotential resulting in the saturation point shown in figure 2.1 b).

S ₈ ²⁻ −−→ S ₆ ²⁻ + 1

4 S ₈ (2.4)

(3) The soluble S ²⁻ ₆ diffuses to the conductive surface and is further reduced at 2.1 V

S ₆ ²⁻ + 2 e ⁻ −−→ S ₄ ²⁻ (2.5)

(4) The soluble S ²⁻ ₄ ultimately disproportionates into the final discharge product Li 2 S.

5 S ₄ ²⁻ + 4 Li ⁺ −−→ 2 Li ₂ S + 3 S ₆ ²⁻ (2.6) The reaction scheme while charging the battery is similar, but the rates of the different steps may vary [21, 22].

It should be noted that the electrochemical reactions, see Equations 2.3 and 2.5, are

well characterized and valid for low donor number solvents but solvents with high donor

numbers may result in radical formation such as S ⁻ ₃ [23]. The chemical disproportionation

reactions, see Equations 2.4 and 2.6, are less well characterized and it is believed that the

Theory 6 intermediates suggested in this scheme are only the most thermodynamically favourable ones. These intermediates are the easiest to observe e.g. with XAS but the true reaction scheme may involve other intermediates that have not yet been successfully detected [21, 22].

Because the insoluble discharge product Li 2 S is insulating, the common approach is to mix in some form of conductive agent into the cathode formulation to decrease the polar- ization, resulting in higher energy efficiency. It can also act as a template to disperse the sulfur into smaller particles which increases the interfacial area between sulfur, lithium ions and electrically conducting carbon resulting in higher sulfur utilization. The higher utilization increases the discharge and charge capacities towards the theoretical maxi- mum of 1672 mAh g ⁻¹ . The conductive agent is usually carbon because carbon is cheap and has good electrical conductivity. However, this inclusion lowers the fraction of sulfur which decreases the capacity normalized against total cathode mass. Furthermore, the cathode structure needs high porosity to allow effective lithium diffusion and to enable the ∼ 80 % volume increase the formation of Li 2 S gives, in relation to solid S 8 , while preventing cathode structure disintegration. Good electrical contact, however, requires intimate contact between the conductive carbon particles/fibres. These tough require- ments for an effective cathode structure are possible reasons for the ongoing surge of research on cathode host materials which, with the advancements in material science, has improved the Li-S battery technology significantly in recent years [3, 5, 22].

Organic electrolytes are typically used since lithium reacts violently with water. High donor number solvents like 1,3-dioxolane (DOL) and 1,2-dimethoxyethane (DME) are popular choices because it has been shown that these stabilize the soluble intermediates increasing the redox reaction conversion [23, 24]. lithium bis(trifluoromethyl)sulfonimide (LiTFSI) or other similar salts are typically used to increase the ionic conductivity.

LiNO ₃ has been introduced as an electrolyte additive which decreases the capacity fading due to irreversible losses at the anode by forming an protective solid electrolyte interphase (SEI) layer between the anode and electrolyte [25].

2.2 Electrochemistry

This section provides concise descriptions of capacity, coulombic efficiency, C-rate and

battery rates, galvanostatic cycling, internal resistance, polarization, overpotential, and

IR drop, which are necessary to understand the electrochemical characterization per-

formed in this thesis.

Theory 7

2.2.1 Internal resistance and iR drop

The internal resistance in a battery consists of contributions from electrical resistance through the bulk materials, ionic resistance in the pores and electrolyte, charge-transfer resistance at interfaces and mass transport (particularly diffusion) [26]. Comparing the time-scales of the different resistances as the battery is operating, mass transport re- sistances appear on a relatively long time-scale and the charge-transfer resistances on a intermediate time-scale whereas the electrical and ionic resistances operate on very short time-scales corresponding to the voltage drops as the current through the battery is turned on or turned off [26]. Internal resistance can be easily estimated by treating the voltage drop, dE [V], following a current interruption as ohmic:

R = dE

dI (2.7)

where dI [A] is the difference in current passing through the battery when turning on/off the current and R [Ω] is the internal resistance of the battery. Since R may not follow Ohm’s law and is measured over an arbitrary time, it is most often not considered the true internal resistance but the "apparent" internal resistance.

2.2.2 Polarization and overpotential

The total voltage deviation from the equilibrium potential, the open-circuit voltage (OCV), is also including activation polarization and concentration polarization which depends on inherent material characteristics and the concentration of ionic species [27, 28], respectively. The material characteristics determine the energy loss that is required to activate electron transfer through the material and this energy loss corresponds to the activation polarization. The concentration of the different species involved in the electrochemical reactions affect the reaction rate, and a concentration polarization arises when the mass-transport of one or more of the involved active species is limited. The total overpotential, η, is defined as the deviation from the OCV and can be written

η = E − E eq (2.8)

η = η _A + η _I + η _C (2.9)

Theory 8 where E [V] is the measured cell potential in the battery during charge or discharge, E _eq [V] is the OCV, η A [V] is the activation overpotential, η I [V] is the overpotential due to the IR drop related to the internal resistance and η C [V] is the concentration overpotential.

2.2.3 Capacity

A reversible redox battery like the lithium-sulfur batteries can be discharged and charged at various rates, see Figure 2.1 and Equations 2.1-2.6. A current is applied during charge or withdrawn during discharge and the magnitude of this current defines the rate of the charge or discharge. The energy input/drawn from the battery is dependent on the reaction kinetics and therefore also the total capacity during charge/discharge.

How much capacity is charged or discharged in the battery is defined for each charge/dis- charge cycle as q charge or q discharge [mAh g ⁻¹ ]. The theoretical capacities, q theoretical , are determined by the number of electrons transferred in the electrochemical reactions and the molecular weight of the active material. They are derived from Faraday’s law; [29]

Q = nzF (2.10)

where Q [C] is the total transferred charge for an electrochemical reaction, n [mole]

is the number of moles of reacting active electrochemical species, z is the number of transferred electrons and F [F = 96485 C mole ⁻¹ ] is Faraday’s constant. Converting equation 2.10 into mAh g ⁻¹ is done by introducing the molar mass and result in the following expression

q theoretical = nzF

3600M w (2.11)

where M w [g mole ⁻¹ ] is the molar mass. Sulfur has the extraordinarily high theoretical capacity of 1672 mAh g ⁻¹ due to the high number of electrons reacting per sulfur atom, z = 2 , while being a light element, M w = 32.065 [mole g ⁻¹ ]. The practical capacities a Li-S battery can deliver are naturally lower than the theoretical value due to incomplete sulfur utilization and the overpotentials required during the different reaction steps.

These losses of theoretical capacity are dependent on factors such as sulfur loading,

surface area, porosity, solvent polarity and concentration, interaction between sulfur

species and functional groups, irreversible side reactions and cathode thickness. They

affect the final time, t, it takes for a discharge/charge cycle to complete with a constant

Theory 9 charge/discharge current, j [A], and varying voltage V (t) [V]. The practical capacities are then defined as

q =

t

Z

0

j(t)E(t)dt = j Z t

0

E(t)dt (2.12)

The capacity is often normalized against sulfur mass to compare how effectively different Li-S batteries are utilizing the sulfur. It should not, however, be forgotten that the sulfur fraction of the total cathode mass may be significantly lower than one and that the capacity normalized against total cathode mass therefore is significantly lower.

2.2.4 Coulombic efficiency

The coulombic efficiency, C E % , is a measure of how large the discharge capacity is in relation to the charge capacity. It can be defined as

C _E = Q _discharge

Q _charge (2.13)

It is a measure of side-reactions or other parasitic processes in the discharge process relative to the charge process. Coulombic efficiencies close to one signifies that the loss of active material or inactivation of current is similar upon charging and discharging the battery. Coulombic efficiencies > 1 can occur if the discharge products formed are reacting electrochemically and irreversibly to side-products, impurities in the cathode structure are irreversibly reduced in the discharge process, or irreversible side-reactions are more severe upon charging. Coulombic efficiencies < 1 signifies that loss of active material or inactivation of current through side-reactions/parasitic processes are larger upon discharge than the charge. This is typically the case for Li-S batteries because of the diffusion of soluble intermediates to the anode during discharge which are reduced to polysulfides and inactivated for the rest of the discharge, see section 2.1. The charging capacity is then larger because most of the reduced polysulfides at the anode is re-oxidized and can be utilized at the cathode along with the discharge products.

2.2.5 Energy efficiency

Energy efficiency, EE %, is a measure of how much energy is needed for the discharge

process relative to the charging process and can be calculated from the following expres-

sion

Theory 10

EE =

q discharge

R

0

j _discharge (q)V _discharge (q)dq

q _charge

R

0

j _charge (q)V _charge (q)dq

(2.14)

where V discharge (q) [V] is the voltage profile during discharge as a function of the capacity and V charge (q) [V] is the voltage profile during charge as a function of the capacity. Ideally the coulombic efficiency is close to one, signifying similar overpotentials for the charge and discharge reaction pathways which means the redox reaction is energy-efficient. For the Li-S system, EE < 1 because V charge (q) is more shifted from the OCV relative to V _discharge (q) because of the greater overpotential needed during the charge to overcome the insulating nature of the discharge products.

2.2.6 Galvanostatic cycling

Galvanostatic cycling is popularly used in the battery-research field to evaluate battery performance. The current drawn/applied is kept constant while the voltage is varying and measured over time as the battery is discharged/charged.

2.2.7 C-rate and battery rates

The C-rate is defined as the current required to fully charge or discharge the battery to its theoretical capacity in 1 hour. Battery rates can be defined from the theoretical capacity and the number of hours, h [hours], per charge/discharge cycle: [28]

Battery − rate = q theoretical

h (2.15)

It is common practice within the battery research field to express these battery rates as fractions of the C-rate. For the lithium-sulfur interconversion redox reaction with the theoretical capacity of 1672 mAh g ⁻¹ , some typically used battery-rates while galvanos- tatically cycling are

1C = ¹⁶⁷² ₁ = 1672 mA g ⁻¹

C/5 = 334.4 mA g ⁻¹

C/10 = 167.2 mA g ⁻¹

Theory 11

2.3 Scanning electron microscopy and Energy dispersive X- ray mapping

Scanning electron microscopy (SEM) is a technique which can be used to study a sample’s surface topography which gives information on the morphology of the sample material [30]. It produces surface topography images of samples by scanning the samples with a focused electron beam which interacts with the sample material and detecting the different types of electrons leaving the sample as a result of the interactions [30]. In secondary electron SEM, electrons that leave the sample as a result from the ionization of atoms hit by the incoming electron beam are detected [30]. The amount of detected secondary electrons at different spatial positions of the sample is proportional to the density of the sample material which makes more dense areas brighter and areas with low density of material darker [30]. This enables imaging of surfaces in the sample which provides information on the morphology of the material in the sample e.g. if it consists of small particles or larger clusters and what the sizes and shapes of those are.

Energy dispersive X-ray (EDX) mapping utilizes the unique transition energies between quantised energy levels in elements. The same focused electron beam that is used in SEM is inelastically scattered when hitting a sample and core-level holes are generated from released electrons. These holes are filled with electrons from higher energy levels and these energy transitions will result in X-ray radiation which can be recorded with a suitable detector. An elemental map on a sample is generated by varying the focused electron beam and varying the detecting angle for each beam position to probe a 2D area. The concentration of detected elements is proportional to the intensity of detected X-rays.

2.4 X-ray photoelectron spectroscopy

X-ray photoelectron spectroscopy (XPS) is a surface-sensitive technique used to detect elemental compositions at surfaces of solid materials which can be used to study phys- ical and chemical processes occurring at surfaces. It measures the binding energies of the different elements present at a surface by utilizing the photoelectrical effect when irradiating the sample with X-rays [31, 32];

E _b = hν − E _k − φ (2.16)

Theory 12 where E b is the binding energy of electrons from atoms present at the surface hit by the incident X-ray, E k is the kinetic energy of the released electrons, hν is the energy of the incident photons and φ is the instrument’s work-function, unique for each instrumental XPS setup. The incident photons are releasing electrons from the core levels of the surface located elements and some of those electrons hit the detector which measures their kinetic energy, see Figure 2.2 [31, 32]. Since the binding energies for the core levels are unique for each element, it is possible to identify the elemental composition at the surface. It is also possible to identify chemical bonds by measuring the shifts in binding energies induced by the chemical environment.

Figure 2.2: Schematic drawing of XPS principles adapted from reference [32]. a) The photoelectrical effect, b) an example of a XPS spectrum and c) a simplified scheme of

the experimental setup.

2.5 Thermogravimetric analysis

In thermogravimetric analysis (TGA), the weight loss of a sample is measured by a sen- sitive calibrated balance while heating the sample in an oxygen or nitrogen atmosphere.

The software calculates the weight loss by sampling the weight at small time-steps and

Theory 13 couples this to a temperature detector, yielding weight loss versus temperature data.

This data is useful to analyze chemical synthesis and chemical compositions since the observed weight-losses can be correlated to the difference in thermal decomposition tem- peratures or oxidation temperatures between the different components in the sample.

2.6 Nuclear magnetic resonance spectroscopy

In nuclear magnetic resonance (NMR) spectroscopy, detection of chemical compounds can be made by exposing the sample to a magnetic field making the present nuclei absorb and re-emit electromagnetic radiation with certain resonance frequencies [33].

It can be used on isotopes with an odd number of protons and/or neutrons such as

1 H which has an intrinsic magnetic moment and angular momentum corresponding to

a non-zero spin [33]. Such isotopes has unique nuclear magnetic resonant frequencies

in different chemical compounds due to the different chemical environmental shifts of

the resonance frequencies resulting from the electronic interaction affecting the magnetic

moment [33]. A key feature of NMR spectroscopy is that the measured intensity of the

emitted electromagnetic radiation at a given resonant frequency is proportional to the

concentration of the isotope in the particular chemical environment that gives rise to the

this signal [33]. The combination of the resonant frequency and the intensity can then

be used to assign chemical structures of the compounds present in the sample.

3 | Experimental information

This chapter contains all relevant experimental information required to reproduce the experiments performed within this project.

3.1 Materials

In this section, all materials used for the different experiments are presented.

3.1.1 Materials used for cathode and electrolyte preparation

Ketjen Black carbon (KB, BET surface area ∼ 1300 m ² g ⁻¹ , pore volume ∼ 3.0 cm ³ g ⁻¹ , HPL Ekzon) and SBR binder (Targray PSBR-100) were used as received. LiNO 3

(Aldrich) and lithium bis(trifluoromethyl)sulfonimide (LiTFSI, Novolyte) were dried overnight at 120 ^◦ C before use. Carboxymethylcellulose (CMC) binder (Leclanché), 1,2- dimethoxyethane (DME, Novolyte), 1,3-dioxolane (DOL, anhydrous, Sigma-Aldrich), poly(ethyleneoxide) (PEO, M w = 4000000 Aldrich), poly(2-ethyl-2-oxazoline) (PEOx, M w =500000, Aldrich), poly(vinyl pyrrolidone) (PVP, M w =360000, Aldrich), Pluronic F-127 (Surfactant, BASF), Carbon nanofibers (CNF, iron-free, D x L 100 nm x 20-200 µ m, Aldrich) and C-Energy super C65 (BET surface area 62 m ² g ⁻¹ , nonporous, TIM- CAL) were used as received.

3.1.2 Materials used for esterification of ketjen black carbon

1-(2-Hydroxyethyl)-2-pyrrolidone (Link agent, Sigma-Aldrich), N,N’-Diisopropylcarbodiimide

(DIC, Acros organics), 4-(Dimethylamino)pyridine (DMAP, Sigma-Aldrich) and N,N-

dimethylformamide (DMF, anhydrous, Sigma-Aldrich) were all used as received.

Experimental information 15

3.2 Functionalization of ketjen black carbon

The functionalization of KB was done by first oxidizing the KB with HNO 3 and then attaching the link agent 1-(2-Hydroxyethyl)-2-pyrrolidone by an Steglich esterification reaction [34].

3.2.1 Oxidation of ketjen black carbon

3.0 g of KB was mixed with 50 ml concentrated HNO 3 in a round flask connected to a reflux condensation tube. The suspension was heated to 110 ^◦ C and vigorously stirred with reflux condensation for 30 minutes. Brown gas was observed continuously during this reaction time. The reaction was then quenched by adding 50 ml of distilled water followed by vacuum-filtrating the suspension and washing the filtration product several times with water. The black filtration product was then re-dispersed in water and sonicated for 10 minutes. The suspension was then vacuum-filtrated again and washed several times with water. This work-up procedure was repeated 3 times. The resulting filtration product was dried at 80 ^◦ C in vacuum in a Buchi-furnace connected to a schlenk-line for 18 hours.

3.2.2 Steglich esterification of oxidized ketjen black carbon

1.0 g of oxidized KB and 0.207 g (1.694 mM) DMAP was dispersed in 50 ml DMF in a round flask with vigorous stirring. The suspension was cooled to 0 ^◦ C before 7.7 ml (68 mM) HEP and 2.9 ml (19 mM) DIC were added. The reaction was then allowed to occur at 0 ^◦ C for 5 minutes before the ice-bath was removed and the suspension was left in room-temperature for 3 hours. The resulting black suspension was then vacuum-filtrated and washed three times with water and three times with ethanol. The filtration product was redispersed in acetonitrile and sonicated for 15 minutes before being vacuum-filtrated again with the same washing procedure. This work-up procedure was repeated three times. The final filtration product was then dried overnight at 70 ^◦ C in a oven.

3.3 Preparation of Batteries

This section includes all the necessary information of how the batteries used in this work

were prepared.

Experimental information 16

3.3.1 Cathode preparation

Cathode slurries with the compositions presented in Table 3.1 were prepared by a two- step procedure for all cathodes except the ones based on functionalized KB which were prepared with a one-step procedure. These procedures are presented in the following sections.

Table 3.1: Compositions and sulfur loadings of the different cathodes prepared includ- ing the cathodes optimized with KB (Z-1, Z-2, Z-3 and S-7-1), cathodes with varying S:KB ratio (V-0, V-1, V-2 and V-3) and cathodes utilizing functionalized KB with the Steglich esterification approach (S-7-C and S-7-A). The largely varying sulfur load- ings were the result of using doctor-blading with varying slit-thickness when coating the cathode films, the slit-thicknesses presented in the column indexed "slit-thickness".

The average cathode thicknesses were measured using an ABSOLUTE micrometer (Mi- tutoyo, Mexico). Some of the cathodes were calendered and the average thicknesses of

these are also included and indexed "Calendered".

Cathodes Compositions (weight-%)

Optimized with KB S KB CNF C65 Binder Binder S loading (mg S/cm

) Cathode thickness (µm) Calendered (µm) slit-thickness (µm)

Z-0 65 21 3.5 3.5 7 2:3 CMC:SBR ∼ 1.2-3.5 ∼ 30-120 45-50

Z-1 65 21 3.5 3.5 7 4:1 PEO:PVP ∼ 1.4-3.1 ∼ 35-90 50-70

Z-1-C 65 21 3.5 3.5 7 4:1 PEO:PVP ∼ 1.1-5.5 ∼ 30-140 50-90

Z-1-cal 65 21 3.5 3.5 7 4:1 PEO:PVP ∼ 1.4-2.7 ∼ 22-45 50

Z-2 65 21 3.5 3.5 7 4:1 Pluronic:PVP ∼ 1.2-2.5 ∼ 35-110 50-65

Z-2-cal 65 21 3.5 3.5 7 4:1 Pluronic:PVP ∼ 1.5-2.1 ∼ 30-60 ∼ 21-33 50-65

Z-PEOx 65 21 3.5 3.5 7 4:1 PEOx:PVP ∼ 1.3-2.5 ∼ 30-62 50

S-7-0 65 21 3.5 3.5 7 PEO ∼ 1.0-1.3 ∼ 25-35 50

Varying S:KB ratio S KB CNF C65 Binder Binder S loading (mg S/cm

) Cathode thickness (µm) Calendered (µm) slit-thickness (µm)

V-0 60 30 3 0 7 4:1 PEO:PVP ∼ 0.8-1.5 ∼ 28-56 22-30 50

V-1 67.5 21 3 0 7 4:1 PEO:PVP ∼ 1.0-1.5 ∼ 29-47 18-33 50

V-2 72 18 3 0 7 4:1 PEO:PVP ∼ 0.7-1.0 ∼ 27-50 17-27 50

V-3 75 15 3 0 7 4:1 PEO:PVP ∼ 1.0-2.5 ∼ 30-67 18-38 50

Functionalized KB S KB CNF C65 Binder Binder S loading (mg S/cm

) Cathode thickness (µm) Calendered (µm) slit-thickness (µm)

S-7-A 65 21 3.5 3.5 7 PEO ∼ 1.0-1.4 ∼ 30-40 50

3.3.1.1 Cathode preparation based on Ketjen Black carbon

KB and sulfur powders were manually grounded together and then heated at 155 ^◦ C

for 20 minutes in a Buchi-furnace. Then CNF, super C65 and the binder were added

to the mixture in the appropriate solvent and mixed by planetary ball milling for 2

hours. Water-ethanol (10.5 ml 20:1 for Z-1, Z-2 and Z-PEOx. 6 ml 5:1 for V-0, V-

1, V-2 and V-3) as solvent was used for 4:1 PEO:PVP and only 7.5 ml water when

using 2:3 CMC:SBR. All slurries were coated by doctor-blading with an adjustable slit-

thickness, onto aluminium foil and dried at ambient conditions. Some cathodes of the

Z-1 compositions were coated onto carbon-coated aluminium foil with ∼ 0.075 mg cm ⁻²

carbon loading to achieve higher sulfur loadings while maintaining a good film quality,

the cathodes are indexed Z-1-C. The Z-1-cal, Z-2-cal, V-0, V-1, V-2 and V-3 cathode

coatings were all calendered three times with a 4 µm setting with a Roll Press HT-200

(HC China Limited) resulting in averaged compressions of 26.2 (%), 16.7 (%), 38.1 (%),

32.9 (%), 42.9 (%) and 42.3 (%), respectively (based on the cathode thicknesses presented

in Table 3.1). Cathodes were cut into 13 mm diameter discs with the resulting sulfur

Experimental information 17 loadings presented in Table 3.1. The cathodes were then transferred to an argon-filled glove box and dried under vacuum at 55 ^◦ C overnight.

3.3.1.2 Cathode preparation based on functionalized Ketjen Black carbon The Steglich-esterificated carbon cathode slurries and references (S-7-0 and S-7-A, see Table 3.1) were prepared by mixing sulfur, KB, CNF, C65 and the binder directly in water-ethanol (20:1) as solvent followed by planetary ball-milling for 2 hours. The re- sulting slurries were then processed to cathodes in the same way as the KB based carbon cathodes.

3.3.2 Battery assembly

The dry cathodes were assembled into CR2025 (20 mm, Toshiba) coin cells, with the procedure presented in reference [35], inside an argon-filled glove box with a 16 mm diameter circular piece of lithium foil (125 µm thick, Cyprus Foote Mineral) as the anode and a polyethylene separator (SOLUPOR, Lydall Performance Materials). The electrolyte used was 1 M LiTFSI, 0.25 M LiNO3 in 1:1 DME:DOL, and the electrolyte amount was fixed at 6 µL per mg of sulfur in the cathode.

3.4 Experimental characterization

The batteries prepared in this work were electrochemically analyzed using galvanostatic cycling, rate-capability tests, cycle-wait test and internal resistance measurements. The morphology and structure of the prepared cathodes were investigated with SEM and the esterification of KB with the link agent was characterized by TGA, XPS, NMR and FTIR.

3.4.1 Galvanostatic cycling

Batteries with all the different cathodes were cycled galvanostatically between 2.6 V and

1.8 V versus Li/Li ⁺ at a rate of C/10 (167.2 mA g ⁻¹ of sulfur) using an Arbin BT-2043

battery tester. All Electrochemical measurements were started within 30 minutes of

battery assembly.

Experimental information 18

3.4.2 Cycle-wait test

The cycle-wait test utilizes galvanostatic cycling with resting periods. During the gal- vanostatic cycling, after every three discharge-charge cycles, the applied current is shut- off and the battery was allowed to rest for a specific time while the open-circuit voltage is still measured. After the specified resting time, the battery is galvanostatically cycled for three more cycles before resting again and this procedure is repeated for an increas- ing resting time for each new loop. The qualitative amount of lost discharge capacity is then calculated from subtracting the extrapolated difference of the last and second discharge capacity for each resting period (the discharge capacity before resting and the second discharge capacity after resting) with the first discharge capacity after rest. The resulting output data is a graph of lost discharge capacity [mAh g ⁻¹ ] versus resting time [days].

The cycle-wait test was conducted on batteries with Z-0, Z-1 and Z-2 cathodes for com- parison. The test was performed with an MPG-2 (Biologic Science Instruments) by galvanostatic cycling between 2.6 V and 1.8 V vs Li/Li ⁺ at a rate of C/10 (167.2 mA g ⁻¹ of sulfur) with increasing resting periods between each three cycles. The resting periods used were 0.5, 1, 3, 7 and 14 days, respectively, for one loop and two such loops were measured.

3.4.3 Scanning electron microscopy and Energy dispersive x-ray map- ping

Scanning electron microscopy figures on cathodes with the Z-1 composition and 58:28:7:7 of S:KB:C65:PEO-PVP cathodes were kindly provided by Carl Tengstedt, Scania, Södertälje, Sweden. The latter composition was from an early stage in the experimental optimiza- tion of the cathode composition before CNF was tested as an additive and were prepared with the same procedure as described in section 3.3.1.1.

Scanning electron microscopy figues with coupled sulfur EDX mapping was performed on Z-0, Z-1, Z-2, and Z-PEOx cathodes using a Phenom ProX desktop scanning electron microscope (Phenom-World). The measurements were all done with 10 kV acceleration voltage.

3.4.4 X-ray photoelectron spectroscopy

XPS was performed on KB powder, oxidized KB powder, functionalized KB powder,

a casted functionalized KB electrode and a HEP film casted on aluminium. The KB

Experimental information 19 powder, oxidized KB powder, functionalized KB powder and casted functionalized KB electrode were all measured without any pre-treatment and charge neutralization. HEP was measured as a thin film on an aluminium foil and charge neutralization was required.

All measurements were done in cooperation with Julia Maibach, department of Structural Chemistry, Uppsala University, Sweden. The graphs and energy calibration were done by Julia Maibach. The instrument used was a PHI 5500 ESCA system (Perkin-Elmer) with an Al anode K α X-ray source (1486.6 eV). The step-time used varied between 50 and 100 s and the detecting angle was 45 ^◦ . To calibrate the binding energy scale for all the samples, the PEO C1s signal (286.4 eV) from the casted electrode was used for calibration and all other KB samples aligned according to the resulting KB peak position. HEP was calibrated towards the hydrocarbon peak at 285 eV. All intensities are presented as measured.

3.4.5 Thermogravimetric analysis

TGA was performed on KB, oxidized KB and functionalized KB using a Q Series TGAQ500 (TA Instruments) instrument. The measurements were done with a ramp heating procedure of 5 ^◦ C/minute between room temperature and 500 ^◦ C in 60 ml/min air-flow. The sample-holder used was an aluminium pan and all samples were in powder- form following the procedures presented in sections 3.2.1 and 3.2.2.

3.4.6 Nuclear magnetic resonance

NMR was performed on functionalized KB in cooperation with Fabian Jeschull, depart-

ment of Structural Chemistry, Uppsala University, Sweden. The instrument used was a

JNM-ECP400 (JEOL) with a TH5 probe. The experiment was done with single pulse

measuring ¹ H of the sample with 400 MHz resonance frequency, 6 kHz sweep and 45 ^◦

angle. The sample was dissolved in DCCl 3 for 3 days and sonicated 30 minutes prior to

the measurement.

4 | Results and Discussion

In this chapter results on optimizing the cathode formulation and preparation, com- parison of different binders and functionalization of KB are presented together with a discussion focused on developing high-energy density cathodes. The subsections included are morphology studies, effects of sulfur loading, varying the sulfur:ketjen black carbon ratio, rate capability test and cycle-wait test. The results on the functionalization of KB involves NMR, TGA, XPS and galvanostatic cycling data.

4.1 Optimizing the cathode formulation and preparation to increase the energy density

The various compositions of the cathodes and water-ethanol solvents, presented in Ta- ble 3.1, were derived from an optimization process balancing high sulfur loading versus mechanical cathode stability. Increasing the cathode film thickness to increase the sul- fur loading is limited by aggregation and detachment of cathode particles, since the in- tegrity of the film is kept by adhesion between the particles and the substrate (aluminum foil/carbon-coated aluminium) surface. A higher surface area of the carbon results in enhanced aggregation of the carbon particles to decrease the surface free energy, result- ing in increased cracking of the cathode film and detachment of carbon, coupled to the decreased contact of the carbon particles and the substrate. The increased aggregation of carbon particles may also result in less physical confinement of soluble intermediates, related to a more open structure as argued extensively in the Li-S literature [36–39], and a possible blocking of channels within these agglomerates, preventing successful diffusion of intermediates in the different reaction steps (see Equations 2.3-2.6 in section 2.1).

Furthermore, the different polarities of PEO, PVP, carbon, and sulfur cause an inho- mogeneous distribution of the components resulting in worsening cycling performance due to reduced interfacial contact between carbon and sulfur. The solutions to these problems employed within this work were to

• introduce CNF as additive which distributes the different components finely and pre-

vents the aggregation of the high surface area KB (see Figure 4.3) leading to an increasing

Results and Discussion 21 adhesion which enables thicker cathode films (see Figure 4.1). Furthermore, the preven- tion of large agglomerates increases the porosity which facilitates deposition of discharge products and possibly the diffusion of intermediates enhancing the reaction kinetics

• introduce the non-porous super C65 carbon black as additive to reduce the capacity fading (see Figure 4.5 and 4.6 in section 4.1.2)

• add ethanol to the water solvent during the cathode preparation to tune the polarity towards PEO, increasing the homogeneity and contact between the different components (see supporting Figure A.1 in the appendix)

• increase the S:KB ratio. A lower KB ratio decreases the aggregation problems while the high surface area and porosity enables a lower ratio to be used while maintaining good battery performance.

• use doctor-blade coating with adjustable slit-thickness to control the thickness of the cathodes which is useful to probe the upper limits before cracks appear for the different cathode compositions, enabling maximized cathode thicknesses while maintaining good film quality. This coating technique also enables good control over the resulting sulfur- loadings which was used to produce cathodes of similar loadings for comparisons of different binders and additives.

• use carbon-coated aluminium as cathode substrate which enhances the adhesion and enables much thicker films without introducing cracks. Much higher sulfur loadings can then be used while maintaining good battery performance (see Table 3.1).

Introducing these components and concepts into the optimization process resulted in well controlled, reproducible, homogeneously dispersed cathode films with less cracks (see Figure 4.1) and allowed sulfur loadings as high as ∼ 5.5 mg S g ⁻¹ for the optimized Z-1-C cathodes (see Table 3.1) which is a major improvement compared with previous work (∼ <1.0 mg S g ⁻¹ ) [18].

Figure 4.1 shows that CNF improves the film quality by preventing cracks to appear which, as earlier argued, may be related to prevention of carbon-sulfur particle aggrega- tion, and increased mechanical integrity resisting the strain by volume shrinkage as the cathode slurry is drying. Furthermore, it shows that the optimized cathodes with the Z-1 composition are homogeneous and reproducible, the deviations in sulfur loading being

<10 %, which is an important attribute for commercializing aspects. Notable is the fact

that the deviation in loading is dependent on the size of the coater. It is positive that

the deviations observed in this work are so small, considering that the coater is less than

50 cm long and that typically coaters of several meters in length are needed to achieve

sufficiently small deviations to reach commercial reproducibility. As Zheng et al. [19]