lithium metal microreference electrodes and their ... · 6.5.3 relationship between + and 94 d li...

Lithium metal microreference electrodes and their applicationsto Li-ion batteriesCitation for published version (APA):Zhou, J. (2007). Lithium metal microreference electrodes and their applications to Li-ion batteries. TechnischeUniversiteit Eindhoven. https://doi.org/10.6100/IR624713

DOI:10.6100/IR624713

Document status and date:Published: 01/01/2007

Document Version:Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can beimportant differences between the submitted version and the official published version of record. Peopleinterested in the research are advised to contact the author for the final version of the publication, or visit theDOI to the publisher's website.• The final author version and the galley proof are versions of the publication after peer review.• The final published version features the final layout of the paper including the volume, issue and pagenumbers.Link to publication

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, pleasefollow below link for the End User Agreement:www.tue.nl/taverne

Take down policyIf you believe that this document breaches copyright please contact us at:[email protected] details and we will investigate your claim.

Download date: 27. Jul. 2020

https://doi.org/10.6100/IR624713

https://doi.org/10.6100/IR624713

https://research.tue.nl/en/publications/lithium-metal-microreference-electrodes-and-their-applications-to-liion-batteries(9fd89d19-6d5f-463a-bd56-17d78d295d22).html

Lithium Metal Microreference Electrodes and their Applications to Li-ion Batteries

Jiang Zhou

Samenstelling van de promotiecommissie: prof.dr. J.W. Niemantsverdriet Technische Universiteit Eindhoven, voorzitter

prof.dr. P.H.L. Notten Philips Research Laboratories Eindhoven Technische Universiteit Eindhvoen, eerste promotor

prof.dr. R.A. van Santen Technische Universiteit Eindhvoen, tweede promotor

prof.dr. J.J. Kelly Universiteit Utrecht

prof.dr. J. Schoonman Technische Universiteit Delft

prof.dr. G. de With Technische Universiteit Eindhoven

prof.dr. P.P.J. van den Bosch Technische Universiteit Eindhoven

dr. F.A. de Bruijn Technische Universiteit Eindhoven

A catalogue record is available from the Library Eindhoven University of Technology Zhou, Jiang Lithium metal microreference electrodes and their applications to Li-ion batteries/by Jiang Zhou. - Eindhoven: Technische Universiteit Eindhoven, 2007. Proefschrift. - ISBN 978-90-386-0919-5 Subject headings: electrochemistry; batteries / Li-ion / reference electrodes / concentration gradient / capacity fade / electrochemical impedance ; EIS Trefwoorden: elektrochemie; batterijen / Li-ion / referentie elektroden / concentratie-gradiënt / capaciteit verloren / elektrochemisch impedantie ; EIS Cover design: Jiang Zhou Printed by Eindhoven University Press Copyright © 2007 by Jiang Zhou This research has been financially supported by the Dutch Government within the framework of the Economy Ecology and Technology (EET) program.

Lithium Metal Microreference Electrodes and their Applications to Li-ion Batteries

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Technische Universiteit Eindhoven, op gezag van de

Rector Magnificus, prof.dr.ir. C.J. van Duijn, voor een commissie aangewezen door het College voor

Promoties in het openbaar te verdedigen op woensdag 11 april 2007 om 16.00 uur

door

Jiang Zhou

geboren te JiangXi, China

Dit proefschrift is goedgekeurd door de promotoren: prof.dr. P.H.L. Notten en prof.dr. R.A. van Santen

In honor of my father

“Real knowledge is to know the extent of one's ignorance.”

(Confucius)

Acknowledgements i

Acknowledgements This thesis is the end of my four-year journey in my Ph.D. work whereby I have been accompanied and supported by many people. It is my great pleasure that I have now the opportunity to express my gratitude for them.

First and foremost, I wish to express my great gratitude to my promoter, prof.dr. Peter H.L. Notten, for his invaluable guidance, his understanding and encouragement throughout the course of my research. This thesis could have never been completed without his immense help in all aspects. His depth and breadth of knowledge and innovative ideas, as well as his attitude towards science, have made a deep impression on me. I am really glad that I have come to get know Peter Notten in my life.

Particular thanks are due prof.dr. J. W. Niemantsverdriet, for interviewing me in Singapore and for being the chairman of my promotion committee. I would also like to thank the other members of my promotion committee who made the effort to read this thesis and provide me with valuable comments: prof.dr. J.J. Kelly, prof.dr. J. Schoonman, prof.dr. R.A.van Santen, prof.dr. G. de With, prof.dr. P.P.J. van den Bosch and dr. F.A. de Bruijn.

My sincere thanks go to my colleague, dr. Rogier Niessen, for his timely help and stimulating discussion. Particularly, I will never forget the company I had from him on the first day I was in Holland. Thank you, Rogier.

My officemates Alfred Dortmans and Peter Kalisvaart deserve warm thanks for their help in both work and life. They taught me a lot about the life in Holland and encourage me in Dutch learning. By this chance, I would like to say “Heel Hartelijk Bedankt!”

The episode of acknowledgement would not be complete without the mention of my colleagues, Afifa Belfadhel-Ayeb, René van Beek, dr. Dmitry Danilov, E. Florentin, Kiem Thanh Le, dr. Alexander Ledovskikh, dr. John Mills, Martin Otten, Bert Op het Veld, dr. Evgeny Verbitskiy, Paul Vermeulen, Jan van den Meerakker and dr. Martine Wehrens-Dijksma. In addition, I am also grateful to all the people from the analysis department. Thank you all.

Special thanks go to special friends who are from the so-called 405 big yard. We grew up together like brothers. Although we are now spreading around the world, I still receive your encouragement and inspiration, not only for my research work, but also for my life.

I believe I am indebt to my sisters, Zhiqin and Zhimin, who have cared and encouraged me since I was born. Thanks also go to my brother-in-law, Xin. They rendered me enormous happiness.

Last, but definitely not least, I want to show my deepest gratitude to my parents, Qigang Zhou and Guanghui Shen, for their endless love and constant care. They formed part of my vision and taught me good things that really matter in life. Regrettably, my father passed away in 2005 and would thus never see the final result. I dedicate this work to him, knowing he would be most proud.

Jiang Zhou Eindhoven, February, 2007

Contents iii

Contents

List of Symbols vii

1 Introduction 1 1.1 General Introduction 2 1.2 Li-ion batteries 3

1.2.1 Chemistry of Li-ion batteries 3 1.3 Scope of this thesis 5

2 Materials and Technology 7 2.1 Introduction 8 2.2 Materials 8

2.2.1 LiCoO2 cathode 8 2.2.2 Graphite anode 10 2.2.3 Electrolyte 14

2.3 Solid Electrolyte Interface 17 2.4 Lithylene technology 19

2.4.1 Principle of the Lithylene technology 19 2.4.2 Polymer rivets 20 2.4.3 Preshaped batteries 21

3 Experimental 27 3.1 Introduction 28 3.2 Definitions 28 3.3 Battery materials 29 3.4 Battery formation regime 30 3.5 Electrochemical impedance spectroscopy (EIS) 31

3.5.1 Fundamentals of EIS 31 3.5.2 Equivalent circuit of an electrochemical process 32 3.5.3 Impedance response of an electrochemical process 33

4 Lithium metal microreference electrodes 37 4.1 Introduction 38 4.2 Substrate material for the microreference electrodes 39 4.3 Experimental 40 4.4 Results and discussion 41

4.4.1 Lithium deposition 41

4.4.2 Stability of lithium metal microreference electrodes 43 4.4.3 Influence of the lithium layer thickness on the stability of reference

electrodes 44 4.4.4 Influence of deposition current density on the stability of reference

electrodes 46 4.4.5 Stripping efficiency of lithium deposits 48 4.4.6 Microreference electrodes in impedance measurements 50 4.4.7 Microreference electrodes in potential measurements 51 4.4.8 Stability of the re-deposited microreference electrodes 53

4.5 Conclusions 54

5 On reaction resistances 57 5.1 Introduction 58 5.2 Experimental 58 5.3 EIS measurements 59

5.3.1 Impedance response of Li-ion batteries 59 5.3.2 EIS vs. electrode potential 60 5.3.3 Values of Rs, RSEI and Rct 62

5.4 Rct 64 5.4.1 Determination of Io from I-η plots 65 5.4.2 Further discussion about Io and Rct 68

5.5 Summary 70

6 Mass transport in the electrolyte of Li-ion batteries 73 6.1 Introduction 74 6.2 Theory 75

6.2.1 General 75 6.2.2 Relationship between concentration gradient and electric potential 77

6.3 Experimental 78 6.4 Results 79

6.4.1 Steady state 79 6.4.2 The linearity of concentration gradient 81

6.4.2.1 3-electrode measurements 81 6.4.2.2 4-electrode measurements 83

6.4.3 Concentration gradient vs. applied current 85 6.4.4 Diffusion coefficient and diffusion overpotential 87 6.4.5 Limiting current 90

6.5 Discussion 91 6.5.1 3-electrode measurements 91 6.5.2 Activity coefficient 92

Contents v

6.5.3 Relationship between and 94 +LidI , +LimI ,

6.6 Conclusions 95

7 Capacity fade of Li-ion batteries 97 7.1 Introduction 98 7.2 Experimental 98 7.3 Battery capacity fade 99 7.4 Loss of active materials 101

7.4.1 Loss of the active material in the positive and negative electrodes 101 7.4.2 Possible causes for the loss of the positive electrode material 102

7.5 Increase in battery impedance 108 7.5.1 Impedance of the positive and negative electrodes 108 7.5.2 Increase of the positive electrode impedance 110

7.6 Battery capacity fade, a complicated process 114 7.7 Concluding remarks 116

Summary 119

Samenvatting 123

摘要 127

List of publications 131

Curriculum Vitae 132

List of symbols vii

List of Symbols Symbol Meaning Value Unit

+Lia activity of Li+ ions mol/m3

refLi

a + activity of Li+ ions in the reference state mol/m3

A geometric surface area of electrode m2

cj concentration of species j, the superscript mol/m3

of numbering refers to different locations in electrolyte

0Li

c + electrolyte concentration 1000 mol/m3

posLi

c + concentration of Li+ ions at the positive mol/m3

electrode/electrolyte interface negLi

c + concentration of Li+ ions at the negative mol/m3

electrode/electrolyte interface co the concentration of Li+ ions in the electrolyte mol/m3

cr the concentration of Li+ ions in the electrode mol/m3

coxbulk the bulk concentration of the oxidized species mol/m3

credbulk the bulk concentration of the reduced species mol/m3

+∆Li

c concentration difference of Li+ ions, superscripts mol/m3

refer to different locations in the electrolyte between which the concentration difference is discussed Cdl double layer capacitance F Cint differential intercalation capacitance F CSEI capacitance of the SEI layer F d thickness of lithium metal deposit m Dj diffusion coefficient of species j m2/s Dox diffusion coefficient of the oxidized species m2/s Dred diffusion coefficient of the reduced species m2/s E electrical potential V Eac a sinusoidal voltage signal V Edc a direct voltage signal V Em maximum amplitude of a sinusoidal voltage signal V Eeq equilibrium potential in the absence of current flow V

List of symbols viii

Symbol Meaning Value Unit Eeq’ quasi equilibrium potential which forms during V current flow Eneg potential of the negative electrode V Epos potential of the positive electrode V Eref potential of microreference electrode, superscript V indicates the numbering of the microreference electrode F Faraday constant 96485 C/molij current density carried by charged ionic species j A/m2

Io exchange current, superscript pos indicates the A positive electrode Ia anodic current A Ic cathodic current A Iac an a.c. current A Idc a d.c. current A Im amplitude of an a.c. current A Idl double layer current A If faradaic current A Ij current carried by charged ionic species j A Id, j diffusion-related current of charged ionic species j A Im,,j migration-related current of charged ionic species j A Jj flux of species j mol/m2.s k rate constant for a charge transfer reaction unit depends on parameter ka rate constant for oxidation (anodic) reaction unit depends on parameter kc rate constant for reduction (cathodic) reaction unit depends on parameter l length of Cu wire mm MLi molecular weight of Lithium 6.94 g/mol n number of electrons involved in reaction - Qdep amount of charge in deposition C Qstr amount of charge in stripping C r radius of Cu wire m

List of symbols ix

Symbol Meaning Value Unit R universal gas constant 8.314 J/mol⋅K R resistance, superscripts pos and neg indicate Ω positive and negative electrodes, respectively Rct charge transfer resistance, superscripts pos and neg Ω indicate positive and negative electrodes, respectively Rs serial resistance, superscripts pos and neg indicate Ω positive and negative electrodes, respectively RSEI resistance for Li+ ions migrating through the SEI Ω layer, superscripts pos and neg indicate positive and negative electrodes, respectively t time s tj transference number of species j - T absolute temperature K uj mobility of species j m/s⋅V vj velocity of species j m/s Vmax the maximum voltage that a Li-ion battery V can reach during charging x distance normal to the electrode surface m zj valence state of ionic species j - Z impedance of a process Ω Zre real component of impedance Ω Zim imaginary component of impedance Ω Zw Warburg resistance Ω α transfer coefficient - φ Galvani potential V or phase shift - γ activity coefficient - η stripping efficiency of lithium metal deposit - or overpotential for reactions, superscripts pos and neg V indicate positive and negative electrodes, respectively ηd diffusion overpotential from electrolyte, superscripts V pos and neg indicate positive and negative electrodes, respectively

+Liµ standard chemical potential of Li+ ions J/mol

+Liµ electrochemical potential of Li+ ions J/mol

ρLi density of lithium metal 0.53 g/m3

σ Warburg prefactor Ω/s1/2

List of symbols x

Symbol Meaning Value Unit τ relaxation time of a process s ω frequency of a sinusoidal signal Hz

jcx

∂

∂ concentration gradient of species j mol/m4

x∂∂φ potential gradient V/m

1 Introduction

Chapter 1 Introduction 2

1.1 General Introduction A battery is a device that enables the energy liberated in a chemical reaction to be converted directly into electricity.1 Batteries can be divided into primary and rechargeable batteries. In primary batteries, the electrochemical reaction is irreversible. Hence, primary batteries are not capable of being easily or effectively recharged electrically. In contrast, rechargeable batteries, often referred to as secondary batteries, can be recharged electrically by passing current through them in the opposite direction to that of the discharge current.2 A rechargeable battery is therefore an energy storage device which is able to store electricity in a chemical manner.

The application of batteries has been rapidly expanding recently. The last decades witness the boom of portable electronic devices, e.g. mobile phones, laptops, etc. The high performance components of the products are designed to be “lighter, thinner, shorter, smaller”. For this reason, urgent demand exists for secondary batteries that are more compact, lighter and higher energy density and longer lifetime, which catalysed the invention and commercialisation of Nickel-Metal Hydride (NiMH) and Lithium ion (Li-ion) batteries.

On the other hand, the traditional energy sources such as fossil fuel and natural gas are running out in the foreseeable future. Meanwhile, the rise of environmental consciousness requires to cut down the emission of automobiles. One solution to this important socio-economic issue is to utilize Hybrid Electric Vehicles (HEVs) and Electric Vehicles (EVs) in which batteries play a crucial and/or decisive role.3-10 This trend in turn ignites researches on large size and high power NiMH11-14 and Li-ion batteries.15-20

Nowadays, the most common rechargeable batteries available on the market are lead-acid, Nickel-Cadmium (NiCd), NiMH and Li-ion batteries. Some characteristics of these rechargeable battery systems are summarized in Table 1.1. In comparison to other existing rechargeable battery systems, Li-ion batteries are very favorable of meeting the requirement of power consumption for portable devices and HEVs/EVs. It is for this reason that studies on Li-ion batteries are one of the hottest topics in the field of battery and energy storage. Table 1.1 Battery characteristics of some rechargeable battery systems. [21]

nominal voltage

(V)

energy density

(Wh/kg)

energy density (Wh/l)

power density (W/kg)

cycle life

(cycles)

price

(US$/kWh)

temperature

(°C)

lead-acid 2.0 15-35 60-90 150-300 <600 300-120 -20-50

NiCd 1.2 27-55 45-110 260 500-2000 1400-300 -40-60

NiMH 1.2 35-50 80-175 180-220 300-600 550-350 -10-40

Li-ion 3.6 60-90 155-200 200-300 500-1000 1000-200 -20-45

Li-ion batteries 3

1.2 Li-ion batteries In 1990 Sony Corp. first revealed data of their lithium batteries.22 In this new battery concept, a carbon anode replaced the formerly used lithium metal anode in rechargeable lithium batteries. When the battery is operating, Li+ ions exchange between the positive and negative electrodes. Such a new type of battery system is later on referred to as Lithium ion batteries, also “rocking chair” batteries. As metallic lithium is not present in the battery, Li-ion batteries are chemically less reactive, safer, and offer longer cycle life than possible with rechargeable lithium batteries that employ lithium metal as the negative electrode material.

Compared to other rechargeable battery systems, Li-ion batteries exhibit advantages of high energy density (see Figure 1.1), high cell voltage and cyclability. This explains why Li-ion batteries receive most attention at both fundamental and applied levels and account for 63% of worldwide sales values in portable batteries.23,24

400

300

200

100

0

Ener

gy d

ensi

ty (W

h/l)

0 50 100 150 200 250 Energy density (Wh/kg)

Figure 1.1 Comparison of the different battery technologies in terms of volumetric and gravimetric energy density. [23]

1.2.1 Chemistry of Li-ion batteries

The positive electrode material in a Li-ion battery is usually a metal oxide (LiMO2, M = Co, Ni, Mn), with either a layered or tunnelled structure. The negative electrode material is commonly a graphitic carbon that has a layered structure similar to graphite.2 A lithium-ion


conducting electrolyte acts as an ionic path between the positive and negative electrodes. During cycling, Li+ ions are reversibly inserted into or extracted from interstitial space between atomic layers within the active material. This process is explained in detail below, based on the active materials used in this thesis, viz, LiCoO2 positive electrode and graphite negative electrode.

When the battery is operating, Li+ ions are extracted from LiCoO2 by electrochemical oxidation during charging and are inserted into LiCoO2 by electrochemical reduction during discharging, as indicated in equation 1.1.

LiCoO2 Li1-xCoO2 + xLi+ + xe- (0 ≤ x ≤ 0.5) (1.1)

charge

discharge For the negative electrode, it has been confirmed that graphite is reduced by electrochemical intercalation with lithium during charging and is electrochemically oxidized by lithium deintercalataion during discharging. This process is described in equation 1.2.

6C + yLi+ + ye- LiyC6 (0 ≤ y ≤ 1) (1.2) charge

discharge Thus, in the lithium cobaltite positive electrode/graphite negative electrode system, the overall cell reaction is that shown in equation 1.3

LiCoO2 + 6C LiyC6 + Li1-xCoO2 (1.3) charge

discharge The principle of the Li-ion rechargeable battery is further illustrated in Figure 1.2. In this figure, the layered active materials are shown on metallic current collectors.

LiyC Li1-xCoO2

Copper negativecurrent collector

Aluminum positivecurrent collector

NegativeElectrode

Positive Electrode

Discharge

Charge

Charge

Discharge

Li+ conducting

electrolyte

Li+

Li+

Li+

Li+

Li+

Figure 1.2 Schematic overview of charging and discharging of an Li-ion battery. [25]

Scope of this thesis 5

1.3 Scope of this thesis The main goal of the research described in this thesis is to obtain some fundamental understanding of the electrochemistry of Li-ion batteries. For this purpose, the positive and negative electrodes and the electrolyte are the objects to be studied separately. The behaviour of the entire battery is also investigated, concentrating on battery capacity fade phenomenon.

In Chapter 2 a brief introduction to battery materials is given to LiCoO2 cathode, graphite anode and electrolyte. The solid electrolyte interface (SEI) is also addressed, focusing on the relevance of SEI to battery capacity loss. The Lithylene technology with which the Li-ion batteries are made is described. To develop Li-ion batteries of different shapes is one task of this project. The results will be represented in this chapter.

Chapter 3 deals with the experimental details. After defining some specific terms, procedures used to form the batteries are explained. The principle of the electrochemical impedance spectroscopy (EIS) and its application in this thesis is outlined.

One objective of this Ph.D. project is to develop a lithium metal microreference electrode that can be embedded in Li-ion batteries to perform in situ measurements, thus gaining deeper insight into the characteristics of Li-ion batteries. This work is presented in detail in Chapter 4. Factors affecting the lifetime of the microreference electrode are discussed. The microreference electrode is proved to be suitable in EIS and electrode potential measurements.

Chapter 5 concerns the reaction resistances of Li-ion batteries. Making use of impedance measurements, the contribution of the positive and negative electrodes to the total battery resistance is determined with respect to battery state-of-charge (SoC). Three components of the electrode resistance, viz serial resistance (Rs), resistance for Li+ ions migrating through the SEI layer (RSEI) and charge transfer resistance (Rct) are evaluated by fitting impedance spectra to equivalent circuit. As the most influencing component of battery resistance, Rct of the positive electrode is comprehensively discussed.

Chapter 6 is devoted to investigations on mass transport in the electrolyte. A set of formulas is developed to describe the concentration gradient in the electrolyte and its relevance to the potential difference between microreference electrodes. The concentration gradient, the diffusion coefficient of Li+ ions and the diffusion overpotential from the electrolyte are obtained. The validity of these formulas is also discussed.

Chapter 7 concentrates on the issue of battery capacity loss. Studies on cycled batteries are presented. It will be shown that the capacity fade is largely caused by the positive electrode which suffers from loss of recyclable Li+ ions and increase in impedance. The mechanism of battery capacity fade is also discussed.

Finally, all major work carried out in this thesis will be addressed in the summary.


References 1. C. A. Vincent, F. Bonino, M. Lazzari and B. Scrosati, Modern Batteries: An

introduction to electrochemical power sources, Edward Arnold, p.1 (1984). 2. D. Linden, in Handbook of Batteries, 3rd ed., D. Linden and T. B. Reddy, Eds., Chap.

1, McGraw-Hill (2001). 3. S. Inman, M. El-Gindy and D. C. Haworth, Int. J. Heavy Vehicle Systems, 10, 167

(2003). 4. C. Fellner and J. Newman, J. Power Sources, 85, 229 (2000). 5. R. F. Nelson, J. Power Sources, 91, 2 (2000). 6. B. Weinstock, J. Power Sources, 110, 471 (2002). 7. K. Morita, JSAE Review, 24, 3 (2003). 8. M. Anderman, J. Power Sources, 127, 2 (2004). 9. J. M. Ogden, R. H. Williams and E. D. Larson, Energy Policy, 32, 7 (2004). 10. O. Bitsche and G. Gutmann, J. Power Sources, 127, 8 (2004). 11. A. Taniguchi, N. Fujioka, M. Ikoma and A. Ohta, J. Power Sources, 100, 117 (2001). 12. U. Köhler, J. Kümpers and M. Ullrich, J. Power Sources, 105, 139 (2002). 13. K. Shinyama, Y. Magari, K. Kumagae, H. Nakamura, T. Nohma, M. Takee and K.

Ishiwa, J. Power Sources, 141, 193 (2005). 14. R. Markolf, D. Ohms, G. Müller, Chr. Schulz, J. Harmel and K. Wiesener, J. Power

Sources, 154, 539 (2006). 15. P. Nelson, I. Bloom, K. Amine and G. Henriksen, J. Power Sources, 110, 437 (2002). 16. D. P. Abraham, J. Liu, C. H. Chen, Y. E. Hyung, M. Stoll, N. Elsen, S. MacLaren, R.

Twesten, R. Haasch, E. Sammann, I. Petrov, K. Amine and G. Henriksen, J. Power Sources, 119-121, 511 (2003).

17. J. R. Belt, C. D. Ho, C. G. Motloch, T. J. Miller and T. Q. Duongc, J. Power Sources, 123, 241 (2003).

18. N. Sato, J. Power Sources, 169-170, 758 (2003). 19. E. P. Roth and D. H. Doughty, J. Power Sources, 125, 308 (2004). 20. J. Arai, T. Yamaki, S. Yamauchi, T. Yuasa, T. Maeshima, T. Sakai, M. Koseki and T.

Horiba, J. Power Sources, 146, 788 (2005). 21. D. Linden, Handbook of Batteries, 2nd ed., McGraw-Hill (1995). 22. T. Nagaura and K. Tozawa, Prog. Batteries Solar Cells, 9, 209 (1990). 23. J.-M. Tarascon and M. Armand, Nature, 414, 359 (2001). 24. H. Takeshita, Proc. Conf. Power 2000, San Diego, 25 September (2000). 25. S. Megahed and B. Scrosati, Interface, 34 (1996).

2

Materials and Technology

Part of this chapter is based on the article by P.H.L. Notten, M. Ouwerkerk, H. van Hal, D. Beelen, W. Keur, J. Zhou and H. Feil published in J. Power Sources, 129, 45 (2004).

Chapter 2 Materials and Technology 8

2.1 Introduction This chapter is devoted to battery materials and battery technology. The brief introduction to battery materials is confined to LiCoO2 cathode, graphite anode and electrolyte, aiming to facilitate the discussions to be given in the following chapters. Section 2.2.1 and 2.2.2 deal with the structure of LiCoO2 and graphite, respectively, as well as their structure related properties and electrochemical behaviour. Section 2.2.3 concentrates mainly on the oxidation/reduction stability and ionic conductivity of the electrolyte and their dependence on electrolyte composition. The solid electrolyte interface (SEI), tightly associating with electrode materials and the electrolyte, plays a crucial role in the performance of Li-ion batteries. Being an important aspect of batteries, though not a material of Li-ion batteries, the SEI layer is addressed in Section 2.3, largely concerning the relevance of SEI to battery capacity fade.

The adopted battery technology focuses on the so-called Lithylene technology with which the Li-ion batteries investigated in this thesis are fabricated. This technology will be described in Section 2.4. It will be shown that this technology enables to make 3-D batteries of different shapes, being able to provide freedom to electronic device design. 2.2 Materials

2.2.1 LiCoO2 cathode

In 1980, Mizushima and coworkers recognized that LiCoO2 had a structure similar to the layered structures of the dichalcogenides and showed that the lithium could be removed electrochemically, thus making it a viable cathode material.1 Despite the high cost and toxicity of Co, LiCoO2 is currently the most common cathode material for commercial Li-ion batteries by virtue of its high working voltage, structural stability and long cycle life.2

LiCoO2 adopts a layered rock-salt structure (see Figure 2.1) that is based on a close-packed network of oxygen atoms with the Li+ and Co3+ ions ordering on alternating (111) planes of the cubic rock-salt structure. This (111) ordering introduces a slight distortion of the lattice to hexagonal symmetry.3

The structural stability of LiCoO2 depends on its composition. Stoichiometric LiCoO2 maintains a stable structure in air even at elevated temperature.4 In contrast, the delithiated LixCoO2 electrode (x < 1) is metastable4 which can undergo phase change to an electrochemically inactive structure and are prone to oxygen loss in the presence of an organic electrolyte solvent, particularly when batteries reach an operating temperature of 50 °C or more.5 Therefore, the theoretical capacity of the LiCoO2 cell is relatively low at 140 mAh/g6 because only around 0.5 Li/Co can be reversibly cycled without causing cell capacity loss due to changes in the LiCoO2 structure.

LiCoO2 cathode 9

Li c O

Co

Figure 2.1 Hexagonal lattice of rhombohedral LiCoO2. The arrow indicates the direction of the c-axis. [7]

-10

10

30

50

70

0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Mon

o.

I I + IIII II

Tem

p. (°

C)

x in LixCoO2

Figure 2.2 The phase diagram for LixCoO2. [3]

The LixCoO2 electrode (0 ≤ x ≤ 1) experiences phase changes during lithium

intercalation and deintercalation. Reimers and Dahn determined the phase diagram of LixCoO2 (see Figure 2.2) by electrochemical means and in situ X-ray diffraction (XRD).3 In Figure 2.2, I and II represent the 1st and 2nd hexagonal phases, respectively. I + II is the two-phase coexisting region while the monoclinic phase is formed when x is around 0.5 in LixCoO2. An in situ XRD study on the complete removal of lithium from a LixCoO2 electrode demonstrated that it is possible to prepare compositions very close to the end-member CoO2 (x = 0).8 The deintercalated phase, which retains its layered character, seems to be of the CdCl2 type with a monoclinic distortion.9 Noticeably, recent TEM studies have identified the presence of spinel phase (LiCo2O4) originating on the surface of severely cycled LiCoO2 electrode.10-12 The electrochemically less active spinel phase is believed to be in part accountable for the capacity loss after high cycle numbers.11 In addition, the


degradation of the particle surfaces may decrease the active volume of the electrode by impeding the electrical contact to particles of LiCoO2, leading to the observed decreased electrochemical performance of the LiCoO2 electrode.12

A feature of LixCoO2 is that its electrical conductivity changes dramatically with composition, behaving like a metal at x ≤ 0.6 and a typical semiconductor at x = 1,13-16 changing by 4 (for the x = 1.0 compound) orders of magnitude at ambient temperatures and up to 6 orders of magnitude at lower temperatures.16,17 Interestingly, Nishizawa et al. reported that the insulating nature of LiCoO2 cannot revert to its original state after successive intercalation/deintercalation.18

Commercial LiCoO2 electrode is commonly made up of LiCoO2 particles as active material, poly(vinylidene fluoride) (PVDF) as binder and conductive carbon to enhance the electrical conductivity.19 Such a mixture is coated on an aluminium foil behaving as a current collector. Figure 2.3 shows a scanning electron microscopy (SEM) picture of the LiCoO2 electrode used in this work, provided by SKC company, South Korea.

Figure 2.3 SEM picture of a LiCoO2 electrode

2.2.2 Graphite anode Graphite has long been known capable of hosting guest species to form the so-called graphite intercalation compounds (GIC). As early as 1938, Rüdorff and Hofmann suggested the application of graphite intercalation compounds for rechargeable electrochemical energy sources.20 About twenty years later, Herold discovered that Li+ ions can be intercalated within graphite to form Li-GIC.21 In 1981, Sanyo reported in a patent the first success in electrochemical lithium intercalation.22 Nowadays, graphite is the most widely used anode material in commercially available Li-ion batteries because of its relatively high specific capacity (theoretically 372 mAh/g), small irreversible capacity and good cycleability.23 In this case, Li+ ions, the guest species, can be reversibly intercalated into graphite by electrochemical means.

Graphite anode 11

The electrochemical lithium intercalation properties of graphite greatly depend on the crystallinity, morphology and orientation of its crystallites. For instance, the graphite material determines both the potential and current characteristics of the intercalation reaction and also the tendency for the solvation of LiCn compounds and therefore the suitability and compatibility with the electrolyte.

A

B

A

C

A

B

A

0.3354 nm

c-axis

(b) (a) (c)

Figure 2.4 (a) The hexagonal structure of a carbon layer, (b) the structure of hexagonal (2H) and (c) rhombohedral (3R) graphite. [24]

Being a carbonaceous material of layered structure, the basic building block for

graphite is a planner sheet of carbon atoms arranged in a hexagonal array (see Figure 2.4a) known as graphene layer. Most commonly, these graphene layers, weakly bonded together by van der Waals forces, stack in an ABAB sequence along the c-axis with an interplane distance of 0.3354 nm as shown in Figure 2.4b. This structure results in the hexagonal graphite also called 2H graphite. In a less common polymorph, ABCABC stacking occurs, termed rhombohedral or 3R graphite (see Figure 2.4c).

Stage-1 Stage-3Stage-2 Stage- 4

Li Layer

Graphene layer

Figure 2.5 Stage structures of graphite intercalated compounds. [23]


A general feature of lithium intercalation into graphite is the staging phenomenon. The intercalated Li+ ions are known to stay in-between the graphene layer(s). The term “stage” essentially refers to the number of graphene layers that lie between alternate lithium layers, which is depicted in Figure 2.5. Figure 2.6 shows the Dahn’s phase diagram of the lithium-graphite system.25 When lithium is intercalated into graphite, the following phases, e.g., dilute stage-1, stage-4, liquid-like stage-2L, stage-2 and stage-1 are successively formed. These stages can be easily monitored and controlled by the electrochemical reduction of carbons in Li+-containing electrolytes (see Figure 2.7). The potential plateaus in Figure 2.7 characterize the phase transitions during lithium intercalation. The fully lithiated graphite displays a potential very close to the potential of lithium metal.

0

20

40

60

0 0.1 0.2 0.3 0.4 0.50

10

20

30

40

50

60

70

3 +

2L

3+2

4,3 1’+41’ 2 + 1

2L+2

Tem

pera

ture

(°C

)

x in LixC6

Figure 2.6 The phase diagram as measured for graphite. [25]

Another important feature of GICs is the in-plane ordering of the guest species with

respect to the adjacent graphene sheets called the “superlattice structure”. Figure 2.8 schematically shows the superlattice structure of stage-1 Li-GIC. The structure of stage-1 Li-GIC thus gives a composition of LiC6, which restricts the theoretical capacity of graphite to 372 mAh/g.

The properties of graphite can be changed upon lithium intercalation. For instance, the stacking order of the graphene layers in the graphite shifts to AAAA during the lithium intercalation reaction and the interplane distance of LiC6 moderately increased from 0.3354 nm to 0.370 nm.26,27 The increase in the interplanar spacing indicates that graphite experiences volume expansion during intercalation and volume contraction during deintercalation. This volume expansion and contraction can cause the disconnection of electrode particles from the current collector and consequently result in irreversible capacity

Graphite anode 13

loss.28 As another example, Li-GICs attain an electrical conductivity of orders of magnitude higher than that of graphite.29 This effect results from the increased electron density in the conductivity band of graphite during lithium intercalation. Therefore, graphite intercalation compounds are occasionally named “synthetic metals”.30

0.0

0.1

0.2

0.3

0.0 0.2 0.4 0.6 0.8 1.00

0 .1

0 .2

0 .30 0 .2 0 .4 0 .6 0 .8 1

1’ to 4

2L to 23 to 2L

2 to 1

Vol

tage

(V)

x in LixC6 Figure 2.7 The voltage profile of a Li/graphite cell illustrating Li staging upon intercalation of graphite. [31]

Figure 2.8 The in-plain structure of LiC6.

Graphite, or more precisely graphitic carbons, used in commercial Li-ion batteries appears in a variety of shapes and morphologies, e.g. beads, fibers, flakes, etc.32 Being mixed with PVDF and conductive carbon, graphitic carbon is coated on copper foil that acts as a current collector to make the graphite electrode for Li-ion batteries.19

The SEM pictures of graphite fibers and flakes are shown in Figure 2.9a and b, respectively. Both samples come from SKC company, South Korea. The graphite electrode studied in this thesis is made of graphite flakes, as shown in Figure 2.9b.


(a) (b)

Figure 2.9 SEM pictures of (a) graphite fiber and (b) graphite flake.

2.2.3 Electrolyte

The electrolyte used in a typical commercial Li-ion battery system is an organic solution that comprises Li-based salt and solvents and acts as an ionic path between the two electrodes. Compared with the positive and negative electrodes, the role of electrolyte is often considered trivial.6 Practically, however, the chemistry of the electrolyte often affects the battery performance very significantly. Battery electrolytes are generally required to have the following fundamental properties: (1) high ionic conductance, (2) thermal and chemical stability, (3) wide potential window (electrochemical stability), (4) low reactivity toward other components in the battery, such as positive and negative electrodes, current collector, separator, etc. and (5) non-toxicity and safety.33 At present, LiPF6 and LiClO4 are often used as salts while propylene carbonate (PC), ethylene carbonate (EC), diethyl carbonate (DEC), dimethyl carbonate (DMC), 1,2-dimethoxyethane (DME), diethylether (DEE) and ethyl methyl carbonate (EMC) are well-known solvents for the electrolyte.

The window of the oxidation/reduction stability is a first requirement for electrolyte systems for Li-ion batteries.34 The adoption of highly oxidizing (> 4 V vs. Li/Li+) positive electrode materials in Li-ion batteries requires electrolyte systems that operate well outside their window of thermodynamic stability (3.5 V).6 Table 2.1 summarizes the oxidation potentials of electrolytes containing LiClO4 or LiPF6 dissolved in different solvents. All the oxidation potentials of the electrolytes are above the maximum operation potential of the conventional positive electrode materials (e.g. ~4.2 V for LiCoO2, ~4.3 V for LiMnO4, vs. Li/Li+), exhibiting outstanding oxidative stability. However, it must be pointed out that this oxidative stability is due to the fact that the electrolyte oxidation reaction is kinetically limited.35 Therefore, for storage at fully charged state and/or long-term cycling, the

Electrolyte 15

electrolyte oxidation can become pronounced and have a significantly negative impact on battery capacity fade.36 Moreover, the oxidation potential of an electrolyte also depends on the positive electrode material. Imhof et al. found that the starting point of CO2 gas evolution (1 M LiN(SO2CF3)2) was 4.2 V at LiNiO2, whereas at LiCoO2 and LiMn2O4, CO2 evolution was observed only above 4.8 V.37

The reduction potentials of several solvents are listed in Table 2.2. Obviously, the reduction potentials of these solvents are higher than the potential of a fully lithiated graphite electrode (see Figure 2.7), indicating that electrolyte reduction is inevitable during lithium intercalation. In fact, it is generally accepted that electrolyte reduction takes place at the surface of graphite during the first charging (intercalation) in the potential range of 1.8 to 0.8 V (vs. Li/Li+, depending on the solution composition).32 Fortunately, the reduction products form the well-known solid electrolyte interface (SEI) that prevents further reduction of electrolyte. This issue will be addressed in the next section.

Table 2.1 Oxidation potentials of solvent electrolytes. [38]

Oxidation Potential vs. Li/Li+ (V) Solvent

1 M LiClO4 1 M LiPF6

EC 5.8* > 6.0

PC 5.8 > 6.0

DEC 5.5 > 6.0**

DMC 5.7 > 6.0

DME 4.9* 4.9*

DEE 4.7* 4.9* * Mixed with PC ** Mixed with EC

Table 2.2 Experimental potential values of solvent reduction. [39]

Solvent Reduction Potential vs. Li/Li+ (V)

EC 1.36

PC 1.00-1.60

DEC 1.32

DMC 1.32

The ionic conductivity of the electrolyte is of practical importance because it affects the

internal resistance and rate performance of the battery. Generally, the conductivity of organic electrolyte solutions based on aprotic solvents, in which lithium salts are dissolved, is at most 2 × 10-2 S/cm, which is much lower than that of aqueous electrolyte solutions (8 ×


10-1 S/cm of H2SO4 for lead-acid batteries, 5 × 10-1 S/cm of KOH for alkaline Ni/Cd and NiMH batteries).33

Many lithium batteries have used electrolyte solutions consisting of mixed solvent systems. This trend is due to the idea that the shortcomings of each solvent will be made up by mixing the solvents with different characteristics.33 This is the now fairly well understood “synergistic” effect on conductivity which results from the blending of a low polarity solvent and a high polarity solvent.40 The resulting electrolyte has a conductivity higher than that of either single solvent system. Cyclic esters like EC and PC have high dielectric constant and provide high ionic conductivity. However, as the viscosities of pure EC and PC are too high for battery application, mixtures of EC or PC with low viscosity solvents such as linear carbonates, e.g. DEC, DMC and EMC are commonly adopted.

The ionic conductivity of the electrolyte depends on the composition of the electrolyte. In practical batteries, rather high concentrations of lithium salts are used in organic solvents in order to achieve high ionic conductivity. Some research groups41,42 investigated the dependence of the conductivity on the electrolyte composition and found maximum conductivities among the studied electrolytes. This is why the salt concentration of commercially available electrolyte lies in 1~1.5 M.

The most popular electrolyte currently used in Li-ion batteries by manufacturers and researchers is LiPF6 salt dissolved in a binary or ternary mixture of EC and linear carbonates like DMC, DEC, and EMC.43 This is based on the following facts of such electrolytes:

i. LiPF6 exhibits outstanding reductive and oxidative stability44,45 and well passivates and protects Al (a material for current collector of the cathode). Additionally, LiPF6 is also a flame retarding material.

ii. EC has a high dielectric constant, supplying a high ionic conductivity46 and does not co-intercalate with lithium ions. Co-intercalation of solvent molecules with lithium into graphite anode is known to take place in PC-based electrolyte, leading to drastic expansion of the graphite matrix, resulting in graphite exfoliation and degradation.47

iii. Linear carbonates decrease the viscosity of the electrolyte and have good penetrating abilities into polyolefin-based separators.

iv. Their presence favors the formation of a stable solid electrolyte interface film on the surface of graphite.

Besides the above-mentioned advantages, LiPF6 has a very limited thermal stability due to its spontaneous thermal degradation.48,49 Due to this shortcoming, researchers are now searching for alternatives to replace LiPF6. Several new salts have been developed in recent years.50-53 This area of new salts may yield a new generation of more stable Li-ion batteries.

Solid Electrolyte Interface 17

2.3 Solid Electrolyte Interface The solid electrolyte interface, named SEI layer by Peled,54 is recognized as a determinant factor on the performance of Li-ion batteries because the nature and behaviour of this layer affect the cycle life, power capability, shelf life and safety of Li-ion batteries.

In Li-ion batteries, the lithiated graphite electrode is thermodynamically unstable towards the organic electrolyte. When the graphite electrode is cathodically polarized to a low potential, solvent, salt anions and impurities in the electrolyte can be reduced to form some insoluble products which deposits on the surface of graphite to form the SEI layer. This process takes place mostly during the first several cycles of a working battery.

A nicely formed SEI layer is electronically non-conductive whereas it is Li+ ion conductive. Li+ ions can migrate through the SEI layer to intercalate into graphite while further side reactions are prevented since electrons cannot go through this layer. Thus, the SEI layer dramatically slows the kinetics of electrolyte reduction.

The SEI layer has a multilayer structure and mainly comprises inorganic components that can be Li2CO3, LiF, LixPF5-x, LixPF3-x, etc., depending on the solution composition.32,55 The formation of the SEI layer no doubt consumes some Li+ ions which can no longer be involved in the main electrochemical storage reactions. This Li+ ion consumption not only happens during the first several cycles but also occurs during both cycling and storage,56,57 resulting in battery capacity loss.

Binder particles

Current collector

Carbon Particles

Initial SEI film Thicker SEI film

Figure 2.10 Evolution of the SEI film on the surface of carbon when the battery is cycled. [58]

The SEI layer can thicken upon cycling, which is illustrated in Figure 2.10. Aurbach59 explained this thickening process as follows: A nicely formed SEI layer is electronically non-conductive. However, the SEI layer can be damaged to some extent during charging. It is well known that graphite expands during lithium intercalation. Since the SEI layer mainly consists of inorganic components, its flexibility is limited so that it is not capable of coping with the expansion of graphite. As a result, some micro-cracks may develop in the SEI layer. Consequently, electrons go through these cracks to the electrolyte and then react with


solvents, salt anions or impurities. As a result, the SEI layer can be ‘repaired’ but also thickened.59 This process repeats during cycling and continuously consumes Li+ ions. The thickened SEI layer also increases the migration resistance to Li+ ions. These two factors are believed to partially account for battery capacity fade.58,59

Recently, it is generally accepted that a complex surface chemistry is also decisive in determining the long-term viability of the commonly used LixMOy cathodes, where M = Mn, Co or Ni, and binary or ternary mixtures thereof.60-63 The surface film on the cathode material is suggested to be formed electrochemically during the first few cycles.64,65

Concerning LiCoO2 cathode material, a surface reaction that is common for LiCoO2 in LiPF6 solutions is the acid-base reaction between the basic LiCoO2 compound and unavoidably present trace of HF.32 Thereby, LiCoO2 cathode always contains LiF on its surface.61 It has been reported that surface films containing a high concentration of LiF are highly resistive (by more than two orders of magnitude compared with surface films comprised of Li2O, LiOH, Li2CO3, ROCO2Li, etc.).66

The importance of the SEI layer in the battery degradation has been noticed. Some researchers claimed that the source of capacity loss was due to a low-conductivity SEI layer on the cathode. 67,68 Aurbach et al. attributed the formation and thickening of surface films on both electrodes to be the most important factor that limits the cycle life of Li-ion batteries.69

Figure 2.11 shows SEM pictures of the SEI layer on LiCoO2 electrode and graphite electrode. The cycled LiCoO2 particles are covered with a thin film although this film is not distinct in comparison with the pristine LiCoO2 particles (see Figure 2.3). In contrast, the cycled graphite electrode is completely covered with a cloud-like film that is clearly seen in Figure 2.11b.

(a) (b)

Figure 2.11 SEM pictures of the SEI layer on (a) a LiCoO2 electrode and (b) a graphite electrode.

Lithylene technology 19

2.4 Lithylene technology

2.4.1 Principle of the Lithylene technology

The principle of the Lithylene technology70 is schematically represented in Figure 2.12, revealing a stack of the individual components. As this technology is stack-based, both single-sided (ss) electrodes and double-sided (ds) electrodes are applied to the exterior and interior of the battery stack, respectively. The electrodes are either mechanically or laser cut according to the desired battery shape. Prior to the electrode assembly, small holes with a diameter of 0.95 and 1.25 mm for the negative and positive electrodes, respectively, are either punched or laser cut. The distance between the rectangularly positioned holes are, in this example, 5 mm from center to center.

separator

punched holes

registration hole

rivet polymer

ss positive electrode

ds negative electrode

ds positive electrode


ds positive electrode


ss positive electrode

rivet polymer

Figure 2.12 Schematic representation of the construction of a 3-stack Lithylene battery.

As shown in Figure 2.12, a single-sided LiCoO2 positive electrode forms the basis of

the stack. A double-sided graphite negative electrode that is followed by a double-sided positive electrode is subsequently positioned on top of the single-sided positive electrode. The stack can then be repeated, ending up with the final single-sided positive electrode and the rivet polymer sheet. The required separators are placed in-between the positive and negative electrodes to prevent short circuit. In order to keep the holes open, the separators locally bridging the holes are melted by hot needles of 365 °C. Afterwards, the stack together with the rivet polymer sheets that are applied to the exterior of both single-sided positive electrodes is placed in-between a hot press at 120 °C for a certain period of time.


During this so-called riveting process, a pressure of 3 atm is applied. The riveting time is dependent on the stack numbers of the battery. As an example, it takes about 1.5 min to rivet a 3-stack battery shown in Figure 2.12. The polymer melts and penetrates each of the holes under the specified heating and pressure conditions. After cooling under 1 atm pressure, the polymer hardens, acting as rivets to keep the cell’s active materials together.

The Lithylene technology uses polymer rivets to keep the active materials in good electric contact and to provide stable battery structure, eliminating the need of metal cans for conventional batteries. In principle, Li-ion batteries made with this technology are lighter than their conventional metal-canned counterparts. Therefore, a higher gravimetric energy density can be achieved. Moreover, since all the battery materials adopted in this technology are commercially available, an additional advantage is that once new advanced electrode materials become available, they can be quickly used by the present Lithylene technology.

(a) (b)

Figure 2.13 (a) Optical macroscopic cross-section of a stack showing a nicely formed polymer rivet and (b) an incomplete polymer rivet marked by arrows.

2.4.2 Polymer rivets

The property of the polymer rivets is very important since they play a key role in the Lithylene technology. An example of a nicely formed rivet polymer is shown in Figure 2.13a. A typical problem that may occur is the formation of incomplete rivets as shown in Figure 2.13b. This problem happens under the following two conditions: (1) the separator covering the holes is not completely removed, leaving the holes only partly opened. As a consequence, insufficient amount of molten polymer flows into these holes during the riveting process, generating incomplete rivets; (2) The rivet technology is stack-based, the capacity of the batteries can be simply extended by increasing the stack number of the

Lithylene technology 21

batteries. Accordingly, more polymer sheets should be used as riveting material and the time needed to rivet the stack should also be extended. Otherwise the holes may not be fully filled with polymer, resulting in incomplete rivets. Such an incomplete rivet may lead to a locally less stable battery structure.

2.4.3 Preshaped batteries The most important characteristic of the Lithylene technology is that the stacks can be preshaped during the riveting step by using shaped riveting tools (Figure 2.14a) and, dependent on the shape of the battery, pre-formed flat electrodes (Figure 2.14b). Figure 2.15a shows an example of a preshaped battery. In this example, the radius of the battery is continuously changing from 1.45 cm to 1.72 cm and, consequently, it fits perfectly well into the housing of an electrical shaver, as Figure 2.15b reveals. In this way, the scarce empty space inside the often small-sized electronic products can be utilized much more efficiently.

(a)

(b)

Figure 2.14 (a) Upper and lower parts of the preshaped riveting tools and (b) examples of flat and pre-formed, laser-cut electrodes, clearly revealing the pattern of 1 mm-diameter holes.

Nowadays, product design is experienced to be one of the leading selling points,

distinguishing products from competing manufacturers. However, manufacturers are currently forced to some extent to fit the product design to the shape of commercially


available batteries which are mainly cylindrical and rectangular. This put some serious limitations to the freedom of product design. With these preshaped batteries, it is possible to adapt the shape of batteries to that of the design of the electronic products, generating a high degree of design freedom.

(a) (b)

Figure 2.15 (a) An example of preshaped battery and (b) its perfect fitness into a shaver housing. References 1. K. Mitzushima, P. C. Jones, P. J. Wiseman and J. B. Goodenough, Mater. Res. Bull.,

15, 783 (1980). 2. M. Winter, J. O. Besenhard, M. E. Spahr and P. Novák, Adv. Mater., 10, 725 (1998). 3. J. N. Reiners and J. R. Dahn, J. Electrochem. Soc., 139, 2091 (1992). 4. J. R. Dahn, E. W. Fuller, M. Obrovac and U. von Sacken, Solid State Ionics, 69, 265

(1994). 5. M. M. Thackeray, in Handbook of Battery Materials, J. O. Besenhard, Ed., Wiley-

VCH, p.300 (1999). 6. J.-M. Tarascon and M. Armand, Nature, 414, 359 (2001). 7. J. B. Bates, N. J. Dudney, B. J. Neudecher, F. X. Hart, H. P. Jun and S. A. Hackney, J.

Electrochem. Soc., 147, 59 (2000). 8. G. G. Amatucci, J. M. Tarascon and L. C. Klein, J. Electrochem. Soc., 143, 1114

(1996). 9. L. Seguin, G. Amatucci, M. Anne, Y. Chabre, P. Strobel, J. M. Tarascon and G.

Vaughan, J. Power Sources, 81-82, 604 (1999). 10. H. F. Wang, Y.-I. Jang, B. Y. Huang, D. R. Sadoway and Y.-M. Chiang, J.

Electrochem. Soc., 146, 473 (1999).

References 23

11. H. Gabrisch, R. Yazami and B. Fultz, J. Power Sources, 119, 674 (2003). 12. H. Gabrisch, R. Yazami and B. Fultz, J. Electrochem. Soc., 151, A891 (2004). 13. J. Molenda, A. Stoklosa and T. Bak, Solid State Ionics, 36, 53 (1989). 14. M. Ménétrier, I. Saadoune, S. Levasseur and C. Delmas, J. Mater. Chem., 9, 1135

(1999). 15. N. Imanishi, M. Fujiyoshi, Y. Takeda, O. Yamamoto and M. Tabuchi, Solid State

Ionics, 118, 121 (1999). 16. S. Levasseur, M. Menetrier, E. Suard and C. Delmas, Solid State Ionics, 128, 11

(2000). 17. M. S. Whittingham, Chem. Rev., 104, 4271 (2004). 18. M. Nishizawa, S. Yamamura, T. Itoh and I. Uchida, Chem. Commun., 1631 (1998). 19. R. J. Brodd and K. Tagawa, in Advances in Lithium-Ion Batteries, W. A. van

Schalkwijk and B. Scrosati, Eds., Chap. 9, Kluwer Academic/Plenum Publishers (2002).

20. W. Rüdorff and U. Hofmann, Z. Anorg. Allg. Chem., 238, 1 (1938). 21. A. Hérold, Bull. Soc., Chim. Fr., 187, 999 (1955). 22. H. Ikeda, S. Narukawa and H. Nakajima, Japanese Patent 1769661 (1981). 23. Z. Ogumi and M. Inaba, in Advances in Lithium-Ion Batteries, W. A. van Schalkwijk

and B. Scrosati, Eds., Chap. 2, Kluwer Academic/Plenum Publishers (2002). 24. D. Linden and T. B. Reddy, Handbook of Batteries, 3rd ed., McGraw-Hill (2001). 25. J. R. Dahn, Phys. Rev. B, 44, 9170 (1991). 26. D. Billaud, E. McRae and A. Hérold, Mater. Res. Bull., 14, 857 (1979). 27. X. Y. Song, K. Kinoshita and T. D. Tran, J. Electrochem. Soc., 143, L120 (1996). 28. A. H. Whitehead, K. Edström, N. Rao and J. R. Owen, J. Power Sources, 63, 41

(1996). 29. N. B. Brandt, S. M. Chudinov and Y. G. Ponomarev, Semimetals 1. Graphite and its

compounds, North-Holland, pp.241-243 (1988). 30. M. Winter and J. O. Besenhard, in Lithium Ion Batteries, Fundamentals and

Performance, M. Wakihara and O. Yamamoto, Eds., Chap. 6, Kodansha and Wiley-VCH (1998).

31. T. Zheng, J. N. Reimers and J. R. Dahn, Phys. Rev. B, 51, 734 (1995). 32. D. Aurbach, in Adcances in Lithium-Ion Batteries, W. A. van Schalkwijk and B.

Scrosati, Eds., Chap. 1, Kluwer Academic/Plenum Publishers (2002). 33. M. Morita, M. Ishikawa and Y. Matsuda, in Lithium ion Batteries, Fundamentals and

Performance, M. Wakihara and O. Yamamoto, Eds., Chap. 7, Kodansha and Wiley-VCH (1998).

34. J.-I. Yamaki, in Adcances in Lithium-Ion Batteries, W. A. van Schalkwijk and B. Scrosati, Eds., Chap. 5, Kluwer Academic/Plenum Publishers (2002).

35. D. Guyomard and J.-M. Tarascon, J. Power Sources, 54, 92 (1995). 36. M. Broussely, Ph. Biensan, F. Bonhomme, Ph. Blanchard, S. Herreyre, K. Nechev and

R.J. Staniewicz, J. Power Sources, 146, 90 (2005). 37. R. Imhof and P. Novak, J. Electrochem. Soc., 146, 1702 (1999).


38. K. Hayashi, Y. Nemoto, S. Tobishima and Y. Sakurai, DENKI KAGAKU, 66, 1307 (1999).

39. X. Zhang, R. Kostecki, T. J. Richardson, J. K. Pugh and P. N. Ross, Jr., J. Electrochem. Soc., 148, A1341 (2001).

40. L. A. Dominey, in Lithium Batteries, New Materials, Developments and Perspectives, G. Pistoia, Ed., Chap. 4, Elsevier (1994).

41. Geoffroy, P. Willmann, K. Mesfar, B. Carre and D. Lemordant, Electrochimi. Act., 45, 2019 (1999).

42. M. S. Ding, K. Xu, S. S. Zhang, K. Amine, G. L. Henriksen and T. R. Jow, J. Electrochem. Soc., 148, A1196 (2001).

43. S. S. Zhang, T. T. Jow, K. Amine and G. L. Henriksen, J. Power Sources, 107, 18 (2002).

44. L. A. Dominey, C. Nanjundiah, J. L. Goldman and V. R. Koch, J. Electrochem. Soc., 135, 2914 (1988).

45. D. Guyomard and J.-M. Tarascon, U.S. Patent 5192269 (1993). 46. G. Pistoia, J. Electrochem. Soc., 118, 153 (1971). 47. O. Besenhard, M. Winter, J. Yang and W. Biberacher, J. Power Sources, 54, 228

(1995). 48. T. Dudley, D. P. Wilkinson, G. Thomas, R. LeVae, S. Woo, H. Blom, C. Horvath, M.

W. Juzkow, B. Denis, P. Juric, P. Aghakian and J. R. Dahn, J. Power Sources, 35, 59 (1991).

49. S. E. Sloop, J. K. Pugh, S. Wang, J. B. Kerr and K. Kinoshita, Electrchem. Solid-Stade Lett., 4, A42 (2001).

50. J. Barthel, M. Schmidt and H. J. Gores, J. Electrochem. Soc., 145, L17 (1997). 51. Y. Sasaki, M. Handa, K. Kurashima, T. Tonuma and K. Usami, J. Electrochem. Soc.,

148, A999 (2001). 52. W. Xu and C. A. Angell, Electrochem. Solid-State Lett., 4, E1 (2001). 53. U. Lishka, U. Witelmann and M. Wegner, German Patent DE19829030C1 (1999). 54. E. Peled, J. Electrochem. Soc., 126, 2047 (1979). 55. E. Peled, D. Golodnitsky, G. Ardel, C. Menachem and D. Bar-Tow, J. Electrochem.

Soc., 144, L208 (1997). 56. R. Yazami and Y. F. Reynier, Electrochimi. Act., 47, 1217 (2002). 57. M. Broussely, S. Herreyre, P. Biensan, P. Kasztejna, K. Nechev and R. J. Staniewica,

J. Power Sources, 97-98, 13 (2001). 58. G. Ning, B. Haran and B. N. Popov, J. Power Sources, 117, 160 (2003). 59. D. Aurbach, J. Power Sources, 89, 206 (2000). 60. D. Aurbach, A. Zaban, Y. Ein-Eli, I. Weissman, O. Chusid, B. Markovsky, M. D. Levi,

E. Levi, A. Schechter, M. Moshkovich and E. Granot, J. Power Sources, 86, 91 (1997).

61. D. Aurbach, M. D. Levi and O. Lev, J. Applied Electrochem., 28, 1051 (1998). 62. D. Aurbach, K. Gamolsky, B. Markovsky, G. Salitra, Y. Gofer, U. Heider, R. Oesten

and M. Schmidt, J. Electrochem. Soc., 147, 1322 (2000).

References 25

63. F. T. Quinlan, K. Sano, T. Willey, R. Vidu, K. Tasaki and P. Stroeve, Chem. Mater., 13, 4207 (2001).

64. D. Aurbach, M. D. Levi, E. Levi, H. Teller, B. Markovsky, G. Salitra, U. Heider and L. Heider, J. Electrochem. Soc., 145, 3024 (1998).

65. T. Uchiyama, M. Nishizawa, I. Itoh and I. Uchida, J. Electrochem. Soc., 147, 2057 (2000).

66. D. Aurbach, E. Zinigrad and A. Zaban, J. Phys. Chem., 100, 3089 (1996). 67. D. Zhang, B. S. Haran, A. Durairajan, R. E. White, Y. Podrazhansky and B. N. Popov,

J. Power Sources, 91, 122 (2000). 68. J. Shim, R. Kostecki, T. Richardson, X. Song and K. A. Striebel, J. Power Sources,

112, 222 (2002). 69. D. Aurbach, B. Markovsky, A. Rodkin, M. Cojocaru, E. Levi and H.-J. Kim,

Electrochimi. Act., 47, 1899 (2002 ). 70. R. A. M. Hikmet and H. Feil, U.S. Patent 09352314 (2000).

3 Experimental

Chapter 3 Experimental 28

3.1 Introduction This chapter deals with some experimental aspects that are commonly applied to the following chapters. Various specific terms are often used in the literature and data sheets to specify the characteristics of different battery systems. These terms are defined in Section 3.2. Section 3.3 summarizes the detailed information of the materials used in this thesis to make the Lithylene Li-ion batteries. Batteries must be activated before being subjected to tests. The battery formation regime used in this thesis is described in Section 3.4. Electrochemical impedance spectroscopy (EIS) is nowadays widely used to characterize the behavior of electrochemical systems and is also employed in this thesis. The principle of EIS and its application to batteries are outlined in Section 3.5. 3.2 Definitions Rated capacity: The capacity of a battery, expressed in Ampere-hours (Ah), which is the

total charge expressed in Ah that can be obtained from a fully charged battery under specified discharge conditions. These conditions are specified by the manufactures.

C-rate: A charge or discharge current equal in Amperes to the rated capacity in Ah.

Multiples larger or smaller than the C-rate are used to express large or smaller currents. For example, the C-rate is 300 mA in the case of a 300 mAh battery, whereas the C/2 (or 0.5 C) and 2C rates are 150 mA and 600 mA, respectively.

Cycle life: The number of cycles that a cell or battery can be charged and discharged under

specific conditions, before a lower limit of the capacity is reached. This value is often set at 80% of the nominal capacity.1

CCCV: CCCV stands for Constant Current-Constant Voltage, a widely adopted charging

method for Li-ion batteries. During the CC part, the battery is charged with a constant current until Vmax is reached. During the CV part, the battery voltage keeps at Vmax while the charge current continues decreasing until the cut-off current is reached.

Vmax: The maximum voltage that a Li-ion battery can reach during charging. Vmax is

usually 4.2 V for Li-ion batteries in order to prevent overcharge side reactions at the positive electrode and phase changes in the positive electrode (see Section 2.2.1).

Cut-off current: The charging current at which charging is terminated.

Battery materials 29

Cut-off voltage: The cell or battery voltage at which discharging is terminated. State-of-Charge: The available capacity remaining in a battery, expressed as a percentage of

the rated capacity, usually referred to as SoC. A fully charged battery is characterized by 100% SoC.

3.3 Battery materials All materials used to construct the Li-ion batteries studied in this thesis are summarized here.

Three kinds of electrodes are used, single- and double-sided cathodes and double-sided anodes. The cathodes consist of LiCoO2 on an aluminum foil and the anodes contain graphite on a copper foil. Single- and double-sided cathodes mean that LiCoO2 is coated on one side and both sides of the aluminum foil, respectively. The coating thickness (one side) for the cathodes and anodes are 79 and 71 µm, respectively. Both the cathodes and anodes are provided by SKC, South Korea.

Commercial battery-grade LP 70 is employed as electrolyte. This electrolyte comprises of 1M LiPF6 salt dissolved in EC, DEC and DMC (2:1:2 w/w) and is provided by Merck, Germany.

The 30 µm thick separator is the Celgard (USA) 2300 microporous flat sheet membrane. This trilayer material consists of one layer of polyethylene (PE) between two layers of polypropylene (PP). The PP layers are designed to maintain the integrity of the film, while the low melting point of PE layer is intended to shutdown the cell if an over-temperature condition is reached.2 A SEM picture of the separator is shown in Figure 3.1 where the porous structure of the separator is visible.

Figure 3.1 SEM picture of the Celgard 2300 separator.


The rivet polymer, 100 µm in thickness, is made of low density polyethylene and has a melting point of 100 °C. This rivet polymer is provided by DSM company, The Netherlands.

The leads provide the electrical connection to the outside world after the stack has been packaged. An aluminum lead is used for the cathode and a nickel lead for the anode. Both are attached with PP sealant and are supplied by Sumitomo Electric Industries, Ltd., Japan.

The triple-layered packaging material provides a soft, hermetic bag for the Li-ion batteries. The packaging film is made of an aluminum foil coated with polyethylene terephthalate (PET) on the exterior as a protective layer and with PP on the interior as the sealant. This material is also from Sumitomo Electric Industries, Ltd.

All the materials are used as-received. 3.4 Battery formation regime After assembling the active materials as described in Section 2.4.1, the stack is dried in an oven (50 °C) under vacuum (0.08 bar) for 48 hours, then sealed in an air-tight Al-polymer package and filled with electrolyte. The as-made Li-ion battery must be activated before being subjected to tests. This process is often referred to as battery formation or battery activation. The formation procedure used in this thesis is described as follows:

• CCCV charge at 0.2 C, Vmax is 4.2 V, cut-off current is 0.05 C • Rest 30 min • Degas • Discharge at 0.2 C to 3.0 V • Rest 30 min • 6 standard cycles at 0.2 C • 0.2 C charge for 2 hours • End of formation

Each standard cycle during formation includes:

• CCCV charge at 0.2 C, Vmax is 4.2 V, cut-off current is 0.05 C • Rest 30 min • Discharge at 0.2 C to 3.0 V • Rest 30 min

During the first charging period, gases are generated due to side reactions such as

electrolyte reduction. These gases are removed from the package by vacuuming (0.08 bar) at the degas step. Subsequently, the package is finally sealed. The side reactions reduce the

Electrochemical impedance spectroscopy 31

charge efficiencyi by irreversibly consuming electrical charge, which occurs mainly during the first cycle. In Figure 3.2, the charge efficiency at the first cycle was 91%, quickly increased to 99% at the second cycle and maintained around 99.5% thereafter. From Figure 3.2 it is possible to evaluate the amount of charge consumed by the formation of the SEI layer during the first few cycles.

80

90

100

110

0 2 4 6 8

Cha

rge

effic

ienc

y (%

)

Cycle number

Figure 3.2 Charge efficiency vs. cycle number during battery formation.

Before the end of formation, the battery is charged at 0.2 C for 2 hours, charging the

batteries to about 40% SoC. Occasionally, the batteries are not immediately subjected to tests after formation but stored on a shelf. It is known that Li-ion batteries suffer from capacity loss on storage at either a fully charged or a fully discharged state. Charging the batteries to 40% SoC is to prevent this possible capacity loss. 3.5 Electrochemical impedance spectroscopy (EIS) 3.5.1 Fundamentals of EIS A given physical process has a certain time constant (or relaxation time), τ, which is the product of the capacitance and the resistance of this process. Such a physical process will give its own response if it is excited by a perturbation, for instance, a sinusoidal signal. The basic idea of EIS is to apply a sinusoidal signal to an electrochemical process. If measurements are made over a wide frequency range, different physical processes involved in this electrochemical process may be separated through their different time constants.3

In general, either current or voltage can be the controlled variable of the applied sinusoidal signal, and the phase and magnitude of the dependent variable (voltage or

i For a rechargeable battery, charge efficiency is the fraction, usually expressed as a percentage, of the

electrical charge put into a battery by charging that is recoverable during discharging.


current, respectively) are determined with respect to this signal. Practically, voltage is most commonly adopted as the controlled variable.

During the impedance spectroscopy measurement, a potential that consists of a constant value (Edc) and a superimposed alternating value (Eac) is applied to a steady-state system according to

E = Edc + Eac = Edc + Emsin(ωt) (3.1)

where Em and ω are the amplitude and the angular frequency of the sinusoidal signal, respectively. Typical values for this frequency are between 10-4 Hz and 106 Hz. The response of the resulting current has a phase shift, φ, with respect to E (see equation 3.2) and is measured by the instrument,

I = Idc + Iac = Idc + Imsin(ωt +φ) (3.2)

The impedance of the process is calculated by Ohm’s law

EZI

= (3.3)

The principle of EIS is to extract valuable information of an electrochemical process by analysing the determined impedance response.

Importantly, the amplitude of the sinusoidal signal must be sufficiently small. In the sense that the rates of electrochemical reactions are exponentially dependent upon the potential (refer to the Bulter-Volmer equation), electrochemical processes are inherently non-linear. The a.c. theories that are most fully developed, however, are all linear theories, which means that in order to use the a.c. theories the amplitude of the sinusoidal signal must be kept small enough that the electrochemical system becomes linear to a close approximation. As a general rule, a peak-to-peak amplitude of 10 mV should not be exceeded.4

3.5.2 Equivalent circuit of an electrochemical process Generally, an electrochemical process can be simply considered as an impedance to a small sinusoidal excitation; its performance can be represented by an equivalent circuit of resistors and capacitors that pass current with the same amplitude and phase angle as the real process does under the given excitation.

An electrochemical process always includes, in the case of battery electrodes, a charge transfer step (a redox reaction) that takes place at the electrode/electrolyte interface. This charge transfer step is a faradaic process and has a resistance Rct, the charge transfer resistance. Meanwhile, at the electrode/electrolyte interface, all electrodes exhibit a capacitance, Cdl, the so-called double layer capacitance which is quite independent of any faradaic reaction.

Impedance response of an electrochemical process 33

When a redox reaction occurs at the electrode/electrolyte interface, the total current will be the sum of the double layer charging current (Idl) and the faradaic current (If). Obviously, the equivalent circuit for this redox process is a resistor and a capacitor connected in parallel. Additionally, all of the current must pass through the solution resistance (Rs); therefore Rs is inserted as a series element to represent this effect in the equivalent circuit, which is depicted in Figure 3.3a.

Apart from the kinetics of the charge transfer reaction (Rct), mass transport of the reactants may also play a role in the electrochemical process. A diffusion impedance thus arises. For a semi-infinite diffusion process, it can be described by the Warburg impedance element (Zw) which is often positioned in series with Rct. Hence, the overall electrochemical process is represented by the equivalent circuit as shown in Figure 3.3b, which is called Randles equivalent circuit.

Cdl Cdl

Rs RsRct Zw

Rct

(b) (a) Figure 3.3 Equivalent circuits of (a) a charge transfer process and (b) a charge transfer process including mass transport.

3.5.3 Impedance response of an electrochemical process A sinusoidal voltage can be expressed in complex notation, consisting of a real and an imaginary part

Eac = Emsin(ωt) = Em e(jωt) (3.4)

where e(jωt) = cos (ωt) + jsin(ωt) and j = √-1, according to Eulers equation. When a sinusoidal voltage is applied to an electrochemical process, the determined impedance response can be given in terms of the real and imaginary parts separately,5 according to

1/ 2

1/ 2 2 2 2 1/ 2 2( 1) (ct

re sdl dl ct

RZ RC C R

σωσω ω σω

−

−

+= +

+ + + ) (3.5)

1/ 2 2 1/ 2 1/ 2

1/ 2 2 2 2 1/ 2 2

( ) (( 1) (

dl ct dlim

dl dl ct

C R CZC C R

ω σω σω ω σσω ω σω

− −

−

+ +=

+ + +1)

)+ (3.6)

where Zre and Zim are the real and imaginary parts of the impedance, respectively; ω is the frequency of the sinusoidal voltage and σ is the Warburg prefactor and is given by


1/ 2 1/ 22 2

1 12 bulk bulk

RTc D c Dn F A

σ⎛ ⎞

= +⎜ ⎟ox ox red red⎝ ⎠

(3.7)

in which R is the universal gas constant (8.314 J/mol⋅K), T the absolute temperature (K), n is the number of electrons transferred in this reaction, F the Faraday constant (96,485 C/mol) and A the electrode surface area (m2). The subscript ox and red represent the oxidized and reduced forms of the species, respectively. The parameter cbulk is the bulk concentration of the species and D is the diffusion coefficient of the species.

Most commonly, -Zim is plotted against Zre for different values of ω in a complex plane, denoted as a Nyquist plot. The impedance response has two limiting cases. At low frequencies when ω approaches zero, it is found that equations 3.5 and 3.6 jointly yield5

Zim = Zre – Rs – Rct + 2σ2Cdl (3.8)

This is the equation of a straight line of unit slop and with an intercept on the real Zre axis of Rs + Rct - 2σ2Cdl. The frequency dependence in this regime comes only from Warburg impedance terms; thus the linear correlation of Zre and Zim is characteristic of a diffusion-controlled electrode process.

At high frequencies where the Warburg impedance is negligible in relation to Rct, equations 3.5 and 3.6 lead to

22 2( )

2 2ct ct

re s imRZ R Z ⎛ ⎞− − + = ⎜ ⎟

⎝ ⎠

R

3.4 Nyquist plot for the circuit of F

(3.9)

which is the equation of a circle centred on Zre = Rs + Rct/2 with a radius of Rct/2. A Nyquist plot of the whole expression for -Zim vs. Zre has the idealized form shown in

Figure 3.4 in which both the kinetically controlled (semicircle) and diffusion controlled (linear, unit slop) regions are displayed. In principle, the value of Rs and Rct can be evaluated from the intercept of the semicircle with the Zre axis. The frequency at the semicircle maximum characterizes the time constant of this process from which the value of Cdl can be calculated.

Rs + Rct - 2σ2Cdl

ω = 1/(Rct⋅Cdl)

-Zim

Rs + RctRs

Zre

Figure igure 3.3b.

References 35

In practice, however, such an ideal spectrum is only rarely found in the experiments. There are two common perturbations which may still lead to at least part of a semicircular arc in the complex plan:6

i. The center of an experimental arc is frequently displaced below the real axis (a depressed semicircle) because of the presence of distributed elements in the material-electrode system. Such distribution arises, among others, from the spatial extension of the electrode/electrolyte interface in rough or porous electrodes. In such a case the relaxation time τ does not have a single value but is distributed continuously or discretely around a mean.

ii. Arcs can be substantially distorted (overlap) by other relaxations whose mean time constants are within two orders of magnitude or less of that for the arc under consideration.

In this thesis, only depressed semicircles are encountered in the impedance spectra because of the porous characteristics of the positive and negative electrodes.

As a conclusion, when EIS is employed, it is possible to determine various parameters as long as a proper equivalent circuit is adopted to accurately describe the system under investigation. However, the adopted equivalent circuit remains always an approximation of reality. References 1. G. Hambitzer, K. Pinkwart, C. Ripp and C. Schiller, in Handbook of Battery Materials,

J. O. Besenhard, Ed., Wiley-VCH, p.16 (1999). 2. P. Arora and Z.-M. Zhang, Chem. Rev., 104, 4419 (2004). 3. C. Ho, I. D. Raistrick and R. A. Huggins, J. Electrochem. Soc., 127, 343 (1980). 4. R. Greef, R. Peat, L. M. Peter, D. Pletcher and J. Robinson, Instrumental Methods in

Electrochemistry, John Wiley & Sons, p.257 (1985). 5. A. J. Bard and L. R. Faulkner, Electrochemical Methods, Fundamentals and

Applications, 2nd ed., Chap. 10, John Wiley & Sons, Inc. (2001). 6. J. R. Macdonald and W. B. Johnson in Impedance Spectroscopy, Emphasizing Solid

Materials and Systems, J. R. Macdonald, Ed., John Wiley & Sons, p.16 (1987).

4 Lithium metal

microreference electrodes

Again, the consideration of the electrical potential in the electrolyte, and especially the consideration of the difference of potential in electrolyte and electrode, involve the consideration of quantities of which we have no apparent means of physical measurement, while the difference of potential in pieces of metal of the same kind attached to the electrodes is exactly one of the things which we can and do measure.

J. W. Gibbs, 1899

This chapter is based on the article by J. Zhou and P. H. L. Notten published in J. Electrochem. Soc., 151, A2173 (2004).

Chapter 4 Lithium metal microreference electrodes 38

4.1 Introduction A reference electrode is a device that maintains a fixed value of its potential relative to the solution phase, facilitating potentiometric measurements of another electrode system relative to the reference electrode.1 An ideal reference electrode should show the following properties: (1) it should be reversible and obey the Nernst equation with respect to some species in the electrolyte; (2) its potential should be stable with time; (3) its potential should return to its initial value after small currents are passed through the electrode (no hysteresis); (4) if it is an electrode of the second kind (e.g., Ag/AgCl), the solid phase must not be appreciably soluble in the electrolyte and (5) it should show a low hysteresis with temperature cycling.2

In lithium-based redox systems, a lithium metal (Li/Li+) electrode is known to have a stable reversible potential. Although there is a spontaneous irreversible reaction between the lithium metal and the Li+-based organic electrolyte solution, resulting in the formation of a solid electrolyte interface onto the metal, this irreversible reaction is proved to be self-limited, stabilizing with time.3 Therefore, a lithium metal electrode is widely used as reference electrode in the studies on lithium-based battery systems.

A conventional lithium metal reference electrode can simply be a piece of lithium foil4,5 or can be made by attaching a piece of clean metallic lithium onto a copper wire,6 a copper grid3 or a stainless steel sheet.7 When conventional lithium metal reference electrodes are introduced inside batteries, due to their relatively big size, the working circumstances inside the batteries are influenced, e.g. by locally disturbing the ionic pathway and shielding the electric field between the two battery electrodes. As a result, what is being measured with the reference electrodes deviates from the real working environment inside the batteries. Another obvious shortcoming of conventional lithium metal reference electrodes is that they have to be prepared and introduced inside the batteries in an inert-gas-filled glove box. In addition, such an electrode setup has sometimes also to be operated in a glove box, creating a lot of inconvenience.

An alternative to the conventional lithium metal reference electrode is the micro-reference electrode made by electrochemically depositing either metallic lithium8 or lithium alloys9 onto nickel8 or copper wires.9 The advantage of this method is that the impact of the reference electrodes on the battery operation conditions is minimized because the wires have diameters in the micrometer range. The experiments can therefore be performed in situ. An additional advantage is that the batteries together with the substrate wires can be assembled in open air. However, the reported metallic lithium microreference electrodes did not reveal stable potentials over a longer period of time,8 which is essential to study degradation effects induced by cycling and storage. Other reference electrodes were based on a Li-Sn9 or a Li-Al alloy,10 whose potential deviates from that of the conventional Li/Li+ reference potential, which is most widely used in lithium-based battery systems.

This chapter is devoted to the development of reliable lithium metal microreference electrodes. Such a reference electrode is expected to attain the following features:

Substrate materials for the microreference electrodes 39

i. Micro: small enough to minimize the geometric disturbance to the battery performance and allow in situ measurements.

ii. Easy to make: convenient to prepare the substrate material and to in situ deposit metallic lithium.

iii. Easy to maintain: once the potential of the reference electrode starts to degrade after long-term measurements, this reference electrode can be easily revived by re-depositing metallic lithium onto it.

The factors that influence the stability of these electrodes are discussed. The reliability of the re-deposited microreference electrodes has been studied. Preliminary in situ electrode potential measurements obtained with microreference electrodes upon cycling will also be described. 4.2 Substrate material for the microreference electrodes Aluminum, copper, nickel and platinum are the candidates as substrate materials for the lithium metal microreference electrodes. Amongst these, copper shows some advantages over the others: does not alloy with lithium metal (aluminum does,11,12 thus the potential changes), electrochemically stable under working conditions (used as current collector for the negative electrode) and cheap (compared to platinum). Based on these advantages, copper wire was chosen to be substrate material for the reference electrodes.

The copper wires used in this work are 80 µm in diameter and are entirely covered with an insulation layer that prevents copper from being exposed to the electrolyte. Thus any possible mixed potential along the wire during electrochemical operation is avoided. The copper wires must be pretreated before being implanted into the batteries.

Copper wires about 25 cm long are cut from the roll. Approximately 15 cm is immersed into formic acid (100%) for 5 minutes. After being washed in rigorously stirred water, the copper wire is further processed with acetone soaked cloth in order to completely remove the partially dissolved insulation layer. Subsequently, the bare copper wire is carefully coiled in order to make a mechanically solid connector outside the batteries. This coil is to ensure a good electric contact between the reference electrode and the measurement equipment, as the copper wire itself is too thin to guarantee this.

Similarly, the tip on the other end of the copper wire is also immersed into formic acid. The same procedure as described above is repeated to remove the insulation layer on the tip. The tip is then cut so that only about 0.5 mm is left uncovered for lithium deposition. The accurate length of the bare copper wire is determined with a scale through an optical microscope (see Figure 4.1). The exposed tip is immersed in a polishing bath (HCl:HNO3:H2SO4:H2O = 1:36:217.5:245.5, v/v) for some 15 seconds to get a clean surface for lithium deposition, followed by washing again in rigorously stirred water. All the acids used in the polishing bath are concentrated. The polishing bath process consumes tiny amount of copper. The diameter of the bare copper wire is sometimes reduced from 80 µm to 79 µm.


Bare copper wire

Insulation layer

0.1 mm

Figure 4.1 Length determination for the bare copper wire by scaling through an optical microscope.

4.3 Experimental The processed copper wire was placed in-between the positive and negative electrode, with its tip located in the central area of the battery and being separated from the electrodes with separators. Figure 4.2 schematically represents one stack of such a three-electrode battery.

LiCoO2 electrode

Cu wire covered with insulation layer

Graphite electrode

Separator

Exposed

Separator

Figure 4.2 Schematic representation of a three-electrode battery.

The three-electrode Li-ion batteries were formed and cycled with an automatic cycling

apparatus (Maccor 4000 lab unit, U.S.A.). The formation process was as described in Section 3.4. A similar CCCV regime was adopted to cycle the batteries. The potential plots

Results and discussion 41

of both the positive and negative electrodes vs. microreference electrodes were recorded during cycling by means of the auxiliary inputs of the Maccor equipment.

The lithium deposition was carried out with an Autolab PGSTAT 20 (Ecochemie, Netherlands) under galvanostatic control after battery formation. Both the lithium layer thickness and the deposition current were varied. In order to check the stability of the microreference electrodes a second microreference electrode was implemented at a distance of approximately 0.5 cm. The mutual voltage difference was regularly measured with the Autolab PGSTAT 20. In the stripping efficiency experiments, the stripping current density was maintained at 0.2 mA/cm2 in all measurements.

The EIS measurements were also performed with the Autolab PGSTAT 20 under potentiostatic control, using the Frequency Response Analyser (FRA) mode. The impedance spectra were obtained by applying a sine wave of 5 mV in amplitude in the frequency range of 10 kHz to 10 mHz. 4.4 Results and discussion 4.4.1 Lithium deposition Lithium deposition, as discussed in this section, was carried out after battery formation. Before Lithium deposition, the batteries were charged to 100% SoC. The deposition layer thickness, d (µm), can be controlled by both the current density, i (mA/cm2) and the deposition time, t (s), according to the Faraday equation

10 Li

Li

M itdnFρ

= (4.1)

where MLi is the molecular weight of lithium (6.94 g/mol), n, the number of electrons involved, is 1, ρLi the density of metallic lithium (0.53 g/cm3) and 10 is used to represent the thickness in micrometers.

For practical applications the absolute current, I (A), is important. Since the current is coupled to the current density via the exposed surface area, A (cm2), A has to be calculated from

A = 2πrl + πr2 (4.2)

where r is the radius of the Cu wire, in our case 40 µm, and l is the length of the exposed tip, which can be accurately determined by optical microscopy.

In order to obtain a uniform lithium deposit, the galvanostatic current is first applied between the copper wire and the positive electrode and, subsequently, between the copper wire and the negative electrode. This process is schematically illustrated in Figure 4.3a and b, respectively. This two-step process is essential as the current distribution at the copper wire cannot be uniform due to the shielding effects of the wire itself. Consequently, during the initial deposition step, lithium will mainly be deposited onto the front-side of the copper wire, facing the positive electrode, whereas the back of the copper wire, facing the negative electrode, will hardly be covered (see Figure 4.3a). This introduces voltage instabilities. It is


well known that metallic lithium spontaneously and irreversibly reacts with the electrolyte to form the SEI layer,13 which prevents lithium from further oxidation. This reaction will therefore partly consume the deposited lithium. In the case of a non-uniform lithium layer, the thin lithium deposits will quickly be consumed under both open-circuit conditions and high impedance, voltage-measuring conditions (see Section 4.4.3), resulting in unstable reference electrodes. The proposed two-step deposition process prevents such rapid degradation.

For a copper wire that has an exposed tip of 0.5 mm long, the amount of lithium required to form a 4 µm thick layer is only about 4×10-8 mol. It can be calculated according to Faraday law that the “loss” of the battery electrode capacities is only 1.1 ×10-3 mAh. This capacity loss is negligible with respect to the 300 mAh nominal capacity of the battery.

nonuniform lithium layer

A

Positive electrode

Cu wire

electric field

Negative electrode

separator

I

(a)

(b)

Negative electrode

uniform lithium layer

A

Positive electrode

separator

I

electric field

Figure 4.3 Schematic representation of the lithium deposition process. The left-hand sides show the electric field lines while the right-hand sides show the morphology of the Li-deposit.

Figure 4.4 gives an example of the development of the potential differences during the

two-step lithium deposition process. For a fully charged Li-ion battery, it is to be expected that the potential difference between LixCoO2 (0.5 ≤ x ≤ 1) and the lithium metal reference electrode is above 4.2 V, while that of LiyC6 (0 ≤ y ≤ 1) and lithium reference electrode is around 0.08 V.14 Lithium deposition should therefore take place at these potentials with respect to the positive and the negative electrode, respectively. In our experiments, it always took some time before the expected lithium deposition potential was reached during the first deposition step (see Figure 4.4a). Charge consumed during this initial period is perhaps due


to copper-oxide reduction.8 This phenomenon was, of course, not observed when lithium was subsequently deposited from the negative electrode (Figure 4.4b). Since no lithium deposition occurs when the potential is below the deposition potential, this initial period is not included in the calculation of time in equation 4.1. That is why the process in Figure 4.4a took somewhat longer compared with Figure 4.4b.

0

1

2

3

4

5

0 40 80 120 160

(Epo

s– E

ref)

(V)

(a) Time (min)

0

0.03

0.06

0.09

0.12

0 40 80 120 160

(Ene

g– E

ref)

(V)

(b) Time (min) Figure 4.4 Potential profile of (a) the reference electrode vs. the positive electrode and (b) the negative electrode. The lithium layer thickness was 4 µm. Deposition current density was 0.2 mA/cm2.

4.4.2 Stability of lithium metal microreference electrodes For a reference electrode to be meaningful, its potential must be stable for a long period of time. The stability of the microreference electrodes was checked by means of continuously monitoring the potential difference between two parallel oriented microreference electrodes in one battery at a distance of 0.5 cm. This method is based on the assumption that in case the two microreference electrodes are unstable, it is rather unlikely that any deviation from the expected reference potential is of the same value for both electrodes. Hence, in the case of unstable reference electrodes the potential difference between the two will change as a function of time. Figure 4.5 shows a typical example of such stability test. As can be seen,


the potential difference between two reference electrodes was constant within 0.4 mV for more than 130 hours, indicating that the two microreference electrodes had a nearly identical potential within the accuracy of the potential measurement.

0

0.2

0.4

0.6

0.8

1

0 30 60 90 120 150

(Ere

f1 – E

ref2 ) (

mV

)

Time (h)

Figure 4.5 Potential difference between two microreference electrodes during 130 hours. The lithium layer thickness was 4 µm. Deposition current density was 0.2 mA/cm2.

4.4.3 Influence of the lithium layer thickness on the stability of reference electrodes As mentioned before, the formation of a SEI layer consumes the deposited lithium metal. When the deposited layer is not sufficiently thick, it might be all consumed by the SEI formation reaction, ultimately resulting in an unstable and not well-defined reference electrode potential. On the other hand, since it is a microreference electrode, the deposited lithium metal layer cannot be too thick with respect to the diameter of the copper wire. In addition, it should be noted that despite the fact that the currents will be very low, all potential-measurement systems are, in principle, current driven, thereby slowly consuming the reference electrode in time.3 The input impedance of the reference electrode terminal of a potentiostat is usually larger than 1010 Ω, which results in a measurement current lower than a few hundred pA. Such a low current will lead to a very small material consumption, i.e. of the order of 0.2 µAh, which corresponds to about 5×10-8 mol of metallic Lithium per month.3 In the case of microreference electrodes, this material consumption is not negligible in a longer term. Therefore, it cannot be expected that the as-prepared microreference electrodes are able to run continuously for several months. All together, the effect of the lithium layer thickness on the electrode stability needs to be examined to find out the optimal deposition layer thickness.

In order to reduce the lithium consumption the potential difference between two microreference electrodes was monitored for only 10 minutes each day. The results are shown in Figure 4.6. The deposition current density was 0.2 mA/cm2 in all cases while the lithium layer thickness of the microreference electrodes was 1 µm, 2 µm and 4 µm in Figures 4.6a to c, respectively. In Figure 4.6a, the potential difference was already 61 mV when the measurements started and increased to 185 mV after only one day. This large and


unstable potential difference evidences that 1 µm-reference electrodes do not retain a stable potential. In contrast, the potential difference of 2 µm thick reference electrodes remained constant within 1 mV and started to fluctuate after some 470 hours (see Figure 4.6b). The electrode using 4 µm lithium layers were found to be the most stable as is evidenced by Figure 4.6c. This layer remains stable within 0.5 mV for more than 1500 hours.

In conclusion, Figure 4.6 proves that the thickness of deposited lithium layers plays a vital role in the stability of microreference electrodes. A layer thickness of 4 µm is recommended to prepare long-term stable microreference electrodes and is adopted in our work hereafter.

0

50

100

150

200

0 10 20 30

Time (h)

(Ere

f1 – E

ref2 ) (

mV

)

(a)

-4

-2

0

2

4

0 500 1000 1500 2000

(Ere

f1 – E

ref2 ) (

mV

)

(b) Time (h)

-4

-2

0

2

4

0 500 1000 1500 2000

(Ere

f1 – E

ref2 ) (

mV

)

Time (h) (c)

Figure 4.6 The stability of microreference electrodes at lithium layer thickness of (a) 1 µm, (b) 2 µm (b) and (c) 4 µm. Deposition current density was 0.2 mA/cm2 in all cases.


Remarkably, the first data points in Figure 4.6b and c exhibited a small deviation from the potential measured after stabilisation. This effect has been studied in more detail by continuously measuring the potential difference during this stabilisation period. Figure 4.7 shows an example of such experiment. The stabilisation effect can be explained by the SEI formation onto the deposited lithium. As a matter of fact, the two microreference electrodes were not deposited with lithium simultaneously but consecutively. When the second microreference electrode was being deposited, the first reference electrode already started to build up an SEI layer, inducing the potential to equilibrate. When the potential difference between the two microreference electrodes was measured directly after the lithium deposition at the second electrode was completed, the potential difference between the passivated and freshly deposited reference electrodes was appreciable, which is clearly shown by the first data point in Figure 4.6b and c. This difference decreases once the SEI layer formation starts to saturate at the second electrode until its potential reaches a steady-state value. The result of Figure 4.7 reveals that the SEI layer formation takes about 1 hour under the present conditions.

0

1

2

3

4

5

0 20 40 60 80 100

(Ere

f1 – E

ref2 ) (

mV

)

Time (min) Figure 4.7 Potential change of a freshly deposited reference electrode vs. a stabilized reference electrode. The lithium layer thickness was 4 µm. Deposition current density was 0.2 mA/cm2.

4.4.4 Influence of deposition current density on the stability of reference electrodes The potential stability of the microreference electrodes is apparently based on the stability of the deposited lithium layer, which depends not only on the deposit thickness but also on the adhesion of the deposit. A crucial factor that affects the quality of a deposit and its adhesion is the current density. An increase in current density may bring about the formation of a more spongy deposit, the adhesion of which is less than that of a dense deposit.15 Furthermore, high current densities may also result in the formation of non-uniform deposits, e.g. lithium dendrites. For a microreference electrode, the thinnest spot of the non-uniform lithium layer might be quickly consumed due to the reaction with the electrolyte, exposing the copper wire to the electrolyte. A mixed potential will consequently be induced and thus, the potential of the microreferences will not be stable any more. On the other hand, small current densities frequently may generate the formation of a dense electro-


deposit. However, this dense electro-deposit can result in the development of internal stresses that may have an adverse effect on the adhesion leading, in extreme cases, to spontaneous flaking of the deposit from the substrate.15 Hence, it is worthwhile to find out an optimal deposition current density for the microreference electrodes.

-4

-2

0

2

4

0 500 1000 1500 2000

-4

-2

0

2

4

0 500 1000 1500 2000

-4

-2

0

2

4

0 500 1000 1500 2000

Time (h)

Time (h)

(Ere

f1 – E

ref2 ) (

mV

)

(a)

(Ere

f1 – E

ref2 ) (

mV

)

(b)

(Ere

f1 – E

ref2 ) (

mV

) (c) Time (h) Figure 4.8 Influence of current density on the stability of microreference electrodes, (a) 1 mA/cm2, (b) 0.5 mA/cm2, (c) 0.2 mA/cm2. The lithium layer thickness was 4 µm in all cases.

4 µm thick lithium layers were deposited onto reference electrodes at different current densities varied in the range of 0.2 to 1 mA/cm2. The stability of these reference electrodes


were checked and the results are presented in Figure 4.8. All microreference electrodes are stable over a long period of time. However, the microreference electrodes generated at 1 mA/cm2 (Figure 4.8a) became unstable after 800 hours. In contrast, Figures 4.8b and c reveal very good stability for electrodes produced at 0.5 and 0.2 mA/cm2, respectively. Both were stable for more than 1500 hours. As a general rule, small deposition current density is preferred. Thus, a current density of 0.2 mA/cm2 together with a layer thickness of 4 µm is concluded to be the most favourable preparation conditions.

4.4.5 Stripping efficiency of lithium deposits In order to figure out the dependency of the remaining amount of lithium metal after SEI formation on the deposition conditions, the microreference electrodes were subjected to stripping 24 hours after the lithium deposition. On galvanostatically stripping, the potential of the microreference electrode was measured with respect to the positive electrode, which is shown in Figure 4.9. The stripping was terminated when the potential started to drop sharply. The lithium stripping was performed only from the positive electrode although the lithium deposition was carried out from both the positive and negative electrode. This is due to the fact that the potential of (Eneg – Eref) becomes negative after the first stripping, indicative of no lithium metal left on the copper wire. The stripping efficiency is calculated as follows:

%100×=dep

str

QQ

η (4.3)

where Qstr and Qdep are the amount of charge (C) calculated for stripping and deposition, respectively. Obviously, the higher the stripping efficiency, the more the lithium metal remained. Figure 4.10 summarizes the results.

1

2

3

4

5

0 20 40 60 80 100 120

(Epo

s – E

ref)

(V)

Time (min)

Figure 4.9 Potential profile of the positive electrode vs. the reference electrode during lithium stripping.

Stripping efficiency of lithium deposits 49

In Figure 4.10a, the lithium layers with different thickness were all deposited at 0.2 mA/cm2. As clearly shown here, the stripping efficiency increased with increasing Li-deposit thickness. This result is not surprising because the loss of lithium metal is known due to the formation of SEI layer. It is expected that for the lithium layer deposited at the same current density, the amount of lithium metal consumed on SEI layer formation is more or less constant, regardless of the deposit thickness. Thus, the more the lithium deposited, the more lithium metal remained after SEI formation and, consequently, the higher the stripping efficiency.

0

20

40

60

80

100

0 1 2 3 4 5

Stri

ppin

g ef

ficie

ncy

(%)

(a) Thickness of Li-deposit (µm)

0

20

40

60

80

100

0 0.2 0.4 0.6 0.8 1 1.2

Stri

ppin

g ef

ficie

ncy

(%)

(b) i (mA/cm2) Figure 4.10 Stripping efficiency of lithium microreference electrodes as a function of (a) Li-deposit thickness and (b) depositing current density. Stripping current density was 0.2 mA/cm2 in all cases.

Figure 4.10b gives the stripping efficiency for the 4 µm thick lithium layer deposited at different current densities. Obviously, the stripping efficiency decreased with increasing current density. This result is consistent with the work of van Beek et al.16 The morphology of the lithium deposit is always of importance. However, it appears to be impossible to take the microreference electrode out of the battery without damaging the lithium deposit. Fortunately, van Beek and his coworker16 did thorough studies on this issue. Since the same results were observed for the correlation between the stripping efficiency and the depositing current density, it is reasonable to refer to their work16 for the morphology of the lithium deposit.


One objective of the microreference electrode is to measure the potential of the individual electrodes. As explained in Section 4.4.3, this potential measurement consumes lithium. The work presented in this section reveals that thick lithium layer deposited at small current density has the best stripping efficiency and is therefore preferable in terms of lithium consumption. This conclusion is in line with the results showed in Sections 4.4.3 and 4.4.4 that the microreference electrode with 4 µm thick lithium layer deposited at 0.2 mA/cm2 had the longest lifetime.

4.4.6 Microreference electrodes in impedance measurements EIS is currently widely used as a powerful technique in the field of electrochemistry to generate much useful information. In the application to Li-ion battery systems, EIS is able to determine the series resistance (Rs), diffusion/migration resistance through the SEI layer (RSEI), charge transfer resistance (Rct) and solid-state diffusion coefficient of lithium ion intercalation/deintercalation processes, which is very helpful for understanding the complex electrochemical processes occurring inside Li-ion batteries.

0

0.05

0.1

0.15

0.2

0 0.1 0.2

Pos.Neg.

0.3

-Zim

(Ω)

(a) Zre (Ω)

0

0.05

0.1

0.15

0.2

0.25

0 0.1 0.2 0.3 0.4

-Zim

(Ω)

(b) Zre (Ω) Figure 4.11 Impedance spectra of (a) the positive and negative electrodes, (b) the battery (circles) and the sum of the impedance spectra of the positive and negative electrodes (solid line) at 45% SoC.

Microreference electrodes in potential measurements 51

In order to perform three-electrode EIS measurements, the validity of the reference electrodes in EIS measurements needs to be examined. The simplest experimental validity test of the three-electrode measurements is to conduct the EIS measurements and to compare the sum of the two three-electrode measurements (positive vs. reference + negative vs. reference) with the result of the two-electrode measurements (positive vs. negative) for the battery.6,17 These results are shown in Figure 4.11. The EIS spectra of the positive and the negative electrodes measured vs. a reference electrode are shown in Figure 4.11a. The circles and the solid line in Figure 4.11b represent the measured EIS spectrum of the battery and the sum of the spectra of the positive and negative electrode in Figure 4.11a, respectively. Less than 1% difference was found between the sum of the three-electrode impedance results and the two-electrode measurement in both the real part (Zre) and the imaginary region (-Zim) of the spectrum, except at very high frequencies (about the first 5 points whose difference were less than 5%), indicating that the measurement set-up is appropriate. Thus, the validity of the measurements with the microreference electrodes is confirmed.

Comprehensive applications of impedance measurements in Li-ion batteries will be presented in Chapter 5.

4.4.7 Microreference electrodes in potential measurements One application of the microreference electrodes is to monitor the electrode potentials during cycling. Figure 4.12 shows, as an example, the measured potential profiles for one cycle at 0.02 C.

The potential profiles of the positive electrode and the battery are plotted in Figure 4.12a. At the beginning of charge, the potential of the positive electrode first showed a plateau at 3.92 V. This plateau has been assigned to a two-phase coexistence region consisting of two hexagonal phases of slightly different sizes (the first and second hexagonal phases, see Figure 2.2).18,19 Afterwards, the potential increaseed monotonically, corresponding to a single-phase reaction of the second hexagonal phase.19 When charging approached the end, small changes in the slope was observed, which is assigned to phase transitions between the second hexagonal phase to the monoclinic phase. Beyond that, the potential again increased monotonically until the end of charging, showing that a single phase reaction of the second hexagonal phase again took place in this region.19 The potential profile of the positive electrode for discharging was just the reversed process for charging, which is evident in Figure 4.12a. This result demonstrates the very good reversibility of the positive electrode for lithium intercalation and deintercalation.

Figure 4.12b shows the potential profile of the negative electrode during charge and discharge. Typically, when charge started, the potential of the negative electrode dropped rapidly to a small plateau around 0.21 V, which is attributed to the formation of the stage-4 lithiated graphite (see Figures 2.5 and 2.6).20,21 As intercalation progressed, the potential of the negative electrode continued declining, accompanied by two voltage plateaus around 0.115 and 0.081 V, respectively. The former plateau corresponds to the formation of stage-2 lithiated graphite while the latter to the formation of stage-1 lithiated graphite.20 In addition


to the inflections and plateaus appearing in the potential profile, the negative electrode also exhibited very good reversibility for Li+ ions intercalation and deintercalation, which can be seen from the potential profile for discharge. These are the typical properties of highly graphitized carbons.21

2.5

3

3.5

4

4.5

0 30 60 90 120 1502.5

3

3.5

4

4.5

dischargecharge

(Epo

s – E

ref)

(V)

Eba

ttery

(V)

(a) Time (h)

0

0.3

0.6

0.9

1.2

0 30 60 90 120 150

dischargecharge

(Ene

g –

Ere

f) (V

)

(b) Time (h) Figure 4.12 Potential profiles of (a) the battery and the positive electrode, (b) the negative electrode during one cycle at 0.02 C.

The potential profiles shown in Figure 4.12a and b disclose that, as a whole, the

potential of the negative electrode contributes the most to the battery voltage change during charge and discharge. The negative electrode manifests itself mainly at the beginning of charge and at the end of discharge. For the rest, it is the potential of the positive electrode that affects the battery voltage most significantly. This is particularly evident at the end of charge and the beginning of discharge where the potential profiles of the battery and the positive electrode are very similar to each other.

Stability of the re-deposited microreference electrodes 53

The potential profiles of the positive and negative electrodes in situ measured with the lithium microreference electrode, including the values of the potential plateaus, are very close to those reported in the literature,18-20 demonstrating that the microreference electrode works nicely in potential measurements.

In addition, by making use of the long-term stability of the microreference electrodes, it is also possible to monitor the electrode potentials during prolonged cycling. Figure 4.13 shows the results for several cycles. A good reproducibility of the potential plots for the positive and negative electrodes was observed, again proving the stability and reliability of the microreference electrodes.

2

2.5

3

3.5

4

4.5

90 110 130 150 1700

0.2

0.4

0.6

0.8

1

Time (h)

(Ene

g –

Ere

f) (V

)

(Epo

s – E

ref)

(V)

Figure 4.13 Potential plots of the positive and negative electrodes measured with one microreference electrode for several cycles, at room temperature and 0.2 C rate for both charge and discharge.

4.4.8 Stability of the re-deposited microreference electrodes As the deposited lithium may be consumed during the electrochemical measurements, the microreference electrode potential will, eventually, degrade after a long period of time and is no longer suitable for proper electrochemical measurements. However, this degradation does not necessarily mean that the reference electrodes can no longer be used. One advantage of the microreference electrodes is that it can be easily “revived” by re-depositing lithium. The re-deposition procedure is exactly the same as the lithium deposition described in Section 4.4.1. Figure 4.13 shows the potential difference as a function of time between a re-deposited and a freshly deposited microreference electrode, which amounted to no more than 0.5 mV, evidencing that the re-deposited microreference electrode had the same potential as the freshly deposited electrode. In addition, the fluctuation of this potential difference was within 0.1 mV for more than 300 hours, which


proves that the re-deposited microreference electrode had a stable potential. Thus, the re-deposited microreference electrode is also reliable in long-term, in situ, electrochemical measurements.

0

0.2

0.4

0.6

0.8

1

0 100 200 300 400

(Ere

f1 – E

ref2 ) (

mV

)

Time (h) Figure 4.14 Potential difference between a re-deposited and a freshly deposited microreference electrodes. The lithium layer thickness was 4 µm. Deposition current density was 0.2 mA/cm2.

Of practical interest is how many times the microreference electrodes can be refreshed. Though no systematic work was done on this point, it has been found in practical operations that microreference electrodes refreshed for more than 5 times still work properly and do not generate any visual differences in impedance and potential measurements compared with freshly prepared microreference electrode.

Finally, it is worthwhile to mention that the insulation layer covering the copper wire is very stable in the electrolyte. We examined a microreference electrode in a Li-ion battery subjected to more than 500 cycles and found that the insulation layer on the copper wire remained intact. The polymeric insulation layer ensures that no mixed potential occurs to the microreference electrode during prolonged measurements. 4.5 Conclusions Lithium metal microreference electrodes are successfully prepared by electrochemically depositing lithium from both the positive and negative electrodes onto a copper wire placed in-between the positive and negative electrodes of Li-based battery systems. Both the depositing current density and the layer thickness are found to have a profound influence on the stability of the microreference electrodes. A current density of 0.2 mA/cm2 and lithium layer thickness of 4 µm is shown to be the most favorable lithium deposition condition. Deposition of lithium from both electrodes indeed results in uniform lithium layers. Advantageously, the potential of the re-deposited microreference electrode is also very stable and has the same value as freshly deposited electrodes.

References 55

The microreference electrodes are proved to be a reliable tool for in situ electrochemical measurements. They can be used to measure the impedance spectra of the individual electrode. Moreover, potential profiles of the individual electrode during cycling are obtained with the microreference electrodes. It has been found that the potential of the negative electrode is mainly responsible for the battery voltage change at the beginning of charge and at the end of discharge whereas the potential of the positive electrode dominants the battery voltage at the end of charge and the beginning of discharge. References 1. R. G. Compton and G. H. W. Sanders, Electrode Potentials, Oxford University Press,

(1996). 2. D. T. Sawyer, A. Sobkowiak and J. L. Roberts, Electrochemistry for Chemists, 2nd ed.,

John Wiley & Sons, Inc., p.184 (1995). 3. A. Blyr, C. Sigala, G. Amatucci, D. Guyomard, Y. Chabre and J-M. Tarascon, J.

Electrochem. Soc., 145, 194 (1998). 4. Ib I. Olsen, J. Barker and R. Koksbang, Solid State Ionics, 83, 125 (1996). 5. J. Y. Song, H. H. Lee, Y. Y. Wang and C. C. Wan, J. Power Sources, 111, 255 (2002). 6. M. Dollé, F. Orsini, A. S. Gozdz and J.-M. Tarascon, J. Electrochem. Soc., 148, A851

(2001). 7. K. Sawai and T. Ohzuku, J. Electrochem. Soc., 150, A674 (2003). 8. C. G. Thurston, J. R. Owen and N. J. Hargreaves, J. Power Sources, 39, 215 (1992). 9. K. Amine, C.-H. Chen, J. Liu, M. Hammond, A. Jansen, D. Dees, I. Bloom, D. Vissers

and G. Henriksen, J. Power Sources, 97-98, 684 (2001). 10. M. W. Verbrugge, D. R. Baker and B. J. Koch, J. Power Sources, 110, 295 (2002). 11. A. N. Dey, J. Electrochem. Soc., 118,1547 (1971). 12. M. W. Verbrugge, D. R. Baker and B. J. Koch, J. Power Sources, 110, 295 (2002). 13. E. Peled, J. Electrochem. Soc., 126, 2047 (1979). 14. D. Aurbach, M. D. Levi, E. Levi, H. Teller, B. Markovsky, G. Salitra, U. Heider and L.

Heider, J. Electrochem. Soc., 145, 3024 (1998). 15. A. T. Vagramyan and Z. A. Solov’eva, Technology of Electrodeposition, Robert

Draper Ltd., p.267 (1961). 16. J. R. van Beek and P. J. Rommers, Power Sources, 7, 595 (1979). 17. J. Y. Song, H. H. Lee, Y. Y. Wang and C. C. Wan, J. Power Sources, 111, 255 (2002). 18. M. Inaba, Y. Iriyama, Z. Ogumi, Y. Todzuka and A. Tasaka, J. Raman Spectrosc., 28,

613 (1997). 19. J. N. Reimers and J. R. Dahn, J. Electrochem. Soc., 139, 2091 (1992). 20. Z. Ogumi and M. Inaba, in Advances in Lithium-ion Batteries, W. A. van Schalkwijk,

and B. Scrosati, Eds., Kluwer Academic/Plenum Publishers, p.84 (2002). 21. D. Aurbach, Y. Ein-Eli, O. Chusid, Y. Carmeli, M. Babai and H. Yamin, J.

Electrochem. Soc., 141, 603 (1994).

5 On reaction resistances

Chapter 5 On reaction resistances 58

5.1 Introduction The battery resistance is of practical importance because it has a great impact on the rate capability of batteries. In this chapter, EIS is used as a tool to investigate the resistance of the individual electrodes.

Section 5.3.1 briefly introduces the impedance response of a Li-ion battery. Section 5.3.2 deals with impedance measurements of individual electrodes. These studies are to find out the dependence of the electrode resistance on the electrode potential and reveal the respective contribution of the positive and negative electrode to the total battery resistance.

The resistance of Li-ion batteries is made up of four components, namely, the series resistance that includes the electrolyte resistance and contact resistance (Rs), resistance for lithium ions migrating through the SEI layer (RSEI), charge transfer resistance (Rct) and solid-state diffusion resistance also known as Warburg resistance (Zw). In Section 5.3.3, values of Rs, RSEI and Rct of individual electrode are obtained by fitting the impedance spectra to an equivalent circuit. Among these values, Rct of the positive electrode (Rct

pos) is found to be the largest component resistance which strongly depends on electrode potential, being the most influencing factor in the battery resistance. For this reason, Section 5.4 is devoted to figure out the cause for the dependence of Rct

pos on the electrode potential. Based on experimental results and a qualitative discussion, it is proposed that the variation of Rct

pos with electrode potential can be assigned to the change of electric conductivity of the positive electrode.

The electrode potential is coupled with the battery SoC. For convenience of the discussion, both electrode potential and battery SoC are used in this chapter. 5.2 Experimental

The EIS measurements were performed with an Autolab PGSTAT 20 under potentiostatic control, using the Frequency Response Analyser (FRA) mode. The impedance spectra were obtained by applying a sine wave of 5 mV in amplitude in the frequency range of 10 kHz to 10 mHz.

The state-of-charge of the batteries were determined with Maccor as follows: the batteries were first fully charged, using a CCCV regime at 0.2 C with a cut off current of 0.02 C. This state-of-charge was assigned to be 100% SoC. 0% SoC was obtained by discharging the fully charged batteries to 3.0 V at 0.2 C, followed by a deep discharge at 0.05 C to 3.0 V. The total charge extracted from the batteries was used as basis to calculate the amount of charge needed to reach the SoC in-between 0 and 100% SoC.

The potential of the positive electrode vs. microreference electrodes was recorded during cycling by means of the auxiliary inputs of the Maccor equipment. The charging/discharging current varied from 0.02 to 1.0 C. At 0.02 and 0.05 C, the battery was galvanostatically charged and discharged with cut off voltages of 4.2 V and 3.0 V, respectively. At 0.1 to 1.0 C the battery was first CC charged to 4.2 V, rest for 1 hour and

EIS measurements 59

then CC charged again at 0.05 C to reach 4.2 V. The latter 0.05 C charging was to ensure that the same amount of charge was inserted into the battery in comparison to charging at 0.02 C and 0.05 C. Similarly, for discharging, the battery was first discharged to 3.0 V and after 1 hour rest was deep-discharged at 0.05 C to 3.0 V. These data were used to calculate the overpotential of the positive electrode and to plot I-η curves.

All the batteries were tested right after battery formation was finished. 5.3 EIS measurements 5.3.1 Impedance response of Li-ion batteries The impedance response of Li-ion batteries is exemplified with an impedance spectrum of the negative electrode presented in Figure 5.1. This impedance spectrum consists of two semicircles and a straight sloping line. These two semicircles clearly reflect the well-separated, different time constants of the lithium intercalation process. It is generally accepted that the semicircle labelled 1 appearing from high to medium frequencies results from the migration process of lithium ions through the SEI layer that covers the surface of the active materials. The semicircle labelled 2 occurring from medium to low frequencies is assigned to the charge transfer processes.1-4 The intercept of the impedance spectra with the Zre axis at very high frequencies represents Rs while the straight line in low frequency range stems from the solid-state diffusion process taking place within the host material. Thus, the impedance spectrum of the negative electrode actually describes the entire lithium intercalation/deintercalation processes. The impedance response of the positive electrode and the battery is similar to that of the negative electrode, also comprising two semicircles and a straight sloping line.

0

0.03

0.06

0.09

0 0.03 0.06 0.09 0.12 0.15

1 2

-Zim

(Ω)

Zre (Ω) Figure 5.1 Nyquist plot of a negative electrode used in this work.

Note that both semicircles in Figure 5.1 are suppressed. Suppressed semicircles may happen when the relaxation time τ of a process is not single-valued but is continuously or


discretely distributed around a mean value (see Section 3.5.3). This effect can be very significant for porous materials, which is exactly the case for both the positive and negative electrode materials used in Li-ion batteries.

5.3.2 EIS vs. electrode potential The impedance spectra of the individual electrodes were measured with the microreference electrode at different SoC of the battery, viz, 0, 24, 45, 68, 77, 87, 95 and 100%. Correspondingly, the potential of the positive electrode increased from 3.839 V to 4.253 V while the potential of the negative electrode decreased from 0.470 V to 0.076 V (see Figure 5.2).

0

0.1

0.2

0.3

0.4

0 0.1 0.2 0.3 0.4 0.5 0.6

3.839 V3.907 V3.932 V4.012 V4.069 V4.160 V4.211 V4.253 V

0

0.03

0.06

0.09

0.12

0 0.05 0.1 0.15

4.012 V4.069 V4.160 V4.211 V4.253 V

Zre (Ω)

-Zim

(Ω)

-Zim

(Ω)

(a) Zre (Ω)

0

0.02

0.04

0.06

0.08

0.1

0 0.03 0.06 0.09 0.12 0.15

0.470 V 0.147 V

0.120 V 0.089 V

0.087 V 0.085 V

0.083 V 0.076 V

-Zim

(Ω)

(b) Zre (Ω)

Figure 5.2 Nyquist plots of (a) the positive electrode and (b) the negative electrode at different electrode potentials. The insert shows the spectra of the positive electrode obtained above 4.012 V.

The impedance spectra of the positive electrode at different potentials are plotted in Figure 5.2a. The positive electrode exhibited very similar impedance response at potentials above 4.012 V. However, when the potential dropped below 4.012 V, substantial increases

EIS vs. electrode potential 61

in the impedance were observed with decreasing electrode potential, particularly at 3.839 V. In contrast, the negative electrode demonstrated almost the same impedance response throughout the potential range with the exception of the impedance spectrum at 0.470 V (see Figure 5.2b).

In order to figure out the contribution of the individual electrodes to the total battery impedance, the impedance spectra of the positive and negative electrodes at different battery SoC are compared in Figure 5.3.

0

0.03

0.06

0.09

0.12

0 0.05 0.1 0.15 0.2 0.25 0.3

Battery

Pos.

Neg.

-Zim

(Ω)

(a) Zre (Ω)

0

0.1

0.2

0.3

0.4

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Battery

Pos.

Neg.

-Zim

(Ω)

(b) Zre (Ω)

Figure 5.3 Impedance responses of the battery and individual electrode at (a) fully charged state and (b) fully discharged state.

Figure 5.3a presents the impedance spectra of the battery, the positive and negative

electrodes when the battery was fully charged. The positive electrode showed a slightly higher impedance than the negative one. Nevertheless, these two individual electrodes had comparable impedance and contributed more or less equally to the battery impedance.

As a comparison, the impedance spectra obtained at fully discharged state are plotted in Figure 5.3b. In this case, the impedance of the positive electrode obviously overwhelms that of the negative electrode, being dominant in total battery impedance.


The results presented above reveal that the impedance of Li-ion batteries decreases with increasing SoC, which results from the change of the impedance of the positive electrode. The impedance of a fully discharged Li-ion battery stems mainly from the positive electrode. As the impedance of the positive electrode decreases with increasing SoC, the positive electrode contributes less and less to the battery impedance whereas the contribution of the negative electrode becomes relatively pronounced. When the battery reaches its fully charged state, the positive and negative electrodes contribute comparably to the battery impedance.

5.3.3 Values of Rs, RSEI and Rct As explained in Section 5.3.1, the impedance spectra of the electrodes describe lithium ion intercalation/deintercalation processes. These processes can be modeled with an equivalent circuit (EQC).5,6 The principle is outlined in Section 3.5.2. Thus, parameters of lithium ion (de)intercalation can be obtained by fitting the impedance spectra with a suitable EQC.

The EQC adopted in this thesis is a modified Randles circuit (see Figure 5.4) and is used by some other researchers.7,8 In this EQC, Rs is the series resistance. RSEI and CSEI are the resistance and capacitance of the SEI layer, respectively, Rct the charge transfer resistance, Cdl the double layer capacitance. Zw is the finite space diffusion impedance, which was derived by Ho et al. and Glarum and Marshall.9 Cint is the differential intercalation capacitance, representing the accumulation of the intercalated lithium ions into the host materials.10 In practice, all the capacitances in Figure 5.4 are replaced by constant phase elements (CPE) to yield an accurate fit. The CPE is commonly used to describe the depressed semicircle resulting from a porous electrode.11 The values of Rs, RSEI and Rct can be readily obtained from the fitted parameters.

Zw RSEI

Cdl

Rs

CSEI Cint

Rct

Figure 5.4 Equivalent circuit used to analyze impedance spectra.

The extracted Rs, RSEI and Rct are summarized in Figure 5.5. In order to make a comparative discussion between the positive and negative electrodes, each resistance is plotted in one figure against the battery SoC instead of electrode potential. The solid-state diffusion process, which is usually characterized by a diffusion coefficient (Ddif), is not presented here, as the diffusion process is not considered in this study.

Values of Rs, RSEI and Rct 63

0

10

20

30

40

50

0 30 60 90 120

Rs (

mΩ

)

(a) SoC (%)

0

100

200

300

0 30 60 90 120

0

20

40

60

80

100

0 30 60 90 120

SoC (%)

RSE

I (m

Ω)

(b)

Rct (m

Ω)

(c) SoC (%)

Figure 5.5 The values of (a) Rs, (b) RSEI and (c) Rct as a function of the battery SoC, positive electrode (◊), negative electrode (∆).

Figure 5.5a shows Rs of the positive (Rspos) and negative (Rs

neg) electrodes. As expected, Rs

pos and Rsneg attained very close values that are almost independent of SoC. The slight

difference between Rspos and Rs

neg can be attributed to the different lead and current collector materials used for the positive and negative electrodes, which was indeed measured experimentally.

RSEI of the positive (RSEIpos) and negative (RSEI

neg) electrodes are plotted in Figure 5.5b. Since RSEI reflects the resistance to Li+ ions migrating through the SEI layer, the properties of the SEI layer, such as thickness and porosity, will certainly affect the value of RSEI. On


the other hand, because the properties of the SEI layer hardly vary substantially during one cycle, it is expected, as well as reported in the literature,1,2,10 that RSEI is invariant with SoC. This analysis is, however, not in line with the results presented in Figure 5.5b. Both RSEI

pos and RSEI

neg showed a large value at a fully discharged state. For the positive electrode, RSEI

pos maintained the large value when battery SoC increased to 24% and started to decline at 45% SoC, finally to become constant at 68% SoC. In comparison, RSEI

neg dropped to a certain value when the battery SoC increased to 24% and stabilized at that value, irrespective of the increasing SoC. At this stage, we have no explanations to the unusual large values of RSEI

pos and RSEIneg observed at zero and low SoC. Remarkably, RSEI

pos and RSEI

neg demonstrated the same value at 68% SoC and higher. This phenomenon strongly suggests that the SEI layers on the surface of the positive and negative electrode materials likely have similar properties.

Regarding Rct, the charge transfer resistance of the negative electrode (Rctneg)

maintained very small values throughout the SoC range, indicative of a facile charge transfer process taking place at the negative electrode. Additionally, Rct

neg showed little fluctuation vs. SoC, evidencing that the kinetics of the charge transfer process is not affected by a change of the staged structure of Li-GICs. This result is consistent with the work of Funabili et al.4

Unlike Rctneg, the charge transfer resistance of the positive electrode (Rct

pos) not only exhibited large values, but also demonstrated a strong dependence on battery SoC. Figure 5.5 clearly shows that Rct

pos is the largest component of battery resistance. Large Rctpos

means that the positive electrode attains a sluggish charge transfer process in comparison to the negative electrode. It has been shown in Section 5.3.2 that the impedance of the positive electrode increases with decreasing electrode potential. Figure 5.5 discloses that this change is mostly due to the variation of Rct

pos. The significant decrease of Rctpos with increasing

battery SoC has been reported by other researchers.1,10,12 However, no solid interpretation has been given to this observation although Rct

pos appears to be the most influencing factor in battery resistance. Therefore, some qualitative discussion is held in the following section, in an attempt to shed some lights on this issue. 5.4 Rct The charge transfer resistance, being an important kinetic parameter for an electrochemical process, is given by: 13

0nFIRTRct = (5.1)

where Io is the exchange current (A). In equation 5.1, R, T, n and F are constants; only Io is a variable that can changes the value of Rct.

Determination of Io from I-η plots 65

5.4.1 Determination of Io from I-η plots For an electrochemical process, wherein O and R are the oxidative and reductive species, respectively,

O + ne- R (5.2) kc

ka there always exist a cathodic current (Ic) and an anodic current (Ia). kc and ka are the reaction rate constants for reduction (cathodic) and oxidation (anodic) reactions, respectively. At equilibrium when no net current is flowing, Ic must equal Ia,

Ic = Ia = Io (5.3)

where Io is an important kinetic characteristic of an electron transfer process known as the exchange current.

When current is flowing, the electrode potential deviates from its equilibrium value and an overpotential arises. At very small overpotentials, the relationship between the net current and the overpotential can be described by:14

ηRTnFII = (5.4)

where η is the overpotential (V), I the net current (A). Equation 5.4 states that the net current is linearly related to the overpotential in a narrow potential range near the equilibrium potential (Eeq). Hence, the value of Io can also be determined from the slope of the I -η plot.

In order to obtain overpotentials, the potential of the positive electrode was measured for both charging and discharging the battery at different C-rates. The recorded potential profiles are plotted in Figure 5.6 as a function of the battery SoC.

In Figure 5.6, curves obtained at currents from 0.1 to 1.0 C did not reach 100% SoC. This is because CC charging and discharging at these currents cannot fully insert and extract all charge from the battery, which has been explained in the experimental section. The mean value of the potential curves obtained at 0.02 C during charging and discharging is employed as Eeq curve. The overpotential of the positive electrode at a certain battery SoC is calculated by:

η = Epos – Eeq (5.5)

where Epos is the positive electrode potential measured at currents other than 0.02 C, Eeq the calculated equilibrium potential. Table 5.1 gives an example of the calculated η at different C-rates when the battery was at 68% SoC.

The obtained η values are plotted against I. Figure 5.7a shows, as an example, the I-η curves at 24, 45 and 68% SoC. The positive electrode exhibited the largest η for all currents


at 24% battery SoC. This result reveals that, with decreasing the battery SoC, the electrochemical reactions taking place at the positive electrode become increasingly resistive so that a larger driving force, viz η, is needed to proceed with the reactions. Noticeably, the discharging process attains larger η than charging, which is particularly pronounced at high currents and suggests that the discharging process is more sluggish than charging.

3.8

3.9

4

4.1

4.2

4.3

0 20 40 60 80 100

Epo

s. vs

. Li/L

i+ (V)

(a) SoC (%)

3.1

3.3

3.5

3.7

3.9

4.1

4.3

0 20 40 60 80 100

Epo

s. vs

. Li/L

i+ (V)

(b) SoC (%) Figure 5.6 Potential profiles of the positive electrode at different C-rates during (a) charge and (b) discharge. Following the direction of the arrows, the applied current changed from 0.02 C, via 0.05 C, 0.10 C, 0.20 C, 0.50 C to 1.0 C.

Equation 5.4 points out that I is proportional to η at small overpotentials. This situation is illustrated in Figure 5.7b. Data points at 0.05, 0.10 and 0.20 C determine a straight line that goes through the origin. The slope of this line is used to calculate Io.

Determination of Io from I-η plots 67

Table 5.1 Calculated η (V) of the positive electrode during charging and discharging at different C rates, battery at 68% SoC. Battery has a nominal capacity of 700 mAh.

0.05 C 0.10 C 0.20 C 0.50 C 1.0 C

Charging 0.006 0.017 0.030 0.053 0.087

Discharging -0.013 -0.017 -0.031 -0.079 -0.198

-0.8

-0.4

0

0.4

0.8

-0.4 -0.2 0 0.2 0.4

I (A)

η (V) (a)

-0.2

-0.1

0

0.1

0.2

-0.1 -0.05 0 0.05 0.1

ηRTnFII =I (A)

η (V)

(b) Figure 5.7 (a) I-η plots for the positive electrode at 24% (∆), 45% () and 68% () SoC and (b) illustration of determining Io at 24% SoC.

Table 5.2 lists the calculated values of Io at different battery SoC. Io increases with an increase of SoC, indicating that Rct

pos decreases with increasing SoC because Rct is


proportional to the reciprocal of Io (see equation 5.1). This is in agreement with and confirms the results of impedance measurements. Table 5.2 Calculated Io at different battery SoC. T is 298 K.

SoC slope Io (A) Rct (Ω) 24% 3.29 0.084 0.31 45% 3.58 0.092 0.28 68% 4.40 0.113 0.23 87% 4.62 0.119 0.22

5.4.2 Further discussion about Io and Rct The exchange current, Io, for the Li+ ion intercalation/deintercalation reaction across the electrolyte/electrode interface can be expressed by:1

Io = nFAk(cr)α(co)1-α (5.6)

where k denotes the rate constant for a charge transfer reaction, cr and co are the concentrations of Li+ ions in the electrode and electrolyte, respectively. The parameter α is the transfer coefficient. Substituting equation 5.6 into 5.1 leads to

2 2 1( ) ( )pos

ctr o

RTRn F Ak c cα α−= (5.7)

where k varies with electrode potential.14 co is constant at equilibrium and independent of electrode potential. cr is coupled with electrode potential. Hence, according to equation 5.7, Rct

pos is a function of A, k and cr.

0

200

400

600

800

1000

3.8 3.9 4 4.1 4.2 4.30

100

200

300

400

Cdl

(µF/

cm2 )

Rct (m

Ω)

Epos vs. Li/Li+ (V)

Figure 5.8 Cdl () and Rct (∆) of the positive electrode as a function of the electrode potential.

Further discussion about Io and Rct 69

The variation of A can be checked by measuring the double layer capacitance, Cdl, as Cdl is proportional to A. Figure 5.8 shows the value of Cdl, which is normalized with respect to the BET surface area of the positive electrode, as a function of electrode potential. The Cdl curve clearly went through a minimum value at 4.012 V, showing a similar trend as reported by Dokko et al.12 Apparently, Rct

pos and Cdl have different dependencies on the electrode potential, indicating that the variation of Rct

pos cannot simply result from the change of electrode surface area. Moreover, as reflected by Cdl, the surface area was more or less invariant, which is inconsistent with the considerable decrease in Rct

pos with increasing Epos. This observation further confirms that the surface area is not responsible for the change of Rct

pos. Equation 5.7 reads that Rct

pos is proportional to the reciprocal of (cr)α. α is less than 1 (usually around 0.5). cr can be considered decreasing from 1 for a fully discharge battery to 0.5 for a fully charged battery (see Section 2.2.1), which means 1/(cr)α increases with increasing battery SoC or Epos, leading to an increase in Rct

pos with increasing Epos. This is, however, entirely opposite to the experimental results shown in Figure 5.8. cr can therefore be excluded as a plausible explanation for the variation of Rct

pos. The remaining factor in equation 5.7 that can affect Rct

pos is k. The physical meaning of k is simply a measure of the kinetic facility of a reaction. A small k means a sluggish reaction while a large k indicates a fast process.

Many researchers noticed that the electrical conductivity of LixCoO2 increases with increasing Epos.2,8,12,15,16 As mentioned in Chapter 2, the electrical conductivity of LixCoO2 changes dramatically with composition, behaving like a metal at x ≤ 0.6 and a typical semiconductor at x = 1. Now the question is: is there any relation between k and the electrical conductivity of LixCoO2?

Herein a hypothesis is proposed to qualitatively correlate k and the electrical conductivity of LixCoO2. The positive electrode used in this work is composed of multi-layer LiCoO2 particles, as depicted in Figure 5.9. To start off with the charge transfer reaction, electrons and lithium ions must meet at the electrolyte/electrode interface. EDS analyses showed that the organic electrolyte well penetrated the entire electrode, which means that each particle was surrounded with electrolyte. This is, however, not the case for electrons. Taking discharging as an example, electrons come from the external circuit via the current collector (see Figure 5.9). In order to reach the outmost LixCoO2 layer to carry out the charge transfer reaction, some electrons have to take the path through the inner layer LixCoO2 particles. If the LixCoO2 particles are highly resistive to electrons, such as LiCoO2, it will certainly take relatively long time for those electrons to reach the outmost LixCoO2 layer. In this situation, lithium ions appear to be waiting for electrons and consequently the time needed to finish the charge transfer process is longer. The average time to charge the entire electrode is therefore longer. As a result, a large Rct

pos is observed. On the other hand, for LixCoO2 particles of high electrical conductivity, e.g. Li0.5CoO2, the charge transfer reaction can be completed in a short period of time due to the fast electronic transport and a large k can be expected, leading to a small Rct

pos.


It is postulated above that the change of k originates from the variation of LixCoO2 electrical conductivity, thus influencing Rct

pos. How to experimentally prove this hypothesis, however, remains an open question.

e-

Li+ Li+ Li+e- e-

binder

e-Li+e-

e-

inner layer

Li1-xCoO2 particlee- current collector

outmost layer

Li1-xCoO2 particle

Figure 5.9 Schematic one-dimensional representation of LixCoO2 electrode during discharging.

5.5 Summary The resistance of individual electrode has been investigated as a function of battery SoC. EIS measurements reveal that the resistance of the positive electrode (Rpos) is always larger than that of the negative electrode (Rneg) at all SoC. Rpos shows a strong dependence on SoC whereas Rneg appears to be invariant. Rpos is comparable with Rneg for a fully charged Li-ion battery. In contrast, when the battery is fully discharged, Rpos becomes so large that it dominates the battery resistance and Rneg is negligible.

Both Rpos and Rneg comprise Rs, RSEI, Rct and solid-state diffusion resistance. The values of Rs, RSEI and Rct are obtained by fitting the impedance spectra to an EQC. It has been found that RSEI

pos and RSEIneg have same values at 68% SoC and higher, strongly suggesting

that the SEI layers covering the positive and negative electrodes may have similar properties. The value of Rct

pos manifests itself as the most significant factor in Li-ion battery resistance, especially for a fully discharged battery. Remarkably, Rct

pos decreases greatly with increasing SoC. This is confirmed by I-η studies that observe an increase in Io with increasing SoC.

Rctpos plays a very important role in battery resistance, which makes it worthwhile to

understand the change of Rctpos with SoC. Based on equations for classical electrochemical

systems, it is known that A, k and cr are the three factors which affect Rctpos. Experimental

results and qualitative analyses rule out A and cr as possible causes for the change of Rctpos

with SoC while k appears to be a plausible explanation. Similar to Rct

pos, the electrical conductivity of LixCoO2 decreases with increasing Epos. It is proposed that the high electrical resistance of LiCoO2 severely impede electrons, for

References 71

instance during discharging, to transport from the current collector, via LiCoO2, to the electrolyte-electrode interface and meet lithium ions there to conduct the charge transfer reaction. Such a sluggish process yields an apparently high process resistance. On the contrary, Li0.5CoO2, corresponding to a fully charged battery, attains metal-like high electrical conductivity, which facilitates electrons movement and consequently the charge transfer process, resulting in a small process resistance. Therefore, the decrease of Rct

pos with increasing SoC can be attributed, at least partly, to the change of electrical conductivity of LiCoO2 particles. More research work is required to experimentally prove this hypothesis. References 1. M. G. S. R. Thoma, P. G. Bruce and J. B. Goodenough, J. Electrochem. Soc., 132,

1521 (1985). 2. D. Aurbach, M. D. Levi, E. Levi, H. Teller, B. Markovsky, G. Salitra, U. Heider and L.

Heider, J. Electrochem. Soc., 145, 3024 (1998). 3. P. G. Bruce and M. Y. Saidi, J. Electroanal. Chem., 322, 93 (1992). 4. A. Funabiki, M. Inaba, Z. Ogumi, S.-I. Yuasa, J. Otsuji and A. Tasaka, J. Electrochem.

Soc., 145, 172 (1998). 5. C. Ho, I. D. Raistrick and R. A. Huggins, J. Electrochem. Soc., 127, 343 (1980). 6. H. J. Bergveld, W. S. Kruijt and P. H. L. Notten, Battery Management Systems: Design

by Modelling, Kluwer Academic Publisher (2002). 7. Y.-C. Chang and H.-J. Sohn, J. Electrochem. Soc., 147, 50 (2000). 8. M. Dollé, F. Orsini, A. S. Gozdz and J.-M. Tarascon, J. Electrochem. Soc., 148, A851

(2001). 9. S. H. Glarum and J. H. Marshall, J. Electrochem. Soc., 127, 1467 (1980). 10. M. D. Levi, G. Salitra, B. Markovsky, H. Teller, D. Aurbach, U. Heider and L. Heider,

J. Electrochem. Soc., 146, 1279 (1999). 11. J. R. Macdonald and M. B. Johnson, in Impedance Spectroscopy: Emphasizing solid

materials and systems, J. R. Macdonald, Ed., Chap. 1, John Wiley & Sons (1987). 12. K. Dokko, M. Mohamedi, Y. Fujita, T. Itoh, M. Nishizawa and M. Umeda, J.

Electrochem. Soc., 148, A422 (2001). 13. A. J. Bard and L. R. Faulkner, Electrochemical Methods, Fundamentals and

Applications, 2nd ed., Chap. 3, John Wiley & Sons, Inc. (2001). 14. R. Greef, R. Peat, L. M. Peter, D. Pletcher and J. Robinson, Instrumental Methods in

Electrochemistry, Chap. 3, John Wiley & Sons (1985). 15. M. Shibuya, T. Nishina, T. Matsue and I. Uchida, J. Electrochem. Soc., 143, 3157

(1996). 16. M. Nishizawa, S. Yamamura, T. Itoh and I. Uchida, Chem. Commun., 1631 (1998).

6

Mass transport in the electrolyte of Li-ion batteries

“Substances are frequently spoken of as being electro-negative, or electro-positive, according as they go under the supposed influence of direct attraction to the positive or negative pole. But these terms are much too significant for the use for which I should have to put them; for though the meanings are perhaps right, they are only hypothetical, and may be wrong; and then, through a very imperceptible, but still very dangerous, because continual, influence, they do great injury to science, by contracting and limiting the habitual views of those engaged in pursuing it. I propose to distinguish such bodies by calling those anions which go the anode of the decomposing body; and those passing to the cathode, cations; and when I have the occasion to speak of these together, I shall call them ions. Thus the chloride of lead is an electrolyte, and when electrolysed evolves two ions, chlorine and lead, the former being an anion, and the latter a cation.”

M. Faraday, 1834

This chapter is based on the article by J. Zhou, D. Danilov and P. H. L. Notten published in Chem. Eur. J., 12, 7125 (2006).

Chapter 6 Mass transport in the electrolyte of Li-ion batteries 74

6.1 Introduction The passage of current through batteries induces ionic mass- and charge transport in the electrolyte. The modes of mass transport are:1

i. Migration: movement of a charged ion under the influence of an electric field (a gradient of electrical potential).

ii. Diffusion: movement of a species under the influence of a gradient of chemical potential (i.e. a concentration gradient).

iii. Convection: stirring or hydrodynamic transport. Generally electrolyte flow occurs because of natural convection (convection caused by density gradients) and forced convection.

For Li-ion batteries, the charged ions are Li+ cations and, in this work, PF6- anions.

Concentration gradients of Li+ cations and PF6- anions are established when current flows

through the electrolyte. Consequently, density gradients are also formed in the electrolyte. Mass transport in the electrolyte plays an important role in battery performance from a

kinetic point of view, as it affects the electrode reaction rate. In some cases, mass transport in the electrolyte can be so sluggish that the electrode reaction rate is determined by it. Bergveld et al. have simulated that the diffusion overpotential from the electrolyte in Li-ion batteries is comparable to those of the kinetic overpotentials.2 Sawai et al. have studied the factors affecting the rate capability of graphite electrodes for Li-ion batteries and have found that the dominantly influential factor is not the diffusion of Li+ ions in the graphite electrode but diffusion of Li+ ions in the electrolyte.3

To our knowledge, studies on mass transport in the electrolyte of sealed Li-ion batteries are very limited. The lack of knowledge might be due to the difficulties in in-situ measuring, for instance, the concentration gradients in sealed batteries. Indeed, some work has been done in this field.4-7 However, this was mainly based on optical methods and is not applicable to sealed battery systems because of the complexity of the measurements and the experimental equipment.

In this chapter, an electrochemical method is presented to in-situ determine the established concentration gradients in the electrolyte of sealed Li-ion batteries under galvanostatic charging and discharging conditions. Based on the measured concentration gradients, both the diffusion coefficient of Li+ ions in the electrolyte and the diffusion overpotential contributed by the electrolyte can be calculated. Consequently, the limiting (dis)charging current of the batteries investigated in this work is estimated. Moreover, the relationship between the diffusion current and the migration current of Li+ ions is also discussed.

Theory 75

6.2 Theory 6.2.1 General For prismatic Li-ion batteries, the ionic transport in the electrolyte can be assumed to be one-dimensional, which is described by the Nernst-Planck equation,1

)()()()( xvcxcD

FzxcDxJ jjj

jjjj +

∂∂

−∂

∂−=

φxRTx

−+−+

(6.1)

where Jj(x) is the flux of species j (mol/m2⋅s) at distance x from the electrode/electrolyte interface, Dj is the diffusion coefficient of j (m2/s), ∂cj(x)/∂x the concentration gradient at distance x (mol/m4), ∂φ(x)/∂x the potential gradient (V/m), zj the valence state (dimensionless) and cj the concentration (mol/m3) of species j, v(x) is the velocity (m/s) with which a volume element in solution moves along the axis. The three terms on the right-hand side of equation 6.1 represent the contribution of diffusion, migration and convection, respectively, to the ionic flux.

The electrolyte employed in this work contains LiPF6 as salt, therefore j is Li+ and PF6-.

The current Ij that is associated with the ionic transport of species j in the electrolyte is given by1

Ij = -zjFAJj(x) (6.2)

where A is the geometrical electrode surface area through which the flux passes. The total current, I, at any location in the electrolyte of a Li-ion battery consists of contributions from all species and can, with the assumption that the contribution of convection flux to the total current is negligible (see equation 6.1), be expressed by

+++= IIIII66 ,,,, PFmLimPFdLid

(6.3)

where , the diffusion-related current carried by Li+LidI ,+ cations, can be represented by

xc

FADzI LiLiLiLid ∂

∂⋅=

+

+++, (6.4)

Similarly, the diffusion-related current of PF6- anions ( ) is given by −

6,PFdI

x

cFADzI PF

PFPFPFd ∂

∂⋅=

−

−−−6

666, (6.5)

The migration-related current of Li+ cations ( ) and PF+LimI , 6- anions ( ) can be

described by

−6,PFm

I


xRTcADFz

I LiLiLiLim ∂

∂⋅=

+++

+

φ22

, (6.6)

xRT

cADFzI PFPFPF

PFm ∂∂

⋅=−−−

−

φ666

6

22

, (6.7)

respectively. Since PF6- anions are not involved in the electrode reactions in Li-ion

batteries, no net transport of PF6- anions takes place under steady state (dis)charging

conditions, which leads to

066 ,,

=+ −− PFdPFmII (6.8)

Consequently, equation 6.3 can be simplified and becomes

⎥⎦

⎤⎢⎣

⎡∂∂

⋅+∂

∂=+=

+++

++++

xRTFcz

xc

FADzIII LiLiLiLiLiLimLid

φ,,

(6.9)

Equation 6.9 indicates that the total current, I, in the electrolyte is fully carried by Li+ ions.

The electro-neutrality condition in the electrolyte states that equals . Thus the

concentration gradient for both ions must be identical, i.e.

+Lic −

6PFc

x

c

xc PFLi

∂

∂=

∂

∂ −+ 6 (6.10)

Introducing equations 6.5, 6.7 and 6.10 into equation 6.8, and considering that

leads to

−+ −=6PFLi

zz

xc

FczRT

xLi

LiLi∂

∂⋅=

∂∂ +

++

φ (6.11)

Substitution of equation 6.11 into equation 6.9 and after rearrangement ultimately yields

++

+

=∂

∂

LiLi

Li

FADzI

xc

2 (6.12)

where , F and A are constants. Taking into account that can be considered to be

concentration-independent over a given concentration range, equation 6.12 reveals that the concentration gradient of Li

+Liz +Li

D

+ ions in the electrolyte is a constant under galvanostatic steady state conditions, indicating a linear concentration gradient in the electrolyte. Based on this linearity, it is possible to electrochemically determine the concentration gradient in the electrolyte.

Theory 77

6.2.2 Relationship between concentration gradient and electrical potential The presence of a concentration gradient and/or an electrical potential gradient results in an electrochemical potential gradient. For Li-ion batteries, the electrochemical potential of Li+

ions in the electrolyte, +Liµ , is given by8

φµµ Fzaa

RT LirefLi

LiLiLi +

+

+

++ ++= ln (6.13)

where is the standard chemical potential (J/mol), and are the activities of Li+Liµ +Li

a refLi

a ++

ions and the activity in the reference state (mol/m3), respectively, and φ is the Galvani potential (V). The difference of the electrochemical potential between two different locations in the electrolyte can then be expressed by

( 212

121 ln φφµµ −+=− +

+

+

++ Fzcc

RT LiLi

LiLiLi

) (6.14)

in which the activities are replaced by concentrations of Li+ ions for simplicity. Equation 6.11 can be rewritten as

xc

FzRT

xc

cFzRT

xLi

Li

Li

LiLi ∂

∂⋅=

∂

∂⋅⋅=

∂∂ +

+

+

++

ln1φ (6.15)

After integration with respect to x, equation 6.15 becomes

2

1

21 ln+

+

+

⋅=−Li

Li

Li cc

FzRTφφ (6.16)

Substitution of equation 6.16 into 6.14 leads to

2

121 ln2

+

+

++ =−Li

LiLiLi c

cRTµµ (6.17)

The electrochemical potential is related to the measurable electrical potential, E, by the relation zFE=µ .9 Hence, the following expression is obtained

2

121

21 ln2+

+

++

++

⋅=−

=−Li

Li

LiLi

LiLi

cc

FzRT

FzEE

µµ (6.18)

Measurement of the electrical potential difference, E1 - E2, can be achieved by placing

two microreference electrodes at different locations in the electrolyte. Since , R, T and F

are all known constants, it is possible, according to equation 6.18, to evaluate the concentration difference between two microreference electrodes in the electrolyte. If the distance between two microreference electrodes is known, the concentration gradient in the electrolyte can be determined, as equation 6.12 manifests a linear concentration gradient built up in the electrolyte under galvanostatic steady state conditions.

+Liz


6.3 Experimental All batteries investigated in this chapter had a nominal capacity of 300 mAh. These 300 mAh batteries consisted of three stacks, each stack included one positive and one negative electrode. The microreference electrodes were prepared following the recipe described in detail in Chapter 4. The substrate material for the microreference electrode was a copper wire 80 µm in diameter and covered with an insulation layer.

The construction of a battery with 3 microreference electrodes is schematically represented in Figure 6.1. Note that only the stack in which the microreference electrodes were placed is depicted. These three microreference electrodes, with their tips located in the central area of the battery, were positioned at an equal interval of 5 mm. The microreference electrodes, the positive and negative electrode, were separated from each other with separators. The distance between two microreference electrodes was assumed to be the thickness of the separator(s) in-between, which is 30 µm in this work. For example, x1, the distance between microreference electrode 1 and 2, was 30 µm while x3, the distance between microreference electrode 1 and 3, was 60 µm (see Figure 6.1b). Thus, the microreference electrodes had different distances perpendicular to the electrode surface. Batteries with 2 or 4 microreference electrodes were also made for different purpose.

Microreferenceelectrode 5mm

(a) Positive electrode

5 mm

separator

xx 2

x 3 1

Negative electrode

321

(b) Figure 6.1 Schematic representation of a battery with 3 microreference electrodes. (a) top view, (b) side view. x is the distance between every two microreference electrodes, perpendicular to the electrode plane. It was assumed that x1=x2=30 µm.

Results 79

The experimental setup for the measurement of the potential differences between the microreference electrodes is shown in Figure 6.2. The batteries were charged or discharged galvanostatically with a 2300 Keithley (USA). Simultaneously, the mutual potential difference between the microreference electrodes was monitored with a computer-controlled Autolab PGSTAT 20 (Ecochemie, The Netherlands). The Autolab was operated at the general purpose electrochemical system (GPES) mode, using the galvanostatic chrono method with the current being set at zero.

Keithley

R.E. 1

R.E. 2

+ −

W.E

C.E./R.E.

Autolab

Figure 6.2 Experimental setup for the measurement of potential differences between two microreference electrodes during (dis)charging. 6.4 Results 6.4.1 Steady state

Equations 6.12 and 6.18 are derived at steady state when (dis)charging the batteries. Thus, the potential differences at steady state between two microreference electrodes should be used to calculate the concentration gradients.

Figure 6.3 shows a typical measurement of the potential differences between two microreference electrodes when discharging a 300 mAh battery with a constant current of 0.12 A from its initial 50 % SoC. Before turning on the discharging current, there was no potential difference between these two microreference electrodes. When the current was switched on, a potential step, resulting from the electrolyte resistance, occurred instantly. After a short period of time, the potential difference entered a horizontal plateau, suggesting that the ionic current flowing through the electrolyte reached its steady state. When the


discharging current was switched off, a potential step was again observed and the potential difference subsequently started to decay, corresponding to the battery relaxation towards the new equilibrium state. As clearly seen here, steady state manifests itself as a horizontal plateau in the curve of potential differences. Hereafter, all the potential differences presented in this work are mean values of each horizontal plateau.

-10

0

10

20

30

0 20 40

current on

current off

60

(Ere

f1 – E

ref2 ) (

mV

)

Time (min) Figure 6.3 The potential differences between two microreference electrodes measured when discharging a 300 mAh battery at 0.12 A.

In addition, the time needed to reach the steady state is certainly different for different (dis)charging currents. As can be expected, the higher the applied current, the longer the time to reach steady state. Such a situation is well illustrated in Figure 6.4. The three vertical lines from left to right marks the time needed to enter the horizontal plateaus at 0.03, 0.12 and 0.30 A, respectively. Obviously, the battery took the shortest time to reach steady state at 0.03 A while the longest time for 0.30 A.

-20

0

20

40

60

0 10 20 30 4

0.30 A

0.12 A

0.03 A

0

(Ere

f1 – E

ref2 ) (

mV

)

Time (min)

Figure 6.4 Measured potential differences between two microreference electrodes obtained with a 300 mAh battery at different discharging currents.

The linearity of concentration gradient 81

6.4.2 The linearity of concentration gradient In order to prove that linear concentration gradients are indeed experimentally established under steady state conditions, the experiments performed with batteries containing 3 and 4 microreference electrodes will be described herein. 6.4.2.1 3-electrode measurements One characteristic of a linear concentration gradient is that

2

32

1

21

xcc

xcc LiLiLiLi ++++ −

=−

(6.19)

where , , are the concentrations of any three locations along this concentration

gradient; x

1+Li

c 2+Li

c 3+Li

c

1 and x2 are the distances between two locations. Figure 6.5 illustrates a linear concentration gradient established when discharging the battery shown in Figure 6.1. In the special case of Figure 6.5, when x1 equals x2, equation 6.19 reduces to

3221++++ −=−

LiLiLiLicccc (6.20)

From Figure 6.5 it is clear that and . Introducing

these concentrations into equation 6.18 yields

2,121+++ ∆+=

LiLiLiccc 3,223

+++ ∆−=LiLiLi

ccc

⎟⎟⎞

⎜⎜⎛ ∆

+⋅=∆+

⋅=−+++

2

2,1

2

2,1221 1ln2ln2 LiLiLirefref

cRTccRTEE⎠⎝ ++++ LiLiLiLi cFzcFz

(6.21.1)

and

⎟⎟⎠

⎞⎜⎜⎝

⎛ ∆−⋅−=

∆−⋅=−

+

+

+++

+

+2

3,2

3,22

232 1ln2ln2

Li

Li

LiLiLi

Li

Lirefref c

cFz

RTcc

cFz

RTEE (6.21.2)

where Eref is the electrical potential of the microreference electrode. Elegantly, ⎟⎟⎠

⎞⎜⎜⎝

⎛ ∆

+

+

2

2,1

Li

Li

cc

and

⎟⎟⎠

⎞⎜⎜⎝

⎛ ∆

+

+

2

3,2

Li

Li

cc

can be determined by measuring the potential differences and

, respectively.

)( 21refref EE −

)( 32refref EE −


2,1+∆ Lic

x x

3,2+∆ Lic

Distance

)( 21refref EE − )( 32

refref EE −

+

Con

cent

ratio

n

−

Figure 6.5 Illustration of a linear co t in the electrolyte. 1+Lic , 2

+c , 3+Li

c are the

Li Li

ncentration gradien

concentrations of the electrolyte at three different locations. c and c

ifferences between two locations. x is the distance between two locations.

1

between (Eref2 – Eref

3) and (Eref1 – Eref

2) increased with the increasing current. This

phenomenon will be explained in Section 6.5.1.

Li

2,1+∆ 3,2

+∆ are the concentration

d

Figure 6.6a shows the potential differences between the microreference electrodes measured when discharging the battery with various constant currents. In Figure 6.6a, two straight lines are observed. Considering equation 6.18, it can easily be shown that linearity is expected for low concentration ratios (ln(1+x) ≈ x when x → 0). Noticeably, (Eref

2 – Eref3)

was found to be always larger than (Eref – Eref2) at all applied currents and the difference

⎟⎟⎠

⎞⎜⎜⎝

⎛ ∆

+

+

2

2,1

Li

Li

cc

and ⎟⎟⎠

⎞⎜⎜⎝

⎛ ∆

+

+

2

3,2

Li

Li

cc

Figure 6.6b. Evidently,

are calculated

according to equation 6.21 and the results are plotted in ⎟⎟⎠

⎞⎜⎜⎝

⎛ ∆

+

+

2

2,1

Li

Li

cc

and ⎟⎟⎠

⎞⎜⎜⎝

⎛ ∆

+

+

2

3,2

Li

Li

cc

are found to be essentially equal at all currents, which verifies that

21++ −

LiLicc is indeed equal to .

) )( 32++ −

LiLicc(

4-electrode measurements 83

0

3

6

9

12

15

0 0.05 0.1 0.150

3

6

9

12

15

(Ere

f1 – E

ref2 ) (

mV

)

I (A)(a)

0

0.1

0.2

0.3

0 0.05 0.1 0.15

2+

+

∆

∆

Li

Li

cc

(b) I (A)

(Ere

f2 – E

ref3 ) (

mV

)

Figure 6.6 (a) M ratios of concentrations (easured potential differences, (b) calculated 22,1++∆ cc ,

LiLi

; 23,2++ LiLi

∆ cc , ×) during discharging a 300 mAh battery as shown in Figure 6.1 at different currents.

6.4.2.2 4-electrode measurements The effect of the distance between two microreference electrodes on the measured potential differences was also examined. A battery with 4 microreference electrodes was constructed, as shown in Figure 6.7. These 4 microreference electrodes were divided into two groups and located in-between different positive and negative electrodes. Microreference electrodes 1 and 2 were separated with 4 separators while 3 and 4 were separated with 1 separator. The thickness of the separator was measured after battery assembly. It was found that the thickness of 1 separator remained 30 µm whereas the total thickness of 4 separators was 105 µm. These values were used as distances in the calculations.


The potential differences between two microreference electrodes, (Eref1 – Eref

2) and (Eref

3 – Eref4), were measured under steady state when discharging the battery at different

currents. The results presented in Figure 6.8a clearly reveal that (Eref1 – Eref

2), the potential differences between microreference electrodes 1 and 2, was larger than (Eref

3 – Eref4), the

potential differences between microreference electrodes 3 and 4 for all currents. Moreover, the differences between (Eref

1 – Eref2) and (Eref

3 – Eref4) increased with increasing current.

5mmMicroreference

electrode

(a)

x 2

Negative electrode

Positive electrode

x 1

1 separator

5 mm

3 4

4 separators1

2

(b)

Figure 6.7 Schematic representation of a battery with 4 microreference electrodes. (a) top view, (b) side view. x is the distance between two microreference electrodes, perpendicular to the electrode plane. Measured values, x1=105 µm, x2=30 µm.

Figure 6.8b shows the calculated concentration gradients, corresponding to Figure 6.8a. In contrast to the significant differences observed between (Eref

1 – Eref2) and (Eref

3 – Eref4),

the concentration gradients calculated from both (Eref1 – Eref

2) and (Eref3 – Eref

4) were almost the same, especially at small currents. According to equation 6.18, for a linear concentration gradient as shown in Figure 6.5, the larger the distance between two microreference electrodes, the larger the measured potential difference is while the calculated concentration gradient remains the same. This prediction is in line with the data presented in Figure 6.8.

Conclusively, the results shown in Sections 6.4.2.1 and 6.4.2.2 jointly prove that a linear concentration gradient is established in the electrolyte at steady state.

Concentration gradient vs. applied current 85

0

10

20

30

40

0 0.05 0.1 0.150

10

20

30

40

I (A)

0

2

4

6

8

10

0 0.05 0.1 0.15

I (A)

(Ere

f1 – E

ref2 ) (

mV

)

(Ere

f3 – E

ref4 ) (

mV

)

(a)

Con

cent

ratio

n gr

adie

nt

× 10

-6 (m

ol/m

4 )

(b) Figure 6.8 (a) Measured potential differences when discharging a 300 mAh battery as shown in Figure 5.7, (b) calculated concentration gradients between reference electrodes 1 and 2 (diamonds) and 3 and 4 (dots).

6.4.3 Concentration gradient vs. applied current Equation 6.12 manifests that the concentration gradient is proportional to the applied current. In order to check the validity of equation 6.12, the relationship between the concentration gradient and the applied current will be examined in this section.

Figure 6.9 depicts a battery with 2 microreference electrodes positioned at equal distance from the center. As the concentration of the electrolyte used in this work is 1 mol/l, it can be easily proven that the concentration at the midpoint of a linear concentration gradient is 1 mol/l due to the symmetry. Therefore, as illustrated in Figure 6.9, it is readily known that the concentration of the electrolyte where the two microreference electrodes are located can be expressed by

++ ∆+=LiLi

cc 11 (6.22.1)

++ ∆−=LiLi

cc 12 (6.22.2)


+∆Li

c can be evaluated by substituting equations 6.22.1 and 6.22.2 into equation 6.18, using

the measured potential differences between these two microreference electrodes. The distance between two microreference electrodes was known to be 30 µm. Subsequently, the

concentration gradient, as well as and , can be worked out. 1+Li

c 2+Li

c

Distance

x/2 x/2

+∆Li

c

2+Li

c

1+Li

c

0+Lic

+ −

Con

cent

ratio

n Figure 6.9 Illustration of a linear concentration gradient in the electrolyte when discharging a

battery. The dash-dotted line in the center gives the concentration in the electrolyte at the

midpoint, which is 1 mol/l in this work. Two microreference electrodes (solid dots) are placed at equal distance from the center.

0+Li

c

Figure 6.10a shows the measured potential differences under steady state conditions between two microreference electrodes for both charging and discharging the battery at various currents. The sign of the measured potential differences is an indication of the direction of the concentration gradient. The curve of charging and discharging were symmetric with respect to the zero voltage level. This result is expected, as the concentration gradients of Li+ ions in the electrolyte must have the same value but different directions for charging and discharging.

The calculated concentration gradients corresponding to Figure 6.10a are plotted in Figure 6.10b. A clear linear relation between the concentration gradient and the applied current was observed for both charging and discharging, which is in agreement with equation 6.12, showing that the concentration gradient is proportional to the applied current. The charging and discharging curves were symmetric with respect to the zero concentration gradient level, revealing the good reliability of the measurements.

Diffusion coefficient and diffusion overpotential 87

I (A) (a)

-40

-20

0

20

40

0 0.1 0.2 0.3

(Ere

f1 – E

ref2 ) (

mV

)

-20

-10

0

10

20

0 0.1 0.2 0.3

Con

cent

ratio

n gr

adie

nt

× 10

-6 (m

ol/m

4 )

I (A) (b)

Figure 6.10 (a) Measured potential differences between two microreference electrodes when charging (dots) and discharging (diamonds) a 300 mAh battery at different currents, (b) The correspondingly calculated concentration gradients. The geometric surface area of this battery was 8.9 × 10-3 m2.

6.4.4 Diffusion coefficient and diffusion overpotential Once the concentration gradient is known, the diffusion coefficient of Li+ ions in the electrolyte can be obtained from equation 6.12. As listed in Table 6.1, the calculated diffusion coefficients were almost constant over the current range from 0.03 to 0.30 A, showing only small fluctuations and values very close to what is reported in the literature.2

Figure 6.11 schematically represents the concentration gradient in the electrolyte of a

conventional Li-ion battery without microreference electrodes. is the electrolyte

concentration under open circuit conditions when there is no concentration gradient, which is represented by the horizontal dashed line. The solid line represents the concentration

0+Li

c


gradient in the electrolyte during discharging. and are the electrolyte

concentrations directly adjacent to the positive electrode surface and the negative electrode

surface, respectively. Obviously, both and deviate from . This deviation, by

the definition of Vetter,

posLi

c +negLi

c +

posLi

c +negLi

c +0

+Lic

10 results in a diffusion overpotential, ηd.

Table 6.1 in the electrolyte and η+LiD d for the individual electrode calculated with the

discharging data in Figure 6.10. T was 298 K.

I (A) 1210×+LiD (m2/s) negLic +

(mol/l) posLic +

(mol/l) negdη (mV) pos

dη (mV)

0.03 9.0 1.03 0.97 0.7 0.8

0.06 8.8 1.06 0.94 1.5 1.6

0.12 9.2 1.11 0.89 2.8 3.2

0.18 9.8 1.16 0.84 3.8 4.5

0.24 10.6 1.20 0.80 4.6 5.7

0.30 10.6 1.25 0.75 5.7 7.3

Con

cent

ratio

n negLic +

posLic +

0+Li

c

+−

Distance Figure 6.11 Schematic representation gradients in-between the positive and negative electrodes of a Li-ion battery.

of the concentration

According to Vetter10, ηd is defined by ηd = Eeq’ – Eeq (6.23)

where Eeq is the equilibrium potential in the absence of current flow and Eeq’ is the quasi equilibrium potential which forms during current flow as a result of the changed

concentrations of the species j directly adjacent to the electrode surface which are and negLi

c +

Diffusion coefficient and diffusion overpotential 89

posLi

c + here , in accordance with the Nernst equation. Hence, using equation 6.23 and the

Nernst equation, the diffusion overpotential is given by

0ln+

+

+

=Li

Li

Lid c

cFz

RTη (6.24)

where is 1 mol/l in this work; is either or . 0+Li

c +Lic neg

Lic +

posLi

c +

Comparing Figures 6.9 and 6.11 it can be seen that the distance between the positive and negative electrodes is the same as that between two microreference electrodes, for which the distance was assumed to be determined by the thickness of the separator. Considering the results described in Section 6.4.3 it can be concluded that the concentration gradient obtained with two microreference electrodes is exactly the same as that between the positive and negative electrodes in a conventional battery, as long as the current is the

same. Therefore, by comparison, and in Figure 6.11 are known to be equal to

and , respectively, in Figure 6.9 at the same current. Since and are obtainable

from equations 6.22.1 and 6.22.2, and are consequently known. Thus, η

negLi

c +posLi

c +1

+Lic

2+Li

c 1+Li

c 2+Li

c

negLi

c +posLi

c + d can be

evaluated according to equation 6.24.

The calculated values for , and ηnegLi

c +posLi

c + d are also summarized in Table 6.1. Notice

that |ηdneg| was always smaller than |ηd

pos| throughout the current range. This observation can be easily explained by equation 6.24 that reads ηd is proportional to the natural logarithm of the ratio of two concentrations. As an example, using the data at 0.30 A in Table 6.1 into equation 6.24 gives

75.01ln∝pos

dη and 125.1ln∝neg

dη

Obviously, 1/0.75 > 1.25/1, hence |ηdpos| > |ηd

neg| under discharging conditions. When the battery is under charging, |ηd

pos| will be smaller than |ηdneg|.

Moreover, Table 6.1 also shows that both ηd and the difference between |ηdpos| and

|ηdneg| were negligibly small at the lowest current. Gradually, ηd and the differences between

|ηdpos| and |ηd

neg| increased with increasing current. At 0.30 A, ηd became appreciable, especially for the positive electrode.

In Chapter 5, the total overpotential for the individual electrode are evaluated (see Table 5.1 and Figure 5.7). The contribution of ηd to η can now be estimated, which is illustrated in Table 6.2. Besides the values of ηpos from Table 5.1, ηneg is also presented. As clearly shown in Table 6.2, ηd/η is less than 10% at different currents for both the positive and negative electrodes. η in Table 6.2 was obtained at 68% SoC. It has been shown in Chapter 5 that ηpos increases with decreasing SoC. ηd

pos/ηpos can therefore be expected to decline with decreasing SoC, for ηd

pos appears to be independent of SoC. In comparison


with ηpos, ηneg varies moderately with SoC, indicating a more or less constant ηdneg/ηneg with

respect to SoC. As ηd is accountable for the electrolyte, it can be concluded that the electrolyte only makes a small contribution to η. However, it must be emphasized that this conclusion only holds over the current range studied. When the applied current approaches the limiting current, ηd will become extremely large, which is discussed in the following section. Table 6.2 Comparison of η and ηd for the individual electrodes during battery discharging.

Values of η obtained at 68% battery SoC.

ηpos (mV) ηneg (mV) ηdpos (mV)

ηdneg

(mV) ηd

pos/ηpos ηdneg/ηneg

0.1 C -17 27.5 -0.8 0.7 4.7% 2.5%

0.2 C -31 45 -1.6 1.5 5.1% 3.3%

1.0 C -198 79.4 -7.3 5.7 3.7% 7.2%

6.4.5 Limiting current A battery attains its limiting current when the electrolyte concentration at the surface of one electrode reaches zero. For the Li-ion batteries studied in this work, the electrolyte concentration is 1 mol/l. When the electrolyte concentration at the surface of, for instance, the negative electrode approaches zero, the electrolyte concentration at the surface of the positive electrode is 2 mol/l because of the mass balance of Li+ ions in the electrolyte and the linearity of the concentration gradient. Therefore, the maximum concentration gradient in this Li-ion battery is given by

75 107.6

100.32000

×=×

=∆

∆−

+

xcLi mol/m4

with a distance between the positive and negative electrode of 3.0 × 10-5 m. Thus, the limiting current is calculated, according to equation 6.12, to be 1.1 A with the data listed in Table 6.3. This limiting current corresponds to 3.7 C-rate for the 300 mAh Li-ion batteries studied in this work.

Table 6.3 Values of z, F, diffusion coefficient and the concentration gradient.

Electrode surface area (m2) z F (C/mol) +LiD (m2/s)

xcLi

∆

∆ +

(mol/m4)

8.9 × 10-3 1 96485 9.7 × 10-12 a 6.7 × 107

a mean value of diffusion coefficient from Table 6.1.

Discussion 91

It is known that a large overpotential and, consequently, a large increase in battery voltage, arise if the applied current is beyond the limiting current. We indeed observed experimentally that a fully discharged Li-ion battery reached 4.2 V instantaneously when being charged at 1.17 A (3.9 C-rate). This observation is a clear indication that the diffusion overpotential from the electrolyte becomes extremely large when the applied current is beyond the limiting current. 6.5 Discussion 6.5.1 3-electrode measurements Figure 6.6 shows that (Eref

1 – Eref2) was smaller than (Eref

2 – Eref3) at all discharging currents.

Under the conditions as illustrated in Figure 6.5, subtracting equation 6.21.2 from equation 6.21.1 leads to

⎥⎥

⎦

⎤

⎢⎢

⎣

⎡

⎟⎟

⎠

⎞

⎜⎜

⎝

⎛ ∆−=−−−

+

+

+

2

2,

3221 1ln2)()(Li

Li

Lirefrefrefref c

cFz

RTEEEE (6.25)

Note that is used in equation 6.25 because it has been proven in Section 6.4.2.1 that

.

+∆Li

c

3,22,1++ ∆=∆

LiLic

According to equation 6.25, for very small values of +∆Li

c , 2

2,⎟⎟

⎠

⎞

⎜⎜

⎝

⎛ ∆

+

+

Li

Li

cc

<< 1,

01ln2

2,

≈⎥⎥

⎦

⎤

⎢⎢

⎣

⎡

⎟⎟

⎠

⎞

⎜⎜

⎝

⎛ ∆−

+

+

Li

Li

cc

. As a result, . This situation occurs when

the applied current is small so that the concentration gradient is gentle. Data points at 0.03 A in Figure 6.6a give an example. In contrast, for a relatively large current, a steep

concentration gradient is generated. In this case,

)()( 3221refrefrefref EEEE −≈−

+∆Li

c is no longer negligible and

⎥⎥

⎦

⎤

⎢⎢

⎣

⎡

⎟⎟

⎠

⎞

⎜⎜

⎝

⎛ ∆−

+

+

2

2,

1lnLi

Li

cc

is negative, resulting in the fact that is smaller than

. Data points at 0.12 A in Figure 6.6a indeed show this case.

)( 21refref EE −

)( 32refref EE −

The assumption that +++ ∆+= LiLiLi ccc 2,1, and +++ ∆−= LiLiLi ccc 2,3, holds as long as the

concentration gradient is linear and the microreference electrodes are placed at equal


intervals. Therefore, the relationship that (Eref1-Eref

2) < (Eref2-Eref

3) can be used to check the correctness and reliability of the measurements, of course, under discharging conditions.

When the battery is being charged, the direction of the concentration gradient reverses, similarly,

echrefrefechrefref EEEE arg23

arg12 )()( −>− (6.26)

Moreover, it is expected that

echrefrefedischrefref EEEE arg23

arg21 )()( −=− (6.27)

and

echrefrefedischrefref EEEE arg12

arg32 )()( −=− (6.28)

with the precondition that the charging and discharging current is the same. The subscripts charge and discharge refer to charging and discharging the battery, respectively.

Table 6.4 Potential differences between two microreference electrodes

when discharging/charging a 300 mAh battery at 0.2 C.

21refref EE − (mV) 32

refref EE − (mV)

Discharging 5.0 < 6.0

∧* ∨

Charging 5.5 > 4.8

*: The arrow indicates the trend of two numbers predicted by equations 6.25 and 6.26. Table 6.4 lists, as an example, the measuring results with three microreference electrodes. The trend given in Table 6.4 is in line with what is predicted by equations 6.25 and 6.26. Furthermore, |Eref

1-Eref2|discharge almost equals |Eref

2-Eref3|charge, as equation 6.27 demonstrates.

Similarly, |Eref1-Eref

2|charge and |Eref2-Eref

3|discharge are also close to each other. All these results further confirm the reliability of this theory and the measurements. 6.5.2 Activity coefficient When deriving equation 6.18, the concentrations of Li+ ions were used, instead of activities for simplicity. For a concentrated electrolyte used in this work, electrolyte activity is usually

not equal to concentration. In this section, the effect of the replacement of by on

the experimental results is checked.

+Lia +Li

c

Adoption of in equation 6.18 leads to +Lia

2

1

21 ln2+

+

+

⋅=−Li

Li

Li aa

FzRTEE (6.29)

Activity coefficient 93

Using a = γ ⋅ c into equation 6.29 gives

2

1

2

121 ln2ln2

+

+

++

⋅=−−Li

Li

LiLi cc

FzRT

FzRTEE

γγ

(6.30)

where γ1 and γ2 are the activity coefficients of each concentration, respectively. The difference between equations 6.18 and 6.30 is

equation 6.30 – equation 6.18 = 2

1ln2γγ

FzRT

Li+

− (6.31)

γ is known to be a function of electrolyte concentration, which means, strictly, γ1 ≠ γ2, and consequently, equation 6.30 does not equal equation 6.18. However, over a small concentration range, γ is approximately independent of concentration, viz, γ1 = γ2. Under this condition, equations 6.30 and 6.18 are equal. This suggests that, for instance in Figure 6.10b, small concentration gradients can be considered as calculated by equation 6.30 whereas large concentration gradients are calculated, as it is, according to equation 6.18.

0

5

10

15

20

25

0 0.1 0.2 0.3 0.4

Con

cent

ratio

n gr

adie

nt

× 10

-6 (m

ol/m

4 ) I (A)

Figure 6.12 Concentration gradients at different currents, using the discharging data in Figure 6.10.

Based on the above analyses, small concentration gradients obtained at 0.03 and 0.06 A in Figure 6.10b are regarded as calculated by equation 6.30. Concentration gradients at 0.03 and 0.06 A, together with the origin, determine a straight line, which is represented by the solid line in Figure 6.12. The dashed line in Figure 6.12 is determined by data points at currents higher than 0.06 A. At 0.12 A (0.4 C), the concentration gradient was still on the solid line. Beyond 0.12 A, the calculated concentration gradients started to deviate negatively from the solid line and clearly showed that the deviation increased with

increasing current. This deviation can be attributed to the replacement of by ,

strongly suggesting that equation 6.18 is likely not suitable for currents higher than 0.30 A (1.0 C), otherwise too much error is introduced to the experimental results.

+Lia +Li

c

Furthermore, the observation that the dashed line was below the solid line corresponds to a positive value of equation 6.31, indicating that γ1 < γ2. Bearing in mind that in this


discussion , it can be concluded from γ21++ >

LiLicc 1 < γ2 that γ decreases with increasing

electrolyte concentration.

6.5.3 Relationship between and +LidI , +LimI ,

Introducing equation 6.5 into equation 6.8 yields

x

cFADzI PF

PFPFPFm ∂

∂⋅−=

−

−−−6

666, (6.32)

Substitution of equation 6.10 into equation 6.32 and using −+ =−6PFLi

zz leads to

xc

FADzI LiPFLiPFm ∂

∂⋅=

+

−+−66,

(6.33)

One important parameter related to the migration current is the transference number, tj, defined as the fraction of the total migration current that a given ion j carries.1 The transference number of Li+ cation is given by

−+

+

+

+=

,,

,

PFmLim

LimLi II

It

6

(6.34)

or

−+

+

−−−+++

+++

+

+=

+=

6666PFLi

Li

PFPFPFLiLiLi

LiLiLiLi uu

u

cuzcuz

cuzt (6.35)

with and −+ =6PFLi

cc 16

== −+ PFLizz , where and are the mobilities of Li+Li

u −6PF

u + cation

and PF6- anion, respectively. The mobility of species j, uj, is defined as the limiting velocity

of the ion in an electric field of unit strength, m2/s⋅V. Substituting equation 6.33 into equation 6.34 and after rearrangement gives

xc

FADzt

tI Li

PFLiLi

LiLim ∂

∂⋅⋅

−=

+

−+

+

+

+61, (6.36)

Dividing equation 6.4 by equation 6.36 gives

−

+

++

+

⋅⎟⎟⎠

⎞⎜⎜⎝

⎛−=

6

11

,

,

PF

Li

LiLim

Lid

DD

tI

I (6.37)

Conclusions 95

Substitution of equation 6.35 into equation 6.37 yields

+

+

−

−

+

+

⋅=Li

Li

PF

PF

Lim

Lid

uD

D

u

I

I

6

6

,

, (6.38)

which can be further reduced to

1,

, =+

+

Lim

Lid

I

I (6.39)

by using the Einstein-Smoluchowski equation,1 RT

Fz

Du j

j

j = .

Equation 6.39 demonstrates that the diffusion-related current equals the migration-related current in Li-ion batteries at steady state. A quicker route to this conclusion is to write down equation 6.12 as

xc

FADzI LiLiLi ∂

∂=

+

++2 (6.40)

Notice that the right-hand side of equation 6.40 is actually the description of (see

equation 6.4), thus giving , viz, the total current, I, is two times the diffusion-

related current, . By using into equation 6.9, it also leads to

+LidI ,

+= LidII ,2

+LidI , += LidII ,2

++ = LimLid II ,, (6.41)

Equation 6.41 has been confirmed by simulation.11 To experimentally verify it, however, appears to be very difficult. Nevertheless, the correctness and reliability of equation 6.12 has been experimentally verified. Thus equation 6.41, being a derivation of equation 6.12, should also be correct. 6.6 Conclusions It has been theoretically proven that the concentration gradients in the electrolyte of sealed Li-ion batteries are linear under galvanostatic steady state conditions. Based on this finding, an in situ electrochemical method has been developed to determine quantitatively concentration gradients in the electrolyte by measuring the potential difference between two microreference electrodes placed in-between the positive and negative electrodes. Experimental results confirmed the linearity of the concentration gradients and proved that the concentration gradient was proportional to the applied current under galvanostatic steady state conditions.


The diffusion coefficient of Li+ ions in the electrolyte has been determined and was found to be almost constant at different currents. The diffusion overpotential related to the concentration gradient was calculated to be appreciable at high charging/discharging currents. Based on these obtained values, the limiting current of the batteries studied in this work was estimated to be 1.1 A, corresponding to 3.7 C-rate. This limiting current will guide to operate the batteries more efficiently.

During theoretical derivation, the activity of the electrolyte was replaced by the electrolyte concentration. The impact of this replacement on the experimental results was analyzed and found to be minor in the given current range. The activity coefficient of the electrolyte appeared to decrease with the increase of electrolyte concentration. It is suggested not to apply this method for currents higher than 1 C-rate.

The method presented in this chapter proves to be very useful for a better understanding of mass transport in the electrolyte of Li-ion batteries. The extracted ηd from the electrolyte provides quantitative information about the total battery overpotential thereby helping to advance the operation of Li-ion batteries. All the information obtained with this method can be used in battery modeling work, which nowadays forms a key part of battery management systems.

Finally, it should be pointed out that, in principle, this method can also be applicable to other battery systems with appropriate modification. References 1. J. Bard and L. R. Faulkner, Electrochemical Methods, Fundamentals and Applications,

2nd ed., Chap. 4, John Wiley & Sons, Inc. (2001). 2. H. J. Bergveld, W. S. Kruijt and P. H. L. Notten, Battery Management Systems: Design

by Modelling, Chap. 4, Kluwer Academic Publisher (2002). 3. K. Sawai and T. Ohzuku, J. Electrochem. Soc., 150, A674 (2003). 4. Y. Awakura and Y. Kondo, J. Electrochem. Soc., 123, 1184 (1976). 5. R. E. Russo, F. R. McLarnon, J. D. Spear and E. J. Cairns, J. Electrochem. Soc., 134,

2783 (1987). 6. J. K. Weaver, F. R. McLarnon and E. J. Cairns, J. Electrochem. Soc., 138, 2572 (1991). 7. A. Eklund and R. I. Karlsson, Electrochimi Act., 37, 681 (1992). 8. W. S. Kruijt, H. J. Bergveld and P. H. L. Notten, J. Electrochem. Soc., 145, 3764

(1998). 9. J. Newman and K. E. Thomas-Alyea, Electrochemical Systems, 3rd ed., John Wiley &

Sons, Inc., p.44 (2004). 10. K. J. Vetter, Electrochemical Kinetics, Theoretical and Experimental Aspects, S.

Bruckenstein and B. Howard, Trans. Eds., Academic Press, p.157 (1967). 11. D. Danilov and P. H. L. Notten, Ageing of Li-ion batteries: mathematical description,

207th Electrochemical Society Meeting, May, Quebec City, Canada (2005).

7 Capacity fade

of Li-ion batteries

Chapter 7 Capacity fade of Li-ion batteries 98

7.1 Introduction Rechargeable batteries unavoidably experience irreversible capacity loss during cycling. This capacity loss is often referred to as capacity fade or degradation and is tightly associated with battery cycle life. Cycle life is an important issue for battery development and application. One major task for researchers working in this field is to prolong the battery cycle life. Besides, accurate prediction of battery life is a key point for any battery application, which is still a great technical challenge. To fulfil these targets fundamental understanding of the mechanism of capacity fade is required.

Numerous researchers have studied the capacity fade in Li-ion batteries.1-6 In general, the possible causes for capacity fade can be classified in two categories:

i. Loss of active material: this can stem from phase changes of active material to inactive material,7 active material dissolution,8 lithium metal deposition,9 loss of contact between active material and current collector, loss of contact between particles of active material.1 Any loss of the positive electrode material means the loss of recyclable Li+ ions. The loss of the negative electrode material results in a decreasing capacity to host Li+ ions, which may consequently lead to lithium metal deposition.

ii. Increase of the overall battery impedance: a high battery impedance can shorten the available discharge time (and capacity) upon cycling.10

This chapter is devoted to the capacity fade of the Li-ion batteries studied in this thesis. The above-mentioned possible causes will be examined. The objective herein is twofold: to identify which electrode, the positive or the negative, contributes mostly to the capacity fade, and to figure out the contribution of the loss of active material and increase in battery impedance to the battery capacity fade. This investigation was carried out mainly under normal operating conditions with microreference electrodes. The batteries were operated at room temperature and were cycled between 3.0 and 4.2 V. 7.2 Experimental The Li-ion batteries studied in this chapter were cycled with the Maccor 4000 automatic cycling equipment. A CCCV cycling regime was adopted. During the CC part the batteries were charged at a constant current until 4.2 V. In the CV mode the voltage was kept at 4.2 V until the cut-off current of 0.05 C was reached. Discharge was conducted galvanostatically at different C-rates with a cut-off voltage of 3.0 V. The rest time between charge and discharge was 30 min. Basically, the batteries were cycled at 0.5 C, but in every 10 cycles there was one cycle at 0.2 C. In this chapter, one cycle at, for instance, 0.5 C means that the battery was CCCV charged at 0.5 C and discharged galvanostatically at 0.5 C. The potential plots of both the positive and negative electrodes measured with the microreference electrodes were recorded during cycling by means of the auxiliary inputs of the Maccor equipment.

When cycling was finished, some batteries were discharged at 0.02 C to 3.0 V twice with a rest of 3 hours in-between each discharge. This was to ensure that the batteries were

Battery capacity fade 99

indeed fully discharged. These fully discharged batteries were dismantled in an Ar-filled glove box. The electrodes were washed in DMC for 24 hours and then dried in the glove box.11 Each cycled positive electrode was reassembled with a pristine negative electrode to construct a new battery. Similarly, batteries each consisting of a cycled negative electrode and a pristine positive electrode were also constructed. These batteries, after being sealed in air-tight Al-polymer package and filled with electrolyte, were subjected to capacity tests.

The EIS measurements were carried out with an Autolab PGSTAT 20, using the Frequency Response Analyser (FRA) mode and under potentiostatic control. The impedance spectra were obtained by applying a sine wave of 5 mV in amplitude in the frequency range of 10 kHz to 10 mHz.

The energy dispersive spectroscopy (EDS) was obtained with a Philips XL 40 FEG scanning electron microscope (SEM) attached with an EDAX Genenis 2000. The accelerating voltage was 10 kV.

The X-ray diffraction (XRD) measurements were performed with a Philips X’Pert MPD diffractometer, equipped with a Cu X-Ray source and a monochromator in the diffracted beam.

Transmission electron microscopy (TEM) samples were made from both the pristine and cycled positive electrodes. Pieces of the positive electrode were ultrasonically de-agglomerated in acetone. Droplets of the suspension were dispersed on an amorphous carbon film supported by a standard TEM Cu grid. The samples were analyzed by an FEI Tecnai F30ST Transmission Electron Microscope operated at 300 kV. Oxide particles were stable during the observation under electron beams. 7.3 Battery capacity fade Li-ion batteries suffer from degradation upon cycling. This effect is illustrated in Figure 7.1 by plotting the discharge capacity of a Li-ion battery against the cycle number. In this case, the battery was basically cycled at 0.5 C, but for every 10 cycles there was one cycle at 0.2 C, which is marked by the open triangles in Figure 7.1. Two trends are clearly observed here. One is that the battery capacity gradually decayed with increasing cycle number, regardless of the different discharge currents. The other is that the battery attained a larger capacity at 0.2 C than at 0.5 C. The second observation demonstrates the impact of the overall battery internal resistance on the discharge capacity.

As explained in the Introduction, the battery capacity fade originates from the loss of active material and the increase in overall battery internal resistance. It is possible to quantify the capacity fade caused by the loss of active material if the impact of the overall battery internal resistance can be excluded. Figure 7.1 suggests that using an extremely low current can minimize the effect of the overall battery internal resistance on battery capacity. In this study, a current of 0.02 C is assumed to be sufficiently low to rule out the influence of the overall battery internal resistance and therefore, the battery capacity obtained at this current is considered to be the maximum capacity of the active materials.

The maximum capacities of the battery before and after cycling were determined by one cycle at 0.02 C. The discharge capacity obtained at different cycle numbers and


different discharging C-rates are compared in Figure 7.2. The discharge capacity determined at 0.02 C before and after cycling showed a decrease of 27 mAh. This capacity loss, accounting for 13% of its original capacity, represents the capacity fade stemming from the loss of active materials.

0

50

100

150

200

250

0 200 400 600

D

isch

arge

cap

acity

(mA

h)

Cycle number Figure 7.1 Discharge capacity vs. cycle number. Battery had a nominal capacity of 200 mAh and was cycled at 0.5 C (), every 10 cycles there was 1 cycle at 0.2 C (∆).

0

100

200

300

176 164

132

200 190 203

Dis

char

ge c

apac

ity (A

h)

1st cycle

0.2 C 2nd cycle

0.5 C 510th cycle

0.5 C 511th cycle

0.2 C bef cycle 0.02 C

aft cycle 0.02 C

Figure 7.2 Discharge capacities of a 200 mAh battery at different cycle numbers and discharging C rates.

Apart from the loss of active material, the battery capacity fade is strongly dependent

on the discharge current, which can also be seen in Figure 7.2. After 510 cycles, the discharge capacity at 0.5 C was 132 mAh, 32 mAh smaller than that at 0.2 C. Note that at the beginning of cycling, the capacity difference between 0.2 C and 0.5 C was only 10 mAh. This increase in capacity difference between 0.2 C and 0.5 C upon cycling is a clear indication of the increase in battery internal resistance.

Loss of active materials 101

Conclusively, at the end of cycling, the active material lost 13% of its original capacity while the battery developed an increasing internal resistance upon cycling. This resistance rise will be a key issue in high power applications. The following part of this chapter will focus on these two aspects: the loss of active material and the increase in battery internal resistance. 7.4 Loss of active materials The loss of active material can result from either the positive or negative electrodes, or both of them. This is the first question to be answered in this section. Afterwards, the possible causes to the loss of active material will be examined. 7.4.1 Loss of the active materials in the positive and negative electrodes

In Li-ion batteries, the positive electrode acts as the source of Li+ ions while the negative electrode hosts Li+ ions coming from the positive electrode upon charging. Apparently, the loss of positive electrode material directly leads to the loss of battery capacity. The loss of the negative electrode material can result in insufficient capacity of the negative electrode to contain Li+ ions. Consequently, lithium metal deposition can take place on the negative electrode during charging. This undesirable effect not only consumes recyclable Li+ ions but also causes serious safety problems.

Cycled graphite electrode Pristine graphite electrode

separator separator

Pristine LiCoO2 electrode Cycled LiCoO2 electrode

(a) (b)

Figure 7.3 Different battery configurations

In order to find out the capacity change of the individual electrode upon cycling, some

cycled batteries were dismantled. The cycled positive electrode was reassembled with a pristine negative electrode to make a new battery (see Figure 7.3a). Similarly, batteries were also made up of cycled negative electrodes and pristine positive electrodes (see Figure 7.3b). The objective of this experiment was to compare the discharge capacity of these newly made batteries. The results are summarized in Figure 7.4.

All the discharge capacities in Figure 7.4 were obtained at 0.2 C. The capacity of the original battery dropped from 100 mAh at the first cycle to 82 mAh after more than 500


cycles. The battery was then dismantled to make two new batteries. The battery consisting of the cycled positive electrode and a pristine negative electrode attained a capacity of 77 mAh that was close to the 82 mAh of the original battery at the end of cycling. Undoubtedly, the cycled positive electrode lost part of its capacity so that the newly made battery can only deliver 77 mAh. The difference between the 77 mAh and 82 mAh was always observed with all the batteries tested in these experiments. The reason remains unclear.

0

0.05

0.1

0.15

100

7782

100

Dis

char

ge c

apac

ity (A

h)

1st cycle 511th cycle cycled graphite + pristine LiCoO2

cycled LiCoO2

+ pristine graphite

Figure 7.4 Discharge capacities of the original battery and the newly made batteries at 0.2 C.

The battery comprising the cycled negative electrode and pristine positive electrode

demonstrated a capacity of 100 mAh which was exactly the same as the capacity the original battery exhibited at the first cycle. Obviously, the cycled negative electrode is still able to host all the Li+ ions from the pristine positive electrode; otherwise the 100 mAh capacity cannot be reached. It should be pointed out that this result does not necessarily mean that the negative electrode dose not lose any active materials at all. Many researches have shown that graphite electrode lost capacity upon cycling.12,13 However, as the role of the negative electrode is to store Li+ ions, the negative electrode can be considered not to contribute to the loss of active material as long as it is still capable of containing all Li+ ions coming from the positive electrode.

The results presented in Figure 7.4 demonstrate that the positive electrode indeed loses capacity upon cycling, which results in battery degradation. Meanwhile, the negative electrode retains its capacity of hosting Li+ ions even after long-term cycling. 7.4.2 Possible causes for the loss of the positive electrode material The positive electrode provides the recyclable Li+ ions in the Li-ion battery. Any processes that consume recyclable Li+ ions can lead to the loss of the positive electrode material.

Possible causes for the loss of the positive electrode material 103

These processes include: i. The formation of the SEI layer,

ii. Lithium metal deposition on the negative electrode, iii. Dissolution of the positive electrode material, iv. Loss of contact between electrode material and current collector, loss of contact

between particles of electrode material, v. Phase change of active material to inactive material.

In this section, these factors will be examined one by one in a qualitative way. (i) It is commonly accepted that the formation of the SEI layer consumes recyclable Li+

ions. In reality, Li+ ions will be continuously consumed during cycling due to some unavoidable parasitic reactions such as continuing surface film formation reaction. As this process is inevitable during battery operation, the discussion about the loss of the positive electrode material due to the SEI layer is skipped. It must be clarified that the formation and growth of the SEI layer takes place on both the positive and negative electrodes. In this sense, the negative electrode is responsible for the loss of recyclable Li+ ions as well.

(ii) Lithium metal deposition happens when the negative electrode is not capable of hosting all the Li+ ions from the positive electrode. This condition, however, does not hold for the batteries investigated in this work because Section 7.4.1 has clearly shown that the negative electrode always has sufficient capacity to store Li+ ions.

Another possible cause for lithium metal deposition is that Eneg drops below the lithium metal deposition potential. In principle, the potential of a fully lithiated graphite electrode is still above 0 V vs. Li/Li+ (see Section 2.2.2), indicating that lithium metal deposition cannot take place from a thermodynamic point of view. Kinetically, however, under high-current charge conditions, the negative electrode can be polarized to such an extent that lithium metal dentrite is formed on the surface of the electrode particles. This concern can be easily checked by measuring Eneg during battery operation.

The potential of the individual electrode was monitored during cycling with the microreference electrode. The potential profiles of the positive and negative electrodes in one cycle are presented in Figure 7.5. It is clear that Eneg was always above 0 V during cycling, demonstrating that lithium metal deposition did not occur under these conditions. Hence, for the batteries cycled at 0.5 C, lithium metal deposition is not a cause for the loss of recyclable Li+ ions.

(iii) It has been noticed by some researchers8,14 that the positive electrode material can dissolve in the electrolyte. For Li-ion batteries employing spinel LiMn2O4 as positive electrode, Mn dissolution in the electrolyte appears to be a serious problem and takes great responsibility for capacity fade both on cycling and storage.14 In contrast to LiMn2O4, LiCoO2, the positive electrode used in this work, does not suffer much from active material dissolution. As reported in the literature,8 when LiCoO2 electrode is cycled up to 4.3 V and above, LiCoO2 dissolution becomes appreciable and cobalt can be detected on the surface of the negative electrode. In order to check this point, the negative electrodes after 510 cycles were analyzed by energy dispersive spectroscopy (EDS).

The EDS spectra of the cycled and pristine negative electrodes are compared in Figure 7.6. As expected, there was no cobalt signal on the EDS spectrum of the pristine negative


electrode. Meanwhile, no cobalt was detected on the cycled negative electrode either. This result strongly indicates that LiCoO2 dissolution is not an appreciable process under these cycling conditions. Checking Figure 7.5 it can be found that the maximum Epos was 4.26 V. This voltage is consistent with the conventional limit of reversibility of 4.3 V for LiCoO2 and, according to Amatucci et. al,8 is not sufficient to dissolve LiCoO2 in the electrolyte. Hence, the dissolution of LiCoO2 hardly contributes to the loss of the positive electrode material during cycling. Aurbach et al. reported similar results on LiCoO2 electrode.15

2

2.5

3

3.5

4

4.5

0 100 200 300 4000

0.2

0.4

0.6

0.8

1

disch. charge

(Ene

g– E

ref)

(V)

(Epo

s– E

ref)

(V)

Time (min)

Figure 7.5 Potential profiles of the positive and negative electrodes during one cycle at 0.5 C.

0 0.2 0.4 0.6 0.8 1

C

Co F O

Inte

nsity

(a.

u.)

keV Figure 7.6 EDS spectra of a pristine (dashed grey curve) and a cycled (black curve) graphite electrode. The straight line marks the position of the cobalt signal.


(iv) The electrode materials in Li-ion batteries are intercalation compounds. When the battery is operating, Li+ ions are reversibly inserted into and extracted from the electrodes. As a result, the electrode materials expand during intercalation and shrink during deintercalation. This continuous volume expansion and contraction can weaken the adhesion of the electrode material particles to the current collector and between the particles. Eventually, after long-term cycling some particles may lose contact with the current collector and other particles. These “isolated” particles cannot be utilized any more during cycling, which leads to capacity loss.1,16

The loss of contact between the active material and the current collector and the loss of contact between the particles of the active material sounds very possible. However, to experimentally prove it appears to be difficult. One possible method is to press the cycled electrode and afterwards re-determine the capacity of the battery.17 This method is based on the idea that pressing the electrode under appropriate pressure can resume the contact of the “isolated” particles with the current collector and other particles, thus regaining some capacities. Table 7.1 The thickness of the positive electrode.

Sample Thickness (µm)

Pristine single-sided positive electrode 110

Single-sided positive electrode after 500 cycles 120

Pressed cycled single-sided positive electrode 110

The volume change of the electrode material can be reflected by the change of the thickness of the electrode. As shown in Table 7.1, the thickness of a pristine single-sided positive electrode was 110 µm and increased to 120 µm (~ 9%) after 500 cycles, showing a clear volume expansion upon cycling. After being subjected to a pressure of 4 bar, the thickness of the cycled positive electrode was reduced to its original value. The pressed positive electrode was assembled with a pristine negative electrode to fabricate a new battery. Meanwhile, a battery consisting of an un-pressed cycled positive electrode and a pristine negative electrode was also constructed. The capacities of these two batteries were determined consecutively at 0.2, 0.5, 1.5 and 0.2 C.

The discharge capacities of the two batteries at 0.2 and 0.5 C were equal, as Figure 7.7 demonstrates. When the discharge current increased to 1.5 C, the battery consisting of the pressed cycled positive electrode attained a larger capacity. As the current dropped back to 0.2 C, the two batteries again delivered identical capacity. The results in Figure 7.7 show that the batteries made up of pressed and un-pressed cycled positive electrodes only had differences in internal resistance, which is shown by the different discharge capacities at 1.5 C. On the other hand, it appears that the capacity of the cycled positive electrode cannot be regained by pressing the electrode, as the two batteries showed identical capacity at 0.2 C. This indicates that the cycled positive electrode does not have any “isolated” particles. Based on these analyses, it can be concluded that, for the cycled positive electrode, the


contact between the particles and the current collector and the contact between the particles remained unaffected after long-term cycling and is not a plausible cause for the loss of the positive electrode.

0

0.01

0.02

0.03

0.04

0.05

0 2 4 6 8 10 12 1

1.5 C

0.2 C 0.5 C 0.2 C

4

Dis

char

ge c

apac

ity (A

h)

Cycle number

Figure 7.7 Discharge capacities of two batteries at different C rates. The batteries consisted of cycled positive electrode and pristine graphite electrode. One cycled positive electrode was pressed () under 4 bar while the other was not ().

(v) As mentioned in Section 2.2.1, a delithiated LixCoO2 (x < 1) electrode is metastable

and can undergo phase change to an electrochemically inactive structure. For a Li-ion battery in operation, the positive electrode is, most of the time, delithiated and thus very likely to experience phase transformation. In order to check the possible phase change upon cycling, XRD measurements were carried out with the positive electrodes.

10 20 30 40 50 60 70 80

(101)

C

(104)

(003)

(b)

(a)

Inte

nsity

(a.

u.)

2 θ Figure 7.8 XRD patterns for the positive electrode cycled at 0.5 C, (a) pristine, (b) 600 cycles.


The XRD patterns of the positive electrode material before and after cycling are presented in Figure 7.8. The peak due to the graphite component in the positive electrode is seen. Comparing the XRD patterns of the pristine and the cycled positive electrodes, it can be noted that the characteristic peak positions of the LiCoO2 phase remain unaltered, showing that no new phase has been formed. However, the relative peak intensities have been substantially changed, indicating a change in the structure of the LiCoO2 phase. As can be seen in Table 7.2, the relative intensity of the main peak (003) decreased for the cycled positive electrode. In the LiCoO2 phase, alternate layers of Li and Co cations occupy the octahedral sites of a compact cubic close packing of oxide anions. A decrease in (003) peak intensity would occur when a cobalt atom occupies part of the octahedral sites of the lithium layer.18,19 This indicates that the cation in the well-layered LiCoO2 structure becomes disordered and that a portion of the Li+ ions in the positive electrode becomes inactive.20

Table 7.2 Change in intensity ratio between (003) and (101) plane of the positive electrode material.

Pristine electrode Cycled electrode

Intensity of (003) peak 6555 3675

Intensity of (101) peak 900 918

I(003)/I(101) 7.3 4.0

2.5

3

3.5

4

4.5

0 30 60 90 120 1500

0.3

0.6

0.9

1.2

1.5

(Ene

g –

Ere

f) (V

)

(Epo

s – E

ref)

(V)

Time (h)

Figure 7.9 The potential profiles of the positive and negative electrodes before (gray) and after (black) cycling obtained at 0.02 C.

Figure 7.9 shows the potential profiles of the positive and negative electrodes cycled at

0.02 C. Before cycling, the potential profile of the positive electrode clearly revealed plateaus during charging and discharging (grey curves in Figure 7.9), which features the phase transition processes upon lithium intercalation and deintercalation (see Section 4.4.7). For the cycled positive electrode, however, no plateaus were observed (black curves in


Figure 7.9), suggesting that no two-phase coexistence region existed in the cycled positive electrode. This observation is in line with the XRD patterns that reveal the disordering takes place in the cycled positive electrode.

As a comparison, the potential profile of the cycled negative electrode still displayed plateaus at given potentials, which was almost identical to the potential profile of the negative electrode before cycling. Obviously, the graphite electrode maintains its structure even after long-term cycling.

Summarily, this section analyzes the possible causes for the loss of the positive electrode material. It has been found that, under the specific conditions, viz, cycling at 0.5 C between 3.0 V and 4.2 V, no lithium metal deposition takes place on the negative electrode. Also, the dissolution of LiCoO2 in the electrolyte during cycling is not feasible. In addition, the loss of contact between LiCoO2 particles and current collector and the loss of contact between LiCoO2 particles was not detected. On the other hand, structure disordering, which makes some Li+ ions in LiCoO2 inactive, was probed by XRD measurements. This effect, together with the consumption of Li+ ions in the SEI layer, is assigned to result in the loss of the positive electrode material upon cycling. 7.5 Increase in battery impedance 7.5.1 Impedance of the positive and negative electrodes One cause for battery degradation is the increase of the internal resistance upon cycling. A high internal resistance shortens the time for the battery to reach the cut-off voltage during discharging, thus reducing the discharge capacity. EIS measurements reveal the remarkable increase in battery impedance upon cycling. After more than 600 cycles, it was found that the impedance became 6 times larger.

With the help of microreference electrodes, the impedance development of the positive and negative electrodes was separately monitored during cycling. The positive electrode demonstrated a considerable increase in impedance upon cycling, which is shown in Figure 7.10a. In contrast, the impedance of the negative electrode exhibited only little variation with respect to cycle number (see Figure 7.10b). Surprisingly, the impedance of the negative electrode started to decrease after 204 cycles. This phenomenon is yet not well understood. Nevertheless, the impedance spectra of the individual electrodes help to understand the source of the increase in battery impedance.

To further clarify this issue, Figure 7.11 compares the impedance spectra of the positive and negative electrodes after 608 cycles. The impedance of the negative electrode was so small that it was barely visible. The total battery impedance almost entirely stemmed from the positive electrode. Undoubtedly, the increase in battery impedance originates from the increase in impedance of the positive electrode. Besides its contribution to the loss of active material as discussed in Section 7.4.1, the positive electrode also plays a decisive role in the battery impedance. The following section will therefore concentrate on the impedance of the positive electrode.

Impedance of the positive and negative electrodes 109

Zre (Ω)

0

0.2

0.4

0.6

0.8

0 0.2 0.4 0.6 0.8 1 1.2 1.4

0

0.02

0.04

0.06

0.08

0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16

0

0.05

0.1

0 0.05 0.1 0.15

-Zim

(Ω)

Zre (Ω)

-Zim

(Ω)

(a)

-Zim

(Ω)

(b) Zre (Ω) Figure 7.10 Impedance spectra of (a) the positive electrode and (b) the negative electrode at 1st cycle (solid line), 204th cycle (+), 406th cycle (∆) and 608th cycle () measured with a fully charged battery. The insert enlarges the impedance spectrum of the positive electrode at 1st cycle.

0

0.2

0.4

0.6

0.8

0 0.2 0.4 0.6 0.8 1 1.2 1.4

0

0.03

0.06

0 0.05 0.1

-Zim

(Ω)

Zre (Ω)

-Zim

(Ω)

Zre (Ω) Figure 7.11 Impedance spectra of the battery (), the positive (∆) and the negative electrodes (solid line) at 608th cycle measured with a fully charged battery. The insert magnifies the impedance spectrum of the negative electrode.


7.5.2 Increase of the positive electrode impedance The impedance spectra of the positive electrodes were fitted with an equivalent circuit (see Figure 5.4) to extract the values of Rs

pos, RSEIpos and Rct

pos at different cycle numbers. As expected, the value of Rs

pos was invariant with cycle number, which is revealed in Figure 7.12. The value of RSEI

pos displayed a gradual increase with increasing cycle number and became more or less stable after 700 cycles. As discussed, RSEI

pos represents the resistance for Li+ ions to migrate through the SEI layer that covers the surface of the positive electrode material. Since the thickness of the SEI layer keeps growing during cycling (see Section 2.2.4), it is easy to understand that RSEI

pos increases upon cycling. Rctpos is always of great

interest as it reflects the kinetics of the charge transfer process. In this case, the value of Rct

pos appeared to increase exponentially during cycling.

0

0.3

0.6

0.9

1.2

0 200 400 600 800 1000

R (Ω

)

Cycle number Figure 7.12 The development of Rs

pos (), RSEIpos () and Rct

pos (∆) upon cycling, data extracted from the impedance spectra measured at battery fully charged state.

0

20

40

60

80

100

0 200 400 600 800 1000

R/R

sum

(%)

Cycle number Figure 7.13 The percentage of Rs

pos (), RSEIpos () and Rct

pos (∆) in the total impedance (Rsum) upon cycling, data from Figure 7.12.

Increase of the positive electrode impedance 111

Figure 7.13 compares the percentage of Rspos, RSEI

pos and Rctpos with respect to Rsum that

is the summation of Rspos, RSEI

pos and Rctpos. At the beginning of cycling, Rs

pos accounted for about 20% to Rsum, but quickly dropped to a negligible level as cycling progressed. This is because RSEI

pos and, particularly, Rctpos increased considerably upon cycling. The percentage

of RSEIpos went through a maximum value around 400 cycles and then started to decline. In

contrast to Rspos and RSEI

pos, the percentage of Rctpos continued to increase with increasing

cycle number. As clearly shown in Figure 7.13, Rctpos became increasingly influencing in

Rsum upon cycling. It has been shown in Section 5.4.2 that Rct

pos is a function of the surface area of the positive electrode. A decrease in surface area during cycling will result in an increase in Rct

pos. The surface areas of the pristine and cycled positive electrodes were first determined by BET (Brunauer-Emmett-Teller) measurements, showing that the surface area did not change upon cycling (see Table 7.3). On the other hand, similar to Rct

pos, Cdl of the positive electrode (Cdl

pos) can be evaluated by fitting the impedance spectra with EQC. The obtained values of Cdl

pos are normalized with respect to the BET surface area and plotted in Figure 7.14. Being more or less independent of the cycle number, the values of Cdl

pos indicate that the surface area indeed did not vary upon cycling as the surface area is proportionally linked to Cdl

pos. Thus, both BET and Cdlpos measurements confirm that the increase in Rct

pos was not due to the change of surface area.

Table 7.3 BET surface areas of pristine and cycled double-sided (D.S.) LiCoO2 electrodes.

Sample BET surface area (m2/g)

Pristine D.S. LiCoO2 2.2

Cycled D.S. LiCoO2 2.0

0

200

400

600

800

1000

0 200 400 600 800 1000

Cdl

pos (µ

F/cm

2 )

Cycle number

Figure 7.14 Double layer capacitance vs. cycle number.


The origin of the Rctpos rise upon cycling has not been clearly figured out. Aurbach et al.

attributed the increase in Rctpos to the SEI layer.21,22 One component of the SEI layer is LiF

which is highly resistive to Li+ ion migration. The increasing RSEIpos has shown that the SEI

layer thickens upon cycling, which makes it more difficult for Li+ ion to migrate through the SEI layer in order to take part in the charge transfer reaction. This, in turn, ends up with an increasing Rct

pos. Abraham et. al assigned the Rctpos rise to the compositional change of the

SEI layer during cycling.23 They argued that these compositional changes apparently increased the complexity of Li+ ion movement through the film, resulting in an increasing resistance. Generally, the SEI layer is most likely responsible for the increase in the charge transfer resistance of the positive electrode.

The particles of the positive electrode were analyzed by TEM. Figure 7.15 shows the pictures of LiCoO2 particles before and after cycling. Compared to the particles before cycling, the cycled particles were severely strained with extensive internal defects and microcracks (indicated by arrows). These defects and microcracks are believed to result from the repeated insertion and removal of Li+ ions into and out of the positive electrode during cycling. This is confirmed by a comparative study on the LiCoO2 particles after long-term storage.

(b)(a)

Figure 7.15 TEM pictures of LiCoO2 particles, (a) pristine, (b) after 500 cycles

It is known that a fully charged battery loses substantial capacity after storage on a shelf. In our study, a 300 mAh Li-ion battery was found to irreversibly lost a capacity of 24 mAh (8%) after 4-month storage at a fully charged state. The measured battery impedance before and after storage exhibited a similar behavior to that of the batteries after cycling. The impedance of the negative electrode was almost invariant after storage whereas the positive electrode disclosed a drastic increase in impedance (see Figure 7.16).

Increase of the positive electrode impedance 113

The LiCoO2 particles of the battery after storage were also checked by TEM. Unlike the cycled LiCoO2 particles, no internal defects and microcracks were observed with the LiCoO2 particles after storage. The LiCoO2 particles appeared to remain intact after storage (pictures not shown here). This finding indicates that the internal defects and microcracks in the cycled LiCoO2 particles originate from the repeated insertion and removal of Li+ ions. However, the role of these internal defects and microcracks in impedance rise and capacity loss during cycling remains unclear.

0

0.1

0.2

0.3

0.4

0.5

0 0.2 0.4 0.6 0.8

-Zim

(Ω)

(a) Zre (Ω)

0

0.03

0.06

0.09

0.12

0 0.05 0.1 0.15 0.2

-Zim

(Ω)

(b) Zre (Ω) Figure 7.16 Impedance spectra of (a) the positive electrode and (b) the negative electrode of a Li-ion battery at fully charged state. Impedance was measured before () and after (∆) storage.

One interesting phenomenon observed with TEM is that not all cycled LiCoO2 particles developed internal defects and microcracks. Some cycled particles were just alike the pristine ones. This observation implies that electrochemical cycling does not occur evenly everywhere within the positive electrode. A wide variation in the depth of charge/discharge of LiCoO2 particles can be expected. This gives a suggestion that the cycling performance of Li-ion batteries can be improved if all the LiCoO2 particles can participate in the electrochemical intercalation/deintercalation reactions uniformly.


7.6 Battery capacity fade, a complicated process Based on what has been presented in the previous part of this chapter, it can already be concluded that the battery capacity fade is a complicated process. In reality, many factors contribute to battery capacity loss. First of all, capacity fade heavily depends on the battery operating conditions. Current and voltage are the two primary parameters to be considered. The effect of the discharge current on the capacity fade has been demonstrated in Figures 7.1 and 7.2. A higher current induces a higher degradation rate. The impact of the working voltage is illustrated in Figure 7.17. Three batteries were all cycled at 0.5 C but with different Vmax. The battery cycled between 3.0 and 4.4 V (curve c) exhibited the highest degradation rate. After about 230 cycles, 80% of its original capacity was lost. In comparison, the battery having a Vmax of 4.2 V (curve a) exhibited the best cycling performance. These phenomena clearly show that cycling Li-ion batteries above 4.2 V greatly accelerates the loss of capacity. Apart from current and voltage, the working temperature is another important factor that has a significant impact on the battery capacity loss and has been widely studied by many researchers.11,24-26

0

20

40

60

80

100

120

0 50 100 150 200 250

(a)

(b)

(c)

Cap

acity

ret

entio

n (%

)

Cycle number Figure 7.17 Effect of Vmax on cycling performance, (a) Vmax = 4.2 V, (b) Vmax = 4.3 V, (c) Vmax = 4.4 V, batteries cycled at 0.5 C.

The geometric shape of batteries also has a significant influence on the battery capacity fade. Rubino et al.27 studied the degradation in cylindrical and prismatic Li-ion batteries and found that, after 300 cycles, the prismatic batteries faded 24% while the cylindrical batteries lost 16% of their capacities. The authors attributed the higher fade rate in the prismatic battery to a lower cell stack pressure which resulted in more anode swelling during cycling. The increased porosity of the anode in this battery reduced the integrity of the electrical conduction network causing many graphite particles to become isolated or less accessible.27 As electrode porosity is concerned, Sikha et al.28 modelled the change of porosity of the

Battery capacity fade, a complicated process 115

active material and showed that a decrease in porosity negatively affected the capacity fade by increasing the ohmic drop and introducing higher concentration overpotentials.

What makes battery degradation more complicated is that the causes to capacity loss vary with variant working conditions. A good example is lithium metal deposition that has been proven not to occur under 0.5 C charging conditions (see Section 7.4.2). However, it was measured in our study that, when the current increased to 1.5 C, Eneg dropped below 0 V (vs. Li/Li+) at which potential lithium metal deposition is expected to take place. The same result was also obtained with commercial Sony US18500 Li-ion batteries.29

For the batteries studied in this work, the positive electrode is mainly responsible for the observed battery capacity fade. On the other hand, the negative electrode was reported to play a major role in degradation as well. Aurbach and coworkers21 measured a remarkable increase in the negative electrode impedance and concluded that the increase in the impedance of both the negative and the positive electrodes was the most important factor that limited the cycle life of the batteries. In the work of Nagasubramanian et al., the authors even found that it was the negative electrode that dominated the battery impedance,30 which is entirely opposite to our observations.

These distinct results lead us to conclude that the mechanism of battery capacity fade is also dependent on the chemistry of the batteries. Battery chemistry takes into account the active materials used, the material balance, etc. LiCoO2 and graphitic carbons are commonly used as the positive and negative electrodes, respectively. However, both LiCoO2 and graphitic carbons appear in different shapes and morphologies (see Chapter 2). In our study, the negative electrode was graphitic flakes and the electrolyte consisted of 1 M LiPF6 dissolved in EC, DEC and DMC (2:1:2 w/w). The negative electrode studied by Aurbach et al. was mesocarbon microbead (MCMB) carbon and the electrolyte comprised of 1 M LiPF6 in EC-EMC (1:2 w/w). Different types of graphitic carbons have different electrochemical behavior and different stabilities towards Li+ intercalation/deintercalation.31 For example, MCMB carbon, which appears in spherical shape, is reported to be well-suited for high-rate application.32 The electrolyte is another important issue. The properties of the electrolyte vary with different electrolyte compositions and significantly affect the formation and composition of the SEI layer.33-37 By this means, the electrolyte can greatly affect the degradation process. In Nagasubramanian’s work, the negative electrode was carbon and the positive electrode was LiNi0.8Co0.15Al0.05O2. The electrolyte was made up of 1.2 M LiPF6 in EC-EMC (3:7 w/w). For this kind of positive electrode materials, researchers have confirmed the effect of positive electrode composition on capacity fade.38 Conclusively, the battery degradation characteristics are very likely to be affected by battery chemistry. More systematic work is needed to unravel the correlation between capacity loss and battery chemistry.


7.7 Concluding remarks The mechanism of capacity fade of Li-ion batteries upon cycling has been investigated. It is found that the positive electrode is mainly responsible for the observed battery capacity loss. After intensively cycling, the positive electrode irreversibly lost 13% of its original capacity whereas the negative electrode retained its capacity to host Li+ ions.

Experimental results disclose that lithium metal deposition and dissolution of the positive electrode did not happen under the investigated conditions. The loss of contact between electrode material and current collector and loss of contact between particles of electrode material are proved not to take place. The lost recyclable Li+ ions were partially consumed by the SEI layer. On the other hand, XRD measurements revealed structure disorder occurring in LiCoO2 and a portion of the Li+ ions in the positive electrode became inactive due to this disorder. This effect also accounts for the loss of recyclable Li+ ions.

As mentioned in Section 2.3, SEI layer forms and thickens on both the positive and negative electrodes. The formation of SEI layer is inevitable and beneficial for Li-ion batteries. For the purpose of prolonging the cycle life of Li-ion batteries, suppressing the thickening of the SEI layer would be a feasible means to reduce the loss of Li+ ions.

The battery impedance is another factor that contributes to capacity fade. EIS measurements show that battery impedance developed dramatically upon cycling and reveal that the battery impedance rise stemmed from the positive electrode, particularly from Rct

pos. The thickened SEI layer may contribute to the increasing Rct

pos during cycling. To well understand this phenomenon requires more experimental work among which surface analyses are particularly important.

The cycled LiCoO2 particles were examined by TEM. Interestingly, some cycled particles were found to develop extensive internal defects and microcracks while the other did not. This observation indicates that electrochemical cycling does not occur uniformly everywhere within the positive electrode. A wide variation in the depth of charge/discharge of LiCoO2 particles is implied. By comparing with the LiCoO2 particles after long-term storage, it is confirmed that these internal defects and microcracks resulted from the repeated Li+ insertion and removal during cycling.

Battery capacity loss is a very complicated process. The complexity is based on the fact that many factors have impact on degradation. The capacity fade rate heavily depends on the battery working conditions. Of importance is that the battery chemistries also affect the mechanism of capacity fade very significantly. References 1. P. Arora, R. E. White and M. Doyle, J. Electrochem. Soc., 145, 3647 (1998). 2. R. S. Rubino, H. Gan and E. Takeuchi, J. Electrochem. Soc., 148, A1029 (2001). 3. J. Shim, R. Kostecki, T. Richardson, X. Song and K.A. Striebel, J. Power Sources,

112, 222 (2002).

References 117

4. D. Aurbach, B. Markovsky, A. Rodkin, E. Levi, Y. S. Cohen, H.-J. Kim and M. Schmidt, Electrochim. Act., 47, 4291 (2002).

5. J. Li, E. Murphy, J. Winnick and P. A. Kohl, J. Power Sources, 102, 294 (2001). 6. P. Ramadass, B. Haran, R. White and B. N. Popov, J. Power Sources, 112, 210 (2002). 7. H. F. Wang, Y.-I. Jang, B. Y. Huang, D. R. Sadoway and Y.-M. Chiang, J.

Electrochem. Soc., 146, 473 (1999). 8. G. G. Amatucci, J. M. Tarascon and L. C. Klein, Solid State Ionics, 83, 167 (1996). 9. P. Arora, M. Doyle and R. E. White, in Lithium Batteries, S. Surampudi and R. Marsh,

Eds., PV 98-16, p.553, The Electrochemical Society Proceedings Series, Pennington, NJ (1998).

10. D. Aurbach, B. Markovsky, A. Rodkin, M. Cojocaru, E. Levi and H.-J. Kim, Electrochim. Act., 47, 1899 (2002).

11. J. Shim, R. Kostecki, T. Richardson, X. Song and K. A. Striebel, J. Power Sources, 112, 222 (2002).

12. D. Zhang, B.S. Haran , A. Durairajan, R.E. White, Y. Podrazhansky and B.N. Popov, J. Power Sources, 91, 122 (2000).

13. G. Ning, B. Haran and B. N. Popov, J. Power Sources, 117, 160 (2003). 14. A. Blyr, C. Sigala, G. Amatucci, D. Guyomard, Y. Chabre and J.-M. Tarascon, J.

Electrochem. Soc., 145, 194 (1998). 15. D. Aurbach, B. Markovsky, A. Rodkin, E. Levi, Y. S. Cohen, H.-J. Kim and M.

Schmidt, Electrochimi. Act., 47, 4291 (2002). 16. A. H. Whitehead, K. Edström, N. Rao and J. R. Owen, J. Power Sources, 63, 41

(1996). 17. J. Zhou and P. H. L. Notten, private communication. 18. M. Yoshio, H. Tanaka, K. Tominaga and H. Noguchi, J. Power Sources, 40, 347

(1992). 19. E-D. Jeong, M-S. Won and Y-B. Shim, J. Power Sources, 70, 70 (1998). 20. J. Li, E. Murphy, J. Winnick and P. A. Kohl, J. Power Sources, 102, 294 (2001). 21. D. Aurbach, B. Markovsky, A. Rodkin, M. Cojocaru, E. Levi and H.-J. Kim,

Electrochim. Act., 47, 1 (2002). 22. D. Aurbach, B. Markovsky, A. Rodkin, E. Levi, Y. S. Cohen, H.-J. Kim and M.

Schmidt, Electrochim. Act., 47, 4291 (2002). 23. D. P. Abraham, E. M. Reynolds, P. L. Schultz, A. N. Jansen and D. W. Dees, J.

Electrochem. Soc., 153, A1610 (2006). 24. H.-P. Lin, D. Chua, M. salomon, H.-C. Shiao, M. Hendrickson, E. Plichta and S. Slane,

Electrochem. Solid-State Lett., 4, A71 (2001). 25. C.-K. Huang, J. S. Sakamoto, J. Wolfenstine and S. Surampudi, J. Electrochem. Soc.,

147, 2893 (2000). 26. C.-S. Wang, A. J. Appleby and F. E. Little, J. Electrochem. Soc., 149, A754 (2002). 27. R. S. Rubino, H. Gan and E. S. Takeuchi, J. Electrochem. Soc., 148, A1029 (2001). 28. G. Sikha, B. N. Popov and R. E. White, J. Electrochem. Soc., 151, A1104 (2004). 29. L. Verneyre and J. Zhou, private communication.


30. G. Nagasubramanian and D. H. Doughty, J. Power Sources, 150, 182 (2005). 31. D. Aurbach, H. Teller, M. Koltypin and E. Levi, J. Power Sources, 119-121, 2 (2003). 32. K. Zaghib, G. Nadeau and K. Kinoshita, J. Power Sources, 97-98, 97 (2001). 33. S. E. Sloop, J. B. Kerr and K. Kinoshita, J. Power Sources, 119-121, 330 (2003). 34. M. C. Smart, B. V. Ratnakumar, J. F. Whitacre, L. D. Whitcanack, K. B. Chin, M. D.

Rodriguez, D. Zhao, S. G. Greenbaum and S. Surampudi, J. Electrochem. Soc., 152, A1096 (2005).

35. T. Zheng, A. S. Gozdz and G. G. Amatucci, J. Electrochem. Soc., 146, 4014 (1999). 36. A. M. Andersson, M. Herstedt, A.G. Bishop and K. Edström, Electrochim. Act., 47,

1885 (2002). 37. K. Kanamura, T. Umegaki, M. Ohashi, S. Toriyama, S. Shiraishi and Z.-I. Takehara,

Electrochim. Act., 47, 433 (2001). 38. A. Bloom, S. A. Jones, V. S. Battaglia, G. L. Henriksen, J. P. Christophersen, R. B.

Wright, C. D. Ho, J. R. Belt and C. G. Motloch, J. Power Sources, 124, 538 (2003).

Summary 119

Summary

Lithium metal microreference electrodes and their applications to Li-ion batteries Li-ion batteries are nowadays widely used as power sources for a wide variety of electronic devices by virtue of their high cell voltage, high energy density and excellent cyclability. Though the performance of Li-ion batteries has been greatly improved during the last decade, it is still, to some extent, lagging behind the increasing power consumption for portable devices. To bridge this gap is a continuous task, which in turn, requires some fundamental understanding of Li-ion batteries.

Prior to understanding the nature of Li-ion batteries, the features of the employed active materials must be clearly known. The physical and electrochemical properties of the active materials are briefly discussed in Chapter 2. This chapter also deals with the Lithylene technology that is used to fabricate the batteries studied in this thesis. With this technology, mechanically stable three-dimensional Li-ion batteries are made with different shapes. These preshaped batteries provide great freedom in the design of portable electronic devices. The experimental implications are explained in Chapter 3.

For the purpose of in situ evaluating materials and developing mechanistic understanding of the behavior and degradation characteristics of Li-ion batteries, so-called lithium metal microreference electrodes have been developed, permitting measurement of individual electrode properties in a non-destructive way. This work has been described in Chapter 4. The substrate of the microreference electrode is a 80 µm copper wire placed in-between the positive and negative electrodes during battery construction. Lithium metal microreference electrodes were fabricated by in situ electrochemical deposition of lithium from both the positive and negative electrodes onto the copper wire. Chapter 4 also studied the influence of the depositing current density and the layer thickness on the stability of the microreference electrodes. A current density of 0.2 mA/cm2 and lithium layer thickness of 4 µm were found to be the most favorable lithium deposition condition.

The microreference electrodes were sufficiently small to minimize the geometric disturbances to the battery performance while allowing in situ measurements. Due to the fact that it only contained a very small amount of lithium metal, the life-time of the microreference electrode was limited to a few weeks if subjected to continuous potential measurements. As an advantage, once the potential of the reference electrode starts to degrade after long-term measurements, the reference electrodes could be easily revived by re-depositing metallic lithium onto them.

The microreference electrodes were adopted to measure in situ the potential profiles of the positive and negative electrodes during cycling. The results demonstrated that the potential of the negative electrode mainly accounts for the battery voltage change at the beginning of charge and at the end of discharge whereas the potential of the positive

Summary 120

electrode dominated the battery voltage at the end of charge and at the beginning of discharge.

The battery impedance is of practical importance as it has great impact on the rate capability of batteries. Generally, battery impedance comprises contributions from the series impedance (Rs), the positive (Rpos) and negative (Rneg) electrodes. With the help of the microreference electrodes, Rs, Rpos and Rneg were separately evaluated by means of electrochemical impedance spectroscopy (EIS) measurements-see Chapter 5.

The battery impedance was studied as a function of battery state-of-charge (SoC), increasing significantly from a fully charge state (100% SoC) to a fully discharge state (0% SoC). EIS measurements revealed that Rpos was strongly dependent on SoC whereas Rneg appeared to be invariant with SoC. Based on the physical processes of Li+ ion intercalation/deintercalation, both Rpos and Rneg can be decomposed into four elements, viz, serial resistance (Rs), resistance for Li+ ions migrating through the solid electrolyte interface (SEI) layer (RSEI), charge transfer resistance (Rct) and solid-state diffusion resistance. The intercalation/deintercalation processes can be described by an equivalent circuit and the values of Rs, RSEI and Rct were extracted by fitting the measured impedance spectra to the equivalent circuit. Rs was proved to be small and independent of SoC. RSEI

pos and RSEIneg

attained the same values at 68% SoC and higher, strongly suggesting that the SEI layers covering the positive and negative electrodes may have similar properties. Rct

neg was much smaller than Rct

pos, the latter being found to be the most influencing element in battery resistance, especially for a fully discharged Li-ion battery. Noticeably, Rct

pos heavily depended upon the battery SoC, declining considerably with increasing SoC. This result was confirmed by I-η studies that revealed an increase in Io with increasing SoC. To sum up, battery impedance originates mainly from Rpos. This effect was extremely pronounced at the battery fully discharged state.

The dependence of Rctpos on the SoC was explained by the varying electrical

conductivity of LixCoO2 with SoC. As charge transfer reactions take place at the electrode/electrolyte interface, electrons must transport between the current collector and the electrolyte/electrode interface, via LiCoO2 particles. The high electric resistance of LiCoO2 may severely impede the electron transport process, thereby yielding a high process resistance. On the contrary, Li0.5CoO2, corresponding to a fully charged battery, attains metal-like high electrical conductivity, which facilitates electrons movement and, consequently, the charge transfer process, resulting in a small process resistance. Based on the above hypothesis, the decrease of Rct

pos with increasing SoC can be attributed, at least partly, to the change of electrical conductivity of LiCoO2 particles. It requires, however, more research work to experimentally prove this hypothesis.

The results presented in Chapter 5 showed that the positive electrode was a key point if the rate and/or power capability is to be improved, as the positive electrode largely accounts for battery internal resistance.

When a battery is in operation, a concentration gradient is built up in the electrolyte due to the passage of current, from which arises the diffusion overpotential. Chapter 6 discussed how to experimentally measure the concentration gradient in sealed battery systems.

Summary 121

Subsequently, the diffusion overpotential can be determined once the concentration gradient is known.

It is known that a concentration gradient of Li+ cations is established in the electrolyte when current flows through the battery. This concentration gradient was theoretically proved to be linear under galvanostatic steady-state conditions. Between any two locations along the concentration gradient an electrochemical potential difference exists from which an electric potential difference arises. Based on these findings, an equation was derived to correlate the concentration difference and the electric potential difference in the electrolyte. Thanks to the adoption of the microreference electrode concept, the electric potential difference in the electrolyte could be measured in situ even in sealed Li-ion batteries. Subsequently, the concentration difference and the concentration gradient could be calculated.

Chapter 6 systematically investigated the concentration gradient over a current range of 0.1 – 1.0 C-rate. Experimental results proved the linearity of the concentration gradient and confirmed that the concentration gradient was proportional to the applied current under galvanostatic steady-state conditions. By using the known concentration gradient, the diffusion coefficient of Li+ ions in the electrolyte was determined and was found to be almost constant at different currents. Furthermore, the diffusion overpotential related to the concentration gradient was calculated to be insignificant in comparison to the overall battery overpotential over the current range studied, increasing from ~ 0.8 mV at 0.1 C to ~ 7 mV at 1.0 C.

It must be pointed out that the diffusion overpotential is not proportional to the applied current and does not always remain small. By using the obtained values of the concentration gradient and the diffusion coefficient of Li+ ions, the limiting current of the batteries studied in this work was estimated to be 1.1 A, corresponding to 3.7 C-rate. When the applied current is approaching or even beyond the limiting current, the diffusion overpotential becomes extremely large, making the battery voltage reach the maximum value rapidly. Indeed, it was observed experimentally that a fully discharged Li-ion battery reached 4.2 V instantaneously when being charged at 1.17 A (3.9 C-rate).

During theoretical derivation, for reasons of simplicity, the electrolyte activity was replaced by the electrolyte concentration, which can lead to experimental errors in the case of high current. This situation was discussed in Chapter 6. It was suggested not to apply this method to currents higher than 1 C-rate. In principle, this method can be applied to other battery systems with appropriate modification.

Battery capacity fade is always of great concern. Chapter 7 focused on the mechanism of battery capacity degradation upon cycling. Basically, the loss of capacity stems from two causes: the loss of active material and the increase in overall battery internal impedance. The positive electrode was found to be the major factor in both aspects and contributed mostly to the loss of battery capacity.

The positive electrode irreversibly lost 13% of its original capacity after 511 cycles whereas the negative electrode retained its capacity to host Li+ ions. The loss of positive active material was attributed to the comsumption of recyclable Li+ ions by the SEI layer on

Summary 122

both the positive and negative electrodes and the formation of disordered structure in the positive electrode material.

The impedance of Li-ion batteries increased upon cycling. EIS measurements showed that the impedance of the negative electrode was almost invariant during cycling. The positive electrode was responsible for the increase in the overall battery internal impedance. Similar to Chapter 5, Rct

pos was again found to be crucial and contributed greatly to impedance rise. However, the cause for this increase in Rct

pos remains a question to answer. It was observed with TEM that some cycled LiCoO2 particles developed extensive

internal defects and microcracks. It is worthwhile to notice that some cycled LiCoO2 particles appeared to be as intact as the pristine ones, which indicates that electrochemical cycling does not occur uniformly everywhere within the positive electrode. This observation suggests that to uniformly utilize the positive electrode material may be a way to improve the cycling performance of batteries.

Battery degradation has been proven to be a very complicated process. The complexity results from various factors. The capacity fade rate heavily depends on battery working conditions. The mechnism of battery degradation alters with different working conditions and battery chemistries. To well understand the mechnism of capacity fade requires more detailed investigations. Nevertheless, attention should be paid to the positive electrode for the purpose of prolonging the cycle life of Li-ion batteries.

Samenvatting 123

Samenvatting

Lithium metaal micro referentie-electrodes en hun toepassingen in Li-ion batterijen Li-ion batterijen worden heden ten dage veelvuldig toegepast als energiebron in draagbare electronica, omdat ze een hoge celspanning combineren met een hoge energie-dichtheid en uitstekende levensduur. Hoewel Li-ion batterijen grote verbeteringen hebben doorgemaakt gedurende de laatste 10 jaar, blijft hun ontwikkeling nog steeds enigszins achter bij de sterk toenemende vraag naar energie van mobiele toepassingen. Dit gat overbruggen is een proces waarvoor voortdurend onderzoek en verbeteringen nodig zijn en welke ook om een fundamenteler begrip van Li-ion batterijen vraagt.

Om de Li-ion batterij als geheel goed te kunnen begrijpen moeten eerst de eigenschappen van de elektrodematerialen grondig bestudeerd worden. De fysische en elektrochemische eigenschappen van de elektrodematerialen worden kort besproken in hoofdstuk 2. Dit hoofdstuk beschrijft ook de ‘Lithylene Technologie’ die wordt gebruikt om de batterijen te maken die in dit proefschrift bestudeerd worden. Met deze techniek is het mogelijk om mechanisch stabiele batterijen met een willekeurige vorm te maken. Deze voorgevormde batterijen zorgen voor een grotere ontwerpvrijheid voor draagbare electronica. De experimentele implicaties worden uiteengezet in hoofdstuk 3.

Om de elektrodematerialen in-situ te kunnen karakteriseren en de degradatiemechanismen van een Li-ion batterij te kunnen begrijpen zijn zogenaamde micro-referentie elektroden van Lithium-metaal ontwikkeld. Deze maken het mogelijk beide elektroden afzonderlijk te karakteriseren op een niet-destructieve manier. Dit werk staat beschreven in hoofdstuk 4. Als substraat voor de micro referentie-elektroden is gekozen voor een koperdraad van 80 µm diameter die tussen elektroden wordt geplaatst tijdens assemblage van de batterij. De referentie-elektrode wordt vervolgens gevormd door in-situ Lithium metaal op de koperdraad af te zetten, zowel vanuit de positieve als de negatieve elektrode. Hoofdstuk 4 beschrijft ook de invloed van de stroomdichtheid tijdens depositie en de laagdiktoep de stabiliteit van de micro-referentie-elektrode. Een stroomdichtheid van 0.2 mA/cm2 en een laagdikte van 4 µm blijkt de beste referentie-elektrodes op te leveren.

De micro-referentie-elektroden zijn klein genoeg om in-situ metingen mogelijk te maken zonder de geometrie van de batterij al te veel te verstoren. Omdat de hoeveelheid lithium in de referentie-elektrode erg klein is, is de levensduur beperkt tot enkele weken wanneer er continue metingen worden gedaan. Een voordeel is wel dat het opnieuw deponeren van een laagje lithium de degradatie van de referentie-elektroden weer geheel ongedaan maakt.

De micro-referentie-elektroden zijn gebruikt om in-situ de potentiaal-respons van de afzonderlijke elektroden te meten tijdens laden en ontladen. Uit de resultaten bleek dat de verandering van de batterijspanning aan het begin van de belading en het einde van de

Samenvatting 124

ontlading wordt veroorzaakt door de negatieve elektrode en dat de positieve elektrode aan het einde van de belading en het begin van de ontlading het batterij-voltage bepaalt.

De interne weerstand van een batterij is van groot praktisch belang, omdat het een grote invloed heeft op de maximaal haalbare stroomdichtheid. De weerstand van een batterij wordt bepaald door bijdragen van het seriële weerstand (Rs) en de positieve (Rpos) en negatieve (Rneg) elektrode. Met behulp van de micro-referentie-elektroden konden Rs, Rpos and Rneg elk afzonderlijk geëvalueerd worden door middel van impedantie-spectroscopie (EIS); zoals beschrevan is in hoofdstuk 5.

De impedantie van een batterij is bestudeerd als functie van de ladingstoestand, de ‘State-of-charge’ (SoC), en neemt aanzienlijk toe gaande van een volledig beladen (100% SoC) naar een volledig ontladen toestand (0% SoC). Uit EIS metingen bleek dat Rpos sterk afhankelijk is van de SoC waar Rneg vrijwel geen afhankelijkheid vertoond. Op basis van alle fysische processen die een rol spelen bij Li (de-)intercalatie, kunnen zowel Rpos als Rneg opgedeeld worden in 4 elementen, nl. een seriële weerstand (Rs), migratieweerstand van Li-ionen door het solid-electrolyte interface (RSEI), de een ladingsoverdrachtsweerstand (Rct) een vaste-stof diffusie van Li in het actieve materiaal. Deze elementen kunnen worden ‘vertaald’ naar een equivalent electronisch circuit waaraan de impedantie-spectra gefit kunnen worden. Hieruit kunnen dan waardes voor Rs, RSEI en Rct worden afgeleid. Rs bleek klein en vrijwel onafhankelijk van de SoC te zijn. RSEI

pos en RSEIneg bereikten dezelfde

waarde bij 68% SoC en hoger, wat een sterke aanwijzing is dat de SEI op depositieve en negatieve electrode dezelfde eigenschappen heeft. Rct

neg was veel kleiner dan Rctpos, welke

de meest dominante factor in de totale weerstand van de batterij bleek te zijn, met name in volledig ontladen toestand. Rct

pos bleek zeer sterk afhankelijk van de SoC en neemt sterk af met toenemende SoC. Dit resultaat werd bevestigd door I-η metingen die een toename in Io lieten zien bij toenemende SoC. Samenvattend kan gesteld en worden dat de batterij-weerstand voornamelijk wordt veroorzaakt door Rpos, dit was overduidelijk, met name wanneer de batterij volledig ontladen is.

De afhankelijkheid tussen Rctpos en de SoC kan wellicht verklaard worden door

veranderingen in de electrische geleiding van LixCoO2 als functie van de Li concentratie x. Omdat de ladingsoverdrachtsreactie plaatsvindt aan het grensvlak met het electrolyt, moeten de electronen door het poreuze netwerk van LiCoO2 deeltjes heen om van het substraat naar het interface met het electrolyt te komen. Een sterk verhoogde electrische weerstand van LiCoO2 kan het electronentransport ernstig belemmeren, waardoor de reactieweerstand sterk wordt verhoogd. In volledig beladen toestand daarentegen, is Li0.5CoO2 juist uitstekend electrisch geleidend, wat het ladingsoverdrachtproces juist gemakkelijker maakt en een lagere reactieweerstand geeft. Het lijkt erop dat de afname van Rct

pos met toenemende SoC tenminste gedeeltelijk wordt veroorzaakt door veranderingen in de electrische weerstand van LixCoO2, maar er is nog meer onderzoek nodig om dit experimenteel te bewijzen.

De resultaten in hoofdstuk 5 laten zien dat de positieve elektrode de belangrijkste factor is om de maximale stroom-en vermogensdichtheid te verhogen, omdat deze electrode het grootste deel van de interne weerstand van de batterij veroorzaakt.

Samenvatting 125

Wanneer er stroom door een batterij loopt bouwt zich een concentratie-gradiënt op in het electrolyt, die ook een bijdrage aan de totale overpotentiaal levert. Hoofdstuk 6 beschrijft hoe deze concentratie-gradiënt in-situ experimenteel gemeten kan worden. Is de concentratie-gradiënt eenmaal bekend, dan kan vervolgens kan de diffusie-overpotentiaal bepaald worden.

Theoretisch kan worden aangetoond dat de concentratiegradiënt lineair is onder steady-state omstandigheden wanneer er een constante stroom door de batterij loopt. Daardoor is er een verschil in elektrochemische potentiaal van Li+ tussen elke 2 willekeurige punten langs deze gradiënt, wat een meetbaar spanningsverschil tussen deze punten veroorzaakt. In hoofdstuk 6 wordt een vergelijking afgeleid die de correlatie weergeeft tussen het concentratieverschil van Li+ en het potentiaalverschil in het electrolyt. Door het gebruik van meerdere micro-referentie-elektroden kan het potentiaalverschil tussen verschillende punten langs de concentratie-gradiënt in-situ gemeten worden, waarna het concentratie-verschil en de concentratie-gradiënt uitgerekend kunnen worden.

De concentratie-gradiënt is systematisch onderzocht als functie van de stroomdichtheid, welke gevarieerd werd tussen 0.1 en 1.0 C. De lineariteit van de concentratie-gradiënt onder steady-state condities werd experimenteel bevestigd en tevens werd aangetoond dat de gradiënt recht evenredig is met de stroomdichtheid. De verkregen waarde voor de concentratie-gradiënt kon vervolgens gebruikt worden om de diffusie-coëfficient van Li+ in het electrolyt te berekenen en deze bleek vrijwel onafhankelijk van de stroomdichtheid. Tevens werd aangetoond dat de overpotentiaal die wordt veroorzaakt door de concentratie-gradiënt klein is vergeleken met de totale overpotentiaal; slechts ~0.8 mV bij 0.1 C tot ~7 mV bij 1 C.

De diffusie-overpotentiaal neemt bij toenemende stroomdichtheid uiteindelijk meer dan lineair toe en blijft daardoor uiteindelijk niet altijd klein. Met de verkregen waarden voor de concentratie-gradiënt en de diffusie-coëfficient van Li+ kon de maximale stroomdichtheid van de in dit werk gebruikte batterijen geschat worden op 1.1 A of 3.7 C. Wanneer de aangelegde stroom deze waarde benadert of zelfs overschrijdt, wordt de diffusie-overpotentiaal zeer groot, waarbij de batterij-spanning de maximaal toelaatbare waarde van 4.2 V zeer snel bereikt. Experimenteel werd inderdaad bevestigd dat een volledig ontladen batterij die wordt beladen met 1.17 A, oftewel 3.9 C, vrijwel instantaan een spanning van 4.2 V bereikt.

Bij de theoretische afleiding is ter vereenvoudiging de activiteit van het electrolyt vervangen door de concentratie, wat kan leiden tot afwijkingen bij hogere stromen. Het wordt daarom aangeraden deze methode niet te gebruiken bij stroomdichtheden hoger dan 1 C. In principe kan deze methode, met de juiste aanpassingen, ook toegepast worden op andere batterij-typen.

Hoofdstuk 7 concentreert zich op het degradatiemechanisme van de batterij tijdens herhaaldelijk laden en ontladen. Capaciteitsverlies heeft 2 hoofdoorzaken: intrinsiek capaciteitsverlies door een verlies aan actief materiaal en de toename in de interne weerstand. In beide gevallen bleek de positieve elektrode de grootste bijdrage aan het capaciteitsverlies te leveren.

Samenvatting 126

13% van de capaciteit van de positieve elektrode was na 511 cycles irreversibel verloren gegaan, waar de negatieve elektrode zijn capaciteit vrijwel volledig behouden had. Het verlies aan actief materiaal van de positieve elektrode werd toegescreven aan het verlies van Li+ door de vorming van een SEI aan zowel de positieve als de negatieve elektrode en aan de vorming van een ongeordende structuur in het actieve materiaal van de positieve elektrode.

De impedantie van Li-ion batterijen neemt toe tijdens cyclen. EIS metingen toonden aan dat de impedantie van de negatieve elektrode vrijwel onveranderd bleef en dat de positieve elektrode voor het overgrote deel verantwoordelijk was voor de toename in de interne weerstand van de batterij. Evenals in hoofdstuk 5 bleek een toename van Rct

pos een grote bijdrage te leveren aan de stijging van de impedantie, maar de precieze oorzaak hiervan blijft nog enigszins onduidelijk.

Door middel van Transmissie Electronen Microscopie (TEM) kon worden aangetoond dat sommige gecyclede LiCoO2 deeltjes een groot aantal defecten en micro-scheurtjes vertonen. Echter, sommige deeltjes zagen er nog net zo gaaf uit als in ongebruikt materiaal, wat erop wijst dat niet al het materiaal in gelijke mate wordt beladen en ontladen. Een efficiënter gebruik van het positieve elektrodemateriaal zou daarom een manier kunnen zijn om de levensduur-prestaties van Li-ion batterijen te verbeteren.

Degradatie van Li-ion batterijen blijkt een erg gecompliceerd proces te zijn. Deze complexiteit heeft een aantal oorzaken. De snelheid waarmee capaciteit verloren gaat is sterk afhankelijk van de omstandigheden, zoals bijv. de laad-snelheid. Het degradatie-mechanisme is ook afhankelijk van deze omstandigheden en van de batterij-chemie. Om het degradatie-mechanisme goed te begrijpen zijn meer gedetailleerde onderzoeken nodig, maar er kan wel met zekerheid gesteld worden dat deze zich op de positieve elektrode dienen te concentreren.

127

金属锂微参比电极及其在锂离子电池的应用

摘要

锂离子电池因为具有电压高、比能量高、循环寿命长等特点而被广泛地应用于各

种便携式电器。自 90 年代初商品化以来，在过去的十几年间，锂离子电池的性能有

了很大的提高。尽管如此，当前锂离子电池的性能仍然无法完全满足各类电器不断增

长的能耗需求。对于锂离子电池研究而言，如何弥补这一差距将是一项持续不断的工

作。本博士研究项目的主要目的在于：通过对锂离子电池基本特性的研究来探索提高

锂离子电池性能的途径和方法。锂离子电池的性能在很大程度上取决于所用的电池活性物质。本论文的第二章简

要地介绍了锂离子电池正、负极活性物质的物理和电化学特性。这一章同时详细地描

述了飞利浦拥有专利的 Lithylene 技术。此技术采用一种称为“聚合物铆钉”

（polymer rivet）的方法将电极材料结合在一起，因此省去了普通锂离子电池所需的

金属外壳，从而减轻了电池的重量。Lithylene 技术的大特点是可以生产出具有不

同形状的锂离子电池，这将给产品外形设计带来很大的灵活性。此博士课题中所研究

的锂离子电池全部采用 Lithylene 技术制成。本论文的第四章详细地介绍了用于锂离子电池原位电化学研究的“金属锂微参比

电极”的制作工艺。这种微参比电极的基体材料为一根直径 80 微米（或 40 微米）的

铜丝。铜丝极小的尺寸使其可以在电池的制作过程中被植入电池中而不影响电池的正

常工作。铜丝置于正、负电极间并由隔膜分隔。在电池化成完成后，通过电化学方

法，分别从电池的正、负极在铜丝的表面镀金属锂以生成金属锂微参比电极。研究表

明，电镀条件控制在电流密度为 0.2 毫安/厘米2，镀锂厚度为 4 微米时可以获得性能

优良的微参比电极。金属锂微参比电极可应用于测量电池工作状态下正、负电极的电位曲线。实验结

果显示，对于一个完全放电的电池，在充电的起始阶段，电池电压的变化主要来自于

电池负极的电位变化，而当电池趋于满荷电态时，正极电位的变化是电池电压的主要

影响因素。当电池放电时，电池正极的电位在放电开始阶段影响着电池电压的变化，

当放电将要结束时，电池电压的变化则由负极电位的变化所决定。在电池的循环寿命

实验中，金属锂微参比电极可用于监测正负极电位曲线随电池循环次数的变化。需要

特别指出的是，由于只有极少量的金属锂（约 2 × 10-8 mol）存在于微参比电极的表

面，当微电极长时间地用于电极电位测量时（超过两周时间），其自身的电位会因为

金属锂的消耗而发生变化。此时的微参比电极将不再适用于电极电位的测量。但是，

微参比电极的电位可以通过重新镀锂而恢复，这也是该参比电极的一大优点。电池的内阻对电池的性能具有显著的影响。在第五章中，应用电化学阻抗法测量

锂离子电池的阻抗得知，电池的阻抗随着电池的电量而变化。电池处于满电荷状态

时，其阻抗有小值；随着电量的减少，阻抗逐渐增大并在电池完全放电时达到大

值。电池的阻抗可以认为分别来自于电池的正、负极、电解液和隔膜。利用金属锂微

摘要 128

参比电极可以分别测量出电池正、负极和电解液的电化学阻抗。测量结果表明，当锂

离子电池处于满荷电态时，正、负极的阻抗基本相当，正极阻抗略大。但是，负极的

阻抗基本不随电池的电量变化，而正极的阻抗随着电池电量的减少而显著增大。当电

池处于完全放空状态时，电池的阻抗几乎完全来自于正极。和正、负极的阻抗相比，

电解液的阻抗可以忽略不计且不随电池电量变化。根据锂离子的嵌入和脱嵌的物理过程，正、负电极的阻抗可以分解为四个部分：

连接阻抗（Rs），锂离子迁移经过电极表面的固/液界面所遇的阻抗（RSEI），电荷迁

移阻抗（Rct）和锂离子在电极内的固相扩散阻抗。这些物理过程可以用一个等效电

路来描述，上述各阻抗的值则可通过将测得的交流阻抗频谱和等效电路相比较而获

得。实验数据显示，电解液的连接阻抗很小，且不随电池的电量而变化。当电池的荷

电态（SoC）高于 68％时，正、负电极的RSEI值相等，这说明覆盖在正、负电极表面

的固/液界面具有相近的性质。对于Rct而言，负极的电荷迁移阻抗远小于正极。值得

注意的是，正极的电荷迁移阻抗（Rctpos）随着电池荷电态的增加而显著减小。这一

结果经由测量不同电流值下的极化曲线所证实。综上所述，锂离子电池的阻抗主要来

自于正极的电荷迁移阻抗，当电池完全放电时，这一效果尤其明显。不少研究人员曾报道过正极的电荷迁移阻抗随电池荷电态而变化这一现象。但

是，对于此现象目前尚无明确的解释。本论文尝试从正极活性材料钴酸锂的电导率随

其化学组成而变化这一特点来进行说明。在一个电化学过程中，电荷迁移反应总是发

生在电极和电解液的界面处。锂离子电池的正极由多层钴酸锂颗粒组成，这些颗粒全

部被电解液溶液所浸湿，这意味着当电荷迁移反应开始时，电解液中的锂离子已经位

于电极/电解液的界面处，随时可以参与反应。另一方面，电荷迁移反应所需的电子

则必须从集流体迁移到电极/电解液界面处。已知化学组成为Li0.5CoO2的钴酸锂（对

应于充满电的锂离子电池）具有接近于金属的导电率，此时电子可以很高的速率迁

移，电荷迁移反应得以在较短的时间内完成；而化学组成为LiCoO2的钴酸锂（对应

于完全放电的锂离子电池）电导率极低（同半导体），相应的，电子在钴酸锂颗粒间

的迁移变得困难，延长了电荷迁移反应完成所需的时间，并相应地表现出较高的电荷

迁移阻抗。这样就从物理意义上解释了正极的电荷迁移阻抗随电池荷电态而变化的现

象。当然，如何通过实验证明上述假设还是一个尚需解答的问题。电池工作时，流经电解液的电流在其间建立起浓度梯度。电解液的浓度梯度直接

与电池的工作性能相关，从中可以获得诸如浓差过电位、锂离子的扩散系数、极限电

流等重要的电化学参数。本论文的第六章讨论了如何应用金属锂微参比电极在密封的

电池系统中测量电解液中的浓度梯度。当电流流经锂离子电池时，电解液中产生一个锂离子的浓度梯度。位于此浓度梯

度不同位置上的任意两点间存在着电化学势差，此电化学势差产生一个可测量的电位

差。如果将两根微参比电极放置于正负极间的不同位置，一旦电池开始工作，两根参

比电极间即存在着电位差。我们首先从理论上证明了锂离子的浓度梯度为线性，在此

基础之上推导出一个用以描述电位差和浓度差之间关系的数学公式。这样，通过测量

微参比电极间的电压并结合推导出的公式，可以容易地计算出电解液中锂离子的浓度

梯度。

摘要 129

我们在 0.1 C -1.0 C的电流范围内系统地研究了浓度梯度和电池工作电流的关

系。实验结果不仅很好地证明了浓度梯度的线性特征，同时证明，当电池达到稳态

时，浓度梯度正比于流经电解液的电流。利用所得的浓度梯度计算出的锂离子在电解

液中的扩散系数基本恒定在 9.7×10-12 m2/s。通过已知的浓度梯度还可进一步求得正、

负电极表面的浓差过电位。计算结果显示，在所研究的电流范围内，浓差过电位在整

个电极过电位中只占很小的一部分，当电流自 0.1 C变化到 1.0 C时，浓差过电位由

~0.8 mV 增长至~7 mV。需要特别强调的是，浓差过电位并不正比于电池的工作电流。当工作电流接近或

者超过电池的极限电流时，浓差过电位将急剧增长。由所得的浓度梯度和锂离子的扩

散系数估算出本论文所研究的锂离子电池的极限电流约为 1.1 A，相应于 3.7 C。在实

验中，给一个完全放电的电池施加 1.17 A(3.9 C)的充电电流后，电池电压在瞬间即达

到 4.2 V。这一结果显示了所求得的极限电流值的可靠性，并且表明，在此条件下，

整个电池的过电位主要由浓差过电位所决定。在上述的理论推导中，锂离子的活度为其浓度所替代。这一简化过程势必给实验

结果带来系统误差。通过对此情况的讨论发现，在应用这一方法时，为了避免过大的

实验误差，电池的工作电流应不大于 1.0 C。此外，从原则上来说，此方法同样适用

于其它的电池系统。第七章主要讨论了锂离子电池循环寿命衰减的机理。电池循环寿命的衰减基本上

可以归于以下两大主要因素：电极活性物质的流失和电池内阻的增加。实验发现，经

过 511 次循环的电池，其正极的嵌锂容量减少了 13％，而负极的储锂容量并没有显

著变化。进一步的研究表明，正极材料的流失主要源于电池正、负极表面固/液界面

的形成和生长以及非活性相在正极材料内的生成。电池的阻抗随着循环次数的增加而增长。通过电化学交流阻抗测量得知，负极的

阻抗基本不随电池的循环次数而变化，电池阻抗的增长主要源于正极材料，其中显

著的增长来自于正极的电荷迁移阻抗（Rctpos）。目前，Rct

pos增大的原因尚不明确。利用透射电子显微镜观察经过循环测试的钴酸锂颗粒发现，部分钴酸锂颗粒上发

展出大量可见的缺陷和微小的裂痕。这些缺陷和裂痕确信是由于锂离子在钴酸锂的晶

格中不断重复的嵌入、脱嵌所造成。值得注意的是，部分经过循环测试的钴酸锂颗粒

并没有出现这些缺陷和裂痕。这一现象说明，在电池循环过程中，整个正极上的钴酸

锂颗粒并没有被均匀地利用。也许，通过对此的改善可以提高电池的循环性能。锂离子电池循环寿命的衰减是一个极其复杂的过程。这不仅因为有许多因素可以

影响电池的循环寿命，而且，这些影响因素还会由于电池循环条件的不同而变化。此

外，电池的化学性质本身也对电池的循环性能有着很大的影响。本论文的研究表明，

正极材料是影响锂离子电池循环寿命的主要因素。因而，如果希望提高锂离子电池的

循环性能，重点应该放在改善正极材料的性能上。

List of publications 131

List of publications J. Zhou and P. H. L. Notten, Development of reliable lithium microreference electrodes for long-term in situ studies of lithium-based battery systems, J. Electrochem. Soc., 151, A2173 (2004). P. H. L. Notten, M. Ouwerkerk, H. van Hal, D. Beelen, W. Keur, J. Zhou and H. Feil, High energy density strategies: From hydride-forming materials research to battery integration, J. Power Sources, 129, 45 (2004). J. Zhou, D. Danivol and P. H. L. Notten, A novel method to in situ determine the concentration gradients in the electrolyte of Li-ion batteries, Chemistry, Eur. J., 12, 7125 (2006). J. Zhou and P. H. L. Notten, Impedance studies on Li-ion batteries with Lithium metal microreference electrode, in preparation (2007). J. Zhou and P. H. L. Notten, On capacity fade of Li-ion batteries, in preparation (2007). Conference contributions J. Zhou, H. Feil and P. H. L. Notten, Preshaped Li-ion batteries (oral presentation), 203rd ECS conference, Paris, France, April (2003). P. H. L. Notten, M. Ouwerkerk, H. van Hal, J. Zhou and H. Feil, High energy density strategies: from materials research to battery integration (oral presentation), High Energy Density Electrochemical Power Sources (HEDES), Nice (2003). J. Zhou and P. H. L. Notten, Development of reliable lithium microreference electrodes for long-term in situ studies of lithium-based battery systems (poster), IBMA symposium, Graz, Austria, April (2004). J. Zhou and P. H. L. Notten, Determination of concentration gradient and diffusion coefficient of Li+ ion in electrolyte (oral presentation), 207th ECS conference, Quebec city, Canada, May (2005).

Curriculum Vitae 132

Curriculum Vitae Jiang Zhou was born on 10th March, 1974 in ShangRao, JiangXi province, China. In 1992 he graduated from JinLing high school in NanJing and started his undergraduate study at the Department of Applied Chemistry, TianJin University. After obtaining the B.Sc. degree in electrochemistry in 1996, he joined HuaFei Color Display System Co. Ltd, working as a process engineer. In 1997 he performed a traineeship in France, Holland and Germany in Philips factories. He attended the National University of Singapore in 2000 to conduct his master study in the Department of Chemical and Environmental Engineering. He received his M. Eng. degree in 2002 on the topic of catalytic removal of pyridine. In the same year he started his Ph.D. work at Eindhoven University of Technology (Inorganic Chemistry and Catalysis group) under the supervision of prof. dr. P.H.L. Notten. This research was focused on the development of lithium metal microreference electrodes in order to carry out in situ electrochemical characterization of Li-ion batteries. All of this work was performed in the System in Package Devices group at Philips Research Laboratories, Eindhoven, The Netherlands. In September 2006 he joined Philips Research Laboratories, Eindhoven, working as a research scientist in the group of System in Package Devices.

lithium metal microreference electrodes and their ... · 6.5.3 relationship between + and 94 d li...

Documents