Lithium-Ion Batteries: Solid-Electrolyte Interphase

Lithium-lon Batteries Solid-Electrolyte Interphase

editors

Perla B. Balbuena Yixuan Wang University of South Carolina

Lithium-Ion Batteries Solid-Electrolyte Interphase

Imperial College Press

This page intentionally left blank

Published by Imperial College Press 57 Shelton Street Covent Garden London WC2H 9HE Distributed by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: Suite 202, 1060 Main Street, River Edge, NJ 07661 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

LITHIUM-ION BATTERIES: SOLID-ELECTROLYTE INTERPHASE Copyright © 2004 by Imperial College Press All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.

ISBN 1-86094-362-4

Printed in Singapore.

CONTENTS

Preface

xiii

Chapter 1. SEI on lithium, graphite, disordered carbons and tin-based alloys Emanuel Peled and Diana Golodnitsky 1. 2.

3.

4.

5. 6.

Introduction SEI formation 2.1 The main principles and routes of the SEI formation 2.2 Structure of the SEI Chemical composition and properties of the SEI on inert substrate and on lithium 3.1 Inert metal substrate 3.2 Lithium covered by a native film 3.3 SEI formation in solid polymer and gel electrolytes Carbonaceous electrodes 4.1 Principles of SEI formation 4.2 SEI composition and morphology 4.2.1 HOPG 4.2.2 SLX20 4.2.3 Disordered carbons 4.2.4 Overview of SEI composition and properties in different carbon/non-aqueous electrolyte systems 4.2.5 Effect of carbon modification on SEI formation SEI formation on lithium-tin alloys Conclusions

Chapter 2. Identification of surface films on electrodes in non-aqueous electrolyte solutions: spectroscopic, electronic and morphological studies Doron Aurbach and Yaron S. Cohen 1.

Introduction 1a. Passivation of surface films: a general phenomenon

v

1 1 3 3 5 7 7 10 11 13 13 16 16 28 32 45 49 53 59

70 70 70

Lithium-Ion Batteries: Solid-Electrolyte Interphase

vi

1b.

2.

3.

4.

5.

Modes of growth of surface film phenomena, and their transport properties 1c. On the effect of the electrolyte solutions 1d. The role of the cation in surface phenomena in non-aqueous electrolyte solutions 1e. On the impact of the electrode’s material 1f. Some comments on applications Methods for identification of surface films on electrodes 2a. Introductory remarks 2b. Fast Fourier Transform Infrared Spectroscopy (FTIR) 2c. Raman spectroscopy 2d. Ultraviolet, Visible Light (UV-Vis) 2e. Extended X-ray Absorption Fine Structure (EXAFS), X-ray Absorption Near-Edge Structure (XANES) 2f. X-ray Photoelectron Spectroscopy (XPS) and Auger (AES) Electron Spectroscopy 2g. Energy Dispersive Analysis of X-rays (EDAX) 2h. Secondary Ion Mass Spectrometry (SIMS) 2i. Electrochemical Quartz Crystal Microbalance (EQCM) 2j. X-ray Diffractometry (XRD) 2k. NMR, ESR spectroscopy 2l. Scanning Probe Microscopies (AFM, STM) 2m. The use of UHV systems for identification of surface films formed on lithium The general structure of surface films on reactive surfaces 3a. Introduction 3b. Surface film formation on active metals 3c. Surface film formation on non-reactive metal and carbon electrodes 3d. On transport properties of surface films Impedance spectroscopy of electrodes covered by surface films 4a. Introductory remarks 4b. Active metal electrodes 4b.1 Lithium 4b.2 Mg electrodes 4c. Non-active metal electrodes polarized to low potentials Identification of surface films formed on lithium and non-active electrodes polarized to low potentials in Li salt solutions

72 73 74 75 76 77 77 78 80 81 82 82 82 83 83 85 86 86 87 87 87 88 90 93 93 93 94 94 95 97 100

Contents

5a. 5b.

The preparation of a library of FTIR spectra Identification of surface films formed on Li electrodes in ether solutions 5c. Identification of surface films formed on Li and non-active electrodes at low potentials in ester solutions 5d. Identification of surface films formed on Li and non-active metals at low potentials in alkyl carbonate solutions 5e. The impact of salt anions and contaminant reactions on the surface chemistry of lithium and noble metal electrodes in non-aqueous Li salt solutions 5f. On surface films formed on Li electrodes in polymeric electrolytes 6. Surface films on lithiated carbon electrodes 6a. Introductory remarks: surface film formation on carbon electrodes, the influence of the type of carbon, and the impact of the surface films on Li insertion processes 6b. On the identification of surface films formed on lithiated graphite electrodes 6c. On the correlation between the performance of lithiated graphite anodes and their surface chemistry 7. Surface studies of lithium and lithiated carbon electrodes by scanning probe microscopy 7a. Imaging of Li electrodes by AFM 7b. Graphite electrodes 8. About surface film formation on transition metal oxide cathodes in non-aqueous salt solutions 9. Identification of surface films on calcium and magnesium electrodes 10. Concluding remarks

Chapter 3. Spectroscopy studies of solid-electrolyte interphase on positive and negative electrodes for lithium ion batteries Zhaoxiang Wang, Xuejie Huang and Liquan Chen 1. 2.

Introduction SEI on tin oxide anode in various electrolytes 2.1 Sample preparation and instrumental 2.2 Capacity loss and electrolyte decomposition in first cycle

vii

100 102 105 107

111 116 116

116 117 120 124 124 126 128 129 131

140 140 141 143 144

Lithium-Ion Batteries: Solid-Electrolyte Interphase

viii

2.3 2.4 2.5 2.6

3.

4.

5. 6.

HRTEM study of SEI structure on nano-SnO surface Identification of Li2CO3 and ROCO2Li on nano-SnO anodes Formation of Li2CO3 and ROCO2Li on nano-SnO anodes Question: What is the reduction sequence of SnO and electrolyte? 2.7 Electrolyte-dependent SEI composition 2.8 Conclusion Surface enhanced Raman scattering (SERS) on rough electrodes 3.1 Normal Raman scattering and SERS studies on battery materials 3.2 Experimental 3.3 Electrochemical performance of Ag electrode 3.4 SERS study of passivating film on Ag electrode in lithium batteries 3.5 Prospects and conclusions on Raman scattering in SEI investigation Infrared absorption and X-ray photelectron spectroscopic investigation on performance improvement of surface-modified LiCoO2 cathode materials 4.1 Sample preparation 4.2 Comparison of EC adsorbed on different substrates 4.3 IR spectra of EC on electrodes charged to different voltages 4.4 XPS study on evolution of electronic structure of cathode materials with charge voltages 4.5 Conclusions Summary and comments Acknowledgements

Chapter 4. Scanning probe microscopy analysis of the SEI formation on graphite anodes Minoru Inaba and Zempachi Ogumi 1. 2. 3. 4.

Introduction Charge and discharge characteristics of graphite anode in ECand PC-based solutions Morphology changes of HOPG basal planes in the initial stage of solvent decomposition SEI formation in EC-based solutions

145 146 150 152 154 157 157 158 159 160 160 166

167 170 172 180 185 189 190 190

198 198 201 203 206

Contents

5. 6.

7.

Effect of co-solvent on solvent co-intercalation in EC-based solutions Additives in PC-based solutions 6.1 Roles of VC, FEC, and ES as additives 6.2 Roles of other additives in PC-based solutions Summary and outlook

Chapter 5. Theoretical insights into the SEI composition and formation mechanism: density functional theory studies Yixuan Wang and Perla B. Balbuena 1. 2. 3. 4.

Introduction Theoretical models and computational details Initial reduction of Li+(EC), Li+(PC), and Li+(VC) Comparison of reductive decomposition between EC and PC: Li+(EC)n and Li+(PC)n (n = 2, 3) 4.1 Reduction potentials and ring opening barriers of EC and PC 4.2 Decomposition products of EC and PC: Li+(EC)2 and Li+(PC)2 The effect of VC on the reductive decomposition of EC and 5. PC: (S)nLi+(VC) (S = EC and PC; n = 1 and 2) clusters 5.1 Initial reduction 5.2 Termination reaction of radical anions 5.3 Summary of reductive decomposition of solvents in the presence of VC 6. Associations of lithium alkyl dicarbonates through O–•••Li+•••O– bridges 6.1 Geometries and energetics 6.2 Vibrational frequencies 7. Adsorption and two-dimensional association of lithium alkyl dicarbonates on graphite surfaces through O–-•••Li+••••π (arene) interactions 7.1 H-truncated cluster models 7.2 Adsorption of lithium alkyl dicarbonates on the basal plane of the neutral graphite surface 7.3 Adsorption of lithium alkyl dicarbonates on the basal plane of negatively charged Gr54– and edge plane of Gr78–

ix

211 214 216 220 221

227 227 229 231 232 233 235 238 238 241 246 247 247 255

260 261 262 267

Lithium-Ion Batteries: Solid-Electrolyte Interphase

x

7.4

8.

Two-dimensional association of lithium alkyl dicarbonates on the basal plane of the graphite surface (Gr96) 7.5 Summary about adsorptions of LVD, LED and LPD on anode surface Remarks on the failure of PC and the efficiency of VC for the SEI layer formation in EC/PC-based solutions

Chapter 6. Continuum and statistical mechanics-based models for solid-electrolyte interphases in lithium-ion batteries Harry J. Ploehn, Premanand Ramadass, Ralph E. White, Diego Altomare and Perla B. Balbuena 1. 2.

3.

Introduction Continuum models for SEI growth 2.1 Overview of previous macroscopic models 2.2 Elements of continuum mechanics 2.2.1 Kinematics 2.2.2 Conservation of mass 2.3 Dynamic continuum models for SEI formation and growth 2.3.1 Growth limited by SEI electronic conductivity 2.3.2 Growth limited by solvent diffusion Statistical mechanics-based model 3.1 Description of the lattice-gas model 3.1.1 Implementation of the model 3.2 Results and discussions 3.3 Remarks with respect to the lattice model

Chapter 7. Development of new anodes for rechargeable lithium batteries and their SEI characterization by Raman and NEXAFS spectroscopy Giselle Sandi 1. 2. 3. 4. 5.

Introduction Carbon as a host in lithium ion cells Alternative anode materials UV Raman spectroscopy of templated-disordered carbons SEI characterization by NEXAFS

269 271 272

276

276 277 277 280 282 285 287 288 291 296 297 299 299 305

308 308 310 319 321 323

Contents

6. 7.

Conclusions Acknowledgements

Chapter 8. The cathode-electrolyte interface in a Li-ion battery Kristina Edström, Torbjörn Gustafsson and Josh Thomas 1. 2.

3. 4. 5.

Background LiMn2O4 2.1 XPS analysis 2.2 Elevated temperature effects for LiMn2O4 LiCoO2, LiNiO2 and LiNi0.8Co0.2O2 LiFePO4 Summary

Chapter 9. Theoretical studies on the solvent structure and association properties, and on the Li-ion solvation: implications for SEI layer phenomena Yixuan Wang and Perla B. Balbuena 1. 2. 3. 4.

5.

6.

Index

Introduction Computational details Geometric structures of various cyclic/linear carbonates, and effective additives and co-solvents to PC-based solutions Self and cross associations of cyclic/linear carbonates via C-H•••O interactions 4.1 Geometric and energetic properties 4.2 C-H bond lengths and vibrational frequencies 4.3 C=O bond lengths and vibrational frequencies 4.4 Characteristics of C-H•••O interactions using AIM Li+ solvation from alkyl carbonates 5.1 Interactions between Li+ and organic solvents 5.2 Solvation mumber of Li+ Conclusion: Implications for SEI layer phenomena

xi

331 331

337 337 338 340 344 353 358 360

365 365 367 368 373 374 380 384 384 387 387 390 393

398

This page intentionally left blank

PREFACE

Advances in science and engineering related to the emerging technologies of lithium-ion batteries (LIBs) have been so spectacular in the past decade that they have become the most popular power source for portable computing and telecommunication equipment. LIBs are simply essential for the constantly increasing demands of our information-rich society. It is expected that LIBs will continue to drive large market shares, and that new applications of LIBs such as powering electric and hybrid electric vehicles will flourish after a series of improvements resulting from current research efforts. A typical commercial lithium-ion battery system consists of a carbonaceous anode, an organic electrolyte that acts as an ionic path between electrodes and separates the two electrode materials, and a transition metal oxide (such as LiCoO2, LiMn2O4, and LiNiO2) cathode. Recently a variety of novel LIB components have been proposed, like tin-based alloys and disordered carbons as anode materials, and modifications to the conventional transition metal-oxide cathode made by coating it with metal-oxide nanoparticles, most of which are discussed in detail in this book. One of the most impressive advancements in the chemistry beneath LIBs is the understanding of the electrodes surface chemistry. It is recognized that a passivating layer between an electrode and the electrolyte arises from the reductive decompositions of a small amount of organic electrolytes mostly during the first several cycles of a working cell. This layer, which behaves similarly to a solid electrolyte interphase, was named SEI layer by Peled (J. Electrochem. Soc., 126, 1979, p. 2047), and it is a determinant factor on the performance of LIBs since the SEI nature and behavior affect the LIBs cyclelife, life time, power capability, and even their safety. Therefore, the build-up of appropriate SEI layers is an essential step in optimizing the combination of anode-electrolyte-cathode for LIBs, either through the screening of existing materials or developing novel ones. To this end, the better understanding of the SEI layer formation and growth for typical LIBs systems is fundamental. This book is designed for this purpose. The importance of the SEI is well recognized in the scientific community, as reflected by numerous special sessions in battery technology meetings. In Chapters 1 through 9, this book presents the latest developments of the SEI layer formation, growth, and characterization, including its morphology features on various anodes and cathodes, identification of SEI layers by spectral methods,

xiii

xiv

Lithium-Ion Batteries: Solid-Electrolyte Interphase

and insights into SEI formation and growth mechanisms by state-of-the-art experimental techniques as well as first-principles-based molecular theories. Continuum and statistical-mechanics-based macroscopic models are also included for the understanding of SEI growth and that of its effect on the thermodynamics of lithium intercalation in carbon materials. Some longstanding problems and especially new phenomena for LIBs are deeply and extensively discussed by the authors, most of whom have worked in this field for long times and have witnessed many problems associated with the development of LIBs, for example, the failure mechanism of the solvent propylene carbonate (PC) for LIBs employing graphite as anode material and the reasons behind the impressive improvement caused by a small amount of a solvent additive on the SEI layer of PC-based electrolytes. Peled and Golodnitsky in Chapter 1 begin with a simple introduction to the SEI concept, and the description of the main principles and routes of SEI formation and identification of possible products. It is followed by a discussion of several issues associated with the mechanisms of SEI formation on inert substrates, lithium metal, carbonaceous materials, and tin-based alloys. Attention is focused on the correlation between composition and morphology of the SEI forming on the various planes of highly ordered pyrolytic graphite (HOPG) as well as on different types of disordered electrodes in LIBs. In Chapter 2, Aurbach and Cohen describe various spectroscopic techniques for SEI layer characterization, classifying these techniques in terms of their ability to providing specific or non-specific identification of SEI species, of enabling in situ or ex situ electrode characterization, and of being destructive or nondestructive to the electrode surface. Based on careful identifications of the SEI components, Chapter 2 suggests reduction mechanisms for several combinations of common solvents (ethylene carbonate, propylene carbonate, ethers, and Ȗ-butyrolactone) and lithium salts. Based on scanning probe microscopy morphological studies, a novel insight is suggested to explain the major reasons for failure mechanisms of graphite electrodes in PC solutions. Chapter 3, by Chen and collaborators, concentrates on the spectroscopic investigation of the SEI layer on anodes as well as cathodes of LIBs, including the nanometer-sized SnO anode, and the nano-MgO modified LiCoO2 cathode. The effect of nano scaled materials on the performance of LIBs is well discussed using combination of spectral techniques, such as scanning electron microscopy (SEM), high-resolution transmission electron microscopy (HRTEM), surface enhanced Raman scattering (SERS), Fourier transform infrared (FTIR), and X-ray photoelectron spectroscopy (XPS).

Preface

xv

Inaba and Ogumi in Chapter 4 focus on the role of solvent co-intercalation in the SEI formation on graphite anodes, and discuss the mechanistic aspects of SEI formation using the images obtained from scanning tunneling microscopy (STM), and atomic force microscopy (AFM). Chapter 5, by Wang and Balbuena, provides a first-principles-based theoretical avenue for exploring the failure mechanism of PC in LIBs employing graphite as anode, and the functioning mechanism of a solvent additive (vinylene carbonate, VC) in PC-based solutions. The reductive decompositions of EC, PC and VC are investigated in parallel using high-level density functional theory, including their reduction mechanisms, identification of the main reduction products, and their adhesion to the basal and edge planes of graphite. Ploehn et al. in Chapter 6 use both macroscopic continuum and statistical mechanics-based models to simulate the SEI growth and to predict capacity loss in LIBs. Specifically the former model deals with the effects of electronic conductivity and solvent diffusion on SEI growth, while the latter is a lattice-gas model, which describes the thermodynamics of lithium-ion intercalation in carbons under the presence of a SEI. In Chapter 7, the Raman and near-edge X-ray absorption fine structure (NEXAFS) techniques have been used by Sandi to investigate the electronic and structural properties of carbonaceous materials and those of electrodes made from the synthesized carbons. The electrochemical performance of the carbon anodes is compared and related to the electronic and structural features of the SEI layer. Thomas and collaborators in Chapter 8 present evidences for the formation of some type of Solid Permeable Interface (SPI) between the electrolyte and the cathode in LIBs. It deals with today’s most commonly used cathode materials, such as LiMn2O4, LiCo2O4, LiNiO2 and LiNi1–xCoxO2 and with the recently introduced LiFePO4. Chapter 9, by Wang and Balbuena, deals with theoretical studies on the solvent structure and association properties, and on the lithium-ion solvation. SEI layer related phenomena are discussed in relation to lithium-ion solvation in commonly used solvents, co-solvents, and solvent additives. Perla B. Balbuena Yixuan Wang

This page intentionally left blank

CHAPTER 1

SEI ON LITHIUM, GRAPHITE, DISORDERED CARBONS AND TIN-BASED ALLOYS *

†

EMANUEL PELED and DIANA GOLODNITSKY School of Chemistry, Tel Aviv University, 69978 Tel Aviv, Israel E-mail: *[email protected]; †golod@ post.tau.ac.il

1

Introduction

It is well known that in contact with both liquid and polymer electrolytes, lithium is thermodynamically unstable toward the solvents and salts and becomes covered by a passivating film that slows the corrosion of the lithium. It is now generally accepted that the existence and successful operation of most lithium battery systems, as primary and secondary power sources, are due solely to this anode-surface layer. In the 1960s and early 1970s it was generally believed that, although some passivating film covers lithium, the metal is kinetically stable to many organic 1 solvents. It was assumed that the rate-determining step (r.d.s.) of the deposition-dissolution process for lithium is the electron charge transfer between the metallic electrode and the lithium cation in solution. In 1970, in a study of the electrochemical stability of propylene carbonate (PC), Dey suggested that lithium is covered by a passivating film, probably composed of lithium carbonate, which protects the metal from further chemical attack and 2 imparts stability. This film was presumed to conduct lithium cations. On the basis of a study of the electrochemical behavior of magnesium 3 electrodes in thionyl chloride (TC) solutions, Peled et al. concluded that it is the 2+ migration of Mg ions through the passivating layer that limits the total rate of deposition/dissolution of magnesium. In addition, it was concluded that the deposition of magnesium on an inert nickel cathode begins only after the nickel is covered by a passivating layer (MgCl2) that blocks the electronic current and 4 enables only ionic current to pass. It was further proposed by Peled et al. that this passivating-layer model is valid for all alkali metals in non-aqueous battery systems.

1

2

Lithium-Ion Batteries: Solid-Electrolyte Interphase

The layer formed instantaneously upon contact of the metal with the solution, consists of insoluble and partially soluble reduction products of electrolyte components. The thickness of the freshly formed layer is determined by the electron-tunneling range. The layer acts as an interphase between the metal and the solution and has the properties of a solid electrolyte with high electronic resistivity. For this reason it was called a “solid-electrolyte 3, 4 interphase” SEI. The batteries, consisting of SEI electrode, were called SEI 3, 4 batteries. SEI determines the safety, power capability, morphology of lithium 5-8 deposits, shelf life, and cycle life of the battery. For high performance of the lithium battery, the SEI must be an electronic resistor in order to avoid SEI thickening leading to high internal resistance, self-discharge and low faradaic efficiency (εf). To eliminate concentration polarization and to facilitate the lithium dissolution-deposition processes, the cation transport number should be close to unity. To reduce overvoltage, the SEI should be highly ion-conductive. In the case of the rechargeable lithium battery, it is very important that there be uniform morphology and chemical composition in order to ensure homogeneous current distribution. The SEI must be both mechanically stable and flexible. Good adhesion to the anode is important as well. As emphasized above, practical primary or secondary alkaline or alkaline-earth batteries can be made only if the dissolution or corrosion of the anode can be stopped. Therefore, the electrolyte must be designed to contain at least one material that reacts rapidly with lithium (or with the alkali-metal anode) to form an insoluble solidelectrolyte interphase — the SEI. The importance of the SEI is well recognized in the scientific community; special sessions are devoted to it in battery-related meetings such as the International Meetings on Li Batteries (IMLB), International Symposium on Polymer Electrolytes (ISPE), and in other meetings, including the Electrochemical Society (ECS) Battery Symposium in Japan and the Materials Research Society (MRS). Hundreds of papers dealing with the SEI study have been published (most of them in the last twenty years) and it is impossible to summarize all of them here. New techniques such as X-ray Photoelectron Spectroscopy (XPS), SEM, X-ray Diffraction (XRD), Surface-Enhanced Raman Spectroscopy (SERS), Scanning Tunneling Microscopy (STM), Energy-Dispersive X-ray Spectroscopy (EDS), FTIR, NMR, EPR, Calorimetry, DSC, TGA, QuartzCrystal Microbalance (QCMB), Atomic-Force microscopy (AFM) and in situ

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

3

Neutron Radiography have been recently adapted to the study of the electrode surface and the chemical and physical properties of the SEI. This chapter addresses several issues dealing with the mechanism of SEI formation on inert substrates, lithium, carbonaceous materials and tin-based alloys. Attention is currently focused on the correlation between the composition and morphology of the solid-electrolyte interphase forming on the different planes of highly ordered pyrolytic graphite (HOPG) and different types of disordered carbon electrodes in lithium-ion cells.

2

SEI Formation Processes and Morphology

2.1 The Main Principles and Routes of SEI Formation The deposition-dissolution process of an electrode covered by an SEI involves three consecutive steps, which are described schematically as follows: Electron transfer at the metal/SEI interface M°- ne
M

n+

(2.1)

M/SEI + M

Migration of cations from one interface to the other when t − migration of anions when tX =1) M

n+ M/SEI


M

=1 (or

n+

(2.2)

SEI/Sol + M

Ion transfer at the solid-electrolyte interphase/solution (SEI/sol). For t =1 m(solv) + M

n+

n+.

SEI/Sol


M m(solv)

(2.3)

In principle, any one of these could be the rate-determining step (r.d.s.). However, it was found, by the use of a variety of experimental techniques, that ionic migration through the SEI is the rate-determining step for many systems. In addition, it was found that the rate of nucleation of the metal deposit is 5, 6 affected by the interfacial resistance. This transport process is a key factor in the operation of non-aqueous SEI batteries. The standard reduction potential of lithium is more negative than that of the solvated-electron system (at least in highly purified ammonia, amines and ethers). This results in the formation of the well known blue solutions of 9, 10 − solvated electrons (e sol). In rechargeable batteries under prolonged dissolution, a process of breakdown and repair may take place. Mechanical breakdown can be caused by both local preferential dissolution of the anode and by stresses in the SEI due to uneven retreat of the anode. The new anode surface, exposed to the electrolyte, immediately reacts with it to form a fresh

4

Lithium-Ion Batteries: Solid-Electrolyte Interphase

thin protective film that slows further local corrosion. Because the solvated electron may take part in the early stages of SEI formation and in the break-andrepair healing processes during lithium plating and stripping, it is necessary that the formation and the healing of the SEI take place very quickly. This is especially important on graphite, during the first intercalation step. In addition, the SEI building materials must have extremely low solubility. Thus the electrolyte must be designed to contain one or more SEI precursors having high 0 standard electrode potential (E ) and high exchange-current density (i0) for reduction. However, the data bank of i0 for such reactions is limited. It was therefore suggested to use the data bank for the bimolecular rate constant (ke) 11 for the reaction:

e−aq +S →product − aq

(2.4)

where e is a hydrated electron and S is an electron scavenger and a candidate − material for a lithium-battery electrolyte. The reactivity of materials toward e aq − (in aqueous solutions) is expected to be quite close to that for esol in organic solutions. The data bank for ke in aqueous solutions contains information on 12, 13 more than 1500 materials. The first factor to take into account is that rate 10 −1 −1 constants higher than 10 M sec relate to diffusion-controlled reactions, which are expected to proceed very quickly at the lithium-electrode potential. Therefore SEI precursors should be chosen from this group or at least from the − 9 −1 −1 group having rate constants higher than 10 M sec . For instance, AsF6 and 14-16 CO2, which are good SEI precursors, have values of ke that approach those for diffusion-controlled reactions. Liquid electrolytes typically used in lithium batteries consist of a lithium salt dissolved in an organic solvent, or a mixture of solvents. The solvents fall into two general classes: ethers or alkyl esters of carbonic acid. “Inert” electrolyte components — for example ethers, which are chosen because of their very slow reaction with lithium (or with the LixC6 anode) must be taken from the group that has the smallest rate constant — 7 5 −1 −1 preferably smaller than 10 (or even 10 ) M sec . In many cases there is a good correlation between the SEI composition and − the reactivity of electrolyte components toward e aq. LiF and As-F-O species are 14, 17, 18 − − found in the SEI formed in electrolytes containing LiAsF6. BF4 and ClO4 − 6 0 are much less reactive toward e aq (ke < 10 ) and LiCl and B are rarely found in 17 the SEI in γ-BL solutions. γ-BL is expected to have a high ke, similar to that of − 7 acetone. Ether is kinetically stable vs e aq (ke <10 ), thus in ether-based solutions, 0 the anion may be reduced. Indeed, in ether-based solutions containing LiBF4, B 17 was found in the SEI. When CO2, which has a high ke, is added to the 19 electrolyte, more Li2CO3 is found in the SEI. EC is so far the best SEI-forming

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

5

precursor. We attribute this (in part) to its high i0, which is expected to be similar to that of DMO. Esters such as ethyl acetate and semicarbonates like 7 8 −1 −1 propylene glycol carbonate have moderate rate constants (10 -10 M sec ). This suggests that semicarbonate is not stable with respect to lithium and that the part of the SEI, which is close to the lithium (or the LixC6) anode cannot 14, 20 consist of semicarbonate as already reported and as was confirmed in Ref. 11. The voltage of SEI formation (VSEI) correlates with the reactivity of the − electrolyte components towards e aq as well; this reactivity, in turn, is directly − related to i0. In the case of reactive components like AsF6 , CO2 and EC, VSEI is more positive. However, for more kinetically stable (lower ke) substances, like − − + ClO4 (and probably PF6 and imide), VSEI approaches the Li/Li potential, i.e. the overpotential of the SEI formation process is higher. In order to estimate the relative contribution of the EMF and i0 to the value of VSEI, let us consider the 21 following example. The OCV of the Li/SO2/C cell and of the Li/EC, PC, DEC, DMC/C cells is about the same, 2.8-3 V. However, the SEI formation voltage on a carbonaceous (LixC6) electrode in the SO2-containing electrolyte is 2.4 V, almost 1 V more positive than that in EC, PC, DEC, DMC solutions (where VSEI varies from 0.6 to 1.5 V). This indicates that the kinetics has a more profound effect on the SEI formation voltage than does the thermodynamic parameter (OCV). On the basis of the data presented in Refs. 11 and 22, it was suggested that the rate constants for the reactions of solvated electrons with electrolyte and solvent components and impurities be used as a first screening tool for the selection of electrolyte components for use in lithium and lithium-ion batteries. A brief overview of experimental data supporting the above suggestion (namely the selection of precursors for rapid SEI formation from a group of materials having high rate constants) is presented below (see sub-chapters 3 and 4). 2.2 Structure of the SEI The transport of ions through the SEI, which consists mainly of polycrystalline 5 material, takes place by mobile point (Schottky or Frenkel) defects. As a result, the contribution of the grain boundaries must be taken into account. In the first 3, 4 models describing the SEI, its structure was represented as comprising two or more separate layers of different composition and properties. The first (the SEI itself), is thin and compact, while the second one (if it exists) on top of the SEI

6

Lithium-Ion Batteries: Solid-Electrolyte Interphase

is a more porous, or structurally open layer, that suppresses the mass transport of ions in the electrolyte filling its pores. According to this model, the SEI is made up of ordered or disordered crystals, which are thermodynamically stable with respect to lithium.

Figure 1 Schematic presentation of Polyhetero Microphase SEI. Reproduced from [8] by permission of The Electrochemical Society, Inc. 23

Later Thevenin and Müller suggested several modifications to the SEI model: (1) the polymer-electrolyte interphase (PEI) model in which the lithium in PC electrolyte is covered with a PEI composed of a mixture of Li2CO3, P(PO)x and LiClO4; P(PO)x is polypropylene oxide, formed by reduction-induced polymerization of PC; (2) the solid-polymer-layer (SPL) model, where the surface layer is assumed to consist of solid compounds dispersed in the polymer electrolyte; (3) the compact-stratified layer (CSL) — in this model the surface layer is assumed to be made of two sublayers. The first layer on the electrode surface is the SEI, while the second layer is either SEI or PEI. The first two

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

7

models do not seem to be relevant to lithium battery systems since the PEIs are not thermodynamically stable with respect to lithium. Perchlorate and fluoro8, 17, 19, 22-25 anions (but not halides) were found to be reduced to LiCl and LiF. 26 Aurbach et al. carried out an intensive electrochemical and spectroscopic study of carbon electrodes in lithium-battery systems. On the basis X-ray photoelectron spectroscopy measurements they suggested multi(3 or 5)-layered SEI structures. In our recent SEI study, we assumed that reduction of salt anions and solvents proceeds simultaneously and both organic and inorganic materials 8, 25 precipitate on the electrode as a mosaic of microphases. These phases may, under certain conditions, form separate layers, but we believe that it is more appropriate to treat them as polyhetero microphases (Fig. 1). The equivalent circuit for a mosaic-type SEI electrode is extremely complex. However, to a first approximation, a single-layer SEI is characterized by at least four RC elements in series and a Warburg impedance. These RC elements represent two interfaces-electrode/SE and SE/solution-SEI ionic resistance and capacitance, and grain-boundary resistance and capacitance. Each additional sublayer adds another three RC elements. The total SEI resistance (RSEI) in battery electrolytes 2 is typically in the range of 10-1000 ohm cm . The expected value for Rgb at 30ºC 2 for a 10 nm SEI is between 10-100 ohm cm , i.e. it cannot be neglected and Rgb and Cgb must be included in the equivalent circuit of the SEI, for both metallic 8 lithium and for LixC6 electrodes. We believe that in polymer electrolytes, lithium-passivation phenomena are similar to those commonly occurring in liquid electrolytes.

3

Chemical Composition and Properties of the SEI on Inert Substrate and Lithium

3.1 Inert Metal Substrate Much information about lithium deposition/dissolution on inert electrodes has been obtained over the past twenty years. Thorough studies of the chemical composition of surface films of lithium deposited on a nickel substrate in γbutyrolactone (γ-BL) and tetrahydrofurane electrolytes, containing various salts, 17 such as LiClO4, LiAsF6, LiBF4 and LiPF6 were carried out by Kanamura et al. With the use of XPS it was found, that the outer and inner layers of the surface film covering lithium in LiClO4/γ-BL involve LiOH or possibly some Li2CO3

8

Lithium-Ion Batteries: Solid-Electrolyte Interphase

and Li2O, as the main products. Chlorine and oxygen were distributed uniformly over the surface film. The authors suggested that the hydrocarbon observed in the C 1s XPS spectrum can be attributed both to a hydrocarbon contaminant and to organic compounds incorporated in the surface film. The chemical composition and the depth profile of the surface film formed in LiAsF6 +γ-BL and LiBF4+γ-BL electrolytes are very similar to what is observed in the case of LiClO4-based electrolyte. The only difference is the lithium halide present. However, the Li 1s spectra of LiPF6 +γ-BL electrolyte were completely different 17 from those observed for the other γ-BL electrolytes. The surface film consists of LiF as the main component while close to the bottom, the SEI consists mainly of Li2O. The depth profiles of the SEI in THF were similar to those obtained in γ-BL, but the distribution of LiOH and Li2O was slightly different. The surface film obtained in LiClO4+THF probably has a layer of mixed 27 LiOH+Li2CO3 that is thicker than that obtained in LiClO4+γ-BL. The chemical species in the surface film of lithium deposited in LiBF4+THF electrolyte are not very different from those in LiBF4+γ-BL. However more LiF and other fluoride compounds are formed and chemical species including elemental boron were − also observed in the B 1s spectra. This means that BF4 ion reacts quite strongly with lithium during electrochemical deposition in THF electrolyte. According to the depth profile, the surface film comprises mainly a mixture of LiF, LiOH and 17 Li2CO3. The carbon content in LiAsF6+γ-BL, probably associated with polymeric compounds, was greater than that in LiAsF6+THF. This indicates high reactivity of γ-BL (high ke) vs Li and is in agreement with the screeningtool approach addressed above. Because of lower volatility and higher flash point, the organic carbonates are the preferred solvent class in commercial lithium batteries. The structure and composition of the surface film on lithium in carbonate-based electrolytes has 17, 19, 23, 27-39 been extensively investigated. High reactivity of propylene carbonate (PC) toward the bare lithium metal is expected since its reduction on an ideal polarizable electrode takes place at potentials much more positive than those for 40 THF and 2-methyl-THF. The electrochemical reduction of five organic carbonates, EC, PC, DEC, DMC and VC were studied by cyclic voltammetry at 41 a gold electrode in THF/LiClO4 supporting electrolyte. The reduction + potentials for all carbonates were above 1V (vs Li/Li ), the EVC being the most positive. It was shown that the preferential reduction of vinylene carbonate 41 appears to favor SEI formation. 16 Aurbach and Zaban have found that the lithium surface deposited on a nickel electrode in carbonate-based electrolytes is covered with Li2CO3, LiOH,

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

9

Li2O, LiOR, LiOCO2R (R=hydrocarbon) and lithium halide. Their recent studies revealed that traces of oxygen, water and PC reduction products form passivating surface films on gold, platinum, silver, nickel and copper electrodes, 42 when these are polarized to low potentials in lithium-salt solutions. These films act as solid electrolyte interphases, i.e., they allow transport of alkali metal ions through them. The study also found that the major constituent in the surface films is the PC reduction product CH3CH(OCO2Li)CH2-OCO2M. The analytical tools used in these studies included cyclic voltammetry, electrochemical quartz-crystal microbalance, surface-sensitive Fourier transform spectroscopy and X-ray photoelectron spectroscopy. When a small amount of HF is added to PC containing 1.0 M LiClO4, the lithium deposited on a nickel substrate is covered with a thin LiF/Li2O surface film. However, without the addition of HF, the surface of electrodeposited lithium is covered with a thick film (mainly LiOH and Li2O). These data also support our previous statement on the importance of relatively constant rates of salt-anion and solvent reduction when choosing an electrolyte for rapid SEI formation. 43 It was confirmed by surface-enhanced Raman scattering (SERS) that the main stable species of the SEI film on the surface of the discharged silver electrode in 1M LiPF6/EC:DEC electrolyte are LiOH•H2O and Li2CO3. However, SERS cannot indicate the presence or absence of LiF in the SEI since this substance is inactive in the Raman spectrum. ROCO2Li was found in 1 M LiClO4 PC:DMC electrolyte. In PC-based solutions the order of the interfacial resistance of the SEI for different salts was as follows: LiPF6 >> LiBF4 > 19 LiSO3CF3 >> LiAsF6 > LiN(SO2CF3)2 > LiBr, LiClO4. The RSEI values for 2 LiPF6/PC and LiN(SO2CF3)2 /PC were about 800 and 23 ohm cm , respectively. Such high RSEI may be caused by the presence of highly reactive HF in the LiPF6 electrolyte. The addition of CO2 to solutions considerably reduced the interfacial resistance. A comparison of the SEI properties on bare lithium in four electrolytes, LiClO4/PC, LiClO4/PC-DME, LiAsF6/EC-2MeTHF and LiAsF6/THF-2MeTHF 34 was made by Montesperelli et al. using impedance spectroscopy. After 10 days of storage, the resistance of the passivating film in LiAsF6-based solutions was 2 found to be twice that in LiClO4 electrolyte. High values of Rfilm (~45 ohm cm ) in THF-containing electrolyte were explained by the high reactivity of this solvent towards lithium, followed by the formation of a thick (~220 Å) surface film. It was found that the SEI in LiPF6/γ-BL electrolyte is much thinner than 17 those formed in LiAsF6, LiClO4or LiBF4/ γ-BL-based electrolytes. The SEI thickness was found to be less than a few tens of angstroms in LiPF6 + γ-BL,

10

Lithium-Ion Batteries: Solid-Electrolyte Interphase

while for other electrolytes it exceeds 200 Å. Moreover, the film formed in the LiPF6-containing electrolyte was very uniform and sufficiently compact. The thickness of the lithium surface layer in a lithium perchlorate/propylene carbonate solution, as calculated from the apparent resistance according to the CSL interface model, was found to increase exponentially with storage time 23 from 100 to 1000 Å. The values obtained are in good agreement with those 35 deduced from ellipsometric measurements.

3.2

Lithium Covered by a Native Film

The presence of a native film on lithium does not significantly affect the surface chemistry of the SEI formation. The outer part of the lithium native film 29 consists of Li2CO3 or LiOH and the inner part is Li2O. As with lithium deposited on an inert substrate, the XPS spectra of lithium electrodes, covered by native film and treated in fluorine-containing salts such as LiAsF6, LiBF4, LiPF6, Li-imide and Li-triflate dissolved in THF, always show fluorine, oxygen 19 and carbon peaks. Methyl formate is the most reactive solvent toward lithium, as compared to other polar aprotic solvents including ethers, BL, PC and EC. Even in the presence of trace amounts of water and methanol contaminants, it is 30 reduced to lithium formate as a major precipitate. The presence of CO2 in MF causes the formation of a passivating film containing both lithium formate and lithium carbonate. In PC solutions, the lithium solid electrolyte interphase was 16 shown to be mainly a matrix of Li alkylcarbonates. In PC-based electrolytes, LiPF6, LiBF4, LiSO3CF3, LiN(SO2CF3)2 were found to be more reactive toward 19 lithium than were LiClO4 and LiAsF6. In LiAsF6 PC/THF electrolyte, the concentrations of As and F in the SEI decrease. This indicates that the addition of reactive PC to the ether suppresses salt reduction by competing with it, and the film becomes more organic in nature, containing less LiF. In the case of EC/PC or EC/ether mixtures, the reduction of EC by lithium seems to be the 33 dominant process, followed by the formation of lithium alkyl carbonates. Addition of cyclic compounds with heteroatoms and conjugated double bonds, such as 2-methyl-thiophene (2MeTp) 2-methylfurane (2MeF), and aromatic compounds like benzene are very effective in electrolyte solutions for 37, 38 rechargeable lithium batteries. In the presence of surfactants like polyethylene glycol dimethylether and a mixture of dimethyl silicone and propylene oxide in EC/DMC solutions, the smooth surface morphology and 44 almost constant thickness of the lithium passivating film was detected. Several techniques have been applied to the electrochemical analysis of lithium

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

11

electrodes in a large variety of electrolyte solutions. These include chrono26 potentiometry, cyclic voltammetry, transient methods, fast OCV measurements and impedance spectroscopy. An important finding as a result of these studies is that the passivation of fresh lithium in polar aprotic electrolytes may be completed less than one second. Using a Voigt-type analog model, Aurbach calculated the thickness and resistivity of the SEI. The average thickness of the interphase next to the lithium is about 30-50 Å and the resistivity is on the order 8 of 0.1-0.2 ·10 ohm cm. The resistivity of the more porous part of the SEI on the 8 solution side is estimated to be 3-4 ·10 ohm cm. The higher resistivity of the SEI in LiPF6 and LiAsF6 solutions as compared to other salts, was explained by the replacement of the ROCO2Li surface species by LiF. The author concluded that it is not an increase in the SEI thickness, but rather resistivity changes that lead to the high interfacial impedance of the lithium anode in LiPF6 and LiAsF6 electrolytes. Li2CO3 is stated to be one of the best passivating agents for the enhancement of lithium cycling efficiency.

3.3

SEI Formation in Solid Polymer and Gel Electrolytes

The major differences between polymer (PE) and liquid electrolytes result from the physical stiffness of the PE. PEs are either hard-to-soft solids, or a combination of solid and molten-phase equilibrium. As a result, wetting and contact problems are to be expected at the Li/PE interface. In addition, the replacement, under OCV conditions, of the native oxide layer covering the lithium by a newly formed SEI is expected to be a slow process. Aside from these differences it seems likely that in polymer electrolytes, especially in the gel types, lithium-passivation phenomena are similar to those commonly occurring in liquid electrolytes. Results obtained with PEGDME electrolytes containing different salts showed that the formation of LiF as a result of the − − reduction of anions like AsF6 or CF3SO3 , plays a key role in the lithium45 passivation mechanism. The authors showed that SEI formation was apparently complete in just 2-3 minutes. The increase in the SEI resistance (RSEI) over hours and days is apparently due to the relaxation of the initially formed passivating films or to the continuation of the reaction at a much slower rate. The formation and properties of the lithium SEI was studied in different types of hybrid and gel electrolytes based on polymers and organic solvents combined 46 with organic or inorganic gelation agents. LiI-tetraglyme-based hybrid electrolytes with PVDF-silica membrane form SEIs that are highly stable for more than 3000 hours. The three-fold increase in

12

Lithium-Ion Batteries: Solid-Electrolyte Interphase

the RSEI of LiImide-HPEs after 300 hours of storage and about an order of magnitude increase after 2000 hours, provides evidence of the thickening with time of the passivating layer on lithium. The effect of salt on the stability of the SEI was similar in PVDF-SiO2 and PVDF-Al2O3 HPEs of different porosities. Low and almost constant resistance of the SEI in lithium iodide-containing electrolytes, may be associated with high thermodynamic stability of the iodide anion towards metallic lithium. In hybrid electrolytes with Tefzel membranes, even the initial RSEI value was twice that in HPEs with PVDF membranes, and 2 the RSEI increased sharply (up to 80 ohm cm ) after 300 hours of storage. These data support previous observations that nano-size ceramic fillers incorporated in a polymer membrane improve interfacial resistance in hybrid as well as in 25, 47 composite solid polymer electrolytes; this is due to their ability to adsorb impurities and traces of water. In addition, inorganic fillers may prevent free diffusion of the liquid electrolyte components to the lithium surface and, as a result, inhibit the growth of the SEI. It is worth noting that the resistance of the lithium passivating film in alumina-containing HPEs is twice that in the silicabased HPEs. The same RSEI difference was detected between Al2O3- and SiO2 highly porous and less porous HPEs. At room temperature, the initial RSEI in the 2 LiI-tetraglyme- and LiI-PEGDME-based HPEs was about 200 ohm cm , while 2 in LiImide it was 280 ohm cm . After 200 hours of storage the RSEI increased by about 10% in the former electrolyte and by about 25% in the latter. The initial interfacial resistance of EC: DMC-PAN-based gel electrolytes at room temperature was about the same order of magnitude as in TG-, PEGDME- and EC:DMC-based hybrid electrolytes. The RSEI stability, however, was much lower than that of the hybrid electrolytes and the RSEI 48 increased up to 1.5 k after 900 hours of storage. Thus, doubts are raised as to the inert nature of the PAN matrix with respect to lithium passivation. The reactivity of PAN may stem from impurities in the commercial product and reactivity of the –CN group possibly leading to the formation of LiCN. Dissolution of LiCN could be followed by the breaking and thickening of the SEI. Contrary to EC:DMC solvents, lithium oxide, carbonate, fluoride compounds and alkoxides, which are the basic compounds found in the anode SEI, are highly insoluble in tetraglyme and polyethylene glycol dimethyl ether 22 solvents similar to solid PEO. This property, as expected, increases the stability of the lithium passivation layer by producing a thinner and more compact film. The morphology of lithium deposits from 1-3 M LiClO4-EC/PC-ethylene 49 oxide (EO)/ propylene oxide (PO) copolymer electrolytes was investigated. It

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

13

was found that as the weight ratio of host polymer to liquid electrolyte increased, fewer lithium dendrites were formed, with no dendrites found in electrolytes containing more than 30% w/w host polymer. The authors emphasized that good contact between the polymer and lithium is also of great importance for the suppression of dendrites. Direct in situ observation of lithium 50 dendritic growth in Li-imide P(EO)20 polymer electrolyte shows that dendrites grow at a rate close to that of anionic drift. The interfacial phenomena in solid LiX/PE systems were extensively 51 studied by Scrosati et al. For the dry PEO-based polymer electrolytes it was shown that the interfacial stability can be significantly enhanced by decreasing 52, 53 the ceramic particle size to the scale of nanometers. The mechanism of the processes leading to improved stability is not well understood and some explanations include scavenging effects and screening of the electrode with the 52, 54 ceramic phase.

4 4.1

Carbonaceous Electrodes Principles of SEI Formation

Lithium-ion batteries occupy a large and increasing share of the rechargeablebattery market as a result of their excellent performance in terms of cycle life, energy density, power density and charge rate. However, for the successful use of carbon electrodes in secondary lithium-ion batteries, much work, such as the selection of high reversible and low irreversible capacity carbons, as well as understanding the complex mechanism of lithium-ion intercalation into lithium, has still to be done. 55-57 Surface structure and chemical composition affect the physicochemical properties of carbon. The most important parameters determining the use of carbons as anode material are particle shape and size, pore-size distribution and pore-opening, BET surface area and content of surface species and impurities. The basic building block of carbons is a planar sheet of carbon atoms arranged in a honeycomb structure (called graphene or basal plane). These carbon sheets are stacked in an ordered or disordered manner to form crystallites. Each crystallite has two different edge sites: the armchair and zigzag sites. The reactivity of carbon atoms at the edge sites (and near lattice defects and foreign atoms) is much higher than that of carbon atoms in the basal 55-57 planes. Consequently, the physical and chemical properties of carbon vary

14

Lithium-Ion Batteries: Solid-Electrolyte Interphase

with the basal-plane to edge-plane area ratio. The surface area of carbon 2 powders varies over a wide range from less than a few m /g for large-particle 2 graphite powders to more than 1000 m /g for high-surface-area carbons. As a result, the content of surface groups or heteroatoms, measured as the ratio of 55-58 foreign atom to C varies from nearly zero up to 1:5 in the case of hydrogen. Carbons may have closed and open pores with a large variety of dimensions from a few Å to several microns. The edge atoms in completely closed pores are 58 actually radicals and are said to have a “dangling” bond. These pores are 59 responsible for the “extra” reversible capacity of disordered carbons and 60-63 oxidized graphite. Physicochemical properties of carbon such as wettability, 55-57 catalysis, electrical and chemical bonding to other materials, are strongly dependent on the surface oxygen species, which can have basic, neutral or acidic nature. Since lithium-ion cells typically operate beyond the thermodynamic stability of the organic electrolytes, there occur, along with lithium intercalation, other electrochemical and chemical reactions during the first few cycles. The reduction products of the these reactions form passivating films on the carbon surface, and these produce electrically insulating layers, similar to the SEI formed on lithium. In lithium-ion batteries the first intercalation capacity is larger than the first deintercalation capacity. This difference is the irreversible 64 capacity loss (QIR). Dahn et al. were the first to correlate QIR with the capacity required for the formation of the SEI. They found that QIR is proportional to the specific surface area of the carbon electrode and, assuming the formation of a Li2CO3 film, calculated an SEI thickness of 45±5 Å on the carbon particles, 4, 5 consistent with the barrier thickness needed to prevent electron tunneling. The SEI not only dramatically slows the kinetics of electrolyte decomposition but also reduces active lithium consumption by forming a physical barrier between the lithiated carbon electrode and the electrolyte. It was concluded that when all the available surface area is coated with a film of the decomposition products, 64 further decomposition ceases. The first lithium intercalation to the carbon, schematically presented in 7, 22, 62, 65 Figure 2, is very complex. The current understanding of this process is based on the principle that solvated lithium ions in the electrolyte lose their solvation shells while penetrating the SEI and are incorporated into the carbon structure in a solvent-free form. Such reactions are to be desired. In some cases, however, lithium intercalates together with its solvate shell, thus causing exfoliation of the electrode; these, of course, are undesirable processes. Exfoliation may result in complete destruction of the structure of graphite, large

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

15

irreversible capacity and almost zero reversible capacity in cases where the reduction of the solvated molecules produces gas. In Refs. 11 and 22, it was shown that in order to slow the co-intercalation of the solvated ion, and to enhance the formation of the SEI at the most positive potential (far from the + Li/Li potential), the solvents appropriate for lithium-ion batteries employing a 0 graphite anode must have high solvation energy, high E and high i0 for reduction.

Figure 2 Schematic presentation of the SEI formation on carbon. Reproduced from [7] by permission of the Materials Research Society.

At the electrode surface there is a competition among many reduction reactions of salts, solvents and impurities, the rates of which depend on i0 and η for each process and on the catalytic properties of the carbon surface. The products of reduction of salt anions are typically inorganic compounds like LiF,

16

Lithium-Ion Batteries: Solid-Electrolyte Interphase

LiCl, Li2O, which precipitate on the electrode surface. Reduction of solvents is followed by the formation of both insoluble SEI components like Li2CO3 and partially soluble semicarbonates and polymers. The voltage at which the SEI is formed depends on the type of carbon, the catalytic properties of its surface (ash content, type of crystallographic plane, basal-to-edge plane ratio), temperature, concentration and types of solvents, salts and impurities, and on the current density. For lithium-ion battery electrolytes, VSEI is typically in the range 1.711 0.5 V vs Li reference electrode, but the SEI continues to form down to 0 V. In 66 some cases, εF is less than 100% during the first few cycles. This means that the completion of SEI formation may take several charge-discharge cycles. In addition to the building of the SEI, QIR may be caused by capacity 7, 62 associated with the formation of soluble reduction products (QSP), with the 59 trapping of lithium inside the structure of the carbon (QT), and with unused capacity under specified experimental conditions (Qu). QSEI as well as VSEI depends on the morphology of the carbon and should increase with the ratio of cross-sectional plane area to basal-plane area. This conclusion stems from the 21 data reported by Besenhard et al. on the penetration of the passivating layer into the graphite galleries through the cross-sectional planes. This is in agreement with Ref. 67, where the thickness of the SEI at the cross-sectional planes of an HOPG crystal was found to be greater than that of the basal plane. Factors that are reported to decrease QIR are: increasing the EC content in 68, 69 21, 33, 69 organic carbonates or dioxolane solutions, addition of CO2 or crown 70, 64, 71 61 ethers and increasing the current density.

4.2 SEI Composition and Morphology The chemical composition of the SEI formed on carbonaceous anodes is, in general, similar to that formed on metallic lithium or inert electrodes. However, the variety of morphologies and chemical compositions of carbon surfaces can affect the i0 value for the various reduction reactions and, therefore, cause compositional differences of the SEI. Solvent co-intercalation must be taken into account as well. 4.2.1

HOPG

It is now established that lithium intercalation takes place through the cross section of the graphite. It has been deduced from many experiments, as reported by the groups of Besenhard, Ogumi, Farrington and Yamaguchi, who used

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

17

STM, AFM and dilatometry to study the early stages of lithium intercalation into HOPG, the SEI functions differently on the different planes of graphite particles. On the basal plane, it is enough to have an electronically nonconducting film, while on the cross section (zigzag and armchair planes) the SEI must also be a good lithium conductor. The difference in functioning of the two SEIs is accompanied by a difference in composition. It is therefore important to study separately the composition and properties of the SEI on these two planes (basal and cross section). Carbon atoms on the cross-section zig-zag and armchair planes were found to be much more active than carbon atoms on the basal plane of glassy carbon and highly ordered pyrolytic graphite 55, 57, 62, 63 (HOPG). Thus HOPG, which is considered a special case of graphite materials, analogous to a single crystal, was used as a model electrode in our 4, 5, 73 recent study. Several results using XPS and TOF SIMS. In order to obtain information on the chemical composition of the SEI and the depth distribution of SEI-forming materials, high-resolution XPS spectra were recorded for different sputtering times. The intensity, the shape and the position of the main peaks in the C1s and O1s spectra of the cross-section and basal SEI were found to change on sputtering, indicating different SEI composition on the solution-side surface, in the bulk and at the bottom. Figure 3 shows carbon, oxygen and fluorine 1s spectra of the SEI formed on the cross-section of HOPG in LiPF6 EC:DEC electrolyte. A strong, broad peak with two shoulders is clearly seen in the initial (zero sputtering time) carbon spectrum. The central part of the peak with a maximum at the binding energy of 284.5 eV, is assigned to polyolefins. The shoulder appearing in the vicinity of 285.5 eV is more likely to be due to C-O-H and/or C-O-C bonds; the latter may be associated with oxygen-containing polymeric species formed on solvent decomposition. The shoulder at 284 eV is related to carbon atoms in the 73 Li-O-C group. This group may be a part of an alcoholate molecule or attached 74 to the graphite surface group. After 20 minutes of sputtering, the shoulder attributed to C-OH, C-O-C and C=O groups disappears and the maximum of the peak shifts toward XPS bond of Li-O-C group. (The peak at 283.5 eV, which appears after 50 minutes of sputtering, may result from the shifting of the 2 carbon sp peak.) No carbonates were found in the XPS spectra of the SEI built on the cross section of HOPG. Figure 4 shows the change in atomic concentration of the elements found as a function of sputtering time. In interpreting the concentration XPS depth profile it should be remembered that the sputtering efficiency depends on the type of material, it may be higher for organic materials and lower for stable inorganic materials. Some material

18

Lithium-Ion Batteries: Solid-Electrolyte Interphase

decomposition and surface chemical reactions are to be expected. These factors may, to some extent affect the concentration depth profile, but in our opinion would not change it drastically.

Semicarbonate Polymer Carbonate Li2C2 C-OH C-O-C Li-O-C

a

Carbonate Semicarbonate Li2O C-OH C-O-C Li-O-C

b

LiF F1s

O1s

C1s

c

0 min

0min 0 min 4min

4 min

4 min 20min

20 min 20 min

50min

294

292

50 min

290

288 286 284 BindingEnergy(eV)

282

280

540

538

50 min

536

534 532 530 528 526 Binding Energy (eV)

692

690

688

686 684 Binding Energy (eV)

682

680

Figure 3a High-resolution XPS spectra (a-c) at different times of sputtering of the SEI formed on the cross section of HOPG. b The SEI formed at 2 mA/cm2. c Sputtering rate 0.5 nm/min calibrated for SiO2 .

The carbon signal decreases sharply after 4 minutes of sputtering. This is accompanied by drop in the atomic concentration of oxygen and may indicate that organic compounds, such as polyolefins and polymers or oligomers containing oxygen, are present only at the SEI surface close to the solution. The further increase in the atomic concentration of carbon can be explained by a signal arising from the HOPG underlying the passivating film. The atomic concentrations of fluorine and lithium are much higher than those of carbon, oxygen and phosphorus. (The concentrations of phosphorus- and oxygencontaining compounds were found to be less than 10%). In addition, it should be mentioned that in the bulk SEI on the cross section of HOPG the Li/F ratio is

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

19

close to one, thus the oxygen may be bound to organic or phosphorus compounds.

d

100 C ross Section

A tom ic Concentration (% )

90 80 70 60 50

Li

F

Li

40 30 20 10 0

C P

0

O

10

20

30

40

50

60

70

80

90

Sputtering tim e (m in)

Figure 4 The depth profile of the SEI formed on the cross section of HOPG.

The distinctive feature of the basal SEI is the presence of 10-30 atomic % Li2CO3 on the surface and in the bulk (Figs. 5 and 6). From the O1s spectra and depth profile it seems likely that carbonates and semicarbonates are the main oxygen-containing species on the solution-side surface of the SEI. The Li2O peak can be seen after 4 minutes of sputtering (i.e. at a smaller SEI depth, as compared to that of the cross section). A dramatic increase in the carbon signal from the basal HOPG matrix was observed at about 2 nm depth while for the cross-sectional matrix, the HOPG carbon signal was seen at about 30 nm. This shows unambiguously that the basal SEI is thinner than the cross-sectional one. The estimated depth-dependent SEI composition profiles were constructed with the use of a least-squares curve-fitting technique and deconvolution procedure.

20

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Semicarbonate Polymer =O Carbonate C=OC Li2C2

a

Carbonate Semicarbonate C=O Li2O C-OH C-O-CLi-O-C

b

C-OH C-O-C Li-O-C

C1s

c

LiF

F1s

O1s

0min

0min 4min

4min

0min 10min

4min 10 min 10min

16min

294

292

16min

290

288

286

284

BindingEnergy(eV)

282

280

16 min

540 538 536 534 532 530 528 526

BindingEnergy(eV)

692

690

688

686 684 Binding Energy(eV)

682

680

Figure 5a High-resolution XPS spectra (a-c) at different times of sputtering of the SEI formed on the basal plane of HOPG. b The SEI formed at 2 mA/cm2. c Sputtering rate 0.5 nm/min calibrated for SiO2 .

Figure 7 depicts the proposed SEI composition as a function of sputtering time. The cross-section SEI (Figure 7a) consists mainly of LiF (about 90 atomic % in the bulk of the SEI) with some polymers at the solution side of the SEI and some LiOC groups. Li2O was found only at the bottom of the SEI. It should be mentioned that no lithium carbonate was found either on the solutionside surface of the cross-sectional SEI, or in the bulk. The basal SEI (Figure 7b) contains much less LiF and much more polymeric material (about 50% at the surface of the SEI). From such compositional difference it was deduced that the solvents are preferentially reduced on the basal plane and the salt anion is preferentially reduced on the cross section.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys





21

 

     

!    








  

 



 



















     

Figure 6 The depth profile of the SEI formed on the basal plane of HOPG.

The XPS study of the composition of the SEI formed on HOPG in the LiAsF6-electrolyte was carried out in the same way. Figure 8 shows the proposed chemical composition of the passivating films. As in the previous case, there is a major difference between the chemical compositions of the basal and cross-sectional SEIs. The solution side of the basal SEI consists mainly (about 65%) of polymers, whereas the solution side of the cross-sectional SEI contains only about half that amount. The LiF concentration on the solution side of the SEI is larger on the cross-section and rises up to 70% of the SEI after between 5 and 20 minutes of sputtering. The main compositional difference between the passivating layers in the two electrolytes under investigation is the presence of carbonates (20% in the CS, and 10% in BS) in the SEI that forms in the LiAsF6 electrolyte. From the elemental depth profiles, it is clear that, as with the LiPF6 electrolyte, the basal SEI in the LiAsF6 electrolyte is thinner than the cross-sectional SEI. Similar SEI compositional and thickness features were found in the SEI formed on the cross-section and basal planes in the LiClO4 electrolyte.73

Lithium-Ion Batteries: Solid-Electrolyte Interphase

22



"#$ %

& '

& *

& <'<

 >





 !

       







"#$



@ 

 !



Solution side

"#$ %

& '

& *

& <'<

 >

O-cont. polymers







Gr side









"#$



Figure 7 Estimated SEI composition on HOPG in LiPF6 EC:DMC electrolyte. Reproduced from [74] by permission of the Journal of New Materials for Electrochemical Systems.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

23

 

"#$ %

& '

& *

& <'<

 >

 !





















"#$

@

"#$ %

& '

& *

& <'<

 >



 !





















"#$

Figure 8 Estimated SEI composition on the cross-section and basal planes in LiAsF6 electrolyte.

Time-of-flight secondary-ion mass spectrometry (TOF SIMS) measurements were carried out for the first time at one-micron lateral resolution for a study of the SEI formed on the different planes of HOPG in LiPF6 EC:DEC electrolyte.74, 75 According to the TOF SIMS spectra, fragments containing lithium, fluorine, oxygen, carbon and hydrogen were found in the SEI on both the cross-section and basal planes. However, the number of counts

Lithium-Ion Batteries: Solid-Electrolyte Interphase

24

recorded at each mass from different HOPG planes was significantly different. For instance, the normalized fluorine intensity (number of F counts divided by the total number of secondary-ion counts and multiplied by 105) was 26,563 for the cross-section SEI and 6,065 for the basal. The normalized intensity for CH species was 3,120 and 12,992, respectively. Thus, it is clear that lithium and fluorine dominate the cross-section SEI spectrum and organic species dominate the basal spectrum in good agreement with the XPS data. The solution-side surface of the basal SEI is very rich in C2H3O. The normalized intensity of C2H3O is four times that of lithium, indicating that most of it cannot be bound to LiOC groups in alkylcarbonates or alkoxides. We suggest assigning the excess of C2H3O2 fragments to polymers containing oxygen groups. Heavy fragments represented by masses 280.023 and 144.707 were found only at the surface of the SEI (Figure 9).

CS











J'

J

*

'









#()#

#()#















%

"'



%% Y

%% Y





%

%

%

%

%

%

%

%

%

BS









%

%

%

%

%% Y

%% Y



"'











%

%





'

J

J'

*

#()#

#()#





%

&'



&'



%

%

%

&'



%

%

%

%



%

%

% % % &'



%

%

%

 $# Z[ \[  ] # ##\ ]  [ Z'* ^_ % % ] [ `^ ] \  [  %%< %  \ >%# #% ] J'"{Y Reproduced from [74] by permission of the Journal of New Materials for Electrochemical Systems.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

25

From the depth profiles we attempt to get a rough estimate of the SEI thickness. As with the XPS tests, in TOF SIMS measurements, molecular species can be damaged and fragmented when sputtered. Thus the depth profile of organic materials may represent their destruction with time and not only their concentration vs depth. The apparent SEI thickness was calculated by taking the depth at half-signal intensity. The intensity of secondary-ion counts of all species falls to one half of its maximum value after about 0.5 to 2.5 nm at the basal plane and about 1.7 to more than 30 nm (for fluorine) at the cross section. The secondary-ion signal of the same species falls 1.7 to 5.5 times (or more) faster at the basal plane than at the cross section. SEI thickness measured by XPS for SEI formed under similar conditions is 7 nm for basal SEI and 35 nm for cross-section SEI. A thick SEI at the cross section was generally explained by solvent cointercalation and exfoliation of the graphite.21, 76-78 This must be followed by the formation of carbonates as a result of solvent reduction. However, carbonates were not found, either on the solution-side surface or in the bulk of the cross-section SEI. As opposed to the cross-section SEI, the bulk SEI contains up to 30 atomic % of lithium carbonate. Therefore its absence in the cross-section SEI cannot be explained only by the reduction of Li2CO3 to alkoxides. XPS data showed that the major cross-section SEI compound is LiF, (up to about 90 atomic % — excluding H atoms — in the bulk). The normalized counts of PO3− species, recorded by TOF SIMS at the cross section, are about three times higher than at the basal plane. These two findings clearly show that the exchange-current density for LiPF6 (and for HF) reduction is much higher at the cross section than it is at the basal plane. We believe that this is the reaction that determines the cross-section SEI composition. The increased thickness of the cross-section SEI as compared to the basal may be explained by the cointercalation of ion aggregates like Li2PF6+ at the beginning of SEI formation. This is followed by Li2PF6+ reduction to the PO3- species detected by the TOF SIMS. Possible cointercalation of protons as products of HF dissociation cannot be excluded either. Partial exfoliation of graphite by Li2PF6+ and hydrogen penetration may also cause an increase in the SEI thickness. Polyolefins do not conduct ions and their content in the SEI is an important issue. On the one hand, their softness can add flexibility to the SEI and fill voids. On the other hand, excessive polymer content can block lithium migration in the SEI. Polymers may cause uneven current distribution and uneven lithium intercalation. At elevated temperatures, polymers and other organic materials (such as alkylcarbonates) may dissolve, leading to discontinuity in the SEI and this can trigger a thermal runaway. Many efforts

26

Lithium-Ion Batteries: Solid-Electrolyte Interphase

were made to detect polymers in the SEI. Continuous production of active alkyl radicals in the electrolyte during the first lithium intercalation was detected by Endo with the use of the electron-spin resonance method (ESR).24 Evidence of hydrocarbons in the SEI was found by SIMS analysis.79 The appearance of large numbers of counts of C and CH species, especially at the basal plane supports previous claims for the existence of polymers in the SEI. The most convincing evidence is the CH2 sequence found on the basal plane. A sequence of masses — 311, 325, 339, 353, 367 — with a common difference of 14 can clearly be assigned to polyolefins (CH2)n (Figure 10).

Figure 10 Mass spectra for the sequence (CH2)n of the basal SEI. Reproduced from [74] by permission of the Journal of New Materials for Electrochemical Systems.


Figure 11 depicts surface TOF SIMS images for the SEI on the basal and on the cross-section planes at 100 micron lateral resolution. The brighter the image, the greater the number of secondary ion-counts recorded for the corresponding mass fragment. The non-homogeneous chemical structure of the SEI can be seen at a glance. The SEI on the cross-section is dominated by lithium and fluorine, with one to several dozen micron-sized regions where lithium and fluorine are almost absent. PO3 species are distributed as a few micron-sized islands. In basal SEI, lithium and fluorine are concentrated in large

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

27

regions (100 microns), with some smaller micron-sized particles. There is little or no correlation with graphite topography. The distribution of C2H (and other CxHy-based fragments), O, C2H3O2 (59), and C2H3O (43) indicates full coverage and is fairly uniform. C2H3O is the major component of the basal SEI. Since the TOF SIMS images for all negative and positive ions were recorded simultaneously from almost the same region, it is fair to say that there is an overlap of lithium and fluoride distributions with those of the organic species; this indicates an intimate mixture of these species at the at the submicron level. This is in agreement with our recently developed polyhetero microphase structure model of the SEI. The chemical structure of the cross-sectional SEI after 30 minutes of sputtering remains unaltered in general, apart of the nonhomogeneous The chemical structure of the cross-sectional SEI after 30 minutes of sputtering remains unaltered in general, apart of the non-homogeneous intensity of the signal due to lithium and fluoride secondary ions in the bulk of the basal SEI was too small to construct an image. To estimate the lateral size of SEI-forming microphases we analyzed several different areas of the crosssection SEI at 1micron lateral resolution. These areas are marked by colored solid lines. The corresponding plots represent the number of secondary-ion counts recorded along each line as a function of distance. As can be seen from the width of the peaks, the true lateral dimension of the SEI microphases does not exceed 2 microns (Figure 12).

CS

BS

Li

F

PO3(CS)/O (BS)

C2H

C2H3O

Figure 11 TOF SIMS ion images of the surface SEI on the cross-section and basal planes of the HOPG (100 µm resolution). Reproduced from [74] by permission of the Journal of New Materials for Electrochemical Systems.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

28

Figure 12 TOF SIMS image and lateral size profiles of the SEI-forming microphases at 1 µm resolution. Reproduced from [75] by permission of Elsevier Science Ltd.

4.2.2

SLX20

It is now generally accepted that the surface chemistry and morphology of the edge planes of graphite play a major role in the chemical and electrochemical reactivity of this material in contact with electrolyte. In order to determine whether there is a correlation between the composition and morphology of the SEI formed on the HOPG and on the real anode in lithium-ion batteries, we

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

29

recently performed XPS and TOF SIMS characterization of the SEI on the SLX20 (TIMREX, TIMCAL Ltd.) based electrode. In graphite and carbon powders as typically used, the proportion of prismatic planes visible to XPS and TOF SIMS is simply too low for any effects to be unambiguously determined. This is true even when one uses SFG 6 with its smaller particle size and thus higher proportion of edge area.80 It seems likely, therefore, that powders are unsuitable as models for SEI studies. To overcome this problem, SLX20 graphite was mixed with Teflon (5%) and pressed to pellets at 1 MPa/cm2. As can be seen from the SEM micrographs (Figure 13), such an electrode has well defined cross-section and basal graphite surfaces. Lithium cells were assembled from these pellets and cycled between 0.00 and 2.00 V vs Li/Li+ in LiPF6 EC:DEC electrolyte.

Figure 13 SEM micrographs of cross-section and basal planes of the composite SLX20-based electrode.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

30

Two strong C1s peaks at 284.6 eV and 286 eV are observed at the crosssection surface before sputtering (0 time) (Figure 14a). These peaks can be attributed to polyolefins (284.6 eV) and to ether carbon in CH3OCO2Li or (CH2OCO2Li)2 compounds (286 eV).81, 82 In the bulk of the SEI the intensity of the peak at 286 eV decreases sharply and the peak associated with polyolefins broadens and shifts toward higher energy. A smeared, low-intensity peak with a maximum at 292 eV, related to the two C-F bonds in CF2, vanishes on sputtering. This may indicate that SEI formation was incomplete during the first cycle. The C1s XPS spectrum of the basal SEI shows the presence of lithium carbonate both on the surface and in the bulk (Figure 15a). The concentration of Li2O (peak at 528.7 eV) was much higher in the basal SEI (Figure 14b vs Figure 15b). The O1s peak detected at 532.5 eV could correspond both to C-O bonding in the carbonyl configuration and to lithium carbonate structure. 












* +



* +





  



  

  



  

  









     

   

  



         

         

 $$+/;<

 $$+/;<

a

b 



 



  



  



  

 





  





















|  } % Z #

c Figure 14 XPS spectra of the SEI on the cross-section of SLX-20 electrode in LiPF6 electrolyte.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

31

The F1s peak at 691 eV of the C-F bonds in CF2 (Figure 14c) supports the presence of binder in the cross-section plane of the electrode. During sputtering, the intensity of this peak falls dramatically and the LiF bond becomes the dominating feature in the F1s spectrum. The thickness of the SEI formed on the basal plane of the carbon electrode seems to be lower than that on the cross section. This conclusion can be drawn from the elemental depth profiles, where concentrations of fluorine and lithium are halved after 20 minutes of sputtering in the basal SEI, but almost does not change in the SEI on the cross section (unpublished data). 









  



     

* +

* +










  









     

  



  













 $$+/;<

b

a 











  





     



  











c Figure 15 XPS spectra of the SEI on the basal plane of SLX-20 electrode in LiPF6 electrolyte.

Non-homogeneous chemical structure is observed in the TOF SIMS images of the cross-sectional SEI, while the distribution of the constituents of the basal-plane SEI is uniform (Figure 16). It is of interest to note that in the regions of highest fluoride concentration, oxygen-containing species are almost

Lithium-Ion Batteries: Solid-Electrolyte Interphase

32

absent. The lateral distribution of the bulk SEI components is non-homogeneous for both CS and BS planes. Similarly with the SEI on HOPG, it is clear from the TOF SIMS image that there is overlapping of the regions of the disposition of salt- and solvent-reduction products, indicating the formation of polyhetero microphase structure. The experimental data show good evidence of the compositional and morphological distinctions between the SEIs formed on the basal and cross-sectional planes of graphite. While it is practically impossible to produce carbon-powder electrodes with unidirectional orientation of graphene planes, the observed distinctions, in our opinion, must be given proper weight when designing high-performance lithium-ion batteries.

Li

F

O

CH

CS

BS

Figure 16 TOF SIMS secondary-ion images of the surface SEI on the cross section and basal planes of the composite. TIMREX (SLX 20)-based electrode (10 µm resolution).

4.2.3

Disordered Carbons

It has been reasonably well established that the intercalation capacity of lithium and operating voltage of the lithium-ion battery depend on the properties of the SEI. The formation of the SEI, in turn, is strongly affected by the crystal structure of graphite. Successful development of negative electrodes for lithium-

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

33

ion batteries requires an understanding of how irreversible-loss processes correlate with carbon-surface activity and structure. Many materials given the name graphite actually have a considerable amount of stacking disorder. For carbons in general, the situation is more complex. Most cokes, petroleum cokes, carbon blacks, carbon fibers, pyrolyzed polymers and mesocarbon microspheres have disordered structures. In such structures the size of the crystallites is small and there is a high probability of random stacking (shifts or rotations) of adjacent carbon layers. This type of disorder is called turbostratic disorder. Hundreds of carbons are commercially available; however, selecting the best carbon for use in lithium-ion cells is a subject for much current research. Electrochemical and spectroscopic methods have been used to investigate irreversible-loss mechanisms of lithium intercalation in disordered polymethacrylo-nitrile carbons.79 Voltammetric measurements show that the solvent readily decomposes at potentials 1.2 V positive of the reversible lithium potential. Evidence for hydrocarbon, carbonate and alkylcarbonate formation in the surface film is found with the help of combined XPS and SIMS analysis. Several carbonaceous materials were produced in our laboratory by chemical vapor deposition (CVD) of ethylene and by pyrolysis of dehydrated sugar.83, 84 Yudasaka et al.85 carried out vacuum CVD of 2-methyl-1,2-naphthyl ketone on a nickel-film substrate at 700-1000€C and observed graphite formation. CVD of 2,5-dimethyl-p-benzoquinone on a silicon matrix above 700°C resulted in a graphite-like lamellar structure.86 The carbon film obtained in this study by CVD of ethylene was found to be highly graphitized, as well. This finding is based on the XRD data (Figure 17a), where the sharp peak at 26.518°, with interplanar space d = 3.359 , is close to the d value of graphite (3.354 ). In addition, after disassembling of the cell that had been discharged to 0.00V, the carbonaceous electrode was found to have the golden color typical of a fully intercalated graphite, LiC6. The XRD pattern of the second sample produced by pyrolysis (Figure 17b) has few broad features; this is typical of hard carbons (see, for example, Ref. 87). XPS measurements of the pristine soft carbon sample (not presented here) revealed a sharp C-1s peak located at 284.4 eV. Broadening of the peak during sputtering was observed. This probably follows the partial destruction of the initial carbon crystallites caused by argon-ion bombardment. A similar phenomenon was described by Lascovich et al.88 The surface was found to contain 93 atomic percent C, about 3% O, and various impurities originating from the glove-box atmosphere.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

34

Soft



* +"





















= $ $&  $

Hard





* +"























= $ $&  $

Figure 17 XRD patterns of disordered carbons. Reproduced from [84] by permission of Elsevier Science Ltd.

Electrochemical cells, assembled from these materials and metallic lithium, were cycled between 0.00 and 2.00 V vs Li/Li+ in ethylene carbonate/diethyl carbonate electrolytes containing LiPF6 or LiAsF6. The solid-electrolyte interphase formed on the carbons was characterized by X-ray photoelectron spectroscopy.84 In the XPS spectra of the SEI formed on soft carbon in LiPF6based electrolyte, a strong C1s peak at 284.6 eV is clearly observed at the surface before sputtering (0 time). (Figure 18a). This peak can be assigned to polyolefins (Figure 18a, 0 min).82 During sputtering, significant broadening of the peak is observed. The shoulder appearing in the vicinity of 285.5 eV is more

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

35

likely to be from C-O-H and/or C-O-C bonds; the latter may be associated with 17, 81, 82 oxygen-containing polymeric species formed on solvent decomposition. The ether carbon in CH3OCO2Li or (CH2OCO2Li)2 typically gives a peak at 286-287 eV, and the carbonate peak of these compounds is expected at 19, 81 14 89 290 eV. The lithium carbonate peak appears at 290.2 eV or at 289.8 eV. Hence, a careful analysis of the C1s spectrum (Figure 18a) shows that no carbonates or semi-carbonates are present in the SEI. The broad peak detected in the O1s spectrum at 532.5 eV (Figure 18b) could correspond to C-O bonding in the carbonyl configuration or to lithium carbonate structure. Since the C1s spectrum does not contain the lithium carbonate XPS line, it seems likely that oxygen-to-carbon bonding is related 74, 82, 90 to Li-O-C groups, that may be part of the alcoholate molecule. The O1s peak assigned to Li2O at 528.7 eV is barely visible after four minutes of sputtering. A broad F1s peak at 686.2 eV is observed at the SEI before sputtering and is assigned to LiF (Figure 18c, 0 min). A shoulder at about 688.8 eV, attributed to the P-F bond, is absent in the bulk of the SEI (Figure 18c, 4-20 min).

O1s

C1s

F1s 0 min

0 min 4 min

0 min 4 min 10 min

4 min

10 min

10 min 20 min

20 min

20 min 538

292

290

288

286

284

Binding energy, eV

a

282

536

534

532

530

528

526

694

280

Binding energy, eV

b

692

690

688

686

684

682

680

Binding energy, eV

c

Figure 18 High-resolution XPS spectra (a-c) at different time of sputtering of the SEI formed on soft carbon in LiPF6 electrolyte. Reproduced from [84] by permission of Elsevier Science Ltd.

Figure 19 represents the depth-dependent elemental composition of the SEI formed in LiPF6 electrolyte on the soft carbon (top figure, 19a) and on the hard

36

Lithium-Ion Batteries: Solid-Electrolyte Interphase

carbon (bottom figure, 19b). While the same elements are detected in both cases, their atomic concentrations are significantly different. The atomic concentrations of fluorine and lithium in the SEI on hard carbon are twice those in the SEI on soft carbon. In the latter, the concentration of fluoride is higher than that of lithium, indicating that excess of fluoride may be bound to phosphorus-containing compounds in agreement with the P-F XPS line (Figure 18c). The small excess of lithium over fluorine in the SEI on hard carbon may be related to Li-O-C groups. It should be noted that the phosphorus content was relatively high both in the surface and bulk SEI on hard carbon. It seems likely that most of it is bound to oxygen, since the P2p peaks were located at about 133-134 eV (spectra are not shown here). Overall, the concentration of inorganic compounds is higher in the SEI on hard carbon. The sharp fall in the carbon signal after two minutes of sputtering was found for both electrodes. This indicates that organic compounds, such as polyolefins are present only at the SEI surface close to the solution. Formation of oxygencontaining polymeric chains similar to polyethylene oxide was suggested on the 79, 90 basis of IR and SIMS measurements. It seems likely that the SEI on hard carbon is considerably thicker than that on soft carbon. This conclusion stems from the fact that after 30 minutes of sputtering, the atomic concentration of carbon, which is associated with the signal arising from the electrode material underlying the SEI layer, is 80% for soft carbon and only 25% for hard carbon.

LiPF6-soft carbon

80 70

40

50

Atomic concentration, %

Atomic concentration, %

60

C O Li F P

LiPF6-hard carbon

50

C O Li F P

40 30 20

30

20

10

10 0

0 0

5

10

15

20

sputtering time, min

25

30

35

0

20

40

60

80

100

120

sputtering time, min

Figure 19 Depth-dependent elemental composition of the SEI on soft and hard carbon in LiPF6 electrolyte. Reproduced from [84] by permission of Elsevier Science Ltd.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

37

Figure 20 depicts the estimated depth profile of the possible SEI compounds. It is evident that solvent-reduction products occupy more than 50% of the SEI on soft carbon. The SEI built on soft carbon in LiPF6-EC:DEC electrolyte consists of 35-50 atomic % LiF with more than 50 atomic % of polymers on the solution side of the SEI as well as some PxFy compounds. No Li2O was found on the surface and only a negligible amount in the bulk. The relative concentration of inorganic constituents is higher in the SEI on hard carbon. No carbonates were detected in the SEI built on either electrode in LiPF6 electrolyte. Soft

Hydrocarbons and O-cont.polymers Li2O LiF P xF y

60

Atomic concentration, %

50

40

30

20

10

0

0

4 sputtering time, min

10

Hard

H ydrocarb and O -cont. p olym ers LiF P xFy ano ther P -com p Li-O -C

60

A to m ic c o n c e n tr a tio n ,%

50

40

30

20

10

0

0

4 s p u tte r in g tim e , m in

8

Figure 20 Proposed content of the SEI on disordered carbons in LiPF6 electrolyte. Reproduced from [84] by permission of Elsevier Science Ltd.

38

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Soft

H yd rocarb on s

C 1s

O 1s

C -O H C -O -C S e m ic a r b o n a te s 0 m in

0 m in

4 m in 4 m in 1 2 m in

1 2 m in

294

292

290

288

286

284

282

280

536

534

B in d in g e n e r g y , e V

532

530

528

526

B in d in g e n e r g y , e V

L iF F 1s

A s xF

y

0 m in

4 m in 1 2 m in

692

690

688

686

684

B in d in g e n e r g y , e V

Hard

C -H , C -O -C , C = O , L i- O -C

H ydrocarb on s O 1 s

C 1s

C -O H C -O -C C a r b o n a te s-2 9 0 .5 , s e m ic a r b o n a te s-2 8 9 0 m in A s 2O

3

- 5 3 1 .6

L i-C 2 8 2 .5 4 m in

0 m in

8 m in

4 m in 8 m in

2 0 m in 2 0 m in

L i2O

4 0 m in

4 0 m in

294

292

290

288

286

284

282

280

53 6

53 4

53 2

53 0

52 8

52 6

B in d in g e n e r g y , e V F 1s

A sF 3-6 8 7 e V 0 m in

4 m in

8 m in

2 0 m in 4 0 m in

6 92

6 90

6 88

6 86

6 84

Figure 21 XPS spectra of the SEI on soft and hard carbon in LiAsF6 electrolyte. Reproduced from [84] by permission of Elsevier Science Ltd.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

39

At first glance, the XPS spectra of the SEI in LiAsF6 electrolyte look like those in the LiPF6 electrolyte. However, close inspection of C1s XPS spectra of the SEI formed on hard carbon in LiAsF6-based electrolyte shows that they are more complex than those of the SEI on soft carbon (Figure 21). A striking feature of the C1s spectra is the presence of the carbonate band on the solution side and in the bulk SEI on the hard carbon. C-H as well as C-O characteristic bands are distinguished in the spectra as well. Sputtering is followed by the shift of the C-H peak toward lower energy and the development of an additional shoulder, which can be assigned to LiC (282.5 eV). After deconvolution of the broad O1s peak at about 531.5 eV it was attributed to compounds containing C-H, C-O-C, C=O, and Li-O-C groups. LiF and AsF bands were found in the SEI on the solution side, the intensity of the latter decreasing significantly on sputtering. Soft

C O Li F As

100

A tom ic concentration, %

90 80 70 60 50 40 30 20 10 0 0

5

10

15

20

25

30

35

spu ttering tim e, min

Hard

C O Li F As

60

A tomic concentration, %

50

40

30

20

10

0 0

5

10

15

20

25

30

35

40

45

50

55

60

65

sputtering time, min

Figure 22 Elemental depth profiles of the SEI on disordered carbon electrodes in LiAsF6 electrolyte. Reproduced from [84] by permission of Elsevier Science Ltd.

40

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Comparison of the elemental depth profiles of the SEI (Figure 22) shows that, as with the LiPF6 electrolyte, the SEI formed on soft carbon is thinner. After 10 minutes of sputtering, the carbon concentration from the underlying HOPG approaches 80%, while for the hard carbon it does not exceed 30%. Lithium content remains relatively constant during sputtering, whereas the concentrations of carbon, oxygen, fluorine, and arsenic decrease monotonically. The similarity of the SEI content in the two electrolytes stands out if one looks at the composition of surface SEI formed on soft carbon in both electrolytes (Figures 20a and 23a). The atomic concentration of LiF is in range of 35-38%, polymers 18-21%, C-O-containing compounds 20-24%.

A tom ic co n ce n tra tio n , %

Soft 80 70

H y d r o ca rb o n s O -co n t. p o ly m ers& su b stra te L i2 O L iF A s xF y

60 50 40 30 20 10 0

0

4 sp u tterin g tim e , m in

12

A tom ic concentra tio n, %

Hard Polymers and sub strate C C arbonates L i-O -C comp ou nds L i2 O L i-F A s comp ound s L i-C

50

40

30

20

10

0

0

4 sputtering tim e, m in

12

Figure 23 Effect of the type of carbonaceous material on the SEI composition in LiAsF6 electrolyte. Reproduced from [84] by permission of Elsevier Science Ltd.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

41

Formation of PFx compounds (~12%), found in the SEI formed in LiPF6 electrolyte might proceed in accordance with reaction (4.1).

PF 6− + 2e − + 3Li + → 3LiF + PF3

(4.1)

The excess of fluorine over lithium found on the surface of the SEI on soft carbon gives rise to the possibility that other fluorine-containing species, like 90 PF3O, may be formed when PF 6− reacts with impurities (4.2):

PF 6− +H2O
2HF+PF3O

(4.2)

LixPFyOz SEI compounds as possible electrolyte reduction products cannot be excluded either. The formation of an As-F (~10%) SEI component may be described as:

AsF 6− + 2e − + 3Li + → 3LiF + AsF3

(4.3)

We found that the highest Li2O content in the SEI accompanied the highest carbonate concentration (hard carbon/LiAsF6 electrolyte). Lithium oxide is suggested as the product of the reactions:

CO3− 2 + 5e − + 7 Li + → 3Li 2O +

1 Li 2C2 2

(4.4)

or

CO3−2 + 4e − + 6Li + → 3Li 2O + C

(4.5)

or as the product of SEI decomposition by argon sputtering. Lithium carbonate, 16, 91 in turn, from solvent reduction according to the reaction: O C O

O

CH2

CH2 + 2e +2Li → Li2CO3 + CH2 = CH2

−

+

(4.6)

42

Lithium-Ion Batteries: Solid-Electrolyte Interphase 24

Polymers are formed by anionic polymerization of CH2=CH2. The striking effect of the type of carbonaceous material on the SEI composition is obvious from Figure 23, where the major constituents of the SEI formed on soft and hard carbon in the LiAsF6 electrolyte are summarized. The concentration of polymers on the solution-side SEI of the soft carbon electrode is higher than that on the hard-carbon SEI. In the present study, carbonates were detected solely in the SEI formed on hard carbon in LiAsF6 electrolyte. The atomic concentration of carbonates was extremely high and reached ~47% on the solution side of the SEI. The total concentration of LiF and AsxFy compounds in the surface of the SEI on hard carbon was half (~20%) that on soft carbon (42%) (Figure 23). In the bulk SEI on hard carbon, the concentration of salt-reduction products increased, while on the soft carbon it dropped by a factor of 5). Similar deviations, while less pronounced, in the content of the SEI on the carbons studied, were observed in the LiPF6 electrolyte. For instance, the concentration of LiF (~25%) and polymers (5%) in the solution side of the SEI on hard carbon is smaller and that of C-O compounds is higher than that on soft carbon. In the bulk SEI on hard carbon the concentration of LiF and PF or P-O compounds reaches 65% (Figure 20b), while on the soft carbon it does not exceed 50% (Figure 20a). All these distinctions are probably due to different catalytic properties of the two carbon materials, favoring different reduction processes. Random stacking of carbon layers is more pronounced in hard (non92 graphitizable) carbon. Franklin proposed that in disordered carbons, between the small organized carbon regions there is non-organized carbon which is highly strained (highly buckled graphitic sheets). These regions are assumed to contain a higher density of edge planes (zig-zag and armchair). High fractional coverage of the carbon surface (and possibly bulk) by edge planes should favor catalytic activity of carbon, followed by decomposition of the electrolyte. As shown above, the SEI on hard carbon was considerably thicker than that on soft carbon, in both electrolytes. This conclusion stems from the fact that the atomic concentration of carbon, which (after 30 minutes of sputtering) is associated with the signal arising from the electrode material underlying the SEI layer, is 80-90% for the soft carbon and 30-40% for hard carbon. The thick SEI, explained by solvent cointercalation and exfoliation of the graphite, must contain carbonates as a result of solvent reduction. In fact, a high concentration of carbonates was detected in the SEI formed on hard carbon in the LiAsF6 electrolyte. On the other hand, carbonates were not found either on the solutionside surface, or in the bulk of the SEI formed on hard carbon in the LiPF6electrolyte. XPS data showed that the major SEI compound is LiF (up to about

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

43

60 atomic percent in the bulk). The concentration of P-F and other phosphorus species recorded by XPS in the SEI on hard carbon is twice that in the SEI on soft carbon. These two findings lead us to the conclusion that, as in the case of SEI formation on the cross section of HOPG, the exchange-current density for LiPF6 (and for HF) reduction is much higher at the disordered, more catalytic, hard-carbon surface than it is at the soft carbon. The greater thickness of the SEI on hard carbon may be related both to the high specific surface area of the 2 substrate (330 m /g) and to its pore-size distribution. Pores ranging from 5 to 20 Å are typical for this kind of carbon, a property that enables deep penetration of electrolyte into the bulk of the electrode and more effective contact with the reductive carbon surface.

HOPG-Basal P olym &s olv en t re d. prod S alt red .p rod C arbon ates

HOPG-Cross-section Polym &solvent red. prod Salt red.prod

69.4%

18.5%

22.8%

58.7% 30.6%

Soft carbon Polym &solvent red. prod Salt red.prod

Hard carbon Polym&solvent red. prod Salt red.prod

60.3%

47.5%

52.5% 39.7%

Figure 24 Estimated composition of the SEI on HOPG and soft and hard carbon in LiPF6 electrolyte.

On the basis of the data presented above, it is suggested that the substrate has a more pronounced influence on SEI formation on carbonaceous materials than does the electrolyte. The disturbed graphite structure of hard carbon is characterized by high catalytic activity, which favors decomposition of the

44

Lithium-Ion Batteries: Solid-Electrolyte Interphase

electrolyte and is followed by increased content of carbonates in the SEI formed in the LiAsF6 electrolyte, and of LiF and phosphorus-containing compounds in the SEI in the LiPF6 electrolyte. A comparison of the composition of the SEI formed on the disordered 67, 73-75, 82, 84 carbons and on HOPG, leads to the conclusion that SEI content on hard carbon is similar to that formed on the cross-section of HOPG, while the composition of the SEI on soft carbon is close to that on the basal plane (Figures 24 and 25). HOPG-Basal (4min)

HOPG-Cross-section (4min) Polym&solvent red. prod Salt red.prod Carbonates

Polym&solvent red. prod Salt red.prod Carbonates

44.6% 72.7%

15.9% 4.82% 11.4%

50.6%

Soft carbon (4min)

Hard carbon (4min) Polym &solvent red. prod Salt red.prod Carbonates

Polym &solvent red. prod Salt red.prod

53.8% 31.1% 20.9%

68.9% 25.3% 4min

Figure 25. Estimated composition of the SEI on HOPG and soft and hard carbon in LiAsF6 electrolyte.

Zaghib et al. analyzed the role of the edge and basal-plane sites on the magnitude of the irreversible capacity loss on natural graphites for Li-ion 93 batteries. Flake-like graphite powders of varying average sizes (2-40 µm) and LiClO4 EC:DMC electrolyte were used to simulate the ideal behavior of graphite structure in the lithium-ion cell. Two morphologies — flake-like and cubic structure — were used as models in order to understand the correlation of

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

45

the fraction of edge sites with particle morphology. Calculating the number of 94 carbon atoms in a crystallite with the equations reported by Fujimoto et al., the authors derived a relationship between the size of graphite crystallites, crystallographic parameters and surface area. In brief, the BET surface area 2 decreased from 12.1 to 2.3 m /g with an increase in the average particle size from 2 to 40 mm. The corresponding thickness of the edge planes increased from 0.21 to 2.85 mm. On the basis of a comparison of experimental electrochemical, IR and SEM data with calculated values, it was deduced that the surface area associated with the edge sites and not the total BET surface area, has the dominant influence on the irreversible capacity loss and coulombic efficiency of carbons in lithium-ion batteries. These findings are in agreement with our results of thicker SEI on the cross section.

4.2.4

Overview of SEI Composition and Properties in Different Carbon/ Non-aqueous Electrolyte Systems

We would like to emphasize that irreversible reactions, including gas evolution and disintegration were mainly observed on that part of the surface occupied by the edge planes of acetylene black, activated carbon and vapor-grown carbon 95 fiber in LiClO4/PC solution. Aurbach et al. extensively studied the electrochemical and spectroscopic characteristics of carbon electrodes in 91, 96 lithium-battery systems. LiClO4, LiAsF6 and LiBF4-based electrolytes with MF, PC, EC, THF, DME, 1,3-dioxolane solvents were tested. The carbons investigated included carbon black, graphite and carbon fibers. It was found that the SEI films are similar in their chemical structure to those formed on lithium in the same solutions. Thus, PC is reduced on carbon to ROCO2Li, ethers are reduced to alkoxides, MF to lithium formate. LiAsF6 is reduced to LiF and AsF3, and further to insoluble LixAsFy. IR spectra of graphite-EPDM electrodes cycled in LiClO4-MF solution seem to indicate the existence of LiClO3, LiClO2 or LiClO. CO2 reacts with LixC6 to form Li2CO3 (and probably CO). Because of the high surface area of graphite particles as compared with the lithium-metal − electrode, the role of contaminants, such as HF in LiPF6 and LiBF4-based 97 electrolytes, is much less pronounced. The beneficial effect of inorganic 2− additives, such as CO2, N2O, Sx , etc., on the formation of the SEI on carbons, 92, 96, 98 was emphasized. 99 Interesting results were obtained by Ein-Eli et al., who showed that the use of SO2 as an additive to LiAsF6/MF or LiAsF6/PC, DEC, DMC solutions offers the advantage of forming fully developed passivating films on graphite at

46

Lithium-Ion Batteries: Solid-Electrolyte Interphase +

a potential much higher (2.7 V vs Li/Li ) than that of electrolyte reduction + + (< 2 V vs Li/Li ) or of lithium intercalation (0.3-0 V vs Li/Li ). These data support our approach to the selection of SEI precursors possessing high constant 11, 22 rates. The major surface species are inorganic lithium salts (LixAsFy, Li2CO3, Li2SO2O4, Li2SO3, Li2S2O5 and Li2S) and organic lithium alkyl carbonates 100 (ROCO2Li). Using cyclic voltammetry, Inaba et al. found that for graphite electrodes an EC+DEC solvent mixture is preferred over EC+DME with respect to the formation of a stable passivating film. When graphite electrodes are charged in PC-based solutions, the solvent decomposes at about 1 V, and this makes SEI formation difficult. Using XRD and electrochemical quartz-crystal-microbalance techniques 101 Morita et al. showed that the cathodic intercalation of lithium is accompanied by electrochemical decomposition of the electrolyte. The mass change per coulomb over the potential range of 0.0-0.2 V vs Li/Li+ was higher in EC-DMC than in EC-PC, indicating different surface reactions. EQCM experiments in 102 LiAsF6/EC-DEC solution clearly indicated the formation of a surface film at + about 1.5 V vs (Li/Li ). However the values of mass accumulation per mole of electrons transferred (m.p.e), calculated for the surface species, were smaller than those of the expected surface compounds (mainly (CH2OCO2Li)2). This was attributed to the poor stability of the SEI and its partial dissolution. An unstable passivating layer on petroleum coke in Li-triflate/EC-PC-DMC, followed by interaction between the electrolyte and the intercalated lithium was observed by 103 Jean et al. It was concluded that on long cycling of the lithium-ion battery, the passivating layer on the carbon anode becomes thicker and more resistive, and 104, 105 is responsible, in part, for capacity loss. The mechanism of formation of the passivating film at the interface between lithiated carbon and a liquid or polymer electrolyte was studied by AC106, 107 impedance. Two semicircles observed in AC-impedance spectra of + 107 LiAsF6/EC-2MeTHF electrolytes at 0.8V vs Li/Li were attributed to the formation of a surface film during the first charge cycle. However, in the cases of LiClO4 or LiBF4/ EC-PC-DME (dimethoxy-ethane), only one high-frequency 106 106, 108 distorted semicircle was found in the impedance spectra. Yazami et al. explained the complicated arc shape by surface-film formation followed by electrode gassing during the decomposition of the electrolyte. This phenomenon is less pronounced in Li-triflate, Li-imide and lithium hexafluorophosphate. However we believe that the depressed high-frequency arc may be due to the overlapping of two, or even more arcs and may be associated with grain8, 22 141 boundary resistance in the SEI. In another investigation, it was found that

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

47

the interfacial resistance of graphite electrodes in LiPF6 and LiBF4/EC-DMC solutions is about one order of magnitude higher than that of LiAsF6-based electrolytes and increases considerably upon storage. This is explained by different surface chemistry, namely by the increased resistance of a passivating film containing LiF. 106, 109 Yazami et al. studied the mechanism of electrolyte reduction on the carbon electrode in polymer electrolytes. Carbonaceous materials such as cokes from coal pitch and spherical mesophase and synthetic and natural graphites, were used. The change of Rfilm with composition on LixC6 electrodes was studied 106 for three ranges of x in a Li/POE-LiX/carbon cell. The first step in the lithium intercalation (0 < x < 0.5) is characterized by a sharp increase of Rfilm and is attributed to the formation of a bond between lithiated coke and POE. Such intercalated lithium is irreversible in the 1.5-0.5 V range. In the second step, (∆x~1), lithium intercalates mainly into the coke and the film does not grow significantly, thus a slow increase of Rfilm is observed. In the third step, excess lithium is formed on the surface of the coke, and this induces a further increase in the film thickness and its resistance. The relation between surface properties, pore structure and first-cycle charge loss of different natural and TIMREX graphites in LiPF6 EC:DMC 110 electrolytes was studied by Joho et al. The graphites studied were as follows: KS6-KS44, SFG6-SF44, T15-T44, SLM44, E-SLX2050 and E-NP15. They found that the geometrical surface area of the natural graphite particle excluding the mesopores contributes more to the BET surface area and to the irreversible capacity of the first electrochemical reduction than that of the synthetic graphites. Obviously, natural graphite differs from the synthetic graphites examined, not only in its mesoporosity but also in its surface morphology, i.e. roughness of the particle surface and defects on the surface. The synthetic graphite samples exhibit more highly developed mesoporosity, while the natural graphite has a rougher surface, so that similar values are recorded for their BET surface areas and for the irreversible capacities in the first reduction of the corresponding graphite negative electrodes. In situ neutron radiography carried out before and after cycling of 111 commercial prismatic lithium-ion cells (ICP-340848) revealed displacements of excess electrolyte, most probably as a consequence of an expansion/ contraction of the electrodes as well as evolution of gases during SEI formation. Lithium-7 nuclear magnetic resonance (NMR) measurements on electrodes 112 from aged and cycled lithium-ion batteries have been performed. The active cathode material was LiNi0.8/Co0.2/O2, and the anode consisted of a commercial

48

Lithium-Ion Batteries: Solid-Electrolyte Interphase

graphite blend Cells were aged by storage under 60% charge at elevated temperatures. One of the primary failure mechanisms believed to occur in these cells is the formation of a passivating layer on the positive electrode that eventually leads to loss of electrical contact between active cathode particles. The NMR spectra show buildup of a solid-electrolyte interphase characterized by a relatively featureless absorption centered at 0 ppm. The relationship between the elevated-temperature performance of Li/graphite half-cells and the composition and morphology of the SEI formed on the graphite (TIMREX KS6) surface has been investigated for two electrolyte systems: 1 M LiPF6/ in ethylene carbonate/dimethyl carbonate 113 EC/DMC (2:1) and 1 M LiBF4/ in EC/DMC (2:1). Precycled cells were stored o at different temperatures up to 80 C, and the graphite electrodes were analyzed by XPS and electrochemically under continued cycling. The morphology and the SEI were found to change on storage at elevated temperatures. The surface of the electrodes also shows an increasing amount of polymeric compounds. Studies of the low-temperature behavior of the MCMB Li ion cells with geltype PVDF based electrolyte prepared with Bellcore technology were 114 conducted. It was found that even at modest (C/5) to low (C/10) rates of charge and discharge, batteries show permanent capacity loss at temperature ≤ 20°C. This loss is attributed to continual growth of the SEI resulting from electrolyte reduction. 115 Lanz and Novak studied gas evolution at thick graphite electrodes in γ-butyrolactone EC:DMC electrolyte by Differential Electrochemical Mass Spectrometry (DEMS). TIMREX SPG 6, SPG 15 and SPG 44 carbons were tested. They found that SEIformation on these thick electrodes was not yet complete after the first charge/discharge cycle. The amount of ethylene and hydrogen gas evolved decreases with increasing percentages of GBL in an EC/DMC electrolyte, indicating that the SEI layer is built up from GBL rather than from EC decomposition products. In order to improve the cycling performance of lithium-ion batteries with nonflammable trimethyl phosphate (TMP)-based electrolytes, amorphous 116 carbon (AC) was tested as the anode material. It was found that the reduction decomposition of TMP solvent, which occurred without limit on a natural graphite anode and concomitantly generated a large amount of methane (CH4) and ethylene (C2H4) gases, was considerably suppressed on an amorphous carbon anode. The charge/discharge data and cyclic voltammetry indicated the formation of a highly stable and passivating surface film on the carbon surface 3 at a potential near 1 V. As a result, an AC/LiCoO2 ion cell with 1 mol/dm LiPF6

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

49

EC:PC:DEC:TMP (30:30:20:20) nonflammable electrolyte exhibited promising cycling performance.

4.2.5

Effect of Carbon Modification on SEI Formation

Surface pre-treatment of graphitic electrode materials for lithium-ion cells has recently been shown to significantly reduce the irreversible consumption of material and charge. This improvement is due to the formation of a more sophisticated SEI. We recently found that mild air-oxidation (burnoff) of two synthetic graphites and natural graphite (NG7) improves their performance in 7, 61-63, 65 Li/LixC6 cells. The reversible capacity of the graphite increased (up to 405 mAh/g at 4-11% burnoff), its irreversible capacity was generally lower and the degradation rate of the LixC6 electrode (in three different electrolytes) was much lower. STM images of these modified graphites show nanochannels with openings of a few nm and up to tens of nm. It was suggested that these nanochannels are formed at the zig-zag and armchair faces between two adjacent crystallites and in the vicinity of defects and impurities. Performance improvement was attributed to the formation of SEI chemically bonded to the surface carboxylic and oxide groups at the zig-zag and armchair faces, better wetting by the electrolyte and to accommodation of extra lithium at the zig-zag, armchair and other edge sites and nanovoids. The mechanism by which partial oxidation increases the reversible lithium capacity is believed to be related to lithium bonding at edge atomic sites, as opposed to intercalation between 7 graphene sheets. By Li NMR measurements it was found that the edge-site 117 population is enhanced in the partially oxidized carbon. In particular, oxidation proceeds most rapidly at the zigzag and armchair sites, and results in the formation of COOH acid groups, which have been detected directly (along 118 with CH, COH and C=O groups) by XPS. XPS studies showed that the surface oxygen atomic concentration of NG7 has a broad minimum at 4-22% 63 burnoff. The oxygen peak maximum shifted monotonically with burnoff time, rising from 531.05 eV for pristine NG7 to 534.0 eV for a 34% burnoff sample. From the analysis of XPS spectra by curve fitting, it was shown that the pristine NG7 surface contains mostly (53%) aromatic carbon, about 20% each of CH and COH groups, only 4.8% CO groups, and no COOH groups. The 34% burnoff sample consists mainly of CO groups (33%), C-OH groups (26.6%) and 7 119 8.9% COOH groups. Solid-state Li NMR measurements revealed two kinds of lithium sites in lithiated pristine graphite: lithium intercalated between graphene planes, with ~40 ppm Knight shift (relative to aqueous LiCl), and

50

Lithium-Ion Batteries: Solid-Electrolyte Interphase

lithium chemically associated with the solid electrolyte interphase, characterized by a chemical shift of about 0 ppm. The burnt graphite also exhibited a feature at about 14 ppm, correlated with the excess lithium and attributed to lithium 7 bonded to edge sites. In addition, the Li signal associated with the SEI was more intense in the burnt graphite, consistent with earlier indications that mild oxidation prior to lithiation results in a thicker and more salt-rich SEI. This graphite modification, following mild burnoff, was found to make the LixC6 electrode performance more reproducible and less sensitive to electrolyte impurities. Decomposition of the SEI is generally seen as being one of the major factors influencing the thermal stability of the graphite electrode in lithium-ion 113, 120-122 cells. The effect of mild oxidation of natural graphite (NG7) and some other parameters on the reaction between a fully lithiated graphite anode (LixC6, x = 1.0-1.1) and 1M lithium hexafluoroarsenate, 1:2 (v/v) ethylene carbonate and diethyl carbonate electrolyte were studied by differential scanning 123 calorimetry (DSC). Figure 26 shows the DSC traces for the washed and fully lithiated graphite/5 µL electrolyte samples. It can be seen that only one exothermic peak exists at about 210°C. In the DSC run of the pristine graphite sample, the exothermic reaction produces a great amount of heat and releases much gas, causing an explosion of the DSC pan. In the case where the graphite was mildly oxidized (although it contains 10% more lithium), the explosion was prevented. In addition, the exothermic peak of the mildly oxidized graphite was depressed and shifted from 210 to 214°C. In order to decrease the surface area of separation the small graphite particles were removed. Modification of the burnt graphite resulted in a dramatic lowering of the peak height (from 225 W/g for the burnt sample to 25 W/g for the modified sample). In addition, the energy of the exotherm and the reaction rate decreased from 3610 to 1460 J/g and from 10 to 2 W/g°C, respectively. The exothermic peak is preceded by a small endothermic peak (A in inset of Figure 26). It is suggested that this endothermic peak can be attributed to the decomposition of some SEI materials such as polymers, ROCO2Li, (CH2OCO2Li)2 and ROLi. Decomposition of such products at 200°C was analyzed by temperature-programmed desorption mass 124 spectography (TPD- MS). Although materials like LiF, Li2CO3, Li2O, As and As2O5 are thermally stable up to 600°C, compounds such as As2O3, some polymers and especially semicarbonates are thermally unstable and decompose at about 200°C. In addition, it is expected that polymers and semicarbonates will dissolve in the electrolyte at high temperatures. The dissolution and decomposition processes can lead to the destruction of at least part of the SEI.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

51

This would be followed by a vigorous reaction of the lithium, released from the lithiated graphite electrode, with the electrolyte. Another factor that must be considered in analyzing the effect of mild oxidation on the thermal behavior of the lithiated graphite is the effect of this oxidation on the chemical composition of the SEI on the cross section of the graphite particle. We found that the chemical composition of the SEI has changed as a result of the oxidation of the 125 graphite. The change in the concentration ratio of the organic materials to the inorganic compounds in the SEI depends on the type of the electrolyte. The formation of the chemically bonded SEI at the cross section may help to, at least partially, avoid or slow the exfoliation of the graphite crystallites during the thermal reaction with the electrolyte. 250

2

150 100 50 0

1 Heat Flow [W/g]

Heat Flow [W/g]

200

0

-1

-2 200

A

220

240

Temperature [ 0C]

15% burn-off (x = 1.1)

Modified 15% burn-off * -50 (x = 1.0) Pristine (x = 1.0) -100 40 80 120 160 200 240 280 320 Temperature [ 0C] Figure 26 DSC thermogramms of fully lithiated graphite-electrolyte samples (the modified and the pristine samples were shifted by −25 W/g and −75 W/g respectively). The electrolyte/graphite ratio is 5 µL/ 2 mg. The heat flow values are in units of Watt per gram of graphite. *Sample exploded. Reproduced from [123] by permission of the J. Solid State Electrochem.

Tibbetts et al. showed that oxidative pretreatment of vapor-grown carbon fibers (VGCF) can reduce the capacity of SEI forming in LiClO4/PC electrolyte 126 by an order of magnitude. Their experiments confirm the idea that air etching removes the more active carbon atoms — those capable of decomposing the electrolyte — and completely alters the fiber morphology.

52

Lithium-Ion Batteries: Solid-Electrolyte Interphase 127

It was recently found that chemical oxidation of graphite powder by strong oxidizing agents such as ammonium peroxysulfate and hot concentrated nitric acid gave similar results, i.e. it suppresses QIR and enhances QR to 410-430 mAh/g. Following this wet oxidation, carboxyl groups were identified on the surface of the graphite. Takamura et al. found that heat treatment at 700ºC in the presence of acetylene black, improved the performance of the graphite-fiber 128 anode. A novel, quite flexible strategy for the surface pre-treatment of graphite 129 anodes for lithium-ion cells has been developed. Their concept primarily involves a change in the surface chemical composition, in particular at the prismatic surfaces of graphite. It comprises two independent steps: First, reactive carbon surface sites (“dangling bonds”) are created by “desorption” of the pristine surface groups in an inert atmosphere of argon. Then the “cleaned” carbon surfaces are exposed to reactants, such as O2, CO2, NH3, N2, SO2, H2S, C2H2 at a temperature lower than the temperature of the cleaning procedure. o Argon cleaning or CO2 treatment at 500 C had no significant influence on anode o o behavior. However, a 15-minute treatment at 1000 C with CO2 and at 420 C with O2 brings about significant reductions in irreversible capacity. It was concluded that a nano-rough morphology of the prismatic surfaces offers favorable anchoring/nucleation sites for SEI formation. These data support our suggestion of the chemically bonded SEI. Graphite-surface modification by silylation in nonaqueous solution has been examined by the same research 130 group. Best results were obtained when pre-oxidized graphites were silylated in a mixture of hexamethyldisilazane and trimethylchlorosilane. Another interesting approach to carbon modification, the creation of a 131 core-shell-structured carbon composite, was first applied by Kuribayashi et al. More recently, other groups have also reported on the improved electrochemical 132 performance of such composites. The preparation is based on mixing the carbon precursors with graphite or graphitized carbon and heating the slurry o mixtures at above 1000 C. Carbon-coated natural graphite was prepared by 133 7 thermal vapor decomposition of natural graphite. Li NMR spectra of the fully lithiated carbon-coated natural graphite shows that there are two types of storage sites for lithium insertion: the graphite core part for lithium intercalation and the soft-carbon-type shell part for lithium storage. This material shows superior electrochemical performance as an anode for lithium-ion cells in both EC- and PC-based electrolytes. The irreversible capacity is inversely proportional to the amount of carbon coating.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

5

53

SEI Formation on Lithium-Tin-Based Alloys

The study of phase equilibrium and transport properties in lithium-alloy anodes for rechargeable ambient-temperature lithium batteries began before 134, 135 carbonaceous materials were seriously considered for these applications. An obvious advantage of lithium insertion into metallic matrices as opposed to lithium intercalation into carbonaceous materials is the high packing density of 136 lithium that can be achieved in lithium alloys. Charge capacity can exceed the capacity of carbonaceous materials by a factor of 2.5. As the operating voltage of lithium alloys may be chosen well above the potential of metallic lithium, the problem of lithium deposition during charging can be minimized. Lithium alloys do not seem to suffer from the drawback of solvent co-intercalation. In addition, the melting points of most of the lithium alloys are high (more than 400°C for Li1-4.4Sn compounds). All these properties are expected to result in improved safety and rapid charging capabilities, and therefore have attracted the attention of research groups, many of which have investigated tin-based 137-146 alloys. On the other hand, there are some disadvantages in the use of lithium alloys as anodes in lithium-ion batteries: the most prominent one is the large changes in volume that occur during cycling. The topotactic intercalation of lithium between the graphene layers of carbons requires only minor changes of interlayer spacing and stacking order, whereas insertion of lithium into metallic matrices causes much more drastic three-dimensional structural rearrangements. As a result, these great mechanical stresses cause the host matrix to crack and eventually disintegrate. The situation is much worse when the metal is thick and more susceptible to stress damage. Several solutions to this problem have been 137, 138 suggested: Fuji developed anodes containing tin-based amorphous composite oxides (TCO) as the active material. One of their products was Sn1.0B0.56P0.4Al0.42O3.6. The capacity at the first insertion of lithium to this product was 1030 mAh/g and the de-insertion capacity was 650 mAh/g (Qir = 37%). 144, 145 Yamaki’s group prepared SnBxPyOz. Qr was 600 mAh/g in the first cycle, and in the 25th cycle dropped to 200 mAh/g, which is a 2.7% loss per cycle. Scrosati’s group used SnO2 powders in gel-type polymer-electrolyte 10 cells. A particle size between 1-5 µm was obtained and the reversible capacity 147 was 450 mAh/g. Schoonman et al. doped the SnO2 with silicon. They synthesized the material by an ultrasonic-spray method. The product has a reversible capacity of 900 mAh/g and a reduced irreversible capacity when

54

Lithium-Ion Batteries: Solid-Electrolyte Interphase

compared with other tin oxides. Several groups introduced the use of tin alloys to solve the problem of aggregation of small tin particles to form larger 148-150 agglomerates. In Besenhard’s group, tin-antimony alloys were prepared by electroplating, in which an Sn:Sb ratio of 72:28 was obtained; and by chemical reduction, in which the ratio Sn:Sb was 88:12. In both cases, the longest cycle life was obtained with a particle size of 0.2 µm. When this material was cycled at 0.5 mA/cm² between 0 and 2.5V, the capacity was 600 mAh/g for 50 cycles. In Ref. 148, it was shown that dimensional changes of SnSbn electrodes during cycling turned out to be quite small after the first insertion of lithium. By contrast, the “ breathing” of pure Sn was much more drastic. Tin-antimony alloy 151, 152 was also used by Yamamoto et al. They prepared a 0.2 µm particle-size alloy (SnSb0.14) by chemical reduction and added 0.03 µm nickel powder in order to improve mechanical stability and conductivity. When cycled in a potential window of 0.1-0.8 V and at a current density of 0.4 mA/cm², the electrode produced 430 mAh/g in the first cycle and in cycle 110 it gave 250 mAh/g — a capacity loss of 0.36% per cycle. XRD patterns show that the degree of crystallinity of the alloys decreases with increase in the cycle number. 153-157 Dahn et al. synthesized tin-iron-carbon alloys by mechanical alloying with the use of a ball mill. The mixture, of which the active phase was the Sn2Fe alloy, contained Sn2Fe:SnFe3:C – 24:72:4 (atomic ratio). 10 nm particles were obtained and Qr was 200 mAh/g with 0.21% capacity loss per cycle, until the 80th cycle. The irreversible capacity (Qir) was 13%, which is relatively low. In Ref. 158, it was found that nanostructured SnMn3C compound, which has a low affinity for lithium, behaves differently from any intermetallic system reported to date. Using XRD, in situ Mossbauer spectroscopy, and electrochemical experiments, they conclude that the grain boundaries apparently act as channels to allow lithium to enter the particles. The lithium atoms then reversibly react with tin atoms at and within the grain boundaries to deliver a capacity of about 159, 160 150 mAh/g up to the 120th cycle. Thackeray et al. investigated the Cu6Sn6, Cu6Sn5, and Cu6Sn4 alloys: the relevant elements were mixed and heated to 400 ºC and the product was ground to 400-mesh particle size. The electrode that was prepared from the best alloy (Cu6Sn4) gave 450 mAh/g in the first cycle and approximately 160 mAh/g in the 20th cycle, when it was cycled between 0 and 1.2 V. This study clearly shows the effect of the presence of an inactive phase 161 on the performance of the anode. Blomgren et al. synthesized nanocomposites containing silicon and titanium nitride by mechanical milling. Crystallite size was 5-7 nm, and the reversible capacity was 300 mAh/g for 20 cycles. Small 162 particle size did not improve cycle life when pure tin was used. Owen et al.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

55

electroplated 1 µm tin particles on a copper foil. The capacity in the first cycle was 700 mAh/g, but dropped to 70 mAh/g in the ninth cycle. The discharge 163 current was 5 mA/cm². Dayalan investigated anodes that were prepared from tin-graphite composites. When the amount of tin in the composite was 44%, the electrode produced 500 mAh/g in the first cycle and 300 mAh/g in the 50th (1% capacity loss per cycle). 164 Dahn et al. were the first to use in situ atomic and optical microscopy to study the extremely large volume changes that occur as lithium is electrochemically added to and removed from Li-Sn-Si film in LiPF6 EC:DEC electrolyte. It was found that when lithium is first added to alloy films on rigid substrates, the films expand perpendicular to substrate. When lithium is removed, the films shrink in directions both perpendicular and parallel to the substrates, a process that leads to crack patterns similar to those found in drying mud. The processes taking place in the first intercalation of lithium into an alloy anode in a lithium-ion battery assembled in the discharged state are expected to be very similar to those in a disordered-carbon anode. The intercalation of lithium into the alloy proceeds in parallel with the reduction of the electrolytes and the building of the SEI. However, because of the dependence of io on the catalytic nature of the alloy, the chemical composition and the morphology of the SEI may vary from alloy to alloy. Extremely large reversible volume changes in lithium alloys during charge-discharge cycles may result in shorter cycle life and lower faradaic efficiency as a result of the formation of cracks in both the alloy and the SEI. Therefore, in this case, the flexibility of the SEI is highly important. Mixed-conducting lithium-ion-doped emeraldine polyaniline (PAni)-PEO blends have been developed in order to achieve optimal electronic-ionic 165 conductivity balance in nano-tin composite anodes. They found that the SEI impedance of the composite anodes increases with a decrease in PEO content and is much lower in pressed than in cast electrodes. Nano-Sn, AlSi0.1 and Li4.4Sn powders were studied by EIS to determine the electrochemical kinetics and intrinsic resistance during initial lithium insertion-extraction. It was shown that the SEI formed on particle surfaces, together with particle pulverization are responsible for the high contact resistance. With the use of a high-speed electroplating method (at currents above the 166-169 limiting current density), Peled and Ulus were able to produce two nanosize tin-antimony alloys. The first, with low antimony content, had higher reversible capacity (up to 700 mAh/g), a lower irreversible capacity, a better rate

56

Lithium-Ion Batteries: Solid-Electrolyte Interphase

capability and a lower average working potential vs Li. The second was an antimony-rich tin alloy that had longer cycle life, but low rate capability and a high average working potential vs Li. The addition of a small amount of copper to the alloy improved the cycle life with little or no penalty in capacity. In Refs. 168 and 169 two sets of composite anodes were produced- Sn65Sb18Cu17 and Sn62Sb21Cu17. The weight ratio between the alloy and the PVDF-graphite material was 70:30. Graphite composite was added in order to enhance the resistance of the fragile porous structure of the alloy to cycling effects (volume changes, particle-to-particle break-off), and to sustain a continuous electronic conductivity. The reversible capacity of the composite anode in a half-cell was 495 mAh/g and the irreversible capacity was 25%. The capacity loss until cycle 35 was 0.48% per cycle, which is comparable to the capacity loss (0.51%) of a similar alloy with no graphite. In a tin alloy/LiCoO2 cell the anode reversible capacity was 532 mAh/g, the irreversible capacity was 35% and the capacity loss (until cycle 20) 0.9% per cycle. The faradaic efficiency (QDe-ins/QIns) in both cells was less than 100%. The amount of irreversible lithium, i.e. the amount of lithium that cannot be released at high current densities (0.2 mA/cm²), was found to increase with cycle number. The dQ/dV curves of charge/discharge showed the broadening of peaks between 300-1400 mV; this may indicate slower kinetics resulting from thickening of the SEI. The SEI thickening, in turn, is due to active material disintegration and exposure of the active material to the electrolyte. As the cycle number increases, the SEI thickens and the migration of ions between particles becomes sluggish. This result is in agreement with Refs. 153-157, in which the XRD patterns show that capacity loss of the electrode is caused by the heavily lithiated alloy Li4.4Sn, which becomes detached during cycling. The degradation mechanism involves not only chemical changes, but also morphological ones. In Figure 27a (the electrode before cycling) one can see the network-like, porous structure that characterizes a metal that is plated under high current density. Figure 27b shows one of the holes in this structure in which the graphite flakes and their distribution are clearly seen. The main particle size of the electrode is about 30-100 nm (Figure 27c). SEM micrographs of the same electrode after cycling, with capacity loss of 50% are shown in Figure 28. The shape of the network is lost on cycling and the material is fractured. This indicates that expansion and contraction of the electrode during reversible insertion/deinsertion of lithium to the alloy causes three-dimensional disintegration of the bulk structure.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

57

Figure 27 SEM micrographs of pristine Sn65Sb18Cu17 electrode. Reproduced from [168] by permission of The Electrochemical Society, Inc.

Figure 28 SEM micrographs of Sn65Sb18Cu17 electrode after cycling, with capacity loss 50%. Reproduced from [168] by permission of The Electrochemical Society, Inc.

164

The micrographs appear almost identical to those shown by Dahn. The larger particle size (300-500 nm) of the material after cycling may be explained by aggregation of alloy particles, and SEI thickening, both of which eventually crack and disintegrate as a result of volume changes. In the second part of our work, we characterized a composite anode material, which contained a tinantimony-copper (Sn60Sb20Cu20) alloy of 100 nm particle size, graphite flakes with a particle size of 3-20 mm and PVDF as a binder. The ratio between the alloy matrix and the graphite-PVDF composite was 70:30 and 80:30. Similarly to the previous case the main degradation mechanism of the composite anode is related to particle-to-particle break-off, and to the disconnection of particles

58

Lithium-Ion Batteries: Solid-Electrolyte Interphase

from the current collector. The secondary degradation mechanism results from the thickening of the SEI layer with cycle number and its subsequent cracking, and is influenced by the electrolyte composition. The XPS spectra of fully deinserted alloy after six cycles are shown in Figure 29.

Carbon - 6 Cycles C1s

2000

Tin - 6 Cycles Sn3d

1600

1000

800

Li2CO3 /PVDF

800 10 min sputt.

Intensity

1200

Intensity

Sn3d5/2 5 1

Sn3d3/2

600

10 min sputt.

400

400 200

4 min sputt.

4 min sputt.

0 300

298

296

294

292

290

288

286

284

282

0 500

280

498

496

Binding Energy (eV)

494

492

Oxygen and Antimony - 6 Cycles

2000

486

484

482

480

1200

Li2O

Sb3d3/2

488

Fluorine - 6 Cycles F1s

1600

O1s

2500

490

Binding Energy (eV)

Intensity

Intensity

1500 Sb3d5/2

1000

800 10 min sputt.

10 min sputt.

400 500

4 min sputt.

4 min sputt.

0 544

542

540

538

536

534

532

Binding Energy (eV)

530

528

526

0 700

698

696

694

692

690

688

686

684

682

680

Binding Energy (eV)

Figure 29 XPS spectra of the SEI formed on Sn65Sb18Cu17 electrode (fully de-inserted after 6 cycles). Reproduced from [168] by permission of The Electrochemical Society, Inc.

82

The asymmetry of the main C1s peak (at 284 eV) can be attributed to COH (284.9, 285.2 eV), C=O (286.7, 286.9, 287.0 eV), and O=C-OH (288.2, 288.5 eV), the last possibly associated with polymers. This asymmetry region is less pronounced after 10 minutes of sputtering. The small C1s peak at 291eV 170 may due to the presence of PVDF or LiCO3. Elemental tin at 485 eV, and the broad peaks of SnO (486, 486.9 eV), SnO2 (486.7 eV), and possibly SnF2 (487 eV) are clearly distinguished in the Sn3d spectra. This may indicate the

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

59

incomplete SEI coverage of the electrode surface. It is noteworthy that after 30 cycles, tin and tin compounds were not identified in an electrode, even after 40 minutes of sputtering. This is compatible with the thickening of the SEI. The elemental depth profile of the electrode after 6 cycles revealed that the concentration of Li decreased from 27 to 10 atomic %, and the concentration of F decreased from 17 to 9%. The oxygen content during the sputtering process was almost constant. Therefore the excess of lithium over fluorine was attributed to Li2CO3 and Li2O. Only low concentrations of hydrocarbons and oxygen-containing polymeric compounds were detected in the SEI. In addition, the LiF content in the SEI increases with cycle number from 20 to 48%. Such SEI composition with high concentrations of inorganic salts and low content of polymeric compounds is expected to be followed by increased brittleness and possible break-off on cycling. We believe that the degradation mechanism of the active material in tinbased alloys involves particle-to-particle separation, and the thickening and cracking of the SEI with increasing cycle number.

6

Conclusions

The anode/electrolyte interphase (the SEI) plays a key role in lithium-metal, lithium-alloy and lithium-ion batteries. Today we have some understanding of the first lithium intercalation step into carbon and of the processes taking place on the lithium-metal anode. A combination of a variety of analytical tools including dilatometry, STM, AFM, XPS, EDS, SEM, XRD, QCMB, FTIR, NMR, EPR, TOF SIMS, Raman spectroscopy, AC impedance measurements and DSC is used in order to gain a comprehensive characterization of the processes occurring at the anode/electrolyte interphase. An understanding of SEI-related phenomena is crucial for the development of safer and better lithium-based batteries. The SEI is formed by parallel and competing reduction reactions and thus its composition depends on i0, η and the concentrations of all the electroactive materials. It has been shown that the rate constants of the reactions of solvated electrons with electrolyte and solvent components (and impurities) are a good measure of the stability of these substances towards lithium. It is suggested the rate constants (ke) for these reactions be used as a tool for the selection of electrolyte components. Good correlation was found between ke and SEIformation voltage and composition. Close to the electrode side, the SEI consists

60

Lithium-Ion Batteries: Solid-Electrolyte Interphase −

2−

2−

of fully reduced (thermodynamically stable) anions such as F , O , S and other elements like As, B, C (or their lithiated compounds). The outer part of the SEI (near the solution) consists of partially reduced materials like polyolefins, polyTHF, Li2CO3, LiRCO3, R-O-Li, LiOH, and LiF, LiCl, Li2O, etc. TOF SIMS measurements provide direct evidence for the existence of polymers in the basal SEI. The SEI on the HOPG and SLX electrodes can be described as nonhomogeneous. The true lateral size of the microdomains is about or less than 1 micron. Since the SEI consists of a mosaic of heteropoly-microphases, the contribution of grain boundaries must be considered. The equivalent circuit of the SEI is extremely complex and should be represented by a very large number of series and parallel distributions of RC elements reflecting bulk ionic conductivity and grain-boundary phenomena aside from the Warburg element. For carbon anodes, the exchange-current density of the reduction reactions also depends on the surface properties of the electrode (content of impurities, surface chemistry and surface morphology). The SEI functions differently on the basal and on the cross-section planes. Consequently, the surface chemistry and morphology of the SEI vary with the basal-plane to edge-plane area ratio of carbon. We suggest that the carbon matrix has a more marked effect on the composition and thickness of the SEI than does the nature of the electrolyte. Mild oxidation of graphite was found to improve anode performance. Improvement was attributed to the formation of SEI chemically bonded to the surface carboxylic and oxide groups at the zig-zag and armchair faces, better wetting by the electrolyte and to accommodation of extra lithium at the zig-zag, armchair and other edge sites and nanovoids. Graphite-surface modification by silylation and the creation of the core-shell-structured carbon composite look promising for the enhancement of electrochemical performance. In lithium-ion batteries, with carbonaceous anodes, QIR can be lowered by decreasing the true surface area of the carbon, using pure carbon and electrolyte, applying high current density at the beginning of the first charge and by using proper electrolyte combinations. The processes taking place in the first intercalation of lithium into an alloy anode in a lithium-ion battery assembled in the discharged state are expected to be similar to those in a disordered carbon anode.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

61

References 1. Jasinski R., in Advances in Electrochemistry and Electrochemical Engineering, Vol. 8, ed. by Tobias C. W. (Wiley-Interscience, New York, 1971). 2. Dey A. N., The Electrochemical Society Meeting, Atlantic City, New Jersey, Paper No. 62. 3. Peled E., Straze H. and Brand M., The 26th ISE Meeting, Vienna, 1975. 4. Peled E., J. Electrochem. Soc. 126 (1979), 2047. 5. Peled E., in Lithium Batteries, ed. by Gabano J. P. (Academic, 1983), p. 43. 6. Funabiki A., Inaba M., Abe T. and Ogumi Z., J. Electrochem. Soc. 146 (1999), 2443-2448. 7. Peled E., Golodnitsky D., Ardel G., Menachem C., Bar-Tow D. and Eshkenazy V., Mat. Res. Soc. Symp. 393 (1995), 209. 8. Peled E., Golodnitsky D. and Ardel G., J. Electrochem. Soc. 144 (1997), L208. 9. Mann C. K., in Electroanalytical Chemistry, ed. by Bard A. J. (Marcel Dekker, 1969), p. 3. 10. Electrons in Fluids, ed. by Jortner J. and Kestner N. R. (Springer-Verlag Berlin, 1973). 11. Peled E., Golodnitsky D., Menachem C. and Bar-Tow D., J. Electrochem. Soc. 145 (1998), 3482-3486. 12. Anbar M. and Neta P., Int. J. Appl. Radiation and Isotopes 18 (1967), 493. 13. Buxton G. V., Greenstock C. L., Helman W. P. and Ross A. B., J. Phys. Chem. Ref. Data 17 (1988), 513. 14. Kanamura K., Shiraishi S. and Takehara Z.-I., J. Electrochem. Soc. 143 (1996), 2187. 15. Odziemkovski M. and Irish D. E., J. Electrochem. Soc. 140 (1994), 1546. 16. Aurbach D. and Zaban A., J. Electroanal. Chem. 348 (1993), 155. 17. Kanamura K., Tamura H., Shiraishi S. and Takehara Z.-I., J. Electroanal. Chem. 394 (1995), 49. 18. Dominey L. A., in Lithium Batteries, New Materials, Development and Perspectives, ed. by Pistoia G. (Elsevier, 1994), p. 137. 19. Aurbach D., Weissman I., Zaban A. and Chuzid O. (Youngman), Electrochim. Acta 39 (1994), 51. 20. Bar-Tow D., Peled E. and Burstein L., Proceedings of the Electrochemical Soc. Meeting (Paris, 1997), 97-18 324.

62

Lithium-Ion Batteries: Solid-Electrolyte Interphase

21. Besenhard O. J., Winter M., Yang J. and Biberacher W., J. Power Sources 54 (1995), 228. 22. Peled E., Golodnitsky D. and Penciner J., Anode/Electrolyte Interface, in Handbook of Battery Materials, ed. by Besenhard J. O. (Verlag Chemie (VCH), 1998), pp. 410-457. 23. Thevenin J. G. and Muller R. H., J. Electrochem. Soc.134 (1987), 273. 24. Endo E., Ata M., Sekia K. and Tanaka K., J. Electrochem. Soc. 146 (1999), 49. 25. Peled E., Golodnitsky D., Ardel G. and Eshkenazy V., Electrochimica Acta 40 (1995), 2197. 26. Aurbach D., in Nonaqueous Electrochemistry (Marcel Dekker, NY, Basel, 1999), pp. 289-411. 27. Aurbach D., Daroux M., McDougal G. and Yeager E. B., J. Electroanal. Chem. 358 (1993), 63. 28. Odziemkowski M., Krell M. and Irish D. E., J. Electrochem. Soc. 139 (1992), 3063. 29. Kanamura K., Shiraishi S. and Takehara Z.-I., Chemistry Letters (1995), 209. 30. Ely Y. E. and Aurbach D., Langmuir 8 (1992), 1845. 31. Kanamura K., Shiraishi S. and Takehara Z.-I., J. Electrochem. Soc. 141 (1994), L108-L110. 32. Nazri G. and Muller R. H., J. Electrochem. Soc. 132 (1985), 1385. 33. Aurbach D. and Gofer Y., J. Electrochem. Soc. 138 (1991), 3529. 34. Montesperelli G., Nunziante P., Pasquali M. and Pistoia G., Solid State Ionics 37 (1990), 149. 35. Schwager F., Geronov Y. and Muller R. H., J. Electrochem. Soc. 132 (1985), 285. 36. Matsuda Y., in Electrochemistry in Transition, ed. by Murply O. J., Srinivasan S. and Conway B. E. (Plenum, New York, 1992), Ch.14. 37. Morita M., Aoki S. and Matsuda Y., Electrochim. Acta 37 (1992), 119. 38. Matsuda Y., J. Power Sources 43 (1993), 1. 39. Tobishima S., Hayashi K., Nemoto Y. and Yamaki J., 37th Battery Symposium, Tokyo, Japan, 1996. 40. Campbell S. A., Bowes C. and McMillan R. S., J. Electroanal. Chem. 284 (1990), 195. 41. Aurbach D., Moshkovich M., Cohen Y. and Schechter A., Langmuir 15 (1999) 2947.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

63

42. Moshkovich M., Gofer Y. and Aurbach D., J. Electrochem. Soc. 148 (2001), E155-167. 43. Li G. F., Li H., Mo Y. J., Chen L. Q. and Huang X. J., J. Power Sources 104 (2002), 190-194. 44. Mori M., Kakuta Y., Naoi K. and Fauteux D., 37th Battery Symposium, Tokyo, Japan, 1996. 45. Xu J. and Farrington G. C., Abstracts, Electrochemical Society Meeting (San Antonio, Texas, October, 1996), paper No. 74. 46. Ardel G., Golodnitsky D., Peled E., Appetecchi G. B., Romagnoli P. and Scrosati B., Journal of Power Sources 110 (2002), 4862. 47. Appetecchi G. B., Croce F., Persi L., Ronci F. and Scrosati B., Electrochim. Acta 45 (2000), 1481-1490. 48. Ryoo H.-J., Kim H.-T., Lee Y.-G., Park J.-K. and Moon S.-I., J. Solid State Electrochem. 3 (1998), 1-6. 49. Matsui T. and Takeyama K., Electrochim. Acta 40 (1995), 2165. 50. Brissot C., Rosso M., Chazalviel J.-N. et al., Electrochim. Acta 43 (1998), 1569. 51. Appetecchi G. B., Alessandrini F., Duan R. G., Arzu A. and Passerini S., J. Power Sources 101 (2001), 42. 52. Kumar B. and Scanlon L. G., Proceedings of the 36th Power Sources Conference (Cherry Hill, N. J., 1994), p. 236. 53. Nagasubramanian G., Attia A. I., Halpert G. and Peled E., Solid State Ionics 67 (1993), 51. 54. Croce F., Appetecchi G. B., Persi L. and Scrosati B., Nature 394 (1998), 456; Appetecchi G. B., Scaccia S. and Passerini S., J. Electrochem. Soc. 147 (2000), 4448-4452. 55. Cookson J. T., in Carbon Adsorption Handbook, ed. by Cheremisinoff P. N. and Ellerbush F. (Ann Arbor Science, 1978), p. 241. 56. Kinoshita K., Carbon, Electrochemical and Physicochemical Properties (Wiley-Interscience, 1987). 57. Pierson H. P., Handbook of Carbon, Graphite, Diamond and Fullerenes (Noyes, Park Ridge, N. J., 1993). 58. Dubinin M. M., Chem. Phys. Carbon, ed. by Walker Jr P.L. ( Marcel Dekker, New York, 1966) 2:251. 59. Xing W. and Dahn J. R., The 190th Electrochemical Soc. Meeting (San Antonio, 1996), 96-2, 1055. 60. Chu A. C., Josefowicz J. X., Fischer J. E. and Farrington G. C., The 190th Electrochemical Soc. Meeting (San Antonio, 1996), 96-2, 1062.

64

Lithium-Ion Batteries: Solid-Electrolyte Interphase

61. Menachem C., Peled E. and Burstein L., Proceedings of the 37th Power Sources Conference (N. J., 1996), 208. 62. Peled E., Menachem C., Bar-Tow D. and Melman A., J. Electrochem. Soc. 143 (1996), L4. 63. Menachem C., Peled E., Burstein L. and Rosenberg Y., 8th International Meeting on Li Batteries (Nagoya, Japan, 1996), 224; J. Power Sources 68 (1997), 277. 64. Fong R., Sacken U. V. and Dahn J. R., J. Electrochem. Soc. 137 (1990), 2009. 65. Peled E., in Rechargeable Lithium and Lithium-Ion Batteries, The Electrochem. Soc. Proceedings Series, ed. by Megahed S., Barnett B. M. and Xie L. (1995), 94-28, 1. 66. Zaghib K., Tatsumi K., Abe H., Sakaebi H., Higuchi S., Ohsaki T. and Sawada Y., Proceedings of the Electrochemical Soc. Meeting (San Francisco, 1994), 94-1. 67. Bar-Tow D., Peled E. and Burstein L., The 190th Electrochemical Society Meeting (San Antonio, 1996), 96-2, 1028. 68. Guyomard D. and Tarascon J. M., U.S. Patent 5192629 (1993). 69. Huang C. K., Surampudi S., Shen D. H. and Halpert G., The 184th Electrochemical Society Meeting (San Antonio, 1993), 93-2, 26. 70. Dahn J. R. et al., in Lithium Batteries, New Materials, Development and Perspectives, ed. by Pistoia G. (Elsevier, 1994), p. 22. 71. Shu Z. X., McMillan R. S. and Murray J., J. Electrochem. Soc. 140 (1993), 922. 72. Cookson J. T., in Carbon Adsorption Handbook, ed. by Cheremisinoff P. N. and Ellerbush F. (Ann Arbor Science, 1978), p. 241. 73. Bar-Tow D., Ph.D. thesis (1999). 74. Peled E., Bar Tow D., Merson A. and Burstein L., Journal of New Materials For Electrochemical Systems 3 (1999), 21. 75. Peled E., Bar Tow D., Merson A., Gladkich A., Burstein L. and Golodnitsky D., J. Power Sources 97-98 (2001), 52-57. 76. Dahn J. R. et al., in Lithium Batteries, New Materials, Development and Perspectives, ed. by Pistoia G. (Elsevier, 1994), p. 22. 77. Simon B., Boeve J. P., Spigi B. and Broussely B., The Electrochem. Soc. and Int. Soc. of Electrochem. Meeting (Aug. 1997), Abstracts, Vol. 97-2, Abstract 156. 78. Inaba M., Siroma Z., Kawatate Y., Funabiki A. and Ogumi Z., J. Power Sources 68 (1997), 221.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

65

79. Zavadil K. R., Guidotti R. A. and Even W. R., Mat. Res. Soc. Symp. Proc. (1998), Vol. 496, 539. 80. Blyth R. I. R., Buqa H., Netzer F. P., Ramsey M. G., Besenhard J. O., Golob P. and Winter M., Applied Surface Science 167 (2000), 99-106. 81. Eriksson T., Gustafsson T. and Thomas J. O., in Lithium Batteries, ed. by Surampudi S. and Marsh R. (The Electrochemical Society Proceedings Series, Pennington, N. J. 1998), PV 98-16, p. 315. 82. Bar-Tow D., Peled E. and Burstein L., J. Electrochem. Soc. 146 (1999), 824. 83. Peled E., Eshkenazi V. and Rosenberg Y., J. Power Sources 76 (1998), 153-158. 84. Eshkenazi V., Peled E., Burstein L. and Golodnitsky D., Solid State Ionics (2002), submitted. 85. Yudasaka M., Kikuchi R., Matsui T., Ohki Y., Baxendale M., Yoshimura S. and Ota E., Thin Solid Films 280 (1996), 117. 86. Yu H., Kaneko T., Otani S., Sasaki Y. and Yoshimura S., Carbon 36 (1998), 137. 87. Xing W., Xue J., Zheng T., Gibaud A. and Dahn J., J. Electrochem. Soc. 143 (1996), 3482. 88. Lascovich J. C., Giorgi R. and Scaglione S., Appl. Surf. Sci. 47 (1997), 17. 89. Takehara Z., Shiraishi S. and Kanamura K., Electrochem. Soc. Proc., Vol. 94-28, 57. 90. Eriksson T., Andersson A. M., Bishop A.G., Gejke C., Gustafsson T. and Thomas J. O., J. Electrochem. Soc. 149 (2002), 1, A69-A78. 91. Aurbach D., Chusid O., Ein-Ely Y., Babai M. and Carmeli Y., J. Power Sources 43 (1993), 47. 92. Franklin R. E., Proc. Roy. Soc. (London) A209 (1951), 196. 93. Zaghib K., Nadeau G. and Kinoshita K., J. Power Sources 97-98 (2001), 97-103. 94. Fujimoto H., Tokumitsu K., Mabuchi A., Kasuh T. and Shiraishi M., Carbon 32 (1994), 1249. 95. Yamamoto O., Takeda Y. and Imanishi N., Proc. Electrochem. Soc. Meeting (Honolulu, 1993). 96. Aurbach D., Ein-Eli Y., Chusid O., Carmeli Y., Babai M. and Yamin H., J. Electrochem. Soc. 141 (1994), 603. 97. Momose H., Honbo H., Takeuchi S., Nishimura K., Horiba T., Muranaka Y., Kozono Y. and Miadera H., Abstracts, 8th International Meeting on Li Batteries, (Nagoya, Japan, 1996), p. 172.

66

Lithium-Ion Batteries: Solid-Electrolyte Interphase

98. Besenhard J. O., Wagner M. W., Winter M., Jannakoudakis A. D., Jannakoudakis P. D. and Theodoridou E., J. Power Sources 43-44 (1993), 413. 99. Ein-Eli Y., Thomas S. R. and Koch V. R., J. Electrochem. Soc. 143 (1996), L195. 100. Inaba M., Siroma Z., Kawatate Y., Funabiki A. and Ogumi Z., J. Power Sources 68 (1997), 221. 101. Morita M., Ichimura T., Ishikawa M. and Matsuda Y., J. Electrochem. Soc. 143 (1996), L26. 102. Huang C. K., Surampudi S., Shen D. H. and Halpert G., The 184th Electrochemical Soc. Meeting (San Antonio, 1993), 93-2, 26. 103. Jean M., Chausse A. and Messina R., Abstracts, 8th International Meeting on Li Batteries (Nagoya, Japan, 1996), p. 186. 104. Tanaka K., Itabashi M., Aoki M., Hiraka S., Kataoka M., Fujita S., Sekai K. and Ozawa K., The 184th Electrochem. Soc. Meeting (New Orleans, 1993), Vol. 93-2, Abs. No. 21. 105. Huang C. K., Smart M., Davies E., Cortez R. and Surampudi S., The 190th Electrochem. Soc. Meeting (San Antonio, 1996), 96-2, 171. 106. Yazami R., Deschamps M., Genies S., Frison J. C. and Ledran J., Abstracts, 8th International Meeting on Li Batteries (Nagoya, Japan, 1996), p. 107. 107. Aurbach D., Markovsky B. and Shechter A., 190th Electrochem. Soc. Meeting (San Antonio, 1996), 96-2, 164. 108. Yazami R., Deschamps M., Genies S. and Frison J. C., J. Power Sources 68 (1997), 110-113. 109. Yazami R. and Deschamps M., Abstracts, 8th International Meeting on Li Batteries (Nagoya, Japan, 1996), p. 190. 110. Joho F., Rykart B., Blome A., Novak P., Wilheim H. and Spahr M., J. Power Sources 97-98 (2001), 78-82. 111. Lanz M., Lehnmann E., Imhof R., Exnar I. and Novak P., J. Power Sources 101 (2001), 177-181. 112. Wang Y., Guo X., Greenbaum S., Liu J. and Amine K., Electrochem. Solid- State Lett. 4 (2001), A68-70. 113. Andersson A. M. and Edstrom K., J. Electrochem. Soc. 148 (2001), A1100-9. 114. Lin H., Chua D., Salomon M., Shiao H-C., Hendrickson M., Plichta E. and Slane S. Electrochem. Solid-State Lett. 4 (2001), A71-3. 115. Lanz M. and Novak P, J. Power Sources, 102(2001), 277-82.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

67

116. Wang X., Yasukawa E. and Kasuya S., J. Electrochem. Soc. 148 (2001), A1066-71. 117. Wang Y., Yufit V., Guo X., Peled E. and Greenbaum S., J. Power Sources 94 (2001), 230-237. 118. Menachem C., Peled E. and Burstein L., 37th Power Sources Conference (N. J., 1996), p. 208. 119. Menachem C., Wang Y., Fowers J., Peled E. and Greenbaum S. G., J. Power Sources 76 (1998), 180-185. 120. Maleki H., Deng G., Anani A. and Howard A., J. Electrochem. Soc. 146 (1999), 3224. 121. Okamoto T., Tsukamoto H., Sasaki T., Komatsu S., Nakamitsu K., Mizutani M. and Yamachi M., Proceedings of the 37th Power Sources Conference (N. J., 1996), p. 216. 122. Zhang Z., Fouchard D. and Rea J. R., The 190th ECS Fall Meeting (San Antonio, Texas, 1996), p. 168. 123. Menachem C., Peled E. and Golodnitsky D., J. Solid State Electrochem. 5 (2001), 81-87. 124. Mori S., Asahina H., Suzuki H., Yonei A. and Yasukawa E., The 8th International Meeting on Lithium Batteries (Nagoya, Japan, 1996), p. 40. 125. Bar-Tow D. and Peled E., Proc. Joint ECS Meeting (Paris, France, 1997). 126. Tibbets G. G., Nazri G.-A. and Howie B. J., Abstracts, The 190th Electrochemical Society Meeting (San Antonio, 1996), 96-2, 117. 127. Ein-Eli Y. and Koch V. R., J. Electrochem. Soc., submitted. 128. Takamura T., Kikuchi M. and Ikezawa Y., in Rechargeable Lithium-Ion Batteries: The Electrochem. Soc. Proceedings, ed. by Megahed S., Barnett B.M. and Xie L. (1995), 94-28, 213. 129. Buqa H., Golob P., Winter M. and Besenhard J. O., J. Power Sources 9798 (2001), 122-5. 130. Buqa H., Grogger Ch., Santis M. V. Alvarez, Besenhard J. O. and Winter M., J. Power Sources 97-98 (2001), 126-8. 131. Kuribayashi I., Yokoyama M. and Yamashita M., J. Power Sources 54 (1995), 1. 132. Qiu W., Zhang G., Lu S. and Liu Q., Chin. J. Power Sources 23 (1999), 7. 133. Yoshio M., Wang H., Fukuda K., Hara Y. and Adachi Y., J. Electrochem. Soc. 147 (2000), 1245-1250. 134. Wang J., Raistrick I. D. and Huggins R. A., J. Electrochem. Soc. 133 (1986), 457.

68

Lithium-Ion Batteries: Solid-Electrolyte Interphase

135. Besenhard J. O., in Progress in Intercalation Research, ed. by MullerWarmuth W. and Schollhorn R. (Kluwer, Dordrecht, 1994), p. 457. 136. Besenhard J. O., Yang J. and Winter M., J. Power Sources 68 (1997), 8790. 137. Idota Y., Nishima M., Miyaki Y., Kubota T. and Miyasaki T., Can. Pat. Appl. 2,134,053 (1994). 138. Idota Y., Matsufuji A., Maekawa Y. and Miyasaki T., Science 276 (1997), 1395. 139. Huggins R. A.,J. Power Sources 81-82 (1999), 13-19 140. Courtney A., McKinnon W. R. and Dahn J. R.,J. Electrochem. Soc. 146 (1999), 59-68. 141. Goward G. R., Leroux F., Power W. P., Ouvrard G., Dmowski W., Egami T. and Nazar L. F., Electrochem. Solid-State Lett. 2 (1999), 367-370. 142. Wang Y., Sakamoto J., Huang C. K., Surampudi S. and Greenbaum S. G., Solid State Ionics 110 (1998), 167-172. 143. Wang Y., Sakamoto J., Kostov S., Mansour A. N., denBoer M. L., Greenbaum S. G., Huang C. K. and Surampudi S., J. Power Sources 89 (2000), 232-236. Machill S., Shodai T., Sakurai Y. and Yamaki J., J. Power Sources 73 (1998), 216-223 145. Machill S., Shodai T., Sakurai Y. and Yamaki J., J. Solid State Electrochem. 3 (1999), 97-103 146. Panero S., Savo G. and Scrosati B., Electrochem. Solid-State Lett. 2 (1999), 365-366. 147. Huang H., Kelder E. M., Chen L. and Shoonman J., J. Power Sources 8182 (1999), 362-367. 148. Besenhard J. O., Yang J. and Winter M.,J. Power Sources 68 (1997), 8790. 149. Yang J., Wachtler M., Winter M. and Besenhard J. O., Electrochem. Solid-State Lett. 2 (1999), 161. 150. Wachtler M., Besenhard J. O. and Winter M., J. Power Sources 94 (2001), 189-193. 151. Yang J., Takeda Y., Imanishi N., Ichikawa T. and Yamamoto O., J. Power Sources 79 (1999), 220-224. 152. Yang J., Takeda Y., Imanishi N. and Yamamoto O., J. Electrochem. Soc. 146 (1999), 4009-4011. 153. Mao O., Dunlap R. A., Courtney I. A. and Dahn J. R., J. Electrochem. Soc.145 (1998), 4195-4202.

SEI on Lithium, Graphite, Disordered Carbons and Tin-Based Alloys

69

154. Mao O., Turner R. L., Courtney I. A., Fredericksen B. D., Buckett M. I., Krause L. J. and Dahn J. R., Electrochem. Solid-State Lett. 2 (1999), 3-5. 155. Mao O., Dunlap R. A., Courtney I. A. and Dahn J. R., J. Electrochem. Soc.146 (1999), 405-413. 156. Mao O. and Dahn J. R., J. Electrochem. Soc.146 (1999), 414-422. 157. Mao O. and Dahn J. R., J. Electrochem. Soc146 (1999), 423-427. 158. Beaulieu L. Y., Larcher D., Dunlap R. A. and Dahn J. R., J. Electrochem. Soc. 147 (2000), 3206-3212. 159. Kepler K. D., Vaughey J. T. and Thackeray M. M., Electrochem. SolidState Lett. 2 (1999), 307-309. 160. Kepler K. D., Vaughey J. T. and Thackeray M. M., J. Power Sources 8182 (1999), 383-387. 161. Kim I., Kumta P. N. and Blomgren G. E., Electrochem. Solid-State Lett. 3 (2000), 493-6. 162. Whitehead A. H., Elliott J. M. and Owen J. R., J. Power Sources 81-82 (1999), 33-38. 163. Dayalan E., th Meeting of the Electrochemical Society (), Abs. 261. 164. Beaulieu L. Y., Eberman K. W., Turner R. L., Krause L. J. and Dahn J. R., Electrochem. Solid-State Lett. 4 (2001), A137-A140. 165. Wu Z., Wang C., Appleby A. J. and Little F. E., Solid State Ionics 150 (2002), 383-389. 166. Peled E. and Ulus A., 194th Meeting of the Electrochemical Society (Boston), Vol. 98-16, pp. 123-127. 167. Peled E., Ulus A. and Rosenberg Yu., New Materials for Batteries and Fuel Cells, ed. by Doughty D. H., Brack H. P., Naoi K. and Nazar L. F., Symposium, 5-8 April 1999 (Materials Research Society Symposium Proceedings, Vol. 575). MRS, Warrendale, PA, USA (2000), pp. 151-161. 168. Ulus A., Rosenberg Yu., Burstein L. and Peled E., J. Electrochem. Soc. 149 (2002), A635-A643. 169. Ulus A., Ph.D. thesis (Tel Aviv University, 2002). 170. Moulder J. F., Stickle W. F., Sobol P. E. and Bomben K. D., Handbook of X-Ray Photoelectron Spectroscopy, ed. by Chastain J. (Perkin-Elmer, Eden Prairie, Minnesota, 1992).

CHAPTER 2



IDENTIFICATION OF SURFACE FILMS ON ELECTRODES IN NON-AQUEOUS ELECTROLYTE SOLUTIONS: SPECTROSCOPIC, ELECTRONIC AND MORPHOLOGICAL STUDIES DORON AURBACH* AND YARON S. COHEN† Department of Chemistry Bar-Ilan University, Ramat-Gan 52900, Israel E-mail: *[email protected]; †[email protected]

1 Introduction 1a Passivation of Surface Films: A General Phenomenon Passivation phenomena in electrochemistry and electrochemical systems controlled by surface films are widely dealt with and have been extensively studied over the years.1 Most of the solid electrode materials, including noble metals and carbons, are covered, at least in part, by surface films or surface groups. Both transition and noble metals develop oxide films in aqueous solutions. Carbonaceous materials are naturally covered by surface groups such as OH, C=O, -(C=O)-O(C=O)- due to the oxidation of reactive carbon atoms at their surfaces.2 Active metals from the left three columns of the periodic table of elements (e.g., alkaline, alkaline earth, aluminum) are naturally covered by surface films due to their spontaneous reactions with atmospheric components such as O2, H2O, CO2, and, in the case of the alkaline metals, with N2 as well. Surface film phenomena are largely connected with corrosion science, since corrosion is pronounced when the surface films that cover the metal are permeable to water molecules and ions (e.g., as is the case for iron).3 This can be largely prevented by electrochemical growth of impermeable, blocking surface films (e.g., anodized aluminum).4 Consequently, surface film phenomena have received a lot of attention over the years. Models describing the growth of surface films on electrodes were developed (e.g., the parabolic low of growth was established in many cases).5 The migration of ions through surface films was widely explored. It was established that surface films such as metal oxides that are apparently 70

Identification of Surface Films on Electrodes

71

insulating, can conduct ions due to defects in their crystal structure.6 Equations describing the relationship between the strength of the electric field applied and the resulting ion migration current through surface films were established.7 The kinetics of the simplest solid electrolyte interphase (SEI) should include three stages: charge transfer across the solution-film interface, ion migration through the surface films, and charge transfer in the film-metal interface. It is reasonable to assume that the ion migration is the ratedetermining step. Thus, it may be possible to use the basic Equation 1 for ionic conductance in solids as the starting point: 4, 8, 1 i = 4zFanυ· exp (-W/RT )· sinh (azFE/RT)

(1)

where a is the jump’s half distance, υ is the vibrational frequency in the lattice, z is the ion’s charge, W is the energy barrier for the ion jump, n is the ion’s concentration, E is the electric field, and F is the Faraday number. When all of the potential falls on the surface films, η = ηSEI = El

(2)

where l is the film’s thickness. At equilibrium η=0, so the net current is zero, and the exchange current is io = 2zFanυ· exp (-W/RT)

(3)

In a high electrical field, azFη· > RTl, and thus a Tafel-like behavior is obtained: i = io· exp(azFη/RTl)

(4)

In a low electrical field, Equation 4 can be linearized, and thus an ohmic behavior is obtained: i = 4.6 io η/b

(5)

where b is the analog of the Tafel slope extracted from Equation 4: b = 2.3RT/lazF

(6)

Hence, the average resistivity of the surface films can be extracted as ρ/A = Rfilm/l = b/4.6 io l = RT/2azFio A = the electrode’s surface

(7)

where Rfilm = η/I is the surface film resistance for ionic conductance, extracted from Equation 7 and I = iA. For example, the average resistivity values of surface films formed on active metals such as lithium, magnesium, and calcium in nonaqueous solutions are in the order of 108, 109, and 1010 Ω · cm2, respectively.5

72

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Hence, in many cases the electrochemical response of electrodes covered by surface films converges to a Butler Volmer-type potential-current dependence, which shows a Tafel-like behavior at a high electrical field, and a linear potential-current behavior (Ohm’s low) at a low applied electrical field. The field of high energy density batteries, which relates to the electrochemistry of highly reactive electrodes, requires an intensive study of surface film phenomena, because the electrochemistry of both anodes and cathodes of the most advanced high energy density batteries (e.g., rechargeable Li, Li-ion batteries) is controlled by surface films. 1b Modes of Growth of Surface Films, and Their Transport Properties We can distinguish between two modes of growth of surface films in electrochemical systems: spontaneous or stimulated growth. Most of the metals used as electrode materials, including commonly used transition metals, can be considered as reactive to a wide variety of electrolyte solutions. The spontaneous development of oxide/hydroxide surface films on metals in aqueous solutions as a function of pH, is described quite precisely in Pourbaix diagrams9 and is beyond the scope of this chapter. Active metals such as alkaline and alkaline earth metals react spontaneously and vigorously in aqueous media. In the case of the alkaline metals, the surface films (mostly hydroxides) dissolve in water, and therefore passivation of these metals cannot be seen in aqueous solutions unless the solubility limit (which is very high) is reached. Alkaline earth metals that also react spontaneously and vigorously with water may develop passivation in aqueous solutions, since their hydroxides are much less soluble in water compared with the MOH species (M=Li, Na, K, etc.). Highly interesting, and also complicated, is the surface chemistry of reactive metals in nonaqueous solutions. When active metals (e.g., Li, Mg, Ca), which are always covered by native surface films, are introduced into nonaqueous, polar aprotic solutions, a large variety of surface reactions takes place, which form highly complicated, multilayer and laterally non-uniform surface films. Active metals seem to be stable in a large variety of nonaqueous solutions because of their passivation by these complicated surface films. In the next sections of this chapter, the surface films on active metals and related phenomena are rigorously dealt with in detail. We should also mention the stimulated growth of surface films. Anodic processes of metals may lead to the growth of surface films in cases where the electrochemically dissolved metal ions interact with solution species, resulting

Identification of Surface Films on Electrodes

73

in precipitation of insoluble metal oxide/hydroxide. An example of this is the anodizing process of aluminum,4 and breaking and repair processes during dissolution of Mg++ and Ca++ 10, 11 from the metal electrodes in nonaqueous electrolyte solutions. Another type of stimulated growth of surface films is obtained by cathodic polarization of non-active metal or carbon electrodes in nonaqueous Li salt solutions, which leads to precipitation of surface films comprising insoluble Li salts (due to reduction of solution species). These systems are also dealt with in detail in the next sections. 1c On the Effect of the Electrolyte Solutions When dealing with surface film controlled electrochemical systems, the nature of the electrolyte solutions is the most interesting and critical factor. The scope of this chapter relates only to highly reactive electrodes, including lithium, lithiated carbons, magnesium, etc., and thereby, only polar aprotic solvents are relevant. On a thermodynamic basis, lithium metal should react with any polar aprotic solvent. In order to be polar, a solvent has to contain C-Cl, C-O, C-N, CS, C-P, S-O, S-Cl bonds, etc. Li reduces such bonds to form Li salts, in thermodynamically favorable reactions. Nevertheless, reactive metals such as lithium and magnesium are apparently stable in a large variety of polar aprotic solvents due to passivation phenomena: initial, spontaneous reactions between active metal and polar aprotic solvents form insoluble salts that precipitate on the reactive surface and passivate the active metal.5 There are some inorganic polar aprotic solvents of interest, such as SO2, SOCl2, SO2Cl2.12 However, most of the attention in nonaqueous electrochemistry is focused on organic solvents. Figure 1 presents structural formulae of several polar aprotic organic solvents relevant to the field of Li batteries, plus formulae of relevant Li salts which form highly conductive and stable solutions with these solvents (provided that the salts are not contaminated by Lewis acids). In addition to the list in Figure 1, there are some other polar aprotic organic solvents that are currently used in nonaqueous electrochemistry. These include CH3N (acetonitrile), (CH3)2SO (DMSO), HCON(CH3)2 (DMF), and CH2Cl2 (methylene chloride).13 However, since the major interest in polar aprotic electrolyte systems in electrochemistry relates to lithium batteries, this chapter concentrates only on solvents and salts from the list in Figure 1.

74

Lithium-Ion Batteries: Solid-Electrolyte Interphase

1d The Role of the Cation in Surface Phenomena in Non-aqueous Electrolyte Solutions The nature of the cation is critical in determining the electrode’s reactions in non-aqueous solutions, especially at low potentials. Polar aprotic solvents are reduced at low potentials. The thermodynamics of their reduction processes (e.g., their reduction potentials) depend on the cation involved because the products of these processes are the cation salts. In addition, the nature of the cation determines the solubility of the reduction products, and hence, the Structures of Organic Solvents for Lithium Batteries O

Carbonates

O

O

O O

O

O

O O

Ethylene Carbonate (EC)

Propylene Carbonate (PC)

O

O

O

Diethyl Carbonate (DEC)

Dimethyl Carbonate (DMC)

Esters O H

O H

O

Methyl formate

O O

O O

Ethyl formate

O

Methyl acetate

Ethyl acetate

Cyclic Ethers O

1,3-Dioxolane (DIOX)

Tetrahydorfuran (THF)

Lactones

O

O

O

O

O Valerolactone

O

2-Methyltetrahydrofuran (2-MeTHF)

2,5-Dimethyltetrahydrofuran (2,5-DiMeTHF)

Aliphatic Ethers

O O

O -Butyrolactone

Diethyl ether (DEE)

O

O

1,2-Dimethoxyether (DME)

Li Salts of Interest LiClO 4, LiAsF6, LiPF6, LiBF4, LiPF3(CF2CF3)3, LiBC4O8, LiSO3CF3, LiC(SO2CF3)3, LiN(SO2CF3)2, LiN(SO2CF2CF3)2

Figure 1 Structure formulae of organic solvents for lithium batteries and Li salts of interest.

Identification of Surface Films on Electrodes

75

passivation processes of the electrodes.14 For instance, cathodic polarization of noble metal electrodes in tetra alkyl ammonium salt solutions in ethers, esters, and alkyl carbonates does not lead to passivation of the electrodes. In ethers, the alkyl ammonium cations are reduced to trialkyl amines (soluble), plus the corresponding alkanes and alkenes. In esters and alkyl carbonate solutions, the solvent molecules are reduced to soluble tetra alkyl ammonium salts.14 In contrast, in Li salt solutions (solvents and anions) the solvent’s reduction potentials are higher (compared with that measured in tetra alkyl ammonium salt solutions). The reduction products, which are insoluble Li salts, precipitate on the electrode’s surfaces as passivating surface films that block the electrode and prevent further reduction of solution species.15 Hence, passivation of electrodes by surface films depends on the nature of the cation since it determines the solubility of surface species. The nature of the cation also determines the transport properties of surface films formed on electrodes in non-aqueous solutions. For example, in Li salt solutions the surface films formed on electrodes behave like a solid electrolyte interphase.5 In magnesium or calcium salt solutions, surface films formed on electrodes (comprising salts of the bivalent metal) block the electrodes because they cannot conduct the bivalent metal cations.10, 11 1e On the Impact of the Electrode’s Material In this chapter we deal with four major electrode surfaces: active metals, carbons, non-active metals (e.g., noble metals), and composite electrodes comprising lithiated transition metal oxide powders as the active mass, plus polymeric binder and conductive additives (usually carbon black or graphite powders at low percentage). In terms of general surface chemistry, we find that the surface reactions on lithium, lithiated carbons, carbon, and noble metals polarized to low potentials in non-aqueous Li salt solutions are very similar. All of these electrodes are covered by surface films comprising insoluble Li salts, which are formed by reduction of solution species. Upon anodic polarization of carbon or noble metal electrodes in non-aqueous solutions, solution species are oxidized. Here, the impact of the cations is negligible. It seems that the species that determine the anodic stability of non-aqueous solutions are the solvents. For instance, ether may be oxidized at potentials below 4 V, while alkyl carbonates may apparently be stable up to 5 V (Li/Li+). However, it should be noted that some minor oxidation reactions of alkyl carbonate solvents on noble metal electrodes (e.g., Pt, Au) can be detected even at a potential below 4 V.16 The

76

Lithium-Ion Batteries: Solid-Electrolyte Interphase

anodic stability of electrochemical systems comprising polar aprotic solutions and electrodes other than noble metals (Au, Pt), or carbons, is usually determined by oxidation of the electrodes’ metal. In contrast to cathodic polarization, anodic polarization of noble metals in non-aqueous solutions is usually not accompanied by passivation phenomena. However, an important electrode material for batteries, especially as a current collector for the cathodes, is aluminum. The anodic stability of aluminum depends on passivation phenomena. When the salts contain halogen atoms, e.g., LiPF6, LiBF4, LiAsF6, and LiClO4, aluminum becomes passivated by species such as AlF3, AlCl3, etc., which precipitate on its surface and prevent Al dissolution, but allow electrical contact with the cathodes’ active mass.17 Hence, due to the above-described passivation phenomena, aluminum current collectors are apparently stable in nonaqueous solutions, even at potentials above 5 V (Li/Li+). In the case of cathodes for Li batteries where the active mass constitutes lithiated transition metal oxides, we discovered that there is a possibility for a variety of spontaneous reactions between LixMOy (M=Co, Ni, Mn, V, etc.) compounds and electrolyte solutions comprising alkyl carbonate solutions (strong electrophiles) and Li salts such as LiPF6, which form surface films.18 1f

Some Comments on Applications

The subject of surface films on electrodes in non-aqueous solutions is mostly important for the field of batteries. The performance of both Li and Li-ion batteries depends strongly on passivation phenomena that relate to surface film formation on both the anodes and the cathodes. Lithium and lithiated carbon anodes reduce all the solvents and salt anions in electrolyte solutions relevant to Li batteries. The products of these surface reactions always contain insoluble Li salts that precipitate on the electrodes as surface films. All charge transfer processes of Li, Li-C, and Li alloy anodes in Li batteries involve the critical step of Li-ion migration through the surface films. Thereby, the composition, structure, morphology, and electrical properties of surface films on Li, Li-C, and Li alloy electrodes were studied very intensively over the years. In contrast, reversible magnesium electrodes can function only in surface film-free conditions.10 As mentioned above, several important cathode materials for Li-ion batteries such as LiCoO2, LiNiO2 and LiMn2O4 react spontaneously with solution species in alkyl carbonate solutions and become covered by surface

Identification of Surface Films on Electrodes

77

layers.18 These surface layers are also important in determining the performance of cathodes for Li batteries. It should be noted that aging processes and capacity fading of Li-ion batteries relate very strongly to surface film formation, secondary surface reactions, and passivation phenomena.19 All of these are dealt with in this chapter.

2 2a

Methods for Identification of Surface Films on Electrodes Introductory Remarks

There are various spectroscopic techniques than can provide surface analysis of electrodes. These can be divided according to several categories: 1.

2.

3.

Techniques that provide specific identification of surface species, such as FTIR, Raman, SIMS, XPS, EXAFS, XRD vs. techniques that provide less specific information, such as AES, EDAX, electron (SEM, TEM) scanning and probe (AFM, STM) microscopies, and solid state NMR. Techniques which enable in situ electrode characterization (e.g., surface films can be identified while the electrode is in solution, under potential control), such as FTIR, RAMAN, XRD, EQCM, EXAFS, STM, AFM vs. ex situ methods that include XPS, SEM, and AES, SIMS and solid state NMR. Techniques that can be destructive to the electrode’s surfaces, such as SIMS, XPS, and AES (the X-ray beam, argon ions during sputtering), and Raman (the laser beam), AFM (in contact mode), SEM (the electron beam) vs. techniques such as FTIR spectroscopy, which is not destructive.

When dealing with surface analysis of highly reactive electrode surfaces, specially designed accessories have to be used in order to probe thin surface layers. In situ measurements require the special design of spectroelectrochemical cells. The use of ex situ techniques also requires transfer systems that can convey electrodes from the electrochemical cells to the spectrometers or to the microscopes without exposure to reactive atmospheric gases (O2, N2, CO2, H2O, etc.). Of special importance is the recent use of ultrahigh vacuum systems for the study of the surface chemistry of Li electrodes.20 A methodology developed by Scherson et al. and others enabled the preparation of Li electrodes in UHV and their surface reactions with highly pure solvents.21

78

2b

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Fast Fourier Transform Infrared Spectroscopy (FTIR)

The application of FTIR in chemistry, its unique features, and the relevant instrumentation are well documented.22, 23 In brief, an FTIR spectrometer is based on a Michelson interferometer that provides a spectrum in the time domain, which is Fourier-transformed by a computer to a spectrum in the frequency domain. The sample can be scanned repeatedly, and the accumulated spectra can be averaged, thus producing a representative IR spectrum of a very high signal-to-noise ratio. This enables the measurement of samples containing a very low concentration of the active materials. FTIR is a non-destructive method that can be used for the study of surfaces. The application of FTIR spectroscopy in electrochemistry is based on the use of specific accessories for each purpose. The simplest mode of operation for the analysis of bulk liquid (thin layer cells,) or solids (pelletized with KBr powder) is the transmission mode. It is possible to analyze surface species on electrodes in this mode using grid-type electrodes. For the study of thin layers adsorbed on reflective electrodes, an external reflectance mode should be used. Of special importance is the application of grazing angle reflectance in which the incident IR beam hits the surface at an angle >80°.24, 25 Improved sensitivity is achieved by filtering the incident beam with a polarizer, which allows only P polarized light to reach the surface (or to hit the detector). Commercial accessories, which provide the appropriate beam alignment and polarization, are available (Harrick, Spectratech, etc.). An internal reflectance mode may also be useful for the study of electrode surfaces, especially in cases of nonreflective surfaces. The ATR mode is particularly important.26, 27 Both external and internal reflectance modes can be used for the in situ studies of electrodes in solutions under potential control.28, 29 For the study of powders, e.g., surface species on powdered active electrode materials, the diffuse reflectance mode is the appropriate tool.30 For intensive studies of sensitive samples by FTIR spectroscopy, it is highly recommended to place the FTIR spectrometer in a glove box. The atmosphere of the FTIR spectrometer must be cleaned of H2O and CO2, which are strong IR absorbers, and thus mask the spectra. In addition, both contaminants should be considered as reactive for nonaqueous systems, especially when active electrodes are involved (e.g., Li, Ca, Mg, Li-carbon, LixMOy, M=transition metal). The experimental aspects of ex situ FTIR spectroscopic studies of sensitive electrodes (e.g., Li and Ca surfaces) have been described in detail in Refs. 31

Identification of Surface Films on Electrodes

79

and 32. Briefly, it is possible to analyze thin surface films on active metal, using a reflectance mode, while the active surface is in contact with a KBr or NaCl polished window, as seen in Figure 2, top.

ex-situ FTIR spectroscopy - reflective mode 



























































































































































Relative lithium surface



 











































KBr window

































































































































KBr window











































Wash and dry

Reactions of lithium with solution

















KBr window pressed on the lithium

IR beam is reflected from the protected lithium surface. KBr

lithium + surface films Detector

IR beam

Lithium pressed at the window's edges forms hermetical seals.

A grazing angle reflectance attachment.

in-situ FTIR spectroscopy ATR mode (multiple internal reflectance)


I.R. beam

Single internal reflectance mode

Li C.E.

Li R.E.














































































































































































































Li C.E.

Li R.E.






















































solution


























































solution KBr Detecetor

Ge, ZnSe, Si

IR beam

WE= Thin film of Pt, Au, or Ni, or which Li is deposited electrochemically

Detector

External reflectance Mode 



Li C.E. 



Glass cell + solution

Li R.E. 







KBr window IR Beam

























































Reflactive metal W.E.

Detecor

Figure 2 Schematic view of the ex situ and in situ techniques for the study of lithium electrodes using FTIR spectroscopy. Reproduced with permission from Elsevier Science. (See [32-34].)

80

Lithium-Ion Batteries: Solid-Electrolyte Interphase

The experimental aspects of the performance of in situ FTIR measurements are described in Refs. 33 and 34. Figure 2 shows a typical layout of cells for an in situ external reflectance mode (e.g., SNIFTIRS type measurements)35 and internal reflectance modes, multiple internal reflectance, ATR, and single internal reflectance mode.33, 34 The use of the ATR mode requires crystals, which have a high refractive index (> 2). The common materials that have such a high refractive index and are transparent to the IR in the 500-4000 cm−1 range, which is the most useful optical window for the characterization of functional groups, are KRS-5, ZnSe and germanium (R.I. = 2.37, 2.4 and 4, respectively). The use of the ATR mode requires the facilities of the metal film deposition under UHV. It should be noted that the ATR crystals of these materials are very expensive. All of the above materials are reactive with nonaqueous systems at low potentials. Hence, a single experiment may be extremely expensive because the crystal surfaces may be damaged during these experiments. Thus, we developed the single internal mode that is described in detail in Refs. 33 and 34 and is presented schematically in Figure 2.

2c

Raman Spectroscopy

Raman spectroscopy provides information comparable to that obtained by FTIR. The sample is illuminated by a laser beam (visible light), and the light dispersed from the sample (Raman effect)36 is analyzed. The frequency differences between the light dispersed and that of the initial laser beam reflects the various functional groups of surface and solution species. This method can also be used in situ for the study of electrode surfaces in solutions under potential control.37 However, it should be noted that the laser beam which heats the electrode surface may be destructive to surface species. Except for unique phenomena such as SERS, in which species adsorbed to metallic surfaces (Ag, Au) provide very strong signals,38 the signal to noise ratio of Raman spectra from surface species on electrodes that are measured in situ is low. Since FTIR and Raman provide similar information, the former method is usually preferable for electrode surface studies (especially in situ). There is also a unique application of surface sensitive Raman spectroscopy in the field of intercalation processes. For instance, intercalation of Li into graphite changes the typical Raman graphite peaks.39 Figure 3 shows a scheme of a suitable cell for in situ electrode surface studies by Raman spectroscopy. The working electrode is embedded in an

Identification of Surface Films on Electrodes

81

insulating piston made of a plastic material such as Teflon or polyethylene. The optical window is made of quartz adhered to the glass cell by an epoxy-based adhesive. The laser beam that hits this surface is reflected and the light that is dispersed perpendicular to the reflected beam is analyzed. The cell operates at a thin layer configuration adjusted by the micrometer, as shown in the figure.

Figure 3 A cell for in situ electrode surface studies by Raman spectroscopy. The working electrode (1) is embedded in an insulating piston that can be moved back and forth for the measurement and the electrochemical process, (2) reference electrode, (3) counter electrode, (4) electrical contacts to the reference and counter electrodes, (5) insulating piston which holds the electrodes (made of polyproplylene, teflon, etc.), (6) glass cell, (7) teflon cell holder, (8) teflon tube for argon, (9) glass optical window, (10) teflon piston, (11) base, (12) micrometer, (13) micrometer shaft, (14) electrical contacts to the working electrode, (15) solution entry (via septum), (16) mirror, (17) focusing lens, (18) detector.

2d

Ultraviolet, Visible Light (UV-Vis)

As is widely known, bulk species which have chromophores that absorb in the UV-Vis can be analyzed quantitatively and qualitatively by this spectroscopy. The study of electrodes or species adsorbed as thin layers or electrodes by UVVis is more difficult, due to sensitivity problems and the availability of the appropriate chromophores.39 Another use of this analysis is the so-called electroreflectance.40 Adsorption of species on reflective electrode surfaces changes their reflectivity. Thus, this method can indicate electroadsorption processes very sensitively in situ, although it does not provide specific information on the structure and composition of surface layers.

82

2e

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Extended X-ray Absorption Fine Structure (EXAFS), X-ray Absorption Near-Edge Structure (XANES)

These Extended X-ray Absorption Fine Structure (EXAFS), and X-ray Absorption Near-Edge Structure (XANES) methods provide unique information on the composition of surface species and their structure. They can be used as an in situ tool. However, EXAFS and XANES require a synchrotron radiation source (X-ray).41

2f

X-ray Photoelectron Spectroscopy (XPS) and Auger Electron Spectroscopy (AES)

The XPS technique is based on the analysis of the energy of electrons emitted due to irradiation of surfaces by an X-ray beam. This energy reflects very specifically the elements present on the surface, as well as their oxidation states. This method requires a vacuum system that provides a background vacuum of 10−9-10−10 mmHg. This is an ex situ technique, and its application for the study of sensitive electrodes requires special transfer arrangements. Any modern XPS system includes the option of depth profiling of the surface studied by sputtering the surface with argon ions, followed by XPS analysis. The information thus obtained is highly specific, both qualitatively and quantitatively, and a completed comprehensive element analysis is provided.42 However, it should be noted that this method might be destructive to surfaces. The sputtering, as well as the X-ray beam, may change the oxidation states of elements and induce surface reactions. Auger electron spectroscopy is somewhat similar to XPS in providing surface element analysis, and involves the analysis of Auger electrons emitted from surfaces due to irradiation with an X-ray beam. It is very useful for quantitative analysis of elements on the surface.43

2g

Energy Dispersive Analysis of X-rays (EDAX)

The EDAX technique involves an analysis of the X-ray radiation emitted from surfaces which are studied by scanning electron microscopy (SEM).44 The surface studied by SEM is hit by the electron beam, emitting X-rays of a limited penetration depth which are specific to the elements present on the surface. This method provides qualitative and quantitative element analysis of electrode

Identification of Surface Films on Electrodes

83

surfaces. Figure 4 shows a scheme of a transfer system of electrodes from a glove box atmosphere (highly pure argon) to the high vacuum chamber of an electron microscope.

Figure 4 Scheme of a transfer system for air sensitive samples from a glove box to a SEM system: (1) SEM inlet, (2) system body, (3) O rings, (4) fixed tray, (5) brass disk that seals the samples on the tray, (6) brass shaft, (7) brass cylinder with two rubber O rings, (8) bridge attached to the edge of the shaft, (9) bolt by which the bridge is pressed down, (10) two bolts by which the bridge is raised up to release the tray when evacuated, (11) sample tray and its O rings, (12), (13) manipulator, (14) cover through which the manipulator is moved with two O rings. Reprinted with copyright from The Electrochemical Society Inc.

2h

Secondary Ion Mass Spectrometry (SIMS)

Secondary ion mass spectroscopy is based on surface bombardment by argon ions in UHV, followed by mass spectrometry of the charged species which are sputtered from the sample's surface. It provides specific information on surface species, high spatial resolution, and depth profiling.45 2i

Electrochemical Quartz Crystal Microbalance (EQCM)

This EQCM method is based on the piazoelectric effect of thin quartz crystals (5-10 µm thick). Two electrodes are deposited on two sides of the quartz crystal

84

Lithium-Ion Batteries: Solid-Electrolyte Interphase

and the resonance frequency of the crystal under an alternating electric field is measured. This depends linearly on the mass accumulating on any of the electrodes which are used as the working electrode in the electrochemical system studied (possible resolution of nanograms per cm2).46 By recording the mass and the charge in an electrochemical process in which adsorption and/or precipitation of species occur, it is possible to estimate the equivalent weight of surface species formed at different experimental conditions, e.g. potential, concentration, temperature. Hence, this in situ method can serve as an attractive electroanalytical tool for the in situ study of adsorption processes. Its use for nonaqueous systems requires the development of special cells, as described in Figure 5.

a.

b.

Figure 5 Cells for EQCM measurements: (1) quartz crystal, (2) gold electrodes deposited on both sides of the quartz crystal, (3) counter electrode, (4) reference electrode, (5) solution, (6) polyethylene body, (7) glass cell parts, (8) O rings, (9) electrical contacts for the working electrode, (10) glass tube. Reprinted with copyright from The Electrochemical Society Inc.

Identification of Surface Films on Electrodes

2j

85

X-ray Diffractometry (XRD)

X-ray diffractometry is widely used for the characterization of electrode materials for the battery field, electrocatalysis, etc. Both areas require the development of new materials whose three-dimensional structure is critical for their electrochemical activity. In brief, XRD is based on a monochromatic X-ray beam that hits the sample and is reflected from it at a variety of scattering angles. Since the X-rays are reflected by the atoms in the sample's lattice, and since the wave length is of the same order of magnitude as interatom distances in the solid state, interference among the reflected X-rays occurs, leading to typical, unique diffraction patterns for each specific material.47 A completed analysis of lattice structures can be obtained from judicious treatment of the data in the XRD patterns. It should be noted that XRD can also be applied as an in situ technique. It requires the use of specific cells with windows which do not absorb the X-ray beam. For instance, polyethylene and Mylar films are suitable. A typical cell for in situ XRD measurements of composite electrodes, e.g., lithiated graphite, is presented in Figure 6. There are already reports on the use of in situ XRD measurements for the study of composite electrodes in nonaqueous systems48-50, and the study of surface layers on electrodes (e.g., a lithium electrode in an aprotic medium).51

Figure 6 A cell for in situ XRD measurements, isometric and section views. Reprinted with copyright from The Electrochemical Society Inc.

86

2k

Lithium-Ion Batteries: Solid-Electrolyte Interphase

NMR, ESR Spectroscopy

There are reports on the use of both NMR and ESR for the study of electrode materials52, 53 and bulk products of electrochemical processes.54 For instance, 7Li NMR may be found to be very useful for the study of Li intercalation into carbonaceous materials55 and transition metal oxides.56 A major advantage of these techniques is that they are applied in situ. The electrochemical cell is, in fact, an NMR tube in which the studied electrode is mounted so that it can be placed within the magnet's cavity. While NMR provides information on the environment of the element studied, within the electrode measured, ESR provides information on the formation and stability of radical ions when formed during the course of an electrochemical process. 2l

Scanning Probe Microscopies (AFM, STM)

The relatively novel method of atomic force microscopy (AFM) can be used both ex situ and in situ for the study of surface morphology of electrodes. It is based on a thin and sensitive cantilever to which a sharp microscopic tip is attached. This tip is raster-scanned along the studied surface, changing the deflection of the cantilever as a result of the topography of the surface. The deflection is measured by a laser beam, which is reflected from the back of the cantilever to a detector that measures the position of the laser beam. The changes in the cantilever, as a function of the tip position with respect to the sample plane, are translated by sophisticated software into a 3D picture of the surface topography.57 The application of this technique for the study of electrodes in nonaqueous systems which are highly sensitive to atmospheric contaminants, and which may be volatile, is difficult and requires the design of a special cell and transfer method. It should be noted that there are several variations in the application of AFM in electrochemistry. These include a non-contact mode in which the tip is not in direct contact with the surfaces,58 friction forces between the tip and the surface species (lateral forces),59 and magnetic force microscopy in which a magnetic tip senses magnetic surface species.60 The major advantage of this technique is its possible application as an in situ tool for electrode surface morphology measurements. Its disadvantages are the possibility that the tip will interfere with the original surface morphology, and the experimental difficulties in applying it to sensitive and reactive systems.

Identification of Surface Films on Electrodes

87

Scanning tunneling microscopy (STM) is also a tool for surface morphological studies, which is widely used in situ.61 It is based on the analysis of a tunneling current between a very sharp microscopic tip and the electrode surface caused by a bias potential applied between the two. This method is well established for the study of electrochemical systems.62, 63 Its advantage over AFM is that it is technically much simpler to use for in situ studies of electrochemical studies, and it obtains better resolution. However, the application of STM to nonaqueous systems is impossible when the electrode surfaces are covered by surface species, which are electrically insulating. In order to perform prolonged in situ AFM and STM studies of very sensitive electrodes and solutions, we built special glove boxes in which the AFM/STM systems were placed for measurements under highly pure argon. The glove box can be fully evacuated so that its atmosphere is replaced before each set of measurements. To prevent vibration, the glove boxes are hung on springs and are provided with accessories and connections which enable their disconnection from the feeding pipe when measurements are being taken. The systems that we developed, including electrochemical cells, glove boxes, accessories, are described in Ref. 64. 2m The Use of UHV Systems for Identification of Surface Films Formed on Lithium Especially elegant is the possibility of preparing a highly clean Li surface in ultra high vacuum, and then to react it with atmospheric gases and with liquid layers of solvents of interest, which are condensed on the clean Li surface at a low temperature from the gas phase.21 There are interesting reports on the study of the reaction products of Li with solvents of interest in UHV systems, using FTIR spectroscopy, XPS, AES, TPD, and mass spectroscopy.65 Some interesting findings will be reported on later in the ensuing sections. 3 3a

The General Structure of Surface Films on Reactive Surfaces Introduction

There are several important aspects that have to be dealt with when describing the general structure of surface films on reactive electrodes: 1.

The initial state of the electrodes (e.g., coverage by “native” films, surface groups, etc.).

88

2. 3. 4. 5. 6.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Surface film formation on freshly prepared active metal surfaces vs. the situation where the electrodes are covered initially by ‘native’ films. Aging of surface films, i.e., secondary reactions between surface films and solution species. The effect of trace water. Spontaneous surface reactions vs. stimulated surface reactions (by polarization). The impact of the process of the electrode’s polarization on its surface chemistry.

Each of the above points has a strong impact on the electrode’s surface chemistry and on the structure and properties of surface films on electrodes. 3b

Surface Film Formation on Active Metals

Active metals (Li, Mg, Ca, etc.) react spontaneously with the main atmospheric gases (N2, O2, H2O, CO2) and with most relevant polar aprotic solvents and salt anions. All active metals are covered initially by native surface films formed during their production by their reaction with atmospheric gases. It should be noted that even a usual glove box atmosphere that officially contains less than 1 ppm of H2O and O2 (but may contain hundreds of ppm of CO2 and N2) should be considered as reactive towards lithium or magnesium surfaces prepared freshly in the glove box. Active metals are usually covered by bilayer surface films. The inner layer is comprised of metal oxide, while the outer layer contains mostly carbonates and hydroxides.66 When an active metal is introduced into a polar aprotic electrolyte solution, several processes take place in parallel. These include dissolution of part of the initial surface species, nucleophilic reactions between metal oxide and hydroxide and electrophilic solvents such as esters and alkyl carbonates, and diffusion of solvent molecules towards the active metalnative film interference and their reduction by the active metal. Water molecules that are unavoidably present in solutions solvate most of the relevant surface species, such as metal oxides, hydroxides, carbonates, halides, and organic salts. Hence, water diffuses through the surface films, partially solvates surface species, and reacts within the surface films with the active metal to form MHx, M(OH)x, MOx, etc.66 This scenario is illustrated in Figure 7. The result of these reactions is the formation of highly complicated, non-uniform multilayer surface films. When a fresh active metal surface is exposed to a polar aprotic solution, the following processes take place. Initially,

,GHQWLILFDWLRQRI6XUIDFH)LOPVRQ(OHFWURGHV



the active metal reduces solution species at low selectivity. Solvent molecules whose concentration is the highest, are the first to be reduced. The surface chemistry may be very complicated. Electrons are injected into the solution, thus forming radical anions. These unstable species further react, rearrange, and may accept more electrons, thus forming dianions. 6XUIDFHVSHFLHVZLWK+LJKR[LGDWLRQVWDWH ORZR[LGDWLRQVWDWH VXUIDFHVSHFLHV FRSPDFWSDUW SRURXVSDUW

/L2 /L&2 /L2+ 



















































































































































































 



















 























































































































































OLWKLXP

VROXWLRQUHDFWLRQV GLVVROXWLRQ UHSODFHPHQWV IXUWKHUUHDFWLRQV ZDWHUGLIIXVLRQ ZDWHUUHGXFWLRQ







































ZDWHU ZDWHUUHGXFWLRQ SURGXFWV









OLWKLXP 



































































 



















 





















 























 























 























/L2/L)/L&O /L+/L1 

















52/L52&2/L

6(, )LJXUH   $Q LOOXVWUDWLRQ RI WKH VXUIDFH ILOPV RQ OLWKLXP DV WKH DFWLYH PHWDO FRYHUHG E\ D SULVWLQH VXUIDFHOD\HULVH[SRVHGWRDQRQDTXHRXVHOHFWURO\WHVROXWLRQ



There are follow-up reactions that may include nucleophilic attacks of the anions on the electrophilic solvent molecules, polymerization, and formation of insoluble salts by ionic reactions of the anions with the metal cations. Insoluble salts and other products (e.g., polymeric species) precipitate on the electrode and form an initial surface layer. Further reduction of solution species requires electron transport through the surface layer, and thereby, the follow-up reduction processes are much more selective than the initial ones. Therefore, the composition of the surface layer changes from the metal-film interface to the film-solution interface, i.e., it comprises multilayer surface films. Further



90

Lithium-Ion Batteries: Solid-Electrolyte Interphase

reduction processes take place close to the metal surface. The first monolayers may contain species of the lowest oxidation state-metal oxides, metal halides, and metal carabides. The upper layers contain species of a higher oxidation state such as carbonates, organic salts, e.g., ROLi, ROCO2Li, RCOOLi, and polymeric species. As the surface film becomes sufficiently thick, it blocks further electron transfer, and the system may approach steady state. In any event, the film-solution interface is very dynamic. Some electron tunneling may take place at certain locations of local, high electrical conductivity, and hence, small-scale reduction of solution species continues. There is also dissolution– precipitation of solution species and secondary reactions between the surface species and solution species (e.g., reaction of water with ROCO2Li, reaction of basic Li salts with acidic species). Thus, the solution side of the surface films is expected to be porous, while the inner part, close to the metal, is compact. The result of the above-described surface film formation processes is that surface films on active metals are very non-uniform. They comprise several layers with a mosaic-type lateral structure containing grains of different surface species. 3c

Surface Film Formation on Non-reactive Metal and Carbon Electrodes

In contrast to active metals that provide a continuous driving force for electrode-solution reactions, non-active metals such as noble metals (Au, Pt), nickel, silver and copper, as well as carbons, may be inert in polar aprotic electrolyte solutions at open circuit potentials (usually around 3.V vs. Li/Li+). It should be noted that both non-reactive metals may be covered, at least in part, by oxides, while carbons contain oxygen-based surface groups (e.g., OH, COOH, C=O). Surface films are formed on these electrodes by cathodic polarization in nonaqueous solutions with salts of active metal (e.g., Li). In potentiostatic polarization in which the potential is dropped from OCV to low potentials close to that of the active metal deposition (e.g., 0. V vs. Li/Li+), the dynamics of the surface film formation may be quite similar to those described in the previous section. Upon a gradual polarization (e.g., linear potential scanning, or upon galvanostatic processes), the scenario is different. Solution species are selectively reduced at different potentials according to their reactivity. This is illustrated in Figure 8, which shows schematically the surface processes occurring on gold electrodes polarized cathodically in Li, Na, and K salt solutions (perchlorate salt in propylene carbonate).67 This figure also

Identification of Surface Films on Electrodes 



a.













b. 

([SHFWHG PSH 























([SHFWHG PSH 

$QRGLF 3URFHVVHV 

































































/L%XON 'LVVROXWLRQ



'HFRPSRVLWLRQ RI/L$X$OOR\



1D%XON 'LVVROXWLRQ



3&2[LGDWLRQ











































)XUWKHU/L2 [ 5HGXFWLRQWR/L2 

6WULSSLQJ RI/L83' 







 

























/L 'HSRVLWLRQ 










 









/L 83' 






























































































































$X +\GUR[LGH 5HGXFWLRQ 
























1D 'HSRVLWLRQ

















/L$X$OOR\ )RUPDWLRQ







&DWKRGLF 3URFHVVHV 






















([SHFWHGPSH











  YV/L/L









 YV$J5(





























1D 83' " 



















































































































































2[\JHQ 5HGXFWLRQ 







:DWHU 5HGXFWLRQ 3& &O2 5HGXFWLRQ




3& &O2 5HGXFWLRQ 


























2[\JHQ :DWHU 5HGXFWLRQ 5HGXFWLRQ 
















)XUWKHU1D2 [ 6WULSSLQJ 5HGXFWLRQWR1D2  RI1D 83'"




$X +\GUR[LGH )RUPDWLRQ





)XUWKHU 1D2+ 5HGXFWLRQ WR1D2 































Current







3&2[LGDWLRQ







$QRGLF 3URFHVVHV

'HFRPSRVLWLRQRI 1D$X$OOR\







)XUWKHU/L2+ 5HGXFWLRQWR /L2  Current

91

&DWKRGLF 3URFHVVHV

1D$X$OOR\ )RUPDWLRQ 





([SHFWHGPSH









  YV/L/L









 YV$J5(

3RWHQWLDO

3RWHQWLDO

c. ([SHFWHG PSH











$QRGLF 3URFHVVHV

3&2[LGDWLRQ 











)XUWKHU.2 [ 5HGXFWLRQWR.2 

Current







































































 













.'HSRVLWLRQ DQG.$XDOOR\ IRUPDWLRQ 







5HDFWLRQV RIWKH. 'HSRVLWV ZLWK 6ROXWLRQ 6SHFLHV 

































































































































































































































































2[\JHQ 5HGXFWLRQ 





























































&DWKRGLF

5HGXFWLRQRI 3URFHVVHV :DWHU3& &O2  

































([SHFWHGPSH





  YV/L/L





 YV$J5(

3RWHQWLDO

Figure 8 A schematic presentation of the various electrochemical processes of gold electrodes in (a) PC/LiClO4, (b) PC/NaClO4, and (c) PC/KClO4 solutions. The expected m.p.e. values of the various surface processes are also presented (equal to the equivalent weight of the expected surface species formed). Reprinted with copyright from The Electrochemical Society Inc. (See [67].)





/LWKLXP,RQ%DWWHULHV6ROLG(OHFWURO\WH,QWHUSKDVH

demonstrates how strongly the surface processes may depend on the cation involved. In such gradual processes of surface film formation in polar aprotic solutions, trace oxygen is reduced at around 2 V (vs. Li/Li+), trace water is reduced at around 1.5 V (Li/Li+), and solvent molecules and salt anions are gradually reduced at a potential below 1.5 V (Li/Li+). When noble metals such as Au, Pt or Ag are involved, active metal under potential deposition (UPD) and stripping may take place.68 (See also Figure 8.) Figure 9 shows a typical FKURQRSRWHQWLRJUDP of a graphite electrode during a first cathodic polarization from OCV (≈3 V vs. Li/Li+) to 0. V (Li/Li+). The plateau around 1 V and the gradual decrease in potential down to 0.3 V (the onset potential for Li intercalation into graphite that forms LiC36, i.e., stage 4) reflects the gradual reduction of solution species: reactive atmospheric contaminants, solvent molecules, and salt anions.

 )LJXUH$W\SLFDOSRWHQWLDOFDSDFLW\FXUYHLQDILUVWDQGDIHZVXEVHTXHQWF\FOHVYROWDJHYVWLPH WUDQVODWHGWRFDSDFLW\LQP$KJU RIDFRPSRVLWHJUDSKLWHHOHFWURGHFRPSULVHGRIV\QWKHWLFIODNHVLQ DQ(&'0&/L$V)VROXWLRQF\FOHGJDOYDQRVWDWLFDOO\DW&KYVDOLWKLXPFRXQWHUHOHFWURGH6HH >@ 



The above-described gradual surface reaction processes also form multilayer surface films, as is illustrated schematically in Figure 7. As the electrode reaches the very low potentials, and/or fully lithiated carbon is formed, the surface layer close to the electrode can be further reduced to form species of very low oxidation states (Li2O, LiF, Li-C, LiH, Li3N, etc.). Hence, we can

Identification of Surface Films on Electrodes

93

conclude that the structure of the surface films on an active metal such as lithium, a non-active metal polarized to low potentials, and lithiated carbons in the same Li salt solutions, is very similar: a compact part near the electrode which has a multilayer structure and a porous part in the solution side. Laterally, the surface films are non-uniform and are comprised of grains of different surface species. (Their identity is described in the ensuing sections.) 3d

On Transport Properties of Surface Films

The surface films discussed in this section reach a steady state when they are thick enough to stop electron transport. Hence, as the surface films become electrically insulating, the active electrodes reach passivation. In the case of monovalent ions such as lithium, the surface films formed in Li salt solutions (or on Li metal) can conduct Li-ions, and hence, behave in general as a solid electrolyte interphase (the SEI model 5). See the basic equations 1-7 related to ion transport through surface films in section 1a above. The potentiodynamics of SEI electrodes such as Li or Li-C may be characterized by a Tafel-like behavior at a high electrical field and by an Ohmic behavior at the low electrical field. The non-uniform structure of the surface films leads to a non-uniform current distribution, and thereby, Li dissolution from Li electrodes may be characterized by cracks, and Li deposition may be dendritic. The morphology of these processes, directed by the surface films, is dealt with later in this chapter. When bivalent active metals are involved, their surface films cannot conduct the bivalent ions. Thereby, Mg or Ca deposition is impossible in most of the commonly used polar aprotic electrolyte solutions. Mg or Ca dissolution occurs at very high over potentials in which the surface films are broken. Hence, dissolution of multivalent active metals occurs via a breakdown and repair of the surface films.

4 4a

Impedance Spectroscopy of Electrodes Covered by Surface Films Introductory Remarks

Charge transfer processes with electrodes covered by surface films are usually complicated and involve several stages in series. Hence, the electrochemical processes of electrodes covered by surface films always reflect several time constants. The processes can be studied by a variety of transient methods such

94

Lithium-Ion Batteries: Solid-Electrolyte Interphase

as chronopotentiometry or chronoamperometry with short input pulses. There are reports on studies of Li electrodes by micropolarization techniques (current pulses).69 However, one of the most appropriate techniques for the electrochemical studies of electrodes covered by surface films is impedance spectroscopy. This technique allows a straightforward analysis of the time constants related to the various stages in the electrochemical response of the electrodes, and therefore deserves a special section in this chapter. It should be emphasized that electrodes measured by impedance spectroscopy have to be in equilibrium in order to obtain meaningful results. Hence, the electrodes discussed in this chapter should be measured only after the surface films are fully developed, at steady state. The time scale of their changes due to aging and secondary reactions should be at least one order of magnitude longer than the measuring duration.

4b 4b.1

Active Metal Electrodes Lithium

Figure 10 shows a typical impedance spectrum of a lithium electrode (propylene carbonate/LiAsF6) in both Nyquist and Bode presentations. The Nyquist plot is characterized by a large, flat, high frequency semicircle and low frequency features, which may appear as a small semicircle and a 'Warburg'-type element. As already discussed in detail,70 the high frequency semicircle relates to Li-ion migration through the multilayer, compact part of the surface films (close to the Li side), and the corresponding films’ capacitance. The low frequency features relate to the Li-ion transport through the porous part (solution side) of the surface films, including finite diffusion of Li-ions in solution (in the pores). It should be emphasized that the spectrum in Figure 10 is very typical of Li electrodes in a large variety of polar aprotic solutions (different solvents and Li salts). A very logical equivalent circuit analog that can be fitted very well to impedance spectra of Li electrodes is a ‘Voight’-type analog71 containing 5 R||C circuits in series.72 A typical fit is also presented in Figure 10. Such an analog relates directly to the fact that the major electrochemical process of Li electrodes is Li-ion transport through a multilayer surface film. The resistance of the surface film to Li-ion migration is dominant, and hence, the charge transfer resistance across the film-metal, film-solution interfaces is negligible. All the film resistances are coupled with film capacitance. It is significant that the capacitance calculated for the low frequency time constants is several orders of

Identification of Surface Films on Electrodes

95

magnitude higher than that calculated for the high frequency time constants. This correlates with the low capacitances expected for compact surface layers and the high capacitance expected for the porous part of the surface films. Since the surface species were identified and their dielectric constant is known (usually around 5), it was possible to estimate the thickness of the compact part of the surface films on Li electrodes (several nanometers on average) and their resistivity (10−8-10−9 Ω ⋅ cm2) on average.70

Figure 10 Typical modeling of experimental results (Li electrode in PC-1.5 M LiAsF6 solution after 24 h) by equivalent circuit of five RC circuits in series. Both the Nyquist and the Bode plots are shown . Dashed line, experimental results. Solid line, calculated response. Reprinted with copyright from The Electrochemical Society Inc. (See [72].)

4b.2

Mg Electrodes

Figure 11 shows typical impedance spectra (presented as Nyquist plots) obtained from Mg electrodes in solutions in which they are covered by surface films (organic and inorganic Mg salts). These spectra reflect blocked electrodes. The surface films comprising Mg salts cannot conduct the bivalent ions.10 In ethereal solutions containing RMgX (Grignard salts), Mg(BR4)2, or Mg(AlCl4-nRn)2 complexes, magnesium electrodes are not covered by stable surface films and behave reversibly (i.e., magnesium can be deposited-dissolved electrochemically at relatively low overpotentials).73 Figure 12 shows families of impedance spectra (Nyquist plots) of two Mg electrodes in Grignard salt/THF solutions, one of which was initially covered by native films (MgO-MgCO3), and the other was prepared freshly in solution. The electrode covered by the native surface films has an initially high impedance, which decreases upon storage, while the impedance of the freshly prepared

96

Lithium-Ion Batteries: Solid-Electrolyte Interphase

electrode increases upon storage. Both electrodes reach a similar steady state impedance that relates to complicated adsorption phenomena which do not interfere badly with Mg deposition/dissolution. Note that the native surface films on Mg dissolve in these solutions. This behavior is typical of all ether solutions (THF, polyethers from the “glyme” family) with the above-mentioned complexes.74

Figure 11 Impedance spectra (Nyquist plots) of Mg electrodes freshly prepared in PC + 1 M LiClO4, PC + 1 M LiBF4, THF + 1 M LiClO4 and in THF + 1 M LiBF4 solutions, as indicated. The spectra were measured after 3 h of storage at ocv. Some frequencies are indicated. The high-to-medium frequency spectra are emphasized in the insert. Reprinted with copyright from Elsevier Science. (See [10].)

Figure 12 Impedance spectra of Mg metal electrodes as a function of storage time (indicated) in BuMgCl 1 M/TBAPF6 0.5 M/THF solution. (a) Pristine electrode, covered by native films (MgO, MgCO3, Mg(OH)2, etc.). (b) Electrode surface was freshly prepared in solution. Reprinted with copyright from The Electrochemical Society Inc. (See [73].)

Identification of Surface Films on Electrodes

4c

97

Non-active Metal Electrodes Polarized to Low Potentials

Figure 13 shows a typical Nyquist plot obtained with a nickel electrode polarized to low potentials in a nonaqueous Li salt solution (LiBF4.PC), the relevant equivalent circuit analog and the related simulated spectrum. Scheme 1 explains the equivalent analog and its relevance to the structure of the surface films. These electrodes, when polarized cathodically in the Li salt solutions, develop surface films very similar in their composition and structure to those formed on Li metal. Thus, their high frequency impedance response relates to Li-ion transport through the surface films. The low frequency, which appears as a large arc, reflects the high charge transfer resistance of these electrodes. In contrast to Li electrodes whose charge transfer resistance related to the Li/Li+ couple is very low, the dominant charge transfer resistance of non-active electrodes at low potentials in these solutions relates to reduction of solution species, and thereby, it is very high.75 As already discussed in detail, lithiated graphite electrodes are covered by surface films similar in composition and structure to those formed on lithium or non-active metal electrodes polarized to low potentials in the same solutions.76-78 Therefore, the high frequency impedance is similar for the three types of electrodes and related to Li-ion transport through multilayer surface films, and can be fitted by the ‘Voigt’-type analog (several R||C circuits in series). Li-carbon electrodes have their unique low frequency features that belong to a potential dependent charge transfer across the interface between the surface films and the carbon, to the solid state diffusion of Li-ions in the carbon, and finally, at the very low frequency, to the capacitive behavior of the electrode-accumulation of charge by the intercalation process (see Figure 14).79, 80 It was very interesting to discover that other Li insertion electrodes, such as LiCoO2, LiNiO2, or LiMn2O4, also have a very similar impedance behavior to that of lithiated carbon electrodes.81 It was found that the above cathode materials also interact with solution species and develop surface films that behave according to the SEI model.5 Their impedance also reflects a serial charge transfer process that includes Li-ion transport through surface films, Li-ion transfer across film-active mass interface, Li-ion diffusion (solid state), and, finally, accumulation of Li in the host material (low frequency-capacitive behavior).

98

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Figure 13 A typical Nyquist plot obtained from a Ni electrode polarized to low potentials (0.2 V vs. Li/Li+) in PC solutions (1 M LiBF4 in this case). The equivalent-circuit analog of 4 R| |C circuits in series and their separated Nyquist plots (four semicircles) are also shown. The frame in the lower right represents a typical fitting between the experimental data and this equivalent circuit analog. Reprinted with copyright from The Electrochemical Society Inc. (See [75].)

Scheme 1 A schematic illustration of layered interphase on active metal and equivalent-circuit analog of 4 RC circuits in series. Reprinted with copyright from The Electrochemical Society Inc.

Identification of Surface Films on Electrodes

99

Figure 14 shows typical impedance spectra (Nyquist plots) of a lithiated graphite electrode (LiC12, stage II, 0.11 V vs. Li/Li+), and partially delithiated LixCoO2 (4.V vs. Li/Li+). A common equivalent circuit analog that simulates these spectra and reflects the serial nature of the Li insertion processes into both Li-C and LiMO2 electrodes is also presented in Figure 14. The features in the spectra which relate to the various elements in the model are marked. An important difference between the impedance of Li-C and LiMOx electrodes is the fact that in the latter electrodes, the charge transfer resistance (surface filmactive mass interface) is very pronounced (note the medium frequency semicircle in the spectrum of the LiCoO2 electrode in Figure 14), and its potential dependence is also strong. At some potential ranges (depending on the electrodes) it may become the dominant electrode’s impedance.81

Figure 14 Typical impedance spectra measured from LiNiO2, LiCoO2, LiMn2O4, and lithiated graphite electrodes in EC-DMC/LiAsF6 solutions (Li as R.E. and C.E. electrodes). The potential of the measurements is indicated near each spectrum. A model that provides an excellent fit with these spectra is also presented. The assignment of its various elements to features of the experimental spectra is also shown. Reproduced with permission from Elsevier Science. (See [95].)

100

5

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Identification of Surface Films Formed on Lithium and Non-active Electrode Polarized to Low Potentials in Li Salt Solutions

5a The Preparation of a Library of FTIR Spectra Surface-sensitive FTIR spectroscopy was used as a major tool for identification of surface species formed on Li, noble metal, and carbon electrodes, because it provides specific information about functional groups and types of chemical bonds. It is a surface sensitive technique with the appropriate accessories, can usually be applied in situ to electrodes in solutions under potential control, and is non-destructive. A first step in the use of this technique was the preparation of a library of FTIR spectra of possible surface species formed on Li electrodes in solutions. These include LiOH, Li2CO3, ROLi, RCOOLi, ROCO2Li, polycarbonates, salts, and solvent spectra. It was of special importance to obtain reference spectra from species prepared as thin films on lithium.82 Figure 15 presents a number of FTIR spectra of several important reference compounds from the above list. The reference spectra of the commonly used salts, LiClO4, LiAsF6, LiBF4, LiC(SO2CF3)3, LiN(SO2CF3)2, and LiSO3CF3 (Figure 15a),83-85 were obtained from their KBr pellets (transmittance mode). Reference spectra of Li alkoxides (Figures 15b,c) were obtained by reacting Li surfaces with vapor of the parent alcohols.82 The spectra were measured by reflectance mode, while the Li surfaces were protected by a KBr window (Figure 2). Reference ROCO2Li spectra could be obtained by reacting the thin ROLi films (on Li) with CO2,85 followed by FTIR spectroscopy (reflectance mode). Reference CH3CH(OCO2Li)CH2OCO2Li and (CH2OCO2Li)2 were produced by electrolysis of PC and EC, respectively, in(C4Hg4) NClO4 ethereal solutions (e.g., THF), followed by precipitation of the ROCO2Li species (addition of a Li salt to the electrolyzed solutions).14, 86 FTIR spectra of these species were measured from their KBr pellets (Figure 15d, transmittance mode). Reacting the ROCO2Li species with water formed a mixture of ROH and Li2CO3.87 FTIR spectra of such a mixture (pelletized with KBr) are also presented in Figure 15d and demonstrate the major IR bands of Li2CO3, which is an important surface species (IR peaks at 1504, 1430, and 870 cm−1).14

Identification of Surface Films on Electrodes

a

c

Salt spectra

(1) Li 1pentoxide

b

d

alkoxides

(a) ROCO22Li

(b) ROCO22Li + H 2 O

pentoxide (2) Li 2pentoxide

(c) (CH22OCO22Li)22

101

102

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Figure 15 (a) FTIR spectra of LiClO4, LiSO3CF3, LiN(SO2CF3)2, LiAsF6, LiC(SO2CF3)3, LiN(SO3CF2)2 (pelletized with KBr, transmitance mode), LiBF4 (as a thin layer on a nickel mirror, reflectance mode), as indicated. (See Refs. 82-85.) (b) FTIR spectra obtained from lithium surfaces treated with DN solutions of four different alcohols (0.01 M): (a) Ethylene glycol solution; (b) methanol solution; (c) ethanol soution; (d) methoxyethanol solution. All the spectra are typical of the corresponding Li alkoxides. (See Ref. 82.) (c) (1) FTIR spectra of Li 1 pentoxide: (a) Pelletized with KBr (transmittance mode); (b) deposited on the lithium surface (reflectance mode); (c) deposited on the lithium surface and derivatized with CO2 (reflectance mode). (2) FTIR spectra of Li 2 pentoxide deposited on lithium surfaces. (a) Mostly alkoxide spectrum; (b) the sample of (a) was derivatized with CO2. (See Ref. 85.) (d) (a) FTIR spectrum obtained from KBr palletized major product of PC electrolysis precipitated as lithium salt. (b) FTIR spectrum obtained from KBr pelletized major product of water contaminated PC electrolysis precipitated as lithium salt. (c) FTIR spectrum of the electrolysis product of EC in THF + 0.5 M TBAP on gold, isolated as Li salt (pelletized with KBr). (See Refs. 14, 86.) Reprinted with copyright from The Electrochemical Society Inc. and from Elsevier Science.

5b Identification of Surface Films Formed on Li Electrodes in Ether Solutions From several experiments in which freshly prepared (in solution) Li electrodes were stored in pure ether (no Li salt), it was clear that the ether linkage is attacked and broken by Li metal.88 The products are ROLi species.85, 88, 89 In Figure 16 we demonstrate a comparison between FTIR spectra measured from Li surfaces stored in ethyl glyme and diglyme (ex situ, external reflectance mode) to library spectra of CH3CH2OLi and CH3OCH2CH2OLi (on Li metal).89 It is clear from this comparison that Li reacts with ethyl glyme to form surface CH3CH2OLi and with diglyme to form surface CH3OCH2CH2OLi (in addition to other Li alkoxides). Figure 17 presents another example: a Li surface was in contact with a thin layer of 1-3 dioxolane (DN), was sealed with a KBr window (see Figure 2), and measured periodically by FTIR spectroscopy (internal reflectance mode).82 The DN IR peaks (spectrum a) disappeared upon storage, and the spectra are characterized by typical ROLi peaks (spectrum b). Upon storage, air diffuses to the Li surface and CO2 reacts with the ROLi to form ROCO2Li,85 and ROCO2Li and trace water react also to form Li2CO387 (spectrum 17d), ROCO2Li peaks around 1630 cm−1, 1300 cm−1, and 822 cm−1, and Li peaks at 2500 cm−1 and 879 cm−1 (compare with the reference spectra in Figure 15).

Identification of Surface Films on Electrodes

103

7 Figure 16 FTIR spectra of lithium electrode prepared and stored for several weeks in diethylglyme (CH3CH2-OCH2CH2-O-CH2CH3) and diglyme (CH3OCH2CH2)2O. Spectra of CH3CH2OLi and CH3OCH2CH2OLi (as thin films on lithium) are also shown for comparison. Reprinted with copyright from Elsevier Science. (See [89].)

Figure 17 Spectra obtained from lithium surfaces covered with a thin layer of pure DN sealed and protected with KBr plates. The samples were measured under dry air atmosphere. (a) The spectrum was measured 0.5 h after sample preparation (mostly a DN spectrum). (b) Same sample as (a), 2 h after preparation. The seals were slightly loosened 10 min prior to the measurement. (c) Same as (b); the spectrum was measured several hours after sample preparation. (d) same as (c), a day after sample preparation. Reprinted with copyright from Elsevier Science. (See [82].

104

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Figure 18 shows FTIR spectra measured in situ from Pt/Au electrodes polarized to 0. V (Li/Li+) in THF solutions with LiAsF6, LiN(SO2CF3)2 and LiC(SO2CF3)3 salts (single internal reflectance mode, see Figure 2).84 The AsF6− is the least reactive anion, and thereby spectrum a reflects mostly the formation of ROLi films (e.g., Li butoxide) formed by THF reduction.84 The other two spectra (18b,c), which are much richer in peaks, reflect reduction processes of the reactive salts that dominate the surface chemistry when the solvent is not too reactive (ether).84 A number of experiments similar to those related to Figures 16-18 brought us to the conclusions summarized in Scheme 2, which shows general routes of ether reduction on lithium surfaces. In contrast to the above conclusions, there are reports on experiments in which ethers were condensed on evaporated Li layers, in UHV, which did not show that Li reacts with ethers to form Li alkoxides.21, 65 These differences may result from the different experimental conditions involved. However, we believe that the formulae in Scheme 2 reflect the reactions of ethers with Li surfaces at ambient conditions and inert atmosphere.

Figure 18 FTIR spectra measured in situ from Pt deposited on NaCl (SIRFTIR mode), polarized to 0 V (Li/Li+) in THF 1M solutions of the three salts, as indicated. Reprinted with copyright from Elsevier Science. (See [84].)

Identification of Surface Films on Electrodes

105

Scheme 2: Ether reaction patterns _ R’OR Li +

(a) a). R-O-R’+e- +Li+ _

(b) b). R’OR Li +

ROLi+R’ or R’OLi+R o R2 or R Li RLi

(c) c). R H RH or 2 R

+ (d) d) b).. For instance, (EG) CH3CH2OCH2CH2OCH2CH3+Li e-

CH3CH2OLi + CH2CH2-OCH2CH3 and CH3CH2 +CH3CH2OCH2CH2OLi (e) e). (DME) CH3OCH2CH2OCH3+2Li++2e-

2CH3OLi +CH2=CH2

o (f) f). THF Li CH3CH2CH2CH2OLi (and/or Li-(CH2)4OLi) CH3 o Li (g) g). 2Me-THF CH (CH ) OLi+CH CHCH CH OLi

3

2 4

3

2

2

(and/or Li-ROLi) CH2 O

O

h). (DN) CH2-CH2 + e- + Li+ (h)

CH2CH2OCH2OLi (major) or CH2OCH2CH2OLi

(i) i). CH2CH2OCH2OLi

H

CH3CH2OCH2OLi

or CH3CH3 +HCO2Li o (j) j). CH2CH2OCH2OLi Li LiCH2CH2OCH2OLi

(k) k). ROLi + nDN

polymerization

R-(OCH2CH2-OCH2)n OLi

Scheme 4: Ethers reactions patterns 5c

Identification of Surface Films Formed on Li and Non-active Electrodes at Low Potentials in Ester Solutions.

Scheme 3 describes reduction mechanisms of two selected esters — methyl formate and γ-butyrolactone on lithium, lithiated carbon or noble metals polarized to low potentials (Li salt solutions).32, 90 FTIR spectra of Li electrodes in contact with ester solutions clearly show absorption bands of surface species which contain Li carboxylate groups (-COOLi).23, 90 This is demonstrated in Figure 19, which shows FTIR spectra of a Li surface covered by a thin layer of

106

Lithium-Ion Batteries: Solid-Electrolyte Interphase

γ-butyrolactone protected by a KBr layer (see Figure 2).32 While the initial spectrum (19a) relates to the thin solvent layer on the lithium surface, as the experiment continues, the spectra measured clearly show growing IR peaks of Li carboxylate species (19b, c, typical peaks around 1890 cm−1 and 1425 cm−1).32 It should be noted that all the suggested reaction paths of the esters in Scheme 3 are based on a rigorous analysis of their isolated reduction products (obtained by electrolysis of the esters with noble metal electrodes).14 Scheme 3: Ester reaction patterns γ Butyrolactone O

O

C

C O

CH2

O CO Li+ O

CH2 Li+

+ Li (Hg)

CH2 CH2

O _

1

CH2

CH2 CH2

CH2

COLi H radical termination

BL CH2

2 _ H 2

CH2 CH2 CH3 H

O

O

O

O

C ( Li+ )

C

C

C

CH

O

O

O

CH2

O O

LiO (CH2)3

Li+

(CH2)3

C

O

(CH2)3

Longer Chain Formation

C

O

3 BL (SN2)

O

O HCOCH3

HCO- + CH3

HCOO- + Li+ CH3

H or CH3

General R + Lio

O

HCOOLi CH4

or C2H6

Scheme 3: Esters reactions patterns

_

O

RLi

O C

C C

CH2 CH2

CH2 CH2

HCOCH3

OLi (CH2)3

Li+ O

methyl formate HCOOCH3 + e-

C C

CH2 CH2 Li+ -H+

O

C

O

O

C

OLi

CH2 CH2

CH2 CH2

CH2 CH2

CH

O

OLi (CH2)3

Identification of Surface Films on Electrodes

107

Figure 19 (a) A FTIR spectrum of lithium surface on which a thin layer of -butyrolactone was laid. The spectrum was measured 25 min after the sample preparation. (b) Same as (a); the spectrum was measured several hours after the sample preparation. (c) Same as (a); the spectrum was measured 24 h after the sample preparation. Reprinted with copyright from The Electrochemical Society Inc. (See [32].)

5d

Identification of Surface Films Formed on Li and Non-active Metals at Low Potentials in Alkyl Carbonate Solutions

The most expected reaction of alkyl carbonate solvents with Li (or on non-active electrodes polarized to low potentials in the presence of Li-ions) is their twoelectron reduction to Li2CO3 and alkanes or alkenes as by-products (e.g., PC + 2e− + 2Li+ → Li2CO3 + CH3CH=CH2).91 However, FTIR spectra of Li or noble metal electrodes treated in alkyl carbonate solutions show a different picture.15, 31 Figure 20 shows FTIR spectra measured from Li electrodes prepared fresh in PC solutions of three different salts soon after their preparation, and after two days of storage.92 Comparing the spectra measured from the freshly prepared electrodes (Figure 20), library spectra (Figure 15), and literature data,87 clearly shows that they belong to ROCO2Li species. Recent calculations showed that the typical peaks of ROCO2Li (e.g., 1650 cm−1, 1350-1300 cm−1, 1090 cm−1, and 850-800 cm−1) belong not to simple ROCO2Li species, but rather to dimers, or even polymers, of these compounds, in which Li-ions bridge between the

108

Lithium-Ion Batteries: Solid-Electrolyte Interphase

negatively charged oxygen of the carbonate groups.93, 94 The spectra of the aged electrodes (dashed lines, Figure 20) reflect a partial conversion of the surface ROCO2Li to Li2CO3. This may result from two processes: 1. 2.

A further reduction of the ROCO2Li in the surface films. Reaction of the ROCO2Li with unavoidable trace water in solutions to form Li2CO3, ROLi and CO2.87

Figure 20 FTIR spectra obtained from lithium surfaces freshly prepared and stored in PC solutions. The surfaces were protected with KBr windows and measured using external reflectance mode at a grazing angle. (a) Pure solvent. (b) LiClO4 1M solutions. (c) LiBF4 1M solutions. Solid line, 2 h of storage. Dashed line, 2 days of storage. Reprinted with copyright from The Electrochemical Society Inc. (See [92].)

Figure 21 shows FTIR spectra measured from Li electrodes prepared and stored in various EC-DMC solutions.95 This figure also presents a reference spectrum obtained from a Li electrode that was prepared and stored in a DMC/methanol solution. This spectrum is a superposition of CH3OLi and CH3OCO2Li spectra. Comparing the other five spectra of Figure 21 with this reference spectrum shows that the surface films formed on Li in EC-DMC solutions do not contain CH3OLi or CH3OCO2Li. The Li surface chemistry is

Identification of Surface Films on Electrodes

109

dominated by EC reduction to form (CH2OCO2Li)2.96 In LiPF6 or LiBF4 solutions (see also Figure 20), salt reduction also contributes to the electrodes’ surface chemistry (see discussion in the next section on salt anion reactions). These salt solutions also contain HF, which reacts with the surface carbonates to form surface LiF and carbonic acids.97, 98

Figure 21 FTIR spectra of lithium electrodes prepared and stored for three days in EC-DMC solutions of 1 M LiAsF6 and 1 M LiPF6 and LiBF4, as indicated. A spectrum of lithium electrode prepared and stored in DMC containing 0.1 M CH3OH is also presented for a comparison. Reprinted with copyright from Elsevier Science. (See [96].)

The intensive spectral studies of Li and noble metal electrodes in these solutions converged to the reduction paths of alkyl carbonate solvents, and their secondary reactions (due to the presence of contaminants) are presented in Scheme 4.86, 95-98 Figure 22 shows FTIR spectra measured from Li electrodes prepared fresh and stored in LiAsF6 and LiPF6 solutions of propyl-methyl carbonate (PMC).99 The attached table provides IR absorptions of all the expected reduction products: CH3OLi, CH3OCO2Li, iPrOLi, and PrOCO2Li. FTIR spectra 22a and 22b (LiAsF6 solutions) show that all of the above compounds are indeed formed on Li in this solvent, as suggested in Scheme 4, path l. The spectrum related to the LiPF6 solution (22c) reflects the secondary reactions with HF, as suggested in Scheme 4, path n.

110

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Scheme 4: Possible reduction patterns of alkyl carbonate on Li

Identification of Surface Films on Electrodes

111

Figure 22 FTIR spectra measured ex situ from Li surfaces prepared fresh (in situ) and stored in o MPC solutions. External reflectance mode at grazing angle (80 ), Li surfaces protected by KBr windows. Peak assignments and a table of reference IR absorptions appears. (a) 1 M LiAsF6, 3 days of storage, (b) same as (a), 3 h of storage, and (c) 1 M LiPF6 solution, 3 h of storage. Reprinted with copyright from The Electrochemical Society Inc. (See [99].)

There are several reports on the study of the reactions between lithium metal and alkyl carbonate solvents in ultra high vacuum. Li layers were prepared by evaporation-deposition, on which solvent layers were condensed at low temperatures from their vapor. The surface chemistry of the Li with the solvent on it was then studied at different temperatures by methods such as XPS, FTIR, and mass spectrometry. The results obtained correlate in general with the reaction paths of Scheme 4.100, 101 5e

The Impact of Salt Anions and Contaminant Reactions on the Surface Chemistry of Lithium and Noble Metal Electrodes in Non-aqueous Li Salt Solutions

All of the commonly used salt anions, such as AsF6−, ClO4−, BF4−, PF6−, N(SO2CF3)2−, etc., react with Li metal and are reduced on non-active metal electrodes at low potentials in nonaqueous solutions in the presence of Li ions. FTIR spectroscopy is not a sufficient tool for the study of salt anion reactions, because not all of the products can be detected by surface sensitive IR

112

Lithium-Ion Batteries: Solid-Electrolyte Interphase

spectroscopy (e.g., Li halides). Hence, it is necessary to use additional tools. XPS is very useful for that purpose. This technique was extensively used by several groups for the study of surface films formed on lithium.103-103 It provides surface element analysis, the oxidation state of the various elements on the surfaces, and depth profiling of surface films (repeated sputtering and analysis). Figure 23 shows typical XPS data for elements such as Li, C, O, F, As, P and S, obtained by surface studies of Li electrodes freshly prepared and stored in a.

b.

c.

Figure 23 XPS spectra measured from Li electrodes prepared and stored in solutions for 3 days: (a) EC-DMC 1:1/ 1 M LiAsF6. (b) EC-DMC 1:1/ 1 M LiPF6. (c) DN 1 M/ LiC(SO2CF3)3. Solid lines, before sputtering; dashed lines, after removal of 30-50 Å of surface layer. The relevant elements are marked in each spectrum. Reprinted with copyright from ACS. (See [104].)

Identification of Surface Films on Electrodes

113

LiAsF6, LiPF6, and LiC(SO2CF3)3 solutions (23a-c, respectively).104 The fluorine peaks in Figures 23a-c and the arsenic, phosphorous, and sulfur peaks in Figures 23a, b, c, respectively, clearly demonstrate the strong involvement of the salt anions’ reduction processes in the surface film formation on Li. Scheme 5 summarizes all these studies and suggests reaction paths for several commonly used salt anions. Scheme 5: Surface reactions of commonly used Li salts (a) (b) (c) (d) (e) (f) (g)

LiAsF6 + 2Li+ + 2e- → 3LiF ↓ + AsF3 (sol) AsF3 + 2xLi+ + 2xe- → LixAsF3-x ↓ + xLiF ↑ PF6- + 3Li+ + 2e- → 3LiF ↓ + PF3 LiF + PF5 LiPF6 PF5 + H2O → PF3O + 2HF PF5 + 2xLi+ + 2xe- → LixPF5-x ↓ + xLiF ↓ PF3O + 2xLi+ + 2xe- → LixPF3-xO ↓ + xLiF ↓

(h) (i) (j) (k) (l) (m) (n) (o) (p) (q) (r)

BF4- → LiF ↓, LixBFy ↓ (in general) LiClO4 + 8Li+ + 8e- → 4Li2O + LiCl LiClO4 + xLi+ + xe- → LiClO(4-½x) + ½xLi2O. (x = 2, 4, 6) LiN(SO2CF3)2 + ne- + nLi+ → Li3N + Li2S2O4 + LiF + C2FxLiy LiN(SO2CF3)2 + 2e- + 2Li+ → Li2NSO2CF3 + CF3SO2Li Li2S2O4 + 10e- + 10Li+ → 2Li2S + 4Li2O LiC(SO2CF3)3 + 2e- + 2Li+ → Li2C(SO2CF3)2 + LiSO2CF3, etc. Li2S2O4 + 4e- + 4Li+ → Li2SO3 + Li2S + Li2O 2LiSO3CF3 + 2Li+ + 2e- → 2Li2SO3 + C2F6 R-CF3 + 2Li+ +2e- → RCF2Li + LiF Li2SO3 + 6Li+ +6e- → Li2S + 3Li2O.

Li + , e −

As seen in Figures 18, 20-22, when the salts are LiAsF6 or LiClO4, the surface chemistry of these systems is dominated by solvent reactions.83, 86, 92 In ethereal solutions of salt such as LiN(SO2CF3)2 and LiC(SO2CF3)2, the salt anion reactions on the active electrodes dominate their surface chemistry (see Figure 18 and related reactions paths in Scheme 5).84 When the salts are LiBF4 and LiPF6, the solutions are always contaminated by HF because these salt decompose to LiF and the PF5 or BF3 Lewis acids.105 These acids react readily with water to form HF, and species such as PF3O or BFO. HF polymerizes ether solvents, and therefore, ethereal solutions of these salts are not stable.89 In alkyl

114

Lithium-Ion Batteries: Solid-Electrolyte Interphase aa. Li2CO3

Li 2CO3

1509

0.006

υC-O 884 υC=O as

0.004

Absorbance

0.002

1680

ROCO2 Li

LiPF3(CF 2CF3)3

δOCO

1309

ROCO2 Li

0.000

2924

1670

2963

1301

υP-O

_ 2

831

1090

υP -F

-0.002 2858

-0.004

υC-H

-0.006 4000

1464

Residual EC

3000

2000

1000

Wavenumbers (cm -1)

bb. 200

40 F 1s

2

120 80 40

3 (CF 2 CF 3 )3

160

LiPF6

C 1s

LiPF3(CF 2CF3)3

30 20

LiPF6

– OCO2 –

LiP F

Intensity (CPS) x 10

1783

LiPF6

10

0

0 688

684

680

292

288

284

280

Binding Energy (eV)

Figure 24 (a) FTIR spectra (ex situ, grazing angle, reflectance) of gold mirror polarized to 0.3 V + (Li/Li ). EC-DEC-DMC 2:1:2/ 1 M lithium salt as indicated. (b) XPS spectra of Pt electrodes, + polarized to 10 mV (Li/Li ). EC-DEC-DMC 2:1:2/ 1 M lithium salt as indicated. (See [108].)

carbonate solutions, HF solubilizes the surface carbonate species (Scheme 4) and LiF precipitates on the surface instead.97-98 A comparison between the carbon spectra in Figures 23a and b, related to Li electrodes prepared and stored in EC-DMC solutions of LiAsF6 and LiPF6 is very significant. The C1S spectrum of Li treated in the former solution shows pronounced carbonate peaks around 291 eV. These peaks are absent in the C1S spectrum of the Li electrode treated in the LiPF6 solution. The C1S peaks in the 284-287 eV range in Figure 23b relate to alkoxy species and alkyl groups.106 This is because the HF reacts readily with the ROCO2Li and Li2CO3, leaving the surface organic species at lower oxidation states. When formed, ROCO2Li may also decompose to ROH and CO2, and the former species can react on the Li surface to form ROLi. It should be noted that removal of the carbonates from the electrodes’ surfaces, as described above, allows further reactions of the salt anions with the active surfaces. As summarized in Scheme 5 and seen in Figure 23b (the P2P spectrum)

Identification of Surface Films on Electrodes

115

and Figures 20-22, LixPF4, LixPOFy and LixBFy, and LixBOFy are present in surface films formed on Li and non-active electrodes in LiPF6 and LiBF4 solutions, respectively. The impact of the salt on the surface chemistry of Li and non-active metal electrodes in connection with the acidic contamination is demonstrated below (Figure 24). The Merck company developed the new salt, LiPF3(CF2CF3)3, LiFAP, as a replacement to LiPF6.107 This new salt does not decompose, and hence does not liberate Lewis acids that hydrolize to HF. Indeed, surface films formed on noble metal (Au, Pt) electrodes polarized to low potentials in LiFAP and LiPF6 solutions in alkyl carbonate mixtures (EC, DMC, DEC) are pronouncedly different, as seen in Figure 24.108 The surface films formed in LiFAP solutions are dominated by carbonates (Figure 24a), while the surface films formed in LiPF6 solutions contain Li-P-F compounds and ROLi species. XPS data (Figure 24b) clearly show that the surface films formed in the LiFAP solution contain much less LiF than those formed in LiPF6 solutions. The last aspect dealt with in this section relates to reactions of other common contaminants (atmospheric gases) on the active electrodes’ surfaces. Trace H2O may be reduced to LiOH, Li2O and LiH,67 trace O2 may be reduced to LiO2, Li2O2 and Li2O,109 and CO2 reacts to form Li2CO3.110 Scheme 6 suggests a mechanism for CO2 reaction with lithium to form Li2CO3 and CO (the latter gas was identified in a few in situ measurements of non-active electrodes polarized cathodically in CO2 containing Li salt solutions110). However, it is possible that CO2 in solution forms surface Li2CO3 due to its reactions with Li2O and LiOH. Scheme 6: Reaction patterns of common contaminants (a) CO2 + e− + Li+ → 2Li Li + CO → O = -O-CO2Li 2 2 O= -O-CO2Li + e + Li+ → CO ↑ + Li2CO3 ↓ 2LiOH ↓+ CO2 → Li2CO3 ↓ + H2O Li2O ↓ + CO2 → Li2CO3 ↓ ROLi ↓ + CO2 → ROCO2Li ↓ (b) H2O + e− + Li+ → LiOH + ½ H2 LiOH + Li+ +e− → Li2O + ½ H2 H + e− + Li+ → LiH (c) N2 + 6e− + 6Li+ → 2Li3N (d) O2 + e− + Li+ → LiO2·





116

Lithium-Ion Batteries: Solid-Electrolyte Interphase

LiO2 + e- + Li+ → Li2O2 Li2O2 + 2e− + 2Li+ → 2Li2O (e) HF + e− +Li+ → LiF + ½ H2

5f

On Surface Films Formed on Li Electrodes in Polymeric Electrolytes

It is very hard to identify surface species formed on lithium in contact with polymeric electrolytes, because it is difficult to remove the polymeric matrices from the Li surfaces for spectroscopic measurements. It is possible to study the surface chemistry of Li electrodes in contact with polymeric electrolytes by in situ FTIR spectroscopy using internal reflectance mode (Figure 2). We recently studied the surface chemistry of Li electrodes in two types of polymeric electrolytes: 111 1.

2.

A gel-type electrolyte, comprising polymers such as polyacrylonitrile or poly vinylidene di fluoride-hexa fluoro propiate (PVdF-HFP) EC-PC as plasticizer and LiClO4.112 A solvent-free polymeric electrolyte comprising derivatives of polyethylene oxide (PEO)113 and several Li salts (e.g., LiAsF6, LiClO4, LiN(SO2CF3)2, using in situ FTIR spectroscopy (single internal reflectance mode).84 With the gel electrolytes, the Li surface chemistry was found to be dominated by EC and PC reduction to form ROCO2Li species. With the solvent-free polymeric electrolytes, the Li surface chemistry was found to be dominated by salt anion reduction (see reaction paths in Scheme 5).111 There was no evidence in the spectroscopic studies of any pronounced surface reactions of the PEO-based polymer with Li up to 60 °C. However, as discussed in section b above, Li attacks C-O bonds of ethers at room temperature. Therefore, we cannot exclude the possibility that lithium reacts with PEO derivatives to form surface alkoxy species.

6

Surface Films on Lithiated Carbon Electrodes

6a

Introductory Remarks: Surface Film Formation on Carbon Electrodes, the Influence of the Type of Carbon, and the Impact of the Surface Films on Li Insertion Processes

The composition of the surface films developed on carbon electrodes in a non-aqueous Li salt solution was studied by FTIR spectroscopy in transmittance

Identification of Surface Films on Electrodes

117

and diffuse reflectance modes,114 and by XPS.115 When carbon electrodes are polarized cathodically in nonaqueous Li salt solutions, they develop surface chemistry similar to that developed on non-active metal electrodes, as discussed in the previous section. There are a large variety of carbons that were studied as the active mass for Li insertion electrodes. The first division is graphitic carbon vs. disordered carbon. The former group includes many types of graphite: natural and synthetic flakes, fibers, and mesocarbon microbeads (MCMB) with round shaped particles. The latter group includes soft, graphitizable carbons and hard, non-graphitizable carbons.116, 117 We should also include single and multiwall carbon nanotubes as electrode materials that were studied recently.118, 119 It is important to note that, in general, the basic surface reactions of all of these carbons are similar in the same Li salt solutions.(80) However, there are pronounced differences in the electrochemical behavior of the different types of carbons as a function of their structure. There are two critical aspects in this respect: 1. 2.

How much charge is involved in the surface film formation until passivation is reached (i.e., the irreversible capacity)? What is the reversible capacity and the stability of the electrodes in prolonged, repeated Li insertion-deinsertion cycling?

Graphite carbons are fragile, and the graphene planes are weakly bound to each other. Thus, in the absence of passivation, solvent molecules can cointercalate with Li-ions into the graphite, which leads to its exfoliation (on a nanoscopic scale).116 In addition, in surface reactions in which gas molecules are formed (e.g., EC, PC reduction; see Scheme 4), an internal pressure may be developed, that cracks the particles.114 Such processes increase irreversible capacity and deactivate graphite electrodes. Thus, the electrochemistry behavior of graphite electrodes is very strongly dependent on the formation of passivating surface films, and hence, on the solutions’ composition. Disordered carbons usually have a more robust structure, and so their passivation by surface films is easier compared with the case of graphites. Hence, their Li insertion-deinsertion processes in terms of irreversible capacity and stability are less dependent on the solution composition compared with graphites.80

6b On the Identification of Surface Films Formed on Lithiated Graphite Electrodes We have no evidence that lithiated graphite reacts with ethers in the same manner as Li metal. However, more reactive solvents, esters, alkyl carbonates,

118

Lithium-Ion Batteries: Solid-Electrolyte Interphase

the commonly used salt anions and atmospheric contaminants such as H2O, CO2, O2, and HF, are reduced on carbon electrodes polarized to low potentials, or by lithiated graphite, in the same manner as they are reduced by lithium metal. We present herein two typical examples: Figure 25 shows FTIR spectra measured by transmittance mode from graphite powders pelletized with KBr which were scraped from electrodes cycled in EC-DEC solutions of LiAsF6, LiClO4, LiPF6, and LiBF4, as indicated.120 The spectra related to LiAsF6 and LiClO4 solutions contain mostly carbonate peaks (compare with the library spectra of ROCO2Li in Figure 15 and with Figures 20-22). As can be concluded from the FTIR spectra, the dominant surface species in these systems are the EC reduction products (CH2OCO2Li)2.120 As seen in Figure 25, the surface chemistry of graphite electrodes in LiBF4 or LiPF6 solutions, which usually contain HF, is different. Their surface films are not dominated by ROCO2Li species. The pronounced peak around 1000 cm-1 in the IR spectrum of the electrode treated in the LiBF4 solution relates to some kind of Li-B-O species (not identified). Hence, the surface chemistry of graphite electrodes in these alkyl carbonate solutions is very similar to that of Li electrodes, as discussed in sections 5d, e, above.

Figure 25 FTIR spectra obtained from graphite electrodes after being cycled in LiAsF6, LiClO4, LiPF6 and LiBF4 1 M solutions (a-d, respectively) in EC-DEC mixtures. Graphite particles were pelletized with KBr, transmittance mode. Reprinted with copyright from The Electrochemical Society Inc. (See [120].)

Identification of Surface Films on Electrodes

119

Figure 26 shows FTIR spectra of graphite powders scraped from graphite electrodes cycled in γ-butyrolactone (BL)/LiAsF6 solutions (KBr pellets, transmittance mode).78 The spectrum related to the additive-free solution (indicated) is typical of the cyclic β-keto ester-Li salt formed by BL reduction by lithium.32 (See the formula in the figure.) The spectrum related to the electrode treated with the solution pressurized with CO2 has Li2CO3 peaks as its dominant features. Again, these results demonstrate the similarity between the surface chemistry of Li and lithiated carbon electrodes in organic ester solutions.

Figure 26 FTIR spectra measured from graphite electrodes after being treated in BL-LiAsF6 solution under argon and under CO2, as indicated. One complete intercalation-deintercalation cycle. Graphite powder from electrodes (after washing and drying) was pelletized with KBr (transmittance mode). Reprinted with copyright from Elsevier Science. (See [78].)

120

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Hence, the discussion in sections 5c-e above is also relevant to carbon electrodes, and Schemes 3-5 also provide the reaction paths of esters, alkyl carbonates, salt anions, and contaminants on carbon electrodes polarized to low potentials, and on lithiated graphite as well. The only difference between lithiated graphite and lithium in terms of surface reactions in esters and alkyl carbonate solutions is that since the active surface of carbon electrodes per geometric area is much higher than that of Li electrodes, the impact of trace contaminants on the surface chemistry of carbon electrodes may be less pronounced.

6c On the Correlation Between the Performance of Lithiated Graphite Anodes and Their Surface Chemistry Graphite is the most important anode material so far for rechargeable Li-ion batteries. It may insert lithium reversibly at capacities close to the theoretical one (372 mAh/gr), at fast kinetics.116, 117 However, as mentioned in the previous sections, the 3D structure of graphite is weak. Graphite particles can be easily cracked or exfoliated. The surface reactions of lithiated graphite with solution species and processes such as co-intercalation of solvent molecules can be detrimental to graphite anodes. The major condition for the good performance of Li-graphite anodes is that the surface reactions dealt with in detail in sections 5a-e and 6b form passivating surface films on the graphite, which precipitate rapidly at high enough potentials before detrimental processes such as the buildup of internal pressure → cracking or co-intercalation of solvent molecules → exfoliation, take place. This depends on the nature of the surface species formed, that is, to what extent they are cohesive to each other and adhesive to the graphite surface. A large variety of surface species were identified as good passivating agents for graphite electrodes. Several examples are listed below: 1. 2. 3. 4. 5. 6.

(CH2OCO2Li)2 formed by EC reduction.96 Li2CO3 formed by CO2 reactions.76-78 Li2O formed by H2O reduction.121 Li2S and Li2SO3 formed by SO2 reduction.122 Reduction products of the complex Li bi salicilato borate.123, 124 The reduction products of vinylene carbonate, which may include (CHOCO2Li)2, poly(CHOCO2Li), and polycarbonate.125

Identification of Surface Films on Electrodes

121

Li2CO3 and organic carbonates with sufficiently small alkyl groups are good because they may form networks in which 2D and 3D Li-ions bridge among the negatively charged oxygen and between the negatively charged oxygen and the carbon atoms.93, 94 In a similar way, we can explain the good passivation properties of SO2 reduction products. In this respect, the PC reduction product CH3CH(OCO2Li) CH2OCO2Li is not a good enough passivating agent, because the methyl group interferes with the cohesion and adhesion of these species by steric hindrance. We expect that the formation of polymeric species such as polycarbonates, polyethylene oxide, and poly ROCO2Li on the graphite surface (e.g., in a solution containing VC26 ) enhances their passivation, and thus, increases their stability. In the following four examples, we demonstrate how the behavior of Ligraphite electrodes can be attenuated by relatively small changes in the solution composition, which modify their surface chemistry. 1.

2.

3.

4.

Graphite electrodes do not behave reversibly in ester solutions, such as methyl formate or BL. By addition of CO2 to these systems, Li-graphite electrodes behave reversibly and are very stable in these solutions because their surface films become dominated by Li2CO3.76-78 The performance of Li-graphite electrodes in DMC solutions is very poor. With the addition of several hundreds of ppm of water, the performance improves considerably.48 The explanation for this is that the surface films on graphite in DMC solutions contain CH3OCO2Li and CH3OLi, which are not good passivation agents. When water is present, it reacts with CH3oCO2Li to form surface Li2CO3 (plus CH3OH and CO287), which is a very good passivating agent, as discussed above. Addition of crown ethers such as 23 crown 4 to a PC solution, even at a concentration much less than that of the salt, causes the graphite electrodes to behave reversibly in these solutions. The crown ether molecules form complexes with the Li-ions. When the electrodes are polarized cathodically, the Li-ions bound to the crown ethers preferentially approach the electrode’s surface and modify their detrimental surface reactions with PC. 77, 127 The last example is THF-PC solutions.121 Figure 27 shows chronopotentiograms of graphite electrodes that were lithiateddelithiated in THF, THF-PC, 3M, and in THF-PC 1M solutions. The difference in behavior is striking. In THF, graphite electrodes fail because THF is not sufficiently reactive to be reduced to passivating

122

Lithium-Ion Batteries: Solid-Electrolyte Interphase

surface species. Thus, THF molecules co-intercalate with Li-ions and graphite particles exfoliate, due to co-intercalation (e.g., the classical failure mechanism of Li-graphite electrodes).121 (a )

(b )

(c )

Figure 27 Chronopotentiograms (V vs. t, constant current) of graphite electrodes (KS-44 Lonza, 10% PVDF) in: (a) THF/ 3 M PC, 1 M LiAsF6, (b) dry THF/ 1 M LiAsF6, and (c) THF/ 1 M PC 1M LiAsF6 solutions. After discharge, the potential was held constant at close to 0 V (Li/Li+) for several hours (as indicated), followed by galvanostatic deintercalation. The charges involved in this last process expressed in molar equivalents Li per C6 are marked. Reprinted with copyright from ACS. (See [121].)

In THF solutions with too high a PC concentration, the electrodes’ surface chemistry is dominated by PC reduction. The PC reduction products of the CH3CH(OCO2Li)CH2OCO2Li type are not good enough passivating agents, as discussed above. Therefore, PC reduction is not inhibited quickly enough by passivation before propylene gas is accumulated in crevices in the graphite particles’ edge planes. The internal pressure thus built up cracks the particles, which leads to electrical isolation of the active mass and deactivation of the electrodes.114 In THF-PC 1M, the behavior of Li graphite electrodes is reversible because the change in solution composition (low PC concentration) attenuates the electrode’s surface chemistry. Figure 28 shows FTIR spectra of graphite electrodes treated in THF/PC 3M and THF/PC 1M solutions. The

Identification of Surface Films on Electrodes

123

spectrum related to the latter solution has pronounced Li2CO3 peaks. Hence, at a low PC concentration, the surface films formed on graphite contain Li2CO3, which is an excellent passivating agent, while at a high PC concentration, only CH3CH(OCO2Li)CH2OCO2Li is formed. The explanation for this is that one of the reduction products of PC by graphite polarized to low potentials is CO3= (and CH3CH=CH2 as a co product) or LiCO3−. At a high enough PC concentration, the carbonate anion attacks nucleophilically another PC molecule (see Scheme 4), which leads to the formation of CH3CH(OCO2Li)CH2OCO2Li as a major surface species (not too good a passivating agent). PC is reduced predominantly on graphite, even when its concentration in THF is low. However, at a sufficiently low PC concentration, CO3= or LiCO3− when formed, react faster with Li-ions than with other PC molecules, and thereby, the surface films become rich in Li2CO3.

Figure 28 FTIR spectra measured using diffuse reflectance mode from graphite particles taken from electrodes after being cycled in THF / 1 M PC LiAsF6 and PC/ 1 M LiAsF6 solutions as assigned. Partial peak assignment appears. Reprinted with copyright from ACS. (See [121].)

The last point raised in this section about the surface chemistry of graphite electrodes relates to their morphology. Since the stabilization and reversibility of

124

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Li graphite electrodes depends on a rapid precipitation of cohesive and adhesive surface films, the particles’ morphology plays an important role. The smoother the edge planes of the graphite particles, through which Li is inserted, so the precipitation of passivating surface films by solution reactions may be faster and more efficient.

7

7a

Surface Studies of Lithium and Lithiated Carbon Electrodes by Scanning Probe Microscopy Imaging of Li Electrodes by AFM

Imaging of Li electrodes by AFM is also possible using the contact mode, because the surface films formed on Li in most of the electrolyte systems of interest are hard enough and do not change as a result of the contact with the tip.128 As already demonstrated, it is possible to differentiate by AFM imaging among the morphologies of surface films formed in different solution compositions, to study the effects of solvents, salts, additives, aging, etc. Especially important is the possibility of following surface processes of Li electrodes on nanometric scales. Thus, their failure mechanisms can be understood. The surface films on Li are very non-uniform laterally on the nanometric scale, as explained in section 3 above. Thus, the current density is never uniform. Li is dissolved or deposited preferentially at certain locations through the surface films, where the local ionic conductivity is relatively high. Figure 29 demonstrates what happens to Li surfaces during Li dissolution. Since the current density is not uniform, there is intensive Li dissolution at points of local, high ionic conductivity of the surface films. These films, which comprise mostly Li salts, are not flexible enough to accommodate the topographic changes of the Li surface. Therefore, the surface films are broken, thus exposing fresh lithium to solution species, and allowing intensive surface reactions of Li with solution components in the holes thus formed. This scenario is clearly imaged by in situ measurements of Li surfaces with AFM, as presented in the figure.129 Figure 30 shows what happens to Li surfaces during Li deposition. As demonstrated in the AFM image and explained in the cartoon, fresh Li deposits emerge and grow at locations of high ionic conductivity at the surface films, out of the passivating surface layer. Hence, fresh Li is exposed to the solutions, reacts with solution species, and both Li and solution species are irreversibly

Identification of Surface Films on Electrodes

125

lost. While dendrite formation on Li electrodes was studied years ago by scanning electron microscopy, this method is inferior to AFM in terms of resolution and the fact that it is mostly an ex situ technique.

a a.

Low c urre nt de ns ities :

H igh c urre nt d e ns itie s :

T he s urface film s a cco m m odate the vo lum e c ha nges.

T he s urface film s a re broke n do w n a nd are repa ired b y s urfac e reac tio ns o f L i w ith so lutio n spe c ies.

b.b

c. c

Figure 29 Breakdown and repair of surface films on lithium. (a) An illustration of the morphological phenomena developed on Li electrodes during Li dissolution in low current densities and in high current densities. (b, c) In situ AFM images of different lithium electrodes under alkyl carbonate solutions. Holes created by dissolution are marked with circles. (b) 3 D, 5 × 5 m. (c) 2 D, 500 × 500 nm. (See [129].)

126

Lithium-Ion Batteries: Solid-Electrolyte Interphase

aa. S ol u ti on S ol u ti on

S ol u ti on

S ol u ti on Li +

Li +

Li m e tal

Low c urre nt de ns ities : L i d epo s ited und er ne ath the s ur face film s.

bb.

Li +

Li + Li +

Li + Li +

S .L.

Li +

Li +

S .L.

S .L. Li m e tal

Li m eta l

H ig h c ur re nt dens ities :

Li m e tal

S .L.

D e ndr ite fo r ma tio n.

V o lum e c ha nges ; the sur face film s c rack.

cc.

dd d. d.

2 00u m m 2 00u

Figure 30 (a) An illustration of the morphological phenomena developed on Li electrodes during Li deposition in low and high current densities. (b, c) in situ AFM images of different lithium electrodes in alkyl carbonate solutions. Dendrites created by deposition are marked. (b) 2 D, 50 × 50 m. (c) 3 D, 25 × 25 m. (d) in situ CCD image of Li electrode. Dendrites can be seen as black spots. A 200 m bar is shown. (See [29].)

7b

Graphite Electrodes

AFM was first applied by Fischer et al. about 6 years ago130 for the study of surface phenomena on graphite electrodes in Li salt solutions. The model graphite electrodes for the first studies was HOPG.131 It was possible to follow in situ the precipitation of surface species of cathodically polarized HOPG, and to identify irreversible morphological changes of graphite surfaces due to surface reactions.(132) We have recently found reports in the literature on the study of composite graphite anodes by in situ AFM.133 Figure 31 demonstrates the power

Identification of Surface Films on Electrodes

127

of this method in the study of failure mechanisms of Li-graphite electrodes. It shows AFM images of a single meso carbon microbead (MCMB) particle on a composite graphite electrode during its cathodic polarization in an EC-PC 2:3/LiClO4 solution. The MCMB are round shaped particles that enable the imaging of the edge planes of the particles, through which Li insertiondeinsertion takes place. The images in Figure 31 clearly show surface film formation on the edge planes of the particle, and cracking of the particle due to the development of internal pressure. Cracking exposes a reactive, unpassivated surface that further reacts with solution species. The images also clearly show how the crack is filled with the solid products of the reactions between the

a.a

b. b

c.c d.d

ee. PC

PC

Figure 31 (a-d) In situ AFM images of MCMB particle in PC/ 1M LiClO4 at constant potential ~0.78 V. (e) An illustration of cracking and passivation of MCMB particle during surface film formation. (See [134].)

128

Lithium-Ion Batteries: Solid-Electrolyte Interphase

active surface inside the crack and solution species. This imaging demonstrates the major failure mechanism of graphite electrodes in PC solutions, namely, cracking of graphite particles due to the build-up of internal pressure (gas formation), as explained in section 6 above.114, 134

8

About Surface Film Formation on Transition Metal Oxide Cathodes in Non-aqueous Salt Solution

The cathode materials and the source of lithium in Li-ion batteries are mostly LiMO2 compounds where M=transition metal or a mixture of transition metals such as Co, Ni, Mn, Co-Ni, Mn-Ni, Mn-Cu, etc.135 The redox potentials of these materials can be as high as 5 V (Li/Li+).136 Studies of the anodic stability of Liion battery electrolyte solutions (Li salt, mostly LiPF6, and a mixture of alkyl carbonates that usually include EC) with noble metal electrodes showed that their intensive oxidation may occur only at potentials >5 V.137 However, there are some low scale oxidation processes that occur at potentials below 4 V (Li/Li+).16 Although oxidation of solvent molecules such as EC produces polymeric species16 of high molecular weight, these studies do not indicate formation of surface films on noble metal electrodes at high potentials in these solutions. It is now generally accepted that LiMO2 cathode materials react with solution species in Li battery electrolyte solutions to form surface films.81, 138 FTIR measurements of LiNiO2 and LiCoO2 electrodes shows that the pristine active mass contains surface Li2CO3.95 LiMn2O4, LiCoO2, and LiNiO2 reacts with HF, which is unavoidably present in LiPF6 solutions.139 Consequently, the surface of LiMO2 always contains LiF in any solution contaminated with HF. FTIR spectra of cycled LiNiO2, LiCoO2 and LiCO-NiO2 electrodes include typical ROCO2Li peaks.95, 139 As discussed in previous sections, ROCO2Li salts are the major reduction products of alkyl carbonate solvents on Li and Li-C surfaces.96, .99 One possible explanation for the existence of such surface species is the reflection from the anode’s side; i.e. ROCO2Li are formed by solvent reduction on Li or Li-C, reach saturation in solution, and are re-deposited on the cathode side. But there is also the possibility that LiMO2 reacts nucleophilically with the surface electrophilic alkyl carbonate molecules, e.g., LiNiO2 + EC → NiOOCH2CH2OCO2Li. From our recent spectroscopic studies, the scale of reactivity of the commonly used LixMO2 cathode materials towards solution species is LiNiO2 > LiCoO2 > LiMn2O4.139 We also have spectral evidence for

Identification of Surface Films on Electrodes

129

the formation of polycarbonate species on the cathodes.139 It is logical to assume that polycarbonates may be formed by nucleophilic reactions of carbonate anions, ROCO2- on EC molecules, thus inducing polymerization via anionic mechanisms. In addition, delithiated MO2 may oxidize molecules such as EC. EC radical cations can undergo several reactions with EC molecules including polymerization to form polycarbonates or polyethyloxide while liberating CO2. There is evidence that additives such as VC also polymerize on cathodes to form polycarbonate species.126 In conclusion, the electrochemical behavior of many LixMOy cathodes is also controlled by surface films, i.e., during the course of Li insertiondeinsertion, there are important stages of Li-ion migration through a surface layer and charge transfer of Li-ions through an interface between a surface layer and the active mass. As discussed in section 4 above, impedance spectroscopy of LixMOy cathodes reflects very clearly the impact of the surface films on the electronic properties of these electrodes. It should be noted that upon prolonged storage, especially at elevated temperatures, surface film formation via the mechanisms described above on the cathodes intensifies, and thus the cathodes’ impedance increases. In fact, a pronounced increase in the cathodes’ impedance during prolonged cycling/storage of Li-ion batteries may be a major reason for their capacity-fading.19, 140

9

Identification of Surface Films on Calcium and Magnesium Electrodes

Magnesium and calcium electrodes were investigated in connection with high energy density, non-aqueous batteries.11, 12 Following the great success in the development and commercialization of Li-SOCl2 batteries, there were also attempts to develop Ca and Mg thionyl chloride batteries.141, 142 Li, Ca, and Mg react with SOCl2 and reduce it to the metal chlorides, metal oxides, and metal sulfides. These active metals become passivated in thionyl chloride by surface films, which comprise mostly the metal chloride.12 In the case of lithium electrodes, the surface films (LiCl) formed on them in SOCl2 conduct Li-ions at relatively fast kinetics.12 This condition enables the successful operation of Li/SOCl2/LiAlCl4/C batteries. The inorganic solvent in this battery is also the cathodic active mass, which is reduced on the carbon current collector at potentials around 3.6 (Li/Li+).12 In the case of calcium electrodes, the CaCl2 films that cover this active metal in SOCl2 solutions can conduct Cl- ions. Thus, it was possible to construct

130

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Ca/SOCl2/Ca(AlCl4)2/C batteries that operate in a manner similar to that of lithium-thionyl chloride batteries.141 In contrast, it was impossible to develop Mg-SOCl2 batteries because of the very poor charge transfer kinetics of the Mg anode in SOCl2 solutions (poor ion transport through the MgCl2 surface films).142 Intensive studies were also carried out on the behavior of Ca and Mg electrodes in organic polar aprotic electrolyte solutions.11, 12 Calcium reacts with esters such as methyl formate and γ-butyrolactone to form Ca-carboxylates in the same manner as Li does (see Scheme 3).11, 90 We do not have evidence of possible reactions of Ca and ethers. Ca surfaces prepared and stored in PC are covered by surface films comprising CaCO3. Ca reacts with salt anions such as ClO4−, BF4−, and AsF6−, and the reduction products include Li-halides (CaCl2, CaF2), which precipitate on the Ca surface. The surface films formed on calcium in ethers are comprised mostly of CaO and Li halide formed by reduction of trace O2/H2O and salt anions, respectively. The surface films formed on calcium in polar aprotic organic solutions do not conduct Ca ions. Thereby, Ca deposition is impossible and Ca dissolution occurs at high over potentials via a breakdown and repair mechanism of the surface films.11 Mg also does not react with ether solvents, but reacts with esters, alkyl carbonates, and salt anions such as ClO4−, AsF6−, BF4−, and PF6−, in a manner similar to Li (Schemes 3-5).10 Mg electrodes are covered initially by a MgO/MgCO3 film that is also formed in a glove box atmosphere. Contact of the Mg surfaces with ether solutions with ClO4−, AsF6−, and BF4 salts, or esters and alkyl carbonate solutions with any salt, cover them with Mg-halide, Mg carboxylates, and Mg alkyl carbonate surface films, respectively. None of these films conduct Mg ions, and hence, Mg electrodes in all of the above-mentioned electrolyte solutions are blocked. Mg deposition is impossible, and Mg dissolution requires a breakdown of the surface films at high over potentials.10 Mg electrodes in an ether solution containing the following type of salts — RMgX (R=alkyl, aryl; X=Cl, Br), Mg(BR4)2 (R=alkyl aryl), and Mg(AlCl4-nRn)2 (R=alkyl groups) — do not develop stable surface films, and therefore they are not passivated in these solutions.143 It should be noted that the above formulae of the salts do not represent the electrolyte’s structure in solutions. All of the salts with the formal formulae above form complicated structures in ether solutions, in which ether molecules are bound to the Mg ions and stabilize them.144 As mentioned in section 4 above, complicated adsorption phenomena take place on Mg surfaces in all the above solutions, which could be followed by in situ FTIR spectroscopy.145 An

Identification of Surface Films on Electrodes

131

exact identification of the adsorbed species was impossible. However, it was possible to conclude that the species adsorbed have Mg-C bands and ether molecules bound to the Mg ions. All of these adsorption phenomena in the ethereal solutions do not interfere badly with the reversible behavior of Mg electrodes in these solutions. Mg(AlCl4-nRn)2/ether solutions (THF, glymes), were found to be suitable as electrolyte solutions for rechargeable Mg batteries that were recently developed.74

10

Concluding Remarks

Surface film formation on solid electrodes is a common phenomenon in polar aprotic electrolyte solutions whenever the salt cations are ions of alkaline and alkaline earth metals, such as Li, Na, Mg, and Ca. All active metals may also spontaneously react with polar aprotic solvents and salt anions that dissolve in them and reach passivation due to surface film formation. Thus, most of the studies of commonly used non-aqueous electrochemical systems have to deal with surface film identification, electrical properties, and their impact on the response measured. Li and Li-ion batteries, which involve a major use of nonaqueous electrochemical systems, operate only because of passivation of the highly reactive anodes of these batteries, by surface films. The behavior of the cathodes (LiMO2) of Li-ion batteries is also controlled by surface films. These surface films and related phenomena were extensively investigated over the years. Most of the relevant surface species were identified, as well as their mechanisms of formation. Highly efficient tools for the study of surface phenomena on active electrodes in non-aqueous solutions were developed. Future studies of surface phenomena in polar aprotic electrolyte solutions are important for understanding the capacity fading of advanced Li-ion batteries and their limitation at high temperatures. It was recently found that what limits the performance of advanced Li-ion batteries at elevated temperatures and prolonged cycling/storage are surface phenomena that increase the batteries’ impedance. Hence, the design of new salts which are more stable and do not contaminate the solution with acidic decomposition products, and the judicious design of additives that control the surface chemistry of Li, Li-C, and LiMO2 electrodes in Li battery electrolyte solutions, are important challenges in the field of nonaqueous electrochemistry in general, and Li batteries in particular. The thermal stability of Li-ion batteries is also related to surface film phenomena and possible reactions between solution species and electrode

132

Lithium-Ion Batteries: Solid-Electrolyte Interphase

materials at elevated temperatures. Thus, in spite of the huge amount of hard work dedicated so far to surface film studies in non-aqueous electrochemical systems, there is still room for a lot of innovative work in light of the points above.

References 1. Vijh A. K., Electrochemistry of Metals and Semiconductors: The Application of Solid State Science to Electrochemical Phenomena (Marcel Dekker, New York, 1973), ed. by Diggle J. W. and Vijh A. K., The Anodic Behavior of Metals and Semiconductors Series: Oxides and Oxide Films (Marcel Dekker, New York, 1976), Vol. 4. 2. Takamura T., Kikuchi M. and Ikezawa Y., in Rechargeable Li-Ion Batteries, ed. by Megahed S., Barnett B. M. and Xie L., The Electrochemical Society, Soft Bound Series PV 94-28 (The Electrochem. Soc., Pennington, N. J., 1998), p. 213. 3. Sato N., Electrochemistry at Metal and Semiconductor Electrodes (Elsevier, Oxford, I.K., 1998), Ch. 2, pp. 15–59; Ch. 3, pp. 61–68; Ch. 5, pp. 119– 199; Ch. 11, pp. 373–389. 4. Wernick S. and Pinner R., The Surface Treatment and Finishing of Aluminum and Its Alloys, Vol. 1, 4th edition (Robert Draper, Teddington, England, 1972). 5. Peled E., in Lithium Batteries, ed. by Gabano J. P. (Academic, London, 1983), Ch. 3, p. 43. 6. Young L., Anodic Oxide Films (Academic, London, New York, 1961). 7. Ashcroft N. W. and Mermin N. D., Solid State Physics (Saunders, USA, 1976), Ch. 30, p. 616. 8. Holding A. D., Pletcher D. and Jones R. V. H., Electrochimica Acta 34 (1989), 1529. 9. Pourbaix M., Atlas of Electrochemical Equilibria in Aqueous Solutions, 2nd ed. (Cebelcor, Brussels, 1974). 10. Lu Z., Schechter A., Moshkovich M. and Aurbach D., J. Electroanal. Chem. 466 (1999), 203. 11. Aurbach D., Skaletsky R. and Gofer Y., J. Electrochem. Soc. 138 (1991), 3536. 12. Schlaikjer C. R., in Lithium Batteries, ed. by Gabano J. P. (Academic, London, 1984), p. 303.

Identification of Surface Films on Electrodes

133

13. Mann C. K., in Electroanalytical Chemistry, ed. by Bard A. J. (Dekker Marcel, New York, 1970), Vol. 3. 14. Aurbach D. and Gottlieb H., Electrochimica Acta 34 (1989), 141. 15. Aurbach D., Daroux M. L., Faguy P. and Yeager E., J. Electroanal. Chem. 297 (1991), 225. 16. Moshkovich M., Cojocaru M., Gottlieb H. E. and Aurbach D., J. Electroanal. Chem. 497 (2001), 84-96. 17. Braithwaite J. W., Gonzales A., Nagasubramanian G., Lucero S. J., Peebles D. E., Ohlhausen J. A. and Cieslak W. R., J. Electrochem. Soc. 146 (1999), 448. 18. Aurbach D., J. Power Sources 79 (1999), 135. 19. Broussely M., Herreyre S., Biensan P., Kasztejna P., Nechev K. and Staniewicz R. J., J. Power Sources 13 (2001), 97-8. 20. Wang K. and Ross P., Jr., J. Electrochem. Soc. 142 (1995), L95; Wang K. and Ross P., Jr., Surf. Sci. 365 (1996), 753. 21. Scherson D., in Nonaqueous Electrochemistry, ed. by Aurbach D. (Marcel Dekker, New York, 1999), Ch. 5, p. 213. 22. Schrader B., ed. Infrared and Raman Spectroscopy Methods and Applications (VCH New York, 1995). 23. Roeges N. P. G., A Guide to the Complete Interpretation of Infrared Spectra of Organic Structures (Wiley, New York, 1994). 24. Li Z. and Liu X., J. Electroanal. Chem. 386 (1995), 83. 25. Faguy P. W., Fawcett W. R., Liu G. and Motheo J., J. Electroanal. Chem. 339 (1992), 339. 26. Bae I. T., Sandifer M., Lee Y. W., Tryk D. A., Sukenik C. N. and Scherson D. A., Anal. Chem. 67 (1995), 4508. 27. Pham M.-C., Adami F. and Lacaze P.-C., J. Electroanal. Chem. 265 (1989), 247. 28. Seki H., Kunimatzu K. and Golden G., Appl. Spectrosc. 39 (1985), 437. 29. Patty D. B. and Harris J. M., Langmuir 6 (1990), 209. 30. Maulhardt H. and Kunath D., Talanta 29 (1982), 237. 31. Aurbach D., Daroux M. L., Faguy P. and Yeager E., J. Electrochem. Soc. 134 (1987), 16ll. 32. Aurbach D., J. Electrochem. Soc. 138 (1989), L6. 33. Aurbach D., Youngman (Chusid) O. and Goren E., J. Electrochem. Soc. 136 (1989), 1610. 34. Chusid (Youngman) O. and Aurbach D., J. Electrochem. Soc. 140 (1983) L1.

134

Lithium-Ion Batteries: Solid-Electrolyte Interphase

35. Christensen P. A., Hammett A. and Trevellick P. R., J. Electroanal. Chem. 242 (1988), 23. 36. Duyne R. P., in Chemical and Biochemical Applications of SERS, ed. by Moore C. B. (Academic, New York, 1979), Vol. 4, Ch. 4. 37. Foley J. K. and Pons S., Analytical Chemistry 57 (1985), 945A. 38. Moskevitz M. and Dillela D. P., in Surface Enhanced Raman Scattering, ed. by Furtak T. E. and Chang R. K. (Plenum, New York, 1982), pp. 264–267. 39. Zaban A., Ferrere S. and Gregg B. A., J. Phys. Chem. B 102 (1997), 452–460. 40. Tolmachev Y. V., Wang Z.H., Hu Y. N., Bae I. T. and Scherson D. A., Anal. Chem. 70 (1998), 1149. 41. Hu Y. N., Bae I. T., Mo Y. B., Antonio M. R. and Scherson D. A., Can. J. Chem. 75 (1997), 1721. 42. Somorjai G. B., in Introduction to Surface Chemistry and Catalysis (John Wiley and Sons, New York, 1994), Ch 5, p. 362 and references threin. 43. Zei M. S. and Ertl G., Z. Phys. Chem. 202 (1997), 5-19. 44. Goodhew P. J. and Humphreys F. J., Electron Microscopy and Analysis (Taylor and Francis, London, 1988). 45. Gouerec P., Savy M. and Riga J., Electrochim. Acta 434 (1998), 743. 46. Buttry D. A., in Electroanalytical Chemistry. A Series of Advances, ed. by Bard A. J. (Marcel Dekker, New York, 1991), Vol. 17. 47. Cheetham A. K. and Day P., eds., Solid State Chemistry Techniques (Oxford Science, New York, 1988). 48. Aurbach D. and Ein-Eli Y., J. Electrochem. Soc. 142 (1995), 1746. 49. Luo J. L. and Cui N., J. Alloys Compounds 264 (1988), 299. 50. Rivera M., Seetharaman R., Girdhar D., Wirtz M., Zhang X. J., Wang X. Q. and White S., Biochem. USA 37 (1998), 1485. 51. Nazri G. and Muller R. H., J. Electrochem. Soc. 132 (1985), 1385. 52. Kranke P., Wahren M. and Findeisen M., Isot. Environ. Health Stud. 33 (1997), 245. 53. MacOmber R. S., A Complete Introduction to Modern NMR Spectroscopy (Wiley, New York, 1997). 54. Raper E. S., Creighton J. R., Clegg W., Cucurull-Sanchez L., Hill M. N. S. and Akrivos P. D., Inorg. Chim. Acta 271 (1998), 57. 55. Tkami N., Satoh A., Oguchi M., Sasaki H. and Ohsaki T., J. Power Sources 68 (1997), 283. 56. Tucker M. C., Reimer J. A. and Cairns E. J., Electrochem. Solid State Lett. 3 (2000), 463.

Identification of Surface Films on Electrodes

135

57. Sarid D., Scanning Force Microscopy with Application to Electric, Magnetic and Atomic Forces (Oxford University Press, New York, 1994), Vol. 5. 58. Hamers R. J., J. Phys. Chem. 100 (1996), 13103. 59. Hugstad G., Gladfelter W. L., Weberg E. B., Weberg R. T. and Weatherill T. D., Langmuir 10 (1994), 4295. 60. Sidles J. A., Garbini J. L., Bruland K. J., Rugar D., Zuger O., Hoen S. and Yannoni C. S., Rev. Mod. Phys. 67 (1995), 249. 61. Stroscio J. A. and Kaiser W. J., eds., Scanning Tunneling Microscopy (Academic, New York, 1993). 62. Nanjo H., Newman R. C. and Sanada N., Appl. Surf. Sci., 121 (1997), 253256. 63. Lebreton C. and Wang Z. Z., Surf. Sci. 382 (1997), 193-200. 64. Cohen Y. and Aurbach D., Rev. of Scientific Instruments 70 (1999), 4668. 65. Croze F. and Scrosati B., J. Power Sources 43-44 (1993), 9-19; Wang K., Chottiner G. S. and Scherson D. A., J. Phys. Chem. 97 (1993), 11075. 66. Zhang G., Ross P. N., Kong F. P. and Mclarnon F., J. Electrochem. Soc. 145 (1998), 159. 67. Aurbach D., Moshkovich M. and Gofer Y., J. Electrochem. Soc. 148 (2001), E155. 68. Kolb D. M., Prasnyski M. and Gerischer H., J. Electroanal. Chem. 54 (1974), 25. 69. Geronov Y., Schwager F. and Muller R. H., J. Electrochem. Soc. 129 (1982), 1423. 70. Aurbach D., Zinigrad E. and Zaban A., J. Phys. Chem. 100 (1996), 3098. 71. Orazem M. E., Argawal P. and Garcia Rubio L. H., Electroanal. Chem. 378 (1994), 51. 72. Aurbach D. and Zaban A., J. Electroanal. Chem. 367 (1994), 15. 73. Aurbach D., Moshkovich M., Schechter A. and Turgeman R., Electrochem. Solid State Lett. 3 (2000), 31. 74. Aurbach D., Zhongua L., Schechter A., Gofer Y., Gizbar C., Turgeman R., Cohen Y., Moshkovich M. and Levi E., Nature 407 (2000), 724. 75. Aurbach D. and Zaban A., J. Electrochem. Soc. 141 (1994), 1808. 76. Aurbach D., Youngman Chusid O., Carmeli Y., Babai M. and Ein-Eli Y., J. of Power Sources 43 (1993), 47. 77. Aurbach D., Ein-Eli Y., Chusid O., Babai M., Carmeli Y. and Yamin H., J. Electrochem. Soc. 141 (1994), 603.

136

Lithium-Ion Batteries: Solid-Electrolyte Interphase

78. Aurbach D., Ein-Eli Y., Markovsky B., Carmeli Y., Yamin H. and Luski S., Electrochimica Acta 39 (1994), 2559. 79. Aurbach D. and Levi M. D., J. Phys. Chem. B 101 (1997), 4630. 80. Aurbach D., Gnanaraj J. S., Levi M. D., Salitra G., Fischer J. F. and Claye A., J. Electrochem. Soc. 148 (2001), A525. 81. Aurbach D., Levi M. D. and Lev O., J. Applied Electrochem. 28 (1998), 1051. 82. Youngman O., Gofer Y., Meitav A. and Aurbach D., Electrochimica Acta 35 (1990), 625. 83. Aurbach D., Zaban A., Chusid O. and Weismann I., Electrochimica Acta 39 (1994), 51. 84. Aurbach D., Chusid O. and Weissman I., Electrochimica Acta 41 (1996), 747 85. Aurbach D., Malik Y., Meitav A. and Dan P., J. Electroanal. Chem. 282 (1990), 73. 86. Aurbach D., Gofer Y., Ben-Zion M. and Aped P., J. Electroanal. Chem. 339 (1992), 451. 87. Behrendt W., Gattow G. and Drager M., Z. Anorg. Allg. Chem. 397, (1992), 451. 88. Aurbach D., Daroux M.L., Faguy P. and Yeager E., J. Electrochem. Soc. 135 (1988), 1863. 89. Aurbach D. and Granot E., Electrochimica Acta 42 (1997), 697. 90. Aurbach D. and Ein-Eli Y., Langmuir 8 (1992), 1845. 91. Dey A. N. and Sullivan B. P., J. Electrochem. Soc. 117 (1970), 222. 92. Aurbach D., Ein-Eli Y. and Zaban A., J. Electrochem. Soc. 141(1994), L1. 93. Matsuta S., Asada T. and Kitaura K., J. Electrochem. Soc. 147 (2000), 1695. 94. Wang Y. X. and Balbuena P., J. Phys. Chem. B 106 (2002), 4486. 95. Aurbach D., Markovsky B., Levi M. D., Levi E., Schechter A., Moshkovich M. and Cohen Y., J. Power Sources 81-82 (1999), 95. 96. Aurbach D., Markovsky B., Schechter A., Ein-Eli Y. and Cohen H., J. Electrochem. Soc. 143 (1996), 3809. 97. Kanamura K., Tamura H. and Takehara Z.-I., J. Electroanal. Chem. 333 (1992), 127. 98. Kanamura K., Tamura H., Shiraishi S. and Takehara Z.-I., Electrochim. Acta 40 (1995), 913. 99. Ein-Eli Y., McDevitt S. F., Markovsky B., Schechter A. and Aurbach D., J. Electrochem. Soc. 144 (1997), L 180.

Identification of Surface Films on Electrodes

137

100. Rendek L. J., Chottiner G. S. and Scherson D. A., Langmuir 17 (2001), 849. 101. Wang K., Ross P. N., Jr., Kong F. and McLarnon F., J. Electrochem. Soc. 143 (1996), 422. 102. Kanamura K., Shiraishi S., Tamura H. and Takehara Z.-I., J. Electrochem. Soc. 141 (1994), 2379. 103. Kanamura K., Tamura H., Shiraishi S. and Takehara Z.-I., J. Electroanal. Chem. 394 (1995), 49. 104. Aurbach D., Weissman I., Schechter A. and Cohen H., Langmuir 12 (1996), 3991. 105. Sloop E., Pugh J. K., Wang S., Kerr J. B. and Kinoshita K., Electrochem. Solid State Lett. 4 (2001), A42. 106. Database, XI SpecMaster, ed. by Vincent C. (XPS International, Kawasaki, Japan, 1998). 107. Schmidt M., Heider U., Kuehner A., Oesten R., Jungnitz M., Ignat’ev N. and Sartori P., J. Power Sources 97-98 (2001), 557. 108. Gnanaraj J. S., Levi M.D., Gofer Y. and Aurbach D., J. Electrochem. Soc. (2002), in press. 109. Sawyer D. T., Chlericato G., Angelis C. T., Nanni E. J. and Tsuchiya T., Anal. Chem. 54 (1982), 1720. 110. Aurbach D. and Chusid O., J. Electrochem. Soc. 140 (1993), L155. 111. Aurbach D., Chusid O., Gofer Y., Watanabe M., Momma T. and Osaka T., J. Power Sources 97-98 (2001), 632. 112. Osaka T., Komaba S. and Liu X., in Nonaqueous Batteries, ed. by Aurbach D. (Marcel Dekker, New York, 1999), Ch. 7, p. 411. 113. Nishimoto A., Watanabe M., Ikeda Y. and Kohjiya S., Electrochim. Acta 43 (1988), 1177. 114. Aurbach D., Markovsky B., Gamolsky K., Levi E. and Ein-Eli Y., Electrochim. Acta 45 (1999), 67. 115. Peled E., Golodnitsky D. and Penciner J., in Handbook of Battery Materials, ed. by Besenhard J. O. (Wiley-VCH, New York, 1999), Ch. 6, p. 419. 116. Winter M. and Besenhard J. O., in Handbook of Battery Materials, ed. by Besenhard J. O. (Wiley-VCH, Weinheim, New York, 1999), Ch. 5, p. 383. 117. Dahn J. R., Sleigh A. K., Shi H., Way B. M., Weydanz W. Y., Reimers J. N., Zhong Q. and von Sacken U., Lithium Batteries: New Materials,

138

118. 119. 120. 121. 122. 123. 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135.

136. 137.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Developments and Perspectives, ed. by Pistoia G. (Elsevier-North Holland, Amsterdam, 1993), p. 1. Fischer J. E., Chem. Innov. 30 (2000), 21. Zang Z. H. and Wu H. Q., Mater. Chem. Phys. 71 (2001), 7. Aurbach D., Ein-Eli Y., Markovsky B. and Zaban A., J. Electrochem. Soc. 142 (1995), 2882. Aurbach D., Levi M. D., Levi E. and Schechter A., J. Phys. Chem. B 101 (1997), 2195. Ein-Eli Y., Thomas S. R. and Koch V. R., J. Electrochem. Soc. 143 (1996), L195. Sasaki Y., Handa M., Sekiya S., Kurashima K. and Usami K., J. Power Sources 97-98 (2001), 561. Aurbach D., Markovsky B., Gamolsky K., Koltypin M., Heider U. and Schmidt M., ITE Battery Letters, Vol. 2, No. 3, B46 (2001). Jehoulet C., Biensan P., Bodet J. M., Broussely M., Moteau C. and Tessier-Lescourret C., Proc. Electrochem. Soc. 97-98 (1997), 974. Aurbach D., Gamolsky K., Markovsky B. and Gofer Y., Electrochim. Acta 47 (2002), 1423. Shu X. Z., McMillan R. S. and Murray J. J., J. Electrochem. Soc. 140 (1993), 922. Aurbach D. and Cohen Y., J. Electrochem. Soc. 143 (1996), 3525. Cohen Y. S., Cohen Y. and Aurbach D., J. Phys. Chem. B 104 (2000), 12282–12291. Inaba M., Siroma Z., Funabiki A., Abe T., Mizutani Y., Asano M. and Ogumi Z., Langmuir 12 (1996), 1535. Jeong S. K., Inaba M., Abe T. and Ogumi Z., J. Electrochem. Soc. 148 (2001), A989. Allliata D., Kotz R, Novak P. and Siegenthaler H., Electrochem. Comm. 2 (2000), 436. Morigaki K., Fujii T. and Ohta O., Denki Kagaku 66 (1998), 1122. Aurbach D., Koltypin M. and Teller H., Langmuir (2002), in press. Wakihara M., Li G. and Ikuta H., Li-Ion Batteries, Fundamental and Performance, ed. by Wakihara M. and Yamamoto O. (J. Wiley-VCH, New York, 1998), Ch. 2, p. 26. Ein-Eli Y., Howard W. F., Lu S. H., Mukerjee S., McBreen J., Vaughey J. and Tackeray M. M., J. Electrochem. Soc. 145 (1998), 1238. Kanamura K., J. Power Sources 81-82 (1999), 123.

Identification of Surface Films on Electrodes

139

138. Quinlan F. T., Sano K., Willey T., Vidu R., Tasaki K. and Stroeve P., Chem. Mater. 13 (2001), 4207. 139. Aurbach D., Gamolsky K., Markovsky B., Salitra G. and Gofer Y., J. Electrochem. Soc. 147 (2000), 1322. 140. Aurbach D., Markovsky B., Rodkin A., Cojocaru M., Levi E. and Kim H. J., Electrochim. Acta 47 (2002), 1899. 141. Peled E., Cohen R., Melman A. and Lavi Y., J. Electrochem. Soc. 139 (1992), 1836. 142. Peled E. and Straze H., J. Electrochem. Soc. 124 (1977), 1030. 143. Aurbach D., Schechter A., Moshkovich M. and Cohen Y., J. Electrochem. Soc. 148 (2001), A1004. 144. Aurbach D., Gizbar H., Schechter A., Chusid O., Gottlieb H. E., Gofer Y. and Goldberg I., J. Electrochem. Soc. 149 (2002), A115. 145. Aurbach D., Turgeman R., Chusid O. and Gofer Y., Electrochem. Com. 3 (2001), 252.

CHAPTER 3

SPECTROSCOPIC STUDIES OF SOLID-ELECTROLYTE INTERPHASE ON POSITIVE AND NEGATIVE ELECTRODES FOR LITHIUM ION BATTERIES ZHAOXIANG WANG,* XUEJIE HUANG AND LIQUAN CHEN† Laboratory for Solid State Ionics, Institute of Physics, Chinese Academy of Sciences, P O Box 603, Beijing 100080, China E-mail: *[email protected]; †[email protected]

1

Introduction

This chapter is focused on the spectroscopic investigation of solid-electrolyte interphase (SEI) layers on negative and positive electrodes for lithium ion batteries in the authors’ laboratory. Based on the spectroscopic observations by Fourier transform infrared (FTIR) absorption, Raman and surface enhanced Raman scattering (SERS), and X-ray photoelectron spectroscopy (XPS) as well as scanning electron microscopy (SEM) imaging and high-resolution transmission electron microscopy (HRTEM), the structures and compositions of the SEI layers on the electrodes are characterized. Some important and interesting experimental phenomena are explained such as the capacity loss of the nanometer-sized SnO (nano-SnO) anode in the first cycle, the improved electrochemical performance of surface-modified LiCoO2 cathode material, and the SERS effect on discharged Ag foil. FTIR and Raman spectroscopy of nano-SnO anodes at different discharge states in rechargeable lithium batteries have been investigated. The structure and the composition of the SEI layer are characterized with HRTEM and FTIR spectroscopy, respectively. It is found that irreversible reduction of SnO and electrolyte decomposition lead to capacity loss of the metal oxide anodes in the first cycle. Similar to the SEI layer on carbonaceous anode materials, the main components in the SEI layer on discharged nano-SnO electrode include Li2CO3 and ROCO2Li. The reduction of SnO anode is determined to occur above 1.2V and last until rather low voltages. The formation of Li2CO3 dominates the + solvent reduction above 0.9V (vs Li/Li ) while the formation of ROCO2Li mainly takes place below 0.9V.

140

Spectroscopic Studies of Solid-Electrolyte Interphase

141

SERS signals are detected on discharged Ag electrodes in Li/Ag cells with micro-Raman spectrometer. The successful observation of the SERS signal is attributed to the formation of nanometer Li-Ag alloy particles on the Ag electrode surface during discharge and the interaction between the electrolyte and the Ag electrode. With SERS technique, Li2CO3 and LiOH·H2O are determined to be the main components of the SEI layer on the Ag electrode with some moisture in the electrolyte. Comparative study has been carried out by FTIR and XPS spectroscopy to understand the electrochemical performance improvement of nanometer-sized magnesium oxide (nano-MgO) modified commercial LiCoO2 (MgO/LiCoO2) cathode materials for lithium ion batteries. It is found that the configurations of the solvent molecules of the electrolyte are very sensitive to the disturbance of nano-MgO and commercial LiCoO2. Modifying the surface of commercial LiCoO2 particles with nano-MgO can suppress electrolyte decomposition on the electrode surface at high potentials significantly. In addition, by hindering the formation of oxygen with higher oxidizing power as well as by physically separating the electrolyte from direct contact with the active cathode material, surface coating suppresses the interaction between LiCoO2 and the electrolyte at the uncharged state and alleviates the electrolyte decomposition at charged states.

2

SEI on Tin Oxide Anode in Various Electrolytes

Commercial lithium ion batteries have been using graphite intercalation 1 compounds (GICs) as the active anode materials. However, the available capacities of these materials are limited to the theoretical value of 372 mAh/g based on the saturated LiC6 structure of GIC. Therefore, composite anode materials with higher theoretical capacities become the research focus of many scientists in recent years. Of these new types of anode materials, various metal 2-6 7-14 oxides, nanometer-sized metal particles and fibers, carbon/non-carbon 15-21 22-27 composite materials and Li-alloyed materials seem more promising to be commercialized and attract more attention. Composite tin oxides were first 2, 3 proposed as an active anode material by Fujifilm Corporation. Their theoretical specific capacity is 875 mAh/g, more than twice that of graphite. The actual reversible capacities of these materials are above 550 mAh/g in the first 6 4 cycle. This laboratory and Dahn’s group studied the anode reaction of SnO2 and found that the lithium storage mechanisms of these materials are completely

142

Lithium-Ion Batteries: Solid-Electrolyte Interphase

different from that of GICs. A two-step mechanism was suggested for the reaction of lithium with various tin oxides based on the Raman experimental results, ex situ and in situ X-ray diffraction (XRD): Lithium first reacts irreversibly with tin oxides to produce amorphous Li2O and metallic Sn; then 4, 6 various Li-Sn alloys are formed. Such mechanism indicates that the reversible capacity of the metal oxide anodes is involved with the alloying process instead of intercalation. Many metal oxide anode materials have similar lithium storage mechanism to this. Part of the irreversible capacity comes from the reaction of the metal oxide electrode, which can be generalized as follows: 2yLi + MxOy +2ye− → yLi2O + xM + zLi + M + ze− → LizM +

(1) (2)

Reaction (1) is irreversible while Reaction (2) is theoretically reversible. Clearly the capacity loss in Reaction (1) is higher for SnO2 than for SnO. Therefore, SnO is superior to SnO2 because less lithium is consumed to reduce the tin oxide and a battery with SnO anode can have a higher capacity. Suppose that lithium reacted with SnO to form Li4.4Sn, the irreversible capacity is calculated to be 392 mAh/g and the reversible capacity can be as high as 875 mAh/g. However, the updated experiments show that the capacity loss in the first cycle of SnO anodes is much more than 392 mAh/g. This means that some side reactions must have taken place in the first cycle. It is well known that graphite is unstable in some aprotic electrolytes. For instance, when propylene carbonate (PC) is used as a solvent, the co+ intercalation of solvent molecules and the Li ions will lead to the exfoliation of 28, 29 graphite layers. Only in some selected electrolyte systems such as LiPF6 in EC/DEC (EC for ethylene carbonate and DEC for diethyl carbonate), can graphite show better cycling behavior. Solvent decomposition on the surface of conductive carbon or lithium electrodes will lead to the formation of a passivating layer. Peled named this layer as solid electrolyte interphase 30, 31 (SEI). It is an ionic conductor but electron insulator, mainly composed of Li2CO3 and various lithium alkylcarbonates (ROCO2Li) as well as small amounts of LiF, Li2O, and nonconductive polymers. These compounds have been 28, 32-44 detected on carbon and Li electrodes in various electrolyte systems. Therefore, it would be an interesting question whether semiconductive nanoSnO anode is also sensitive to electrolyte and electrolyte decomposition takes place on it. This section will characterize the structures and compositions of the

Spectroscopic Studies of Solid-Electrolyte Interphase

143

SEI layers on nano-SnO anodes in various electrolytes and the reduction sequence of SnO and electrolyte in the first cycle.

2.1 Sample Preparation and Instrumental Nano-SnO was obtained by ball-milling commercial SnO powder in Ar atmosphere (the weight ratio of agate balls to SnO was 10:1) in the shear mode. After about 12 hours the size of the SnO particles was reduced to about 200 nm, appropriate for the HRTEM observation and helpful for the electrode reaction. Nano-SnO can also be prepared by decomposing the sol-gel reaction (SnCl2+H2C2O4 in alcohol) product, superfine SnC2O4 powder, at 400°C for 6 hours in argon. All the solvents for the electrolytes, PC, EC, DEC, dimethyoxy ethane (DME), and dimethyl carbonate (DMC), were purified by the traditional 45 method. The electrolyte solutions, 1 M LiClO4 in PC/DMC, 1 M LiClO4 in PC/DME, 1 M LiClO4 in EC/DMC and 1 M LiPF6 in EC/DEC (all by 1:1 v/v), were purified prior to use. The working electrodes for the electrochemical performance evaluation of the test cells were prepared by pressing into pellets the mixture of the ballmilled SnO (95% w/w) and polyvinylidene difluoride (PVDF, 5% w/w), and drying them at 80°C in vacuum for 24 hours. The counter electrode was lithium foil. Celgard 2400 microporous membrane was used as the separator. Li/SnO cells were assembled in an argon-filled glove box (MBraun). The cells were 2 cycled at a constant current density of 0.2 mA/cm . The voltage window was set + between 0.0 V and 2.0 V (vs Li/Li ). Cyclic voltammometry tests were carried out on CHI660A Electrochemical Workstation. The assembly of the cells for the IR spectroscopic measurements was the same as for the above cells except that their working electrodes were prepared by pressing the nano-SnO powder into pellets without any additives. Then, these cells were discharged to the preset 2 voltages galvanostatically (0.1 mA/cm ) and kept there (potentiostatically) for 48 hours. After that, the reacted nano-SnO particles was ground together with KBr and pressed into pellets in the glove box. The pellets were then sealed airtightly in containers respectively. All the above operations were carried out in argon atmosphere unless specified. When everything was ready for the Bio-Rad FTS 6000 FTIR spectrometer, the container was opened and the sample was transferred into the vacuum chamber of the instrument. The chamber was vacuumed immediately. The exposure time of the sample to air was less than 10 seconds.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

144

The discharged nano-SnO powder was scratched off the electrode and stored in benzene in a glass tube. The tube was sealed air-tightly and the mixture was dispersed ultrasonically. A droplet of the mixture was placed on a copper grid and transferred into the vacuum chamber of JEOL2010 transmission electron microscope within 1 min. The microscope was operated at 200 KV for observing the microstructure of the discharged nano-SnO particles.

2.2 Capacity Loss and Electrolyte Decomposition in First Cycle Figures 1 and 2 show typical cycling profiles of nano-SnO anode in different electrolytes. The discharge capacity is 1060 mAh/g but its charge capacity is only 840 mAh/g. Considering that the capacity loss due to SnO reduction is + only 392mAh/g according to Eq (1), some Li ions must have been consumed on the formation of SEI layer. This suggestion is supported with the irreversible + voltage plateau at around 1.0V vs Li/Li in the first cycle. In the subsequent + cycles, the irreversible plateau at 1.0V vs Li/Li becomes very short and gradually disappears with cycling.

2.0

Voltage (V)

1.5

1.0

0.5

0.0 0

200

400

600

800

1000

1200

Capacity (mAh/g) Figure 1 Voltage profiles of the first five cycles of a Li/nano-SnO cell with 1 M LiPF6, EC-DEC (1:1 v/v) as the electrolyte.

145

Spectroscopic Studies of Solid-Electrolyte Interphase

1.4 1.2

Voltage (V)

1.0 0.8 0.6 0.4 0.2 0.0 0

200

400 600 800 1000 Capacity (mAh/g)

1200

1400

Figure 2 Voltage profiles of the first five cycles of a Li/nano-SnO cell with 1 M LiClO4, PC-DME (1:1 v/v) as the electrolyte.

2.3

HRTEM Study of SEI Structure on Nano-SnO Surface

Figure 3 shows the HRTEM images of nano-SnO particles before (Figure 3a) and after (Figure 3b) discharge. The diameters of the nano-SnO particles are about 200nm with slight agglomeration. After discharge, a perfect shell structure is observed on the surface of each nano-SnO particle. The thickness of 28 the shell is estimated 30-40 Å, consistent with the calculated value of 20- 45 Å. Such a passivating layer is thick enough to prevent electron tunneling 28 28 effectively and protect the electrolyte from further reduction. Dahn et al. and 39 Aurbach et al. used FTIR to show that the SEI layer on the surface of carbon consists of Li2CO3 and ROCO2Li. Here our observation indicates that SEI layer can be formed not only on the surface of a conductive electrode, but also on a semiconductive electrode.

146

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Figure 3 The high resolution transmission electron microscopy (HRTEM) images of nano-SnO particles before (a, left) and after (b, right) discharge. Reproduced from [46] with permission of The Electrochemical Society Inc.

2.4

Identification of Li2CO3 and ROCO2Li on Nano-SnO Anodes

In order to determine the composition of the SEI layer on nano-SnO anode, FTIR spectra of nano-SnO (Figure 4), the electrolyte (Figure 5) and the nanoSnO electrode at various discharge states (Figure 6) are recorded. Two peaks are 1 1 observed at 515 cm− and 334 cm− in nano-SnO before electrochemical 47 1 treatment, consistent with the previous report. Peaks at 1020, 849, 559 cm− in 36 Figure 5 are attributed to LiPF6 according the litereature. Other labelled peaks have been attributed to EC and DEC and listed in Table 1.

147

Spectroscopic Studies of Solid-Electrolyte Interphase

1.0

515 1620

0.4

334

0.6 3435

Absorbance

0.8

0.2

0.0 5000

4000

3000

2000

1000

-1

Wavenumber (cm ) Figure 4 IR spectrum of nano-SnO before electrochemical treatment.

0.0 4000

3000

2500

2000

1500

1000

559

849

1173

3500

1020 974 903 777 719

0.2

1967

3001

2965

0.4

1636 1558 1483 1396 1263

0.6

2836

Absorbance

0.8

1080

1807 1776

1.0

500

-1

Wavenumber (cm ) Figure 5 Typical FTIR spectrum of the electrolyte (1M LiPF6 in EC/DEC (1:1 v/v).

148

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Table 1 Comparison of FTIR peaks of nano- SnO anodes at different discharge states in lithium rechargeable batteries: Nano-SnO /1M LiPF6, EC-DEC(1:1)/Li.

Positions (cm− ) and Assignments 1

States of anode 1 M LiPF6, EC-DEC (1:1)

Discharged to 1.2 V

EC, DEC

Li2CO3

ROCO2Li

1020w, 849s, 559s

3001w, 2965w, 2836w, 1967w, 1807s, 1776s, 1636w, 1558w, 1483m, 396m, 1263m, 1173s, 1080s, 974m, 903w, 777m, 719m 1809s, 1782s, 1483m, 396m, 1193m, 1078s, 977w, 903w, 783m, 719w

LiPF6

1508m, 1431m, 867s

2935w, 2852vw 1643w, 1408w,

847s, 559s

1251w Discharged to 0.9 V

1807s, 1778s, 1483m, 1195s, 1082s, 975m, 781m, 719w

1519m, 1427m, 868m

2935m, 2885w, 2850w, 1635m, 1408m, 309m,

843s, 560s

Discharged to 0.7 V

1805w, 1752m, 1397m 1195w, 1080m, 979m,

1449w

2980m, 2929m, 2874w, 2850w

848s, 559m

869m

780m

1642s, 1303m, 1253m, 826m 2985m, 2922m 2850w, 1652s, 1404s, 1305s 1246m, 826m

845s, 560s

1510m, 1435m, 868m

2981w, 2920s, 2850w, 1635s, 1406m, 1298s, 1247m, 826m

851s

1510w, 1435w, 863m

2980w, 2945w, 2889w, 2850w 1665s, 1653s, 1404s, 1319s, 1306s, 1067s, 1005m, 831s

851s, 561s

Discharged to 0.2 V

1805s, 1772s, 1192s, 1080s, 975m, 898w, 779m, 719m

1506w,

Discharged to 0.0 V

1805m, 1772s, 1193m, 1080s, 975m, 781m, 719w

Discharged to 0.0 V, then recharged to 2.0 V

1807m, 1778s, 1194m, 1082s, 978m, 901w, 783m, 723w

862m

Note: w: weak; m: middle; s: strong; vw: very weak.

560s

149

Spectroscopic Studies of Solid-Electrolyte Interphase

1.0

1.0

Nanometer SnO discharged to 1.2V

1500

1000

500

1427

486

1082 1195

1500

1000

559 525 862 851

1082

1005 978 901

1115 1194

851

783 723

0.6

1067

1807 1778

2980 2945 2889

2850

895

0.7

863

0.8 871

Absorbance

780

0.9

0.7 0.6

0.5

0.55 920 900 880 860 840 820 800

0.5

2000

500

1404 1319 1306

508 559

1123 1080

979

0.8

0.6

2500

1309 1408

975 Nanometer SnO discharged to 0.0V then charged to 2.0V

0.9

851

1449 1397 1303 1273 1253 1195

1642 1752 1805

848

826

0.60

3000

2000

-1

1.0

2980 2929 2874 2850

869

Absorbance

0.70

3500

719

2500

Wavenumber (cm )

Nanometer SnO discharged to 0.7V

0.65

559

1778 1807

3000

1665 1653

2000

-1

0.7

842

843

Nanometer SnO discharged to 0.9V

0.0 3500

Wavenumber (cm )

0.8

781

783

0.2

1510

2500

1635 1519 1483

0.4

831

3000

868

0.0 920 900 880 860 840 820 800 2935 2885 2850

Absorbance

559 501 433

866 851

0.2

719

1193 1078 977

847

0.4

0.6

0.0 920 900 880 860 840 820 800

0.0 3500

890

0.6

0.8 1643 1508 1483 1431

1809

2935 2852 867

0.4 890

Absorbance

0.4

0.2 0.2

0.8

1782

0.6

1500

1000

500

920 900 880 860 840 820 800

3500

-1

3000

2500

Wavenumber (cm )

2000

1500

1000

500

-1

Wavenumber (cm )

0.35 920

0.35 3500

559 525 1005 978 901

1067

862 851

1082

1404 1319 1306 1194

851

783 723

0.6

863

0.7

1115

1510

1807 1778

2980 2945 2889

2850

0.8

0.6

781

0.9 831

0.7

895

719

0.8

871

Absorbance

862

1247 1193 1080 975

851

0.9

559

1510 1435 1406 1298

1772 1805

1635

2981 2920 2850

0.40

826

0.40

0.45

868

0.45

880

Absorbance

0.50

Nanometer SnO discharged to 0.0V then charged to 2.0V

1.0

Nanomweter SnO discharged to 0.0V

1665 1653

0.55

0.5 900

3000

880

860

840

2500

820

0.5

800

2000

1500 -1

Wavenumber (cm )

1000

500

3500

920 900 880 860 840 820 800

3000

2500

2000

1500

1000

500

-1

Wavenumber (cm )

Figure 6 FTIR spectra of nano-SnO anode discharged to various voltages in Li/nano-SnO cells. The cell was kept at the shown voltage for over 48 hours. In the insets, the scattered dots represent the experimental points and the lines are Gaussian fit components. 1 M LiPF6 in EC/DEC (1:1 v/v) are used as the as the electrolytes for all the cells.

In each of the above spectrum, Li2CO3 can easily be recognized by 1 1 1 2, 31, 32, 47 observing its characteristic bands at 1510cm− , 1435 cm− and 868cm− . However, it is difficult to identify ROCO2Li exactly because ROCO2Li actually represents a series of lithium alkyl carbonates. Their peak positions depend on the structure of the R group and are determined by the composition of the electrolyte, the surface properties of the electrode and the reduction processes on it.

150

Lithium-Ion Batteries: Solid-Electrolyte Interphase

As the electrodes were prepared with pure SnO pellet, all vibrational bands of the organic functional groups should be attributed to the SEI or the electrolyte residue on the electrode surface. Therefore, comparing the FTIR spectra of nano-SnO discharged to different voltages (Figure 6), the bands at 1 1 1 1 1 2980 cm− , 2920-2935 cm− , 2850 cm− , 1635-1665 cm− , 1408-1404 cm− , −1 −1 −1 −1 1319-1306 cm , 1246-1253 cm , 1067 cm and 826-830 cm should also be 32, 36, 38-41 attributed to ROCO2Li, based on the references. It is also found that the peaks attributed to ROCO2Li above 0.9 V are different from those below 0.9 V. As seen in Table 1, a new band appears 1 around 2980-2985 cm− as well as the peaks related to the C-H stretching bands 1 at 2935, 2885 and 2852 cm− above 0.9 V. In addition, band of CO3 bending of −1 ROCO2Li at 826 cm is present below 0.9 V. This implies that EC and DEC may be reduced at different potentials and different ROCO2Li are formed though the compounds cannot be exactly identified based on the available knowledge at present.

2.5

Formation of Li2CO3 and ROCO2Li on Nano-SnO Anodes

Li2CO3 can be detected at all discharge stages as shown in Figure 7. However, the relative intensities of the peaks related to Li2CO3 become weak when the electrode is further discharged to 0.7 V. In contrast, the relative intensities of the peaks belonging to ROCO2Li become stronger than between 1.2 V and 0.9 V. This means that the formation of Li2CO3 begins above 1.2 V and lasts until about 0.70 V. The formation of ROCO2Li occurs mainly below 0.9 V. These results suggest that Li2CO3 can be a direct reduction product of EC and DEC. 40 Aurbach et al. studied the reduction products of electrolyte on graphite electrodes. They detected ROCO2Li and traces of Li2CO3 on the electrode with 1 M LiPF6 dissolved in EC-DEC(1:1). They suggested that ROCO2Li is formed by radical termination reactions while Li2CO3 is by the reaction of ROCO2Li with trace water. However, they did not exclude the possibility of a direct 32, 35, 39, 40 reduction from EC or DEC. They also showed that Li2CO3 is one of the 38 CO2 reduction products. Based on the results of electron energy loss (EEL) and 41, 42 FTIR spectroscopy, Naji et al. proposed a mechanism of EC reduction on the surface of graphite electrode in the presence of LiClO4. They believed that Li2CO3 is formed above 0.8 V by a direct two-electron reduction of EC. Then free radical termination reactions lead to the formation of lithium alkyl carbonate below 0.8 V.

151

Spectroscopic Studies of Solid-Electrolyte Interphase 41, 42

Our results seem to support the reaction scheme of Naji et al. This clearly indicates that the two-electron reduction and free radical termination reaction of EC and DEC may occur on SnO surface at the same potential. Above 0.9 V, two-electron reduction becomes more competitive and the solvent reduction is controlled by the formation of Li2CO3. Below 0.9 V, the free radical termination reaction controls the main reduction process. The variation of the peaks related to Li2CO3 and ROCO2Li may imply that the inner layer of the SEI film is Li2CO3 while the outer layer that contacts the electrolyte is composed of ROCO2Li. 0.5

394

0.2

0.25 700 650 600 550 500 450

0.44 0.42

0.7

527

0.8

483

417

(f)

587

460

0.46

0.9

520

450

0.30

0.48

589 560

601

0.35

560 533

0.40

700 650 600 550 500 450

561

0.1 0.0 700 650 600 550 500 450 700 650 600 550 500 450 0.45 0.50 (e) (d) Absorbance

466

552 604

0.2

560

614

559

0.3

0.4

512

502

(c)

0.4 605

Absorbance

(b)

414

512

0.6 (a)

0.6

0.40 0.38 700 650 600 550 500 450 -1

0.5 700 650 600 550 500 450

Wavenumber (cm )

Figure 7 Selected FTIR spectra of nanometer SnO anode at different discharge states in lithium batteries: (a) Discharged to 1.2 V; (b) Discharged to 0.9 V; (c) Discharged to 0.7 V; (d) Discharged to 0.2 V; (e) Discharged to 0.0 V; (f) Discharged to 0 V and then recharged to 2.0 V. The open circles are experimental points, dotted lines are fit curves by Gaussian function.

The above discussion shows that the SEI film is a mixture of Li2CO3 and ROCO2Li. Its formation leads to the extra capacity loss as well as that due to nano-SnO reduction in the first few cycles. After that, the formation of the SEI layer is completed and the subsequent alloying and de-alloying reactions are theoretically reversible. However, most of the alloying reactions will actually

152

Lithium-Ion Batteries: Solid-Electrolyte Interphase

lead to significant volume variation and probably electrode cracking and crumbing during cycling. As a result, many new surface sites are created and react with lithium to form a surface passivating layer, resulting in further capacity fading. In addition, depending on the properties of the electrolyte and the surface properties of the electrode, the SEI layer is probably not stable and new dissolution and deposition may occur, resulting in continuous capacity fading.

2.6

Question: What Is the Reduction Sequence of SnO and Electrolyte?

Characteristic peaks of SnO are observed at 515 cm− and 334 cm− in the IR 1 spectrum (Figure 4). However, the 515 cm− peak becomes unrecognizable after 1 discharge due to the strong background around 500 cm− (Figure 7). It has been 32, 48 reported that the Li-O stretching vibration is also located in this area. Thus it is difficult to determine the origin of the peak due to the coexistence of SnO, Li2O, Li2CO3 and ROCO2Li in the anode at present. Cyclic voltammogram of SnO anode indicates that only one irreversible + reduction peak (1.0-0.35 V vs Li/Li ) is detected in the first cycle (Figures 8 and 9). So it is difficult to recognize which reaction occurs before the other, the replacement reaction or the electrolyte decomposition reaction, based on these results. Some other experimental methods are necessary for an accurate answer to this interesting question. 1

1

1.0 1st cycle 2nd cycle 3rd cycle

Current (A)

0.5

0.0

-0.5

-1.0 0.0

0.5

1.0

1.5

Potential (V) Figure 8 Cyclic voltammogram of nano-SnO electrode in 1 M LiPF6, EC/DEC (1:1 v/v) at scanning rate of 0.1 mV/s.

153

Spectroscopic Studies of Solid-Electrolyte Interphase 1.5 1st cycle 2nd cycle 3rd cycle

1.0

Current (mA)

0.5 0.0 -0.5 -1.0 -1.5 -2.0

0.0

0.5

1.0

1.5

Potential ( V) Figure 9 Cyclic voltammogram of nano-SnO electrode in 1 M LiClO4, PC/DME (1:1 v/v) at scanning rate of 0.1 mv/s.

Figure 10 shows the Raman spectra of pure nano-SnO and nano-SnO discharged to different voltages. Three bands are observed in pure nano-SnO 1 (the left panel of Figure 10). The 112 and 210 cm− bands are tentatively (1) 49 attributed to the Eg and A1g vibrations of α-SnO according to Lebellac et al. −1 The 85 cm bands might be due to instrumental error as it appears in other samples on the same instrument. During discharging, the relative intensities of 1 the 112 and 210 cm− bands decrease with lowering voltages (the middle panel of Figure 10). This means that the decomposition or reduction of nano-SnO anode begins at about 1.2 V and completes at about 0.9 V. However, the Raman bands of the reduction product(s) cannot be detected due to the weak signal of the sample and the strong interference of the window glass of the test cell. Taking the discharged nano-SnO sample out of the cell, ex situ Raman spectrum 1 is recorded (the right panel of Figure 10). The 110 cm− band should not be 1 assigned to nano-SnO because the other characteristic band (at 210 cm− in pure −1 nano-SnO) cannot be detected in this sample. Therefore, the 110 cm band is attributed to metallic Sn, agreeing with the Raman spectrum of pure metallic Sn. 1 1 1 The bands at 222 cm− , 252 cm− and 352 cm− are due to the interference of the + 1 plasma lines of the Ar laser (488nm excitation). All the other bands, 132 cm− , −1 −1 −1 158 cm , 182 cm and 200 cm , are attributed to LixSn (x≤4.4) alloys. Based on the above experimental results, it can be determined that the replacement reaction of nano-SnO comes before the decomposition of the electrolyte during discharge.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

154 60

200000 0.016

50

132

112

0.014

0.012

6 5

100000

4 0.006 3 210

2 1

850

900

950 1000

0.000 1000

Wavenumber (cm-1)

800

600

400

200

252

0.002

0 0 50 100150200250300350 800

110

50000

0.004

10

0 100

200

352

0.008

158

20

0.010

Intensity

Intensity

Intensity

30

150000

7

182 200 222

210

8

85

40

110

0.018

300

400

500

600

Wavenumber(cm-1)

-1

Wavenumber (cm )

Figure 10 Raman spectra of pure nano-SnO (left, 1064 nm excitation 1), in situ Raman spectra of nano-SnO discharged to various voltages (middle, 1064 nm excitation; 1. nano-SnO; 2-6. nanoSnO discharged to 1.2 V, 0.9 V, 0.7 V, 0.2 V and 0.0 V, respectively; 7. discharged to 0.0 V and then recharged to 2.0 V; 8. window glass of the cell) and ex situ Raman spectrum of nano-SnO discharged to 0.0 V (right, 488 nm excitation).

2.7

Electrolyte-Dependent SEI Compositions

The reversible capacities of SnO anode and its capacity losses in the first and the fifth cycles in different electrolytes have been listed in Table 2. The capacity losses in electrolyte LiPF6+EC/DEC on the first and fifth cycles are much lower than in the other electrolytes.

Table 2 Comparison of the reversible and irreversible capacities of SnO anode in the first and the fifth cycles in four kinds of electrolytes.

Electrolyte

1st cycle capacity (mAh/g)

5th cycle capacity (mAh/g)

reversible

irreversible

reversible

irreversible

LiPF6+EC/DEC

840

220

648

44

LiClO4+EC/DMC

765

450

453

60

LiClO4+PC/DME

690

660

244

74

LiClO4+PC/DMC

650

485

361

114

155

Spectroscopic Studies of Solid-Electrolyte Interphase

The cyclic voltammograms (CVs) of SnO electrodes in 1M LiPF6+EC/DEC and in 1 M LiClO4+PC/DME have been shown in Figure 8 and + Figure 9 respectively. A reduction peak is observed around 0.8 V vs Li/Li in the first cycle in both figures. This peak becomes weak obviously in the second cycle. In the third cycle, it almost disappears. However, significant differences can also be observed in these two figures. Firstly, the CV profiles of the second and the third cycles in Figure 8 almost overlap with each other, implying good reversibility of the anode after the first cycle. In Figure 9, however, the CV curves of the second and the third cycles are still quite different, reflecting the poor reversibility of the nano-SnO anode in 1M LiClO4 + PC/DME electrolyte. Secondly, the staged LixSn alloying processes (after the first cycle) in Figure 8 is very obvious. In Figure 9, nevertheless, these stages become unrecognizable 32 in the third cycle. Aurbach et al. reported that the R of ROCO2Li on carbon varies from methyl to butyl, depending on the electrolyte. Therefore different solvents, such as PC, EC, DEC, DMC and DME, lead to different reduction 28, 32-44 products on the surface of carbon or lithium electrodes. For example, EC 37, 39-41 can be reduced to (CH2OCO2Li)2 and DEC can be reduced to 39, 40 CH3CH2OCO2Li. As both cells use the same nano-SnO anode material, their electrochemical performance difference should definitely be attributed to the different properties of the SEI layers, including composition, stability, density, thickness, and conductivity.

471

509

559 837 # 806

* 868

1084

1030

# 1261

* 1431

# 1308

* 1508

# 1651

0 .8

1771

0 .9 1805

A b s o rb a n c e

1 .0

0 .7 * L i2C O 3 # R O C O 2L i 0 .6 1800

1600

1400

1200

1000

W a v e n u m b e r (c m

800 -1

600

400

)

Figure 11 FTIR spectrum of nano-SnO discharged to 0.7 V in Li/(1M LiPF6 , EC-DEC) /SnO cell. Reproduced from [50] with permission of Elsevier Science.

1800

1600

1400

1200

1109

505

1000

W a v e n u m b e r (c m

629

* 864 # 803

0 .3 5

* L i2C O 3 # R O C O 2 L i ( p o s s ib le )

1047

# 1261

1527 * 1512 1443 * 1433

# 1630

0 .4 0

1788

A b s o rb a n c e

0 .4 5

# 991

1121

Lithium-Ion Batteries: Solid-Electrolyte Interphase

156

800 -1

600

400

)

Figure 12 FTIR spectrum of nano-SnO discharged to 0.8 V in Li/(1M LiClO4, PC/DME) /SnO cell. Reproduced from [50] with permission of Elsevier Science.

Figure 11 shows the IR spectrum of SnO anode discharged to 0.7 V in 1 LiPF6 + EC/DEC. The peaks at 1508, 1431 and 868 cm− are attributed to Li2CO3 39, 40, 42, 47, 51, 52 −1 and the peaks at 1651, 1308 and 837 cm are attributed to ROCO2Li. The other peaks are related to the residual EC, DEC and LiPF6 on the electrode. The FTIR spectrum of SnO anode discharged to 0.8 V in LiClO4+PC/DME is 1 shown in Figure 12. The peaks at 1121, 1109, 1090, 629 cm− are attributed to 47 51 −1 LiClO4 while the peaks at 1630 and 1261 cm belong to DME. The peak at 1 49 1786 cm− is characteristic of PC. Although a shoulder peak of PC is present at 39, 47 −1 51 −1 849 cm , the 864 cm band of Li2CO3 can still be well recognized. Other 2− 2 −1 peaks corresponding to Li2CO3 (CO3 stretching at 1433 cm and CO3 − −1 stretching at 1512 cm ) are coupled with the bands of PC at 1443 and 1 1 1 1527 cm− . Due to strong peaks at 1630 and 1261 cm− for DME, 991 cm− for PC, it is difficult to determine the existence of lithium alkylcarbonates 51 (ROCO2Li). Anyway, Li2CO3 has already been identified to be one of the main reduction products at 0.8V in the first discharge. It indicates that a passivating film can also be formed on the SnO anode surface in PC-based electrolytes. As the reduction of the SnO anode has been finished in the first discharge, the capacity loss in the subsequent cycles should be attributed to the compatibility of the electrode with the electrolyte and the stability of the SEI layer towards the electrolytes. Since it is not possible for the PC molecules and + the Li ions to co-intercalate into the oxide anode, it can be suggested that the SEI layer on the surface of SnO is better developed and more stable with cycling in LiPF6+EC/DEC electrolyte than in the PC-based ones according to the above electrochemical behaviors of SnO anode. Larger capacity loss of SnO

Spectroscopic Studies of Solid-Electrolyte Interphase

157

anode in PC-based electrolyte at 1.0V plateau (Fig.2) means thicker surface film than in EC-based electrolyte. As a result, the performance of the battery degrades with cycling.

2.8

Conclusions

The above results and discussion indicate that the extra capacity loss in the first cycle in tin oxide anode is mainly due to the electrolyte decomposition in the first cycle. HRTEM imaging demonstrates the existence and structure of the SEI layer on nano-SnO anodes. In addition, FTIR shows that the formation of Li2CO3 controls the electrolyte reduction above 0.9V while the formation of ROCO2Li mainly occurs below 0.9V. The properties of the SEI layer depend strongly on the composition of the electrolytes and influence the cycling performance of the nano-SnO anode material. In addition, a joint study of Raman and FTIR spectroscopy determines the reduction sequence of nano-SnO and the electrolyte. SnO is reduced to metallic Sn before electrolyte is decomposed during charging in the first cycle.

3

Surface Enhanced Raman Scattering (SERS) on Rough Electrodes

Many techniques have been developed to characterize the properties of the SEI 53, 54 layer on the anodes, such as X-ray photoelectron spectroscopy (XPS), EELS 41, 42 and selected area electron diffraction (SAED) as well as FTIR and HRTEM. Most of these techniques provide ex situ information on both the electronic and crystalline structural variations of the electrode. Electrochemical impedance spectroscopy (EIS) and electrochemical quartz crystal microbalance (ECQCM) can provide in situ information of macro-scale properties of the SEI layers. Reflectance FTIR techniques and atomic force microscopy (AFM) have been used in situ to study the surface of metal lithium and electrochemically non33, 34, 55 active electrodes, such as Pt, Au and Ni as well. Nevertheless, it is still difficult to study rough electrode surfaces of composite materials in lithium ion batteries with these techniques. In addition, none of the above techniques, except for FTIR spectroscopy, can provide structural information at the molecular levels.

158

Lithium-Ion Batteries: Solid-Electrolyte Interphase

3.1

Normal Raman Scattering and SERS Studies on Battery Materials

Raman spectroscopy is sensitive to both the chemical and the structural 56-58 variations of a material, liquid or solid. As an in situ technique, Raman spectroscopy has been used to characterize the crystalline structural variation of graphite anodes and LixV2O5 and LiMn2O4 cathodes in lithium ion batteries 59-61 during lithium ion insertion and extraction. In the authors’ laboratory, Raman spectroscopy was used to extensively study the strong interactions between the 62-68 components of polyacrylonitrile (PAN)-based electrolytes, the competition + 69 between the polymer and the solvent on association with the Li ions, the ion transport mechanisms of both “salt-in-polymer” and “polymer-in-salt” 70-73 + electrolytes. Based on the Raman spectroscopic study, Li ion insertion and extraction mechanisms in low-temperature pyrolytic carbon anode have also 74-76 been proposed. In many cases, Raman spectroscopy is used as compensation to the IR spectroscopy to give a complete understanding to the structure of a substance though there are as many cases that Raman spectroscopy is used independently. However, Raman spectroscopy is rarely employed to investigate SEI layers. An important reason is that the thickness of the SEI layer is usually very 28, 30, 42 thin. Therefore its normal Raman scattering signals are too weak to be 77 detected. Fleischmann et al. discovered an interesting phenomenon in 1974 that was later called surface enhanced Raman scattering (SERS), on electrochemically roughened silver surface. With the SERS effect, the Raman 6 scattering signal from a proper substrate can be as strong as 10 times that of the normal Raman scattering (without an SERS substrate), making the intensity of the SERS signal comparable to that of the normal Raman scattering signal from a macro-sized bulk material. Now many types of SERS substrates have been found and many applications have been developed for the SERS technique though its enhancement mechanism is still not clear. The SERS spectrum is sensitive to the variation of the surface chemical species and their microstructures on the surface and interface. Currently the sensitivity of the SERS technique is sufficiently high to detect chemical species at a single 78 molecular level. Therefore, it can also provide configuration information of 79, 80 molecules adsorbed on an electrode. As a result, SERS is an effective technique in examining the SEI layer on the electrode surfaces in lithium (ion) batteries. A shortcoming of the SERS technique is that this effect has only been 81-83 observed on a limited number of noble metals such as Ag, Au and Cu. Most

Spectroscopic Studies of Solid-Electrolyte Interphase

159

of the anode materials for lithium ion batteries, such as carbonaceous materials, oxides and alloys, do not show obvious SERS effect. Fortunately, as an anode material with similar electrochemical behaviors to other anodes for lithium ion 84 batteries, silver is an excellent SERS substrate. As the surface chemistry of LiC electrode for lithium batteries is similar to that developed on the noble metal 85 electrode polarized to low potentials in the same solutions, it is believed that SERS is a powerful tool to study the interfacial phenomena in lithium ion batteries, especially in understanding the formation mechanism of the SEI film at early stages. This section will investigate the SERS effect on discharged Ag anodes in Li/Ag cells with micro-Raman spectroscopy. We wish that such a study be a precursor for in situ characterization of the SEI layer and other interesting surface processes on real electrodes for lithium ion batteries.

3.2

Experimental

Commercial Ag foils (99.99%) were ultrasonically cleaned in acetone prior to use. Electrochemical cells were assembled with the cleaned Ag foil as working electrode, metallic lithium foil as counter electrode, and Celgard 2400 microporous membrane as separator. For comparison, 1M LiClO4 in PC+ DMC (1:1 v/v) and 1M LiPF6 in EC/DEC (1:1 v/v) were used as the electrolytes. The 2 cells were discharged to 0.0 V at a constant current density of 0.05mA/cm and short-circuited for 24 hours. Then the Ag foil was taken out of the cell and rinsed with DEC or DMC respectively to remove the electrolyte residue. Finally, the Ag foil was fixed into an optical cell with a quartz window for Raman measurements. The optical cell was sealed in vacuum to avoid air interference during testing. All the above operations were performed in the argon-filled MBraun glove box. The CV plot was recorded on the CHI660 Electrochemical Workstation. The Raman signals were collected on Renishaw 1000 micro-Raman spectrometer. The magnification of the objective lens was 10 and backscattering geometry was used. The resolution of the Raman 1 spectrometer was set to 2 cm− . The integrated time for collecting the Raman signal was 60 s. The laser power on the sample was 7 mW. For comparison, Ag 86 foils originally covered with Ag island film by chemical deposition and commercial Al foils (99.99%) were also used as electrodes.

160

3.3

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Electrochemical Performance of Ag Electrode

Figure 13 shows a CV plot of the Ag electrode in a Li/Ag cell in the first cycle. + An irreversible reduction peak is observed between 1.2 V and 0.8 V vs Li/Li . 87 88 Compared with the cyclic voltammograms of Pt, carbon and the above tin oxide anodes, this reduction peak should be attributed to the electrolyte decomposition. This means that an SEI film is also formed on the Ag electrode surface during discharge of the Li/Ag cell. 0.02

Current (mA)

0.00

-0.02

A -0.04

-0.06

-0.08 0.0

0.5

1.0

1.5

2.0

2.5

3.0

+

Potential (vs Li/Li )

Figure 13 Cyclic voltammogram of Ag electrode in a lithium cell: Li/1M LiClO4, PC-DMC (1:1 v/v)/ Ag in the first cycle at a scanning rate of 0.1mV/sec. The irreversible reduction peak marked as A represents an electrolyte decomposition that leads to the formation of a passivating layer on Ag electrode. Reproduced from [89] with permission of Amer. Chem. Society.

3.4

SERS Study of Passivating Film on Ag Electrode in Lithium Batteries

The Raman spectrum of the Ag anode after discharge in LiClO4-based electrolyte is shown in Figure 14A. The Raman bands of the electrolyte are not 90 observed in the spectrum. It has been reported that crystalline Li2CO3 has four 1 molecular vibrational modes at 712(w), 748(w), 1090(s), 1460 cm− (w) and five 1 lattice vibrational modes at 96(s), 127(m), 156(s), 193(s), 272 cm− (m) in the 90, 91 Raman spectrum. As seen in Figure 14A, the detection of the molecular vibrational bands of Li2CO3 demonstrates the existence of Li2CO3 on the Ag anode. However, the peaks corresponding to the lattice vibration are not observed, implying that the passivating layer on the Ag electrodes is probably amorphous, consistent with the HRTEM observations on, for example, the

161

Spectroscopic Studies of Solid-Electrolyte Interphase 92

above nano-SnO anode, and other anodes. It has been reported that an SEI film 90, 93 can also be formed on discharged Al foil. However, no obvious bands can be observed on the discharged Al anode (Figure 14B). This is understandable 94 considering that Al is not a SERS-active material and that the SERS layer is usually very thin. The observed Raman bands are definitely due to the significant SERS effect on the discharged Ag electrode. That is, discharging the Li/Ag cell changes the surface morphology of the Ag electrode and makes it SERS active.

240 2000

**

* L i2 C O 3

442

A 565

*

159

2500

934

3000

744 713

* 40000

B

3500

1015

1156

1091

4000

1248

1519

50000

2130

2938 2863

1390

A x10

1448

In t e n s ity (a .u .)

60000

B

30000 1800

1600

1400

1200

1000

800

-1

600

400

200

R a m a n s h if t (c m )

Figure 14 Raman spectra of discharged Ag electrode (A) and Al electrode (B) in 1M LiClO4, PC/DMC (1:1 in v/o) electrolyte. The inset is the selected spectra of the corresponding samples. The spectra are recorded on Renishaw 1000 micro-Raman spectrometer with an excitation of 632.8 nm from a He-Ne laser.

There are still some bands that do not belong to Li2CO3 in Figure 14A, such as the bands at 442, 565, 1015, 1156, 1248, 1390, 1519, 2130, 2863 and 1 2938 cm− . Due to lack of experimental data, it is difficult to assign these bands 83 exactly. However, based on the report of Aurbach et al. on the PC+DMC system and Ref. 97, ROCO2Li should be also one of the main SEI components in the current LiClO4-based electrolyte. Similar experimental results are obtained on discharged Ag electrode with a chemically deposited Ag island film. This indicates that the SERS effect is independent of the initial morphology of the Ag electrode. In fact, the bright and smooth surface of Ag electrode becomes black and roughened after discharge (not shown), meaning at least that the surface morphology of the electrode has been changed.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

3566

6700

30 0 0

25 0 0

20 0 0

248

35 0 0

2700

*

195

393 368

519

841

* L i2C O 3

3700

748 715

1091

4700

A

* *

119

B

x1 0 40 0 0

214

A

5700

1493

In te n s it y ( a .u .)

146

7700

159

162

B

1700 700 1800

1600

1400

1200

1000

800

600

400

200

-1

R a m a n s h ift (c m )

Figure 15 Raman spectra of discharged Ag electrode (A) and Al electrode (B) in 1M LiPF6, EC/DEC (1:1 in v/o) electrolyte. The inset is the selected spectra of the corresponding samples. The spectra are recorded on Renishaw 1000 micro-Raman spectrometer with an excitation of 632.8 nm from a He-Ne laser.

In order to find out the influence of the solvent on the SEI properties, LiPF6-based electrolyte is used. Figure 15A shows the Raman spectrum of discharged Ag anode in LiPF6-based electrolyte. Clearly the SERS spectrum on the discharged Ag electrode in LiPF6-based electrolyte is quite different from that in LiClO4-based electrolyte and is much simpler. This indicates that the SERS spectrum is sensitive to the surface components in the SEI layer. Again no signals are detected on the Al electrode in LiPF6-based electrolyte (Figure 15B). Based on the previous results of FTIR spectroscopy, Li2CO3, LiOH, LiF, LixPOFy and various ROCO2Li should have been the main components of the SEI layer in LiPF6-based electrolytes. Figures 16 to 18 show the Raman spectra of dried commercial lithium hydroxide monohydrate (LiOH·H2O, 96%), lithium fluoride (LiF, >98.5%) and lithium carbonate (Li2CO3, 99%). Table 3 lists the observed bands of the SEI layer on discharged Ag electrode and those of pure Li2CO3 and LiOH·H2O. It is easy to recognize that Li2CO3 and LiOH·H2O are the main components of the SEI layer in the LiPF6-based electrolyte. As LiF is not active in the Raman spectrum, the existence of LiF cannot be excluded though no traces of LiF are detected in Figure 15A. It is surprising that no ROCO2Li is detected though it has been proved an important component of the SEI layer. An explanation is that there are some traces of water in the electrolyte and/or that the sample has been exposed to moistures during preparation and

163

Spectroscopic Studies of Solid-Electrolyte Interphase

transferring for some reasons. In this case, no ROCO2Li can be detected as it becomes Li2CO3 and LiOH and then LiOH·H2O according to the following reactions: 2H2O +2e−+2Li+→2LiOH↓+H2↑ 38 + 32 DEC +2e−+2Li →CH3CH2CH2CH3↑+ Li2CO3↓ + 32 DEC + e−+Li →CH3•H2 +CH3CH2OCO2Li↓ + 32 − DEC + e +Li →CH3CH2O•O+CH3CH2OLi↓ + 95 − 2EC+2e +2Li → (CH2OCO2Li)2↓+CH2CH2↑ + 85 − 2EC+2e +2Li →LiCH2CH2OCO2Li↓ 37 ROLi +CO2 → ROCO2Li↓ 98 2RCO3Li + H2O→Li2CO3↓+2ROH + CO2↑ + 38 − 2CO2+2Li + 2e →CO↑+ Li2CO3↓ 87 LiPF6 ⇔ LiF↓+ PF5 ; PF5 + H2O→2HF +PF3O + 87 PF3O +Li +e− → LiF↓+ LixPOFy ↓

1090

96

60000 50000

30000 193 157

Intensity

40000

274

748 711

1459

10000

127

20000

0 1800 1600 1400 1200 1000

800

600

400

200

-1

Raman Shift (cm )

Figure 16 The Raman spectrum of LiOH·H2O excited by 488 nm excitation from Ar+ laser.

Generally, identification of LiOH·H2O in the SEI film by FTIR spectroscopy is based on the presence of the O-H stretching mode around 1 1 3650 cm− and the Li-O stretching mode around 600 to 500 cm− . The highfrequency band is usually weak and often overlaps with the strong O-H stretching mode of water. The low-frequency bands are broad and overlap with 98 Li-O stretching mode of ROLi, Li2O and ROCO2Li. So it is difficult to find out whether LiOH or LiOH·H2O exists by FTIR spectroscopy. Obviously, the SERS technique is helpful to determine it.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

164

0.07 0.06

Intensity

0.05 0.04 0.03 0.02 0.01 0.00 3500

3000

2500

2000

1500

1000

500

-1

Raman shift (cm )

Figure 17 The Raman spectrum of LiF under 488 nm excitation. from Ar+ laser.

3563

(B)

50000

144

60000

10000

40000

8000

30000

10000

0 1400

1200

1000

800

600

400

192 154

393 367

518

840

696

(A)

1090

2000

85 95

4000

118

212

0 3600 3590 3580 3570 3560 3550 3540 3530 3520

245

Intensity

20000

6000

200

-1

Raman Shift (cm ) Figure 18 The FT-Raman spectrum of pure Li2CO3 excited by the 1064 nm line.

165

Spectroscopic Studies of Solid-Electrolyte Interphase Table 3 Assignments of SEI film and comparison to reference compounds.

SEI layer

LiOH·H2O

Li2CO3

Assignments ν4

99

86

85

96

95

95

Li2CO3 lattice mode

119

118

126

LiLiO stretching 100

145

144

156

154

156

Li2CO3 lattice mode

192

192

192

Li2CO3 lattice mode

213

212

Ag 101

245

245

O-Li 101

368

367

OH-vibration along x-axis 96

394

393

OH-vibration “average” over all directions 96

519

518

T’ Li+ vibration 102

713

696

748 840

840

1090

1090

3563

ν6

99

711

ν4 (Ag+Bg)

748

ν4 LiO2Li stretching103

1091

Ag

1459

ν3(Ag+Bg)

3563

Symmetric OH-stretching 104

Note: T for the translational lattice mode.

85, 87, 90

Comparing these results with those by FTIR spectroscopy, SERS is more sensitive in detecting the main components in the SEI film, including Li2CO3, LiOH·H2O and ROCO2Li, except for LiF. This will be helpful to find out the formation mechanism of the SEI film. Considering that SERS may be used as an in situ technique, it should be a proper technique for studying the interfacial phenomena in lithium ion batteries. Figure 19 compares the SERS spectra of the same SEI layer on discharged Ag electrode in 1M LiPF6 + EC/DEC electrolyte with different excitations. Clearly the positions of the bands in these three spectra agree to each other very well. However, their relative intensities depend strongly on the excitation line. This is obviously due to the resonance effect of the Raman scattering which results from the frequency-dependent optical absorbance of the SEI layer. When

Lithium-Ion Batteries: Solid-Electrolyte Interphase

166

the frequency of the excitation line coincides with the adsorption frequency of the subject, resonance Raman scattering will occur. Such an effect is helpful to characterize the optical properties of the subjects and identify some of their vibrational modes. This is called surface enhanced resonance Raman scattering (SERRS).

9000 6000 3000 0 1250

245 213 192 156 145 119 96 86

394 368

519

C

748

12000 840

Intensity (a.u.)

15000

1090

18000

B A 1050

850

650

450

250

50

-1

Raman Shift (cm ) Figure 19 SERS spectra of the SEI film on Ag discharged to 0 V with various excitations: (A) 632.8 nm, (B) 514.5 nm and (C) 488.0 nm.

3.5

Prospects and Conclusions on Raman Scattering in SEI Investigation

Obvious SERRS effect has been observed on discharged Ag electrodes in the Li/Ag cells. The SERS effect is very sensitive to the surface chemistry of the SEI layer that is mainly composed of ROCO2Li and Li2CO3 when the water content in the electrolyte is very low. However ROCO2Li disappears and LiOH·H2O and Li2CO3 become two of the main components of the SEI layer when there is some trace of water in the electrolyte or atmosphere. Thus, SERS is a proper technique to characterize the chemical species of the SEI film in lithium ion batteries. As seen in the above sections, normal Raman scattering are very effective in characterizing the structural variation of the electrode while the SERS effect is helpful in detecting the surface components in the SEI layer on the electrode,

Spectroscopic Studies of Solid-Electrolyte Interphase

167

especially at the early stages of SEI formation. As an important in situ technique, Raman scattering is as effective as the IR spectroscopy. However, the usage of Raman scattering stops at the bulk structural characterization of the electrode and the electrolyte in most cases to date. We expect that Raman spectroscopy be used independently to characterize the in situ structural variation of the electrodes as well as compensation to the IR spectroscopy during charge and discharge. As to the SERS technique, its application should be extended to the real electrode processes for lithium ion batteries as well as other aspects, especially the SEI investigation on the cathodes.

4

Infrared Absorption and X-ray Photoelectron Spectroscopic Investigation on Performance Improvement of Surface-Modified LiCoO2 Cathode Materials

The importance of the surface of an electrode and its interface with the electrolyte cannot be overstated for the performance of a lithium ion battery. The nature of an electrode surface is critical for the electrochemical functionality of the material. Electrochemical (e.g., charge transfer) and 3+ chemical (e.g., Mn disproportion in LiMn2O4) reactions occur at or near the surface, followed by mass transport into the bulk of the electrode, with structural changes as a result. Undesired side reactions can take place as the + electron meets the Li ion at or near the surface of a cathode particle. Spontaneous reactions such as self-discharge and decomposition of the cathode material and electrolyte can also create a reactive surface, where solvent and salt can participate in reactions, resulting in further electrolyte decomposition. It is thus vital to obtain a basic understanding of the electrolyte/electrode interface during electrochemical storage and cycling. The above sections and other authors’ investigations have shown that the SEI on the anode contains various organic and inorganic decomposition products from the electrolyte. However, confirmation of SEI layer on the cathode has proved elusive though the presence of SEI film on the cathode has long been proposed. Recently, correlation between the surface chemistry and surface reactions of a cathode and its electrochemical performance for lithium ion batteries becomes much more concerned than before. Polyether chain from solvent reactions and salt derived compounds, e.g., LiF, LixPFy and LixPFyOz have been detected in the cathode SEI of uncharged LiNiyCo1−yO2 in LiPF6-based 105 electrolytes by FTIR and Raman studies in various electrolytes. An

168

Lithium-Ion Batteries: Solid-Electrolyte Interphase

intermediate migration step occurs through a surface film between electron transfer at the particle surface and diffusion into the bulk, for the most commonly used lithium ion battery cathode materials (LiCoO2, LiNiO2 and 106, 107 LiMn2O4). This surface film is suggested to be formed electrochemically 108, 109 during the first few cycles. Several papers have been published concerning 110 the identification of the interface species on cathode, the reaction of organic 111, 112 carbonates on charged LiCoO2 and LiMn2O4 cathode and on inert (Au or 113 114 glass) electrode, and the formation of SEI on the cycled cathode. These investigations show that the reactions occurring at the cathode surface are chemical and/or electrochemical in origin, and take place both under storage and cycling. With decades of extensive study, it is realized that the surface chemistry, morphology and surface species of the cathode and its interface with the electrolyte have significant influence on the electrochemical performances of a lithium ion battery, such as its reversibility and safety. Surface modification has proved effective in improving the electrochemical performances of the cathode 115-117 materials. However, investigations of why the coating layer and the interface of coating/coated materials can improve the electrochemical 118 performance of the cathode material have been rare. Amatucci et al. improved the elevated temperature performance of Li1+xMn2O4 spinel by applying a layer of inorganic (lithium borate glass) or acetylacetone complexing agent to its surfaces. They attributed the performance improvement to the formation of a physical barrier that separates the electrolyte from the electrode and the reduction of electrolyte oxidation. They proposed that controlling the surface chemistry of electrode materials and better design of solid/solid or inorganic/organic interfaces could minimize side reactions within the 119 rechargeable batteries in general. Endo et al. modified LiCoO2 surface by coating a layer of diamond-like carbon (DLC) plasma film and improved the higher voltage performance of the cathode. They believed that the improved electrochemical performance of the cathode was due to the suppression to the electrolyte decomposition at higher charge voltages. Nevertheless, they failed to 120 give any direct experimental evidence to support their suggestions. Cho et al. studied the 55°C cycling behaviors of orthorhombic LiMn2O4 cathodes coated with Al2O3 at various temperatures. They attributed the suppressed capacity loss 3+ and Mn dissolution to the accumulation of Al ions at the surface of the cathode material. 121 In our previous study, the surface of commercial LiCoO2 was modified by coating its surface with a thin layer of amorphous magnesium oxide (nano-

Spectroscopic Studies of Solid-Electrolyte Interphase

169

MgO, Figure 20). It is shown that surface modification is effective in improving the structural stability of commercial LiCoO2 cathode materials for lithium ion batteries. Cells based on nano-MgO coated LiCoO2 (MgO/LiCoO2) cathode can be cycled between 2.5 V and 4.7 V. A high specific capacity of 210mAh/g can be obtained without degrading the stability of the material (Figure 21). These improvements were mainly attributed to the protective role of the coating layer. The coating layer was supposed to keep the active core material from direct 3+ contact with the acidic electrolyte and prevent the dissolution of the Co ions that have important functions in suppressing the phase transition. However, further experimental evidence has not been given concerning the influence of the coating layer on the oxidation state of the cathode and on the electrolyte decomposition at various charge voltages.

Figure 20 SEM images of pristine LiCoO2 (a, upper left), LiCoO2 coated with 1.5 mol% of MgO (b, lower left), the surface of MgO-coated LiCoO2 electrode after 70 cycles (c, upper right), and an MgO-coated LiCoO2 particle (d, lower right) in the crack in (c).

Lithium-Ion Batteries: Solid-Electrolyte Interphase

170

5.0 4.5

Voltage (V)

4.5 4.0 3.5 3.0

4.0

3.5

3.0 MgO05A26, 0.06mA, Cyc.No.1-13

2.5

2.5

0

40

80

120

160

200

Capacity (mAh/g)

240

280

0

20

40

60

80 100 120 140 160 180 200

Capacity (mAh/g)

Figure 21 Comparison of the electrochemical performances of commercial LiCoO2 (left) and MgO surface-modified LiCoO2 (right) cycled between 2.5 and 4.7 V at 0.1 mA/cm2 (approx. 0.1 C).

This section will compare the interactions between electrolyte and the LiCoO2 surfaces under various conditions, pristine or nano-MgO modified, charged or uncharged, and find out the reasons for the improved electrochemical performances of MgO/LiCoO2 cathode materials.

4.1

Sample Preparation

The surface-modification process to commercial LiCoO2 (Cellseeds, C-5, average particle size: 5-6 µm), the electrode preparation and the assembly of test 121 cells have been described in detail in our previous paper. The surface of commercial LiCoO2 particles was first coated with Mg(OH)2 through the reaction of MgCl2 and NaOH in distilled water and the co-deposit of LiCoO2 with Mg(OH)2. Mg(OH)-coated LiCoO2 was separated from the solution by repeated rinsing and filtering. Mg(OH)2 was dehydrated by heating the coated material at 600°C for 2 hours in air and hence nano-MgO coated LiCoO2 was obtained. Nano-MgO coated LiCoO2 was mixed with carbon black (CB) and a polymer binder at a weight ratio of MgO/LiCoO2:CB:binder = 87:9:4 in a solvent to form a slurry. The slurry was uniformly spread on Al foil by doctor’s blade technique. Test cells were assembled and sealed in the Ar-filled glove box with 2 (MgO-coated) LiCoO2 (approx. 0.8 cm ) as the working electrode, fresh lithium foil as the counter electrode, 1 M LiPF6 in EC/DMC (1:1 v/v; EC for ethylene carbonate and DMC for dimethyl carbonate) as the electrolyte and Celguard

Spectroscopic Studies of Solid-Electrolyte Interphase

171

2400 polypropylene as the separator. In order to have the electrode charged to the required open circuit voltages (OCVs), the cells were first charged to preset 2 voltages at a constant current (<0.1 mA/cm ) and then potentiostatically charged until the nominal charge current faded below 0.1 µA. The voltage values shown in this chapter were picked up after the cell was disconnected from the cycler and stored for more than 24 hours. The test cell was then dissected and the cathode was removed from the cell in the glove box. The electrode was soaked in and then rinsed repeatedly with pure DMC to remove the electrolyte. The washed electrode was dried by being stored in a vacuumed mini-chamber of the glove box for DMC to evaporate at room temperature. Nano-MgO powder was prepared by filtering and washing the reaction product (Mg(OH)2 gel) of MgCl2 with NH4OH in distilled water three times. Then the gel was heated at 600°C for 2 hours in air to get the nanoMgO powder. The washing and drying processes for nano-MgO powder, uncharged commercial LiCoO2 and uncharged surface-modified MgO/LiCoO2 electrodes were the same as for the charged electrodes, after they were soaked in the electrolyte for an hour. Commercial LiCoO2, nano-MgO, MgO/LiCoO2 powders and the electrode material scratched from the aluminum current collector were thoroughly mixed with KBr and pressed into pellets respectively. Liquid EC and fresh electrolyte were cast on KBr pellets. A droplet of the binder dissolved in dimethyl formamide (DMF) was also cast onto KBr pellets and then heated at 150°C for over 24 hours in air. All these samples were separately stored in hermetically sealed containers, ready for the Fourier transform infrared (FTIR) measurements. Charged electrode samples for the X-ray photoelectron spectroscopic (XPS) study were fixed on cleaned sample holders (copper) with a piece of conductive tape and stored in a sealed container. All the above operations were carried out in argon atmosphere unless specified. When everything was ready for the XPS and FTIR instruments, the containers were opened and the samples were transferred into the vacuum chambers of the instruments and the chambers were vacuumed immediately. The exposure time of the sample to air was less than 10 seconds. The FTIR spectra were the average of 200 scans on a BIO-RAD FTS-60 spectrometer. XPS spectra were collected on an ESCALAB5 (VG Scientific; energy resolution: 0.1 eV) with a non-monochromatic Mg Kα radiation (1253.6 eV). Before measurements, the + XPS samples were sputtered with Ar beam (2 KeV, 40 µA) for 10 minutes to remove the SEI layer.

172

4.2

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Comparison of EC Adsorbed on Different Substrates

The IR absorption spectra of commercial LiCoO2, nano-MgO and nano-MgO coated LiCoO2 have been shown in Figure 22. It is seen that LiCoO2 has two 1 strong absorption peaks at 522 and 610 cm− while nano-MgO has a broad hump at 1 1 around 1483 cm− . Another two strong bands are observed at 640 cm− and −1 420 cm in nano-MgO. These broad peaks are characteristic of nanometer sized MgO. However, when the surface of LiCoO2 particle is coated with nano-MgO, no obvious spectral variation is observed. The two peaks of pristine LiCoO2 are still there and their relative intensities remain unchanged. The reason for the spectral features is that the content of nano-MgO on LiCoO2 is very low (about 121 1.5 mol% ). Therefore some surface-sensitive characterization techniques are necessary for the identification of the MgO/LiCoO2 interlayer properties. The other peaks observed in Figure 22 are attributed to the instrumental error (the sharp peak 1 at around 1400 cm− , for example) or some contamination to the KBr pellets because these weak peaks reappear in all these samples. Figure 23 compares the IR absorption spectra of pure EC, fresh electrolyte and electrolyte-soaked substrates including commercial LiCoO2, nano-MgO and uncharged MgO/LiCoO2 electrode. The IR spectrum of the binder (f) is also shown for reference. The four characteristic (also the strongest) bands of the 1 binder (at 2243, 1732, 1697 and 1454 cm− ) cannot be detected in the MgO/LiCoO2 electrode. Therefore the interference of the binder (4% w/w) on the IR spectrum of all the binder-containing electrodes, charged or uncharged, is negligible and will not be considered in the following discussion. LiPF6 has 1 1 three characteristic bands in the mid-IR spectrum, 1020 cm− (weak), 849 cm− 1 (strong) and 559 cm− (mid). These bands have been marked with asterisks in Figure 23. Due to the prolonged evacuation and its volatility, even the strongest C=O band of DMC cannot be detected in Figure 23. Therefore the interference of DMC on the spectrum can also be ignored. An overview to the spectra shows that the absorption spectrum of the electrolyte residue depends strongly on the surface properties of the adsorbent substrate. These spectral features will be respectively discussed in the following. The above electrolyte residue on nano-MgO and commercial LiCoO2 is in fact a mixture of EC, LiPF6 and their decomposition products. Figure 24 shows the CH stretching bands of the fresh electrolyte and the electrolyte residue on different surfaces. Clearly the spectrum of electrolyte residue on MgO/LiCoO2 is somewhat similar to that of the electrolyte, but the spectra of the electrolyte

Spectroscopic Studies of Solid-Electrolyte Interphase

173

40

30 25

nano-MgO

20 610

15

MgO/LiCoO2

522

Intensity (a.u.)

35

10

LiCoO2

5 0 400

800

1200

-1

1600

2000

Wavenumber (cm )

Figure 22 FTIR spectra of commercial LiCoO2, nano-MgO and nano-MgO coated LiCoO2 powders. Reproduced from [122] with permission of The Electrochemical Society Inc.

140

Absorption (a.u.)

120 100 80

binder (f)

f

on MgO/LiCoO2 (e)

d

*

e

on nano-MgO (d)

c

60 40 20

on LiCoO2 (c) electrolyte (b)

*

* b

EC (a)

a

0 2500 2250 2000 1750 1500 1250 1000 750

500

-1

Wavenumber (cm ) Figure 23 Comparison of the IR spectra of pure EC (a), the electrolyte (b), and electrolyte residue on different substrates: (c) commercial LiCoO2; (d) nano-MgO; (e) nano-MgO coated LiCoO2; (f) the binder for cathode preparation (the strong bands around 600 cm−1 in c are characteristic bands of commercial LiCoO2). Reproduced from [122] with permission of The Electrochemical Society Inc.

residue on nano-MgO and on commercial LiCoO2 are quite different from that of the electrolyte. These indicate that the configuration of EC is disturbed or some new substances are produced. Comparison of these spectra shows that nano-MgO and commercial LiCoO2 interact with EC the strongest while the surface of MgO/LiCoO2 affects EC the weakest. The slight spectral difference between the electrolyte residue on MgO/LiCoO2 and the fresh electrolyte might

174

Lithium-Ion Batteries: Solid-Electrolyte Interphase

be simply due to the different concentrations of LiPF6 in them. In addition, the 1 disappearance of the intense 2989 cm− band in the fresh electrolyte and the 1 presence of the 2849 cm− band for the electrolyte residue on nano-MgO and commercial LiCoO2 imply the decomposition of the electrolyte and the formation of some new species during the soakage of the materials in the electrolyte. It is believed that strong chemical adsorption and decomposition of electrolyte coexist in the electrolyte residue. 1 Two intense bands are observed at 1796 and 1771 cm− with similar −1 intensities in pure EC (Figure 25). The 1796 cm band has been attributed to 1 the C=O stretching and the 1771 cm− band to the overtone of the ring breathing 1 124 mode at 889 cm− . Clearly the surface properties of the adsorbent influence the spectral features of these two bands drastically in comparison with their counterparts in pure EC: the widths of these two bands are obviously broadened 1 1 and the position of the 1796 cm− band moves to 1804, 1806 and 1810 cm− on commercial LiCoO2, nano-MgO and MgO/LiCoO2, respectively. The relative intensities of these two bands become reversal as well. A simple overlapping of the IR spectra of electrolyte residue on commercial LiCoO2 and on nano-MgO (by a weight ratio of 1:1) cannot explain any of these spectral changes. Therefore, an interface layer must have been formed on commercial LiCoO2 during surface modification, with properties distinct from that of nano-MgO or commercial LiCoO2 and affects the configuration of EC on MgO/LiCoO2. 150

Absorption (a.u.)

120 (c+d) 90 e 60

d c

30

b

a 0 1850 1820 1790 1760 1730 1700 -1

Wavenumber (cm ) Figure 24 Selected IR spectra of the CH stretching modes of the electrolyte residue adsorbed on various substrates. Reproduced from [122] with permission of The Electrochemical Society Inc.

Spectroscopic Studies of Solid-Electrolyte Interphase

175

150

Absorption (a.u.)

120 (c+d) 90 e d 60 c 30 b a 0 1850 1820 1790 1760 1730 1700 -1

W avenum ber (cm ) Figure 25 Comparison of the C=O stretching (approx. 1800 cm−1) and overtone of the ring breathing (approx. 1770 cm−1) bands of pure EC (a), the electrolyte (b) and electrolyte residue on different substrates: (c) commercial LiCoO2, (d) nano-MgO, (e) MgO-modified LiCoO2, and a simple overlapping of (c+d). Reproduced from [122] with permission of The Electrochemical Society Inc.

Figure 26 shows the IR spectra of some ring stretching modes of EC in the −1 electrolyte residue. As marked with asterisk (at 1020 cm ), DMC and LiPF6 are almost undetectable in the electrolyte residue on commercial LiCoO2 and nanoMgO. Compared with the spectrum of pure EC and the electrolyte, a new −1 component is observed at approx. 1197 cm in the electrolyte residue on the three substrates. Another two new components are detected at 1210 and −1 1088 cm in the residue on MgO/LiCoO2. As the content of the decomposed products on nano-MgO coated LiCoO2 is very low, the appearance of these intense bands is believed to be due the variation of configuration (the molecular orientation and the bonding strain) of the EC molecules on the substrate. EC molecule has a planar skeleton with C2v symmetry in its liquid 133, 134 Therefore the plane of an EC molecule tends to be parallel to the phase. adsorbent surface. With this, it can be understood that the spectral features of EC, especially those related to the ring structure of the molecule, will be different from that of pure EC arranged randomly in the electrolyte solution, due to the strong interaction between the adsorbent and the molecules. This can explain the above spectral changes of the C=O stretching mode and the overtone

176

Lithium-Ion Batteries: Solid-Electrolyte Interphase

of the ring-breathing mode. Interaction (chemical adsorption) between the adsorbent and the EC molecule makes the C-O bonding weak but intensifies the C=O bonding when the molecular plane of EC becomes parallel to the adsorbent. Strong adsorption will promote the breaking-down of the C-O bonding and enhances electrolyte decomposition. Nevertheless, regular arrangement of EC molecules on the adsorbent should have made the C=O stretching and the ring-breathing bands narrower than in pure EC. Most probably some of the EC molecules are decomposed and polycarbonate species 124, 125 are formed by polymerization of EC initiated by EC oxidation. Another explanation is that the increasing strain of the EC ring makes the EC molecules more reactive to the adsorbent. Either of these two reasons will broaden the above bands and is indeed supported by the spectral variation of other C=O or C-O related bands described in the following (see Figures 27 and 28). Further study is necessary to identify these possible reasons. 140

Absorbance (a.u.)

120 100

e

80 d 60 c

*

40 b 20 a 0 1250

1200

1150

1100

1050

1000

950

-1

Wavenumber (cm ) Figure 26 Selected IR spectra of the C=O stretching modes of EC (a), the electrolyte (b) and the electrolyte residue on commercial LiCoO2 (c), nano-MgO (d) and MgO-coated LiCoO2 (e) (the peak with an asterisk overhead is from the overlapping of the DMC and LiPF6 bands). Reproduced from [122] with permission of The Electrochemical Society Inc.

Figure 26 shows the IR spectra of some ring stretching modes of EC in −1 the electrolyte residue. As marked with asterisk (at 1020 cm ), DMC and LiPF6 are almost undetectable in the electrolyte residue on commercial LiCoO2 and nano-MgO. Compared with the spectrum of pure EC and the electrolyte, −1 a new component is observed at approx. 1197 cm in the electrolyte residue on the three substrates. Another two new components are detected at 1210 and −1 1088 cm in the residue on MgO/LiCoO 2. As the content of the decomposed

Spectroscopic Studies of Solid-Electrolyte Interphase

177

products on nano-MgO coated LiCoO2 is very low, the appearance of these intense bands is believed to be due the variation of configuration (the molecular orientation and the bonding strain) of the EC molecules on the substrate. 140

Absorbance (a.u.)

120 100

e

80 d 60 c

*

40 b 20 a 0 1250

1200

1150

1100

1050

1000

950

-1

Wavenumber (cm ) Figure 27 Selected IR spectra of EC (a), the electrolyte (b), and the electrolyte residue on (c) commercial LiCoO2, (d) nano-MgO; and (e) MgO-modified LiCoO2 (note the dramatic spectral difference around 850 cm−1 on different substrates). Reproduced from [122] with permission of The Electrochemical Society Inc.

60

40

1512

1564

1644 1647

on nano-M gO

30 20 10

1630

Absorption (a.u.)

50

1665 1670

on M gO/LiCoO 2

on LiCoO 2

electrolyte EC

0 1720 1675 1630 1585 1540 1495 1450 1405 1360 -1

W avenum ber (cm ) Figure 28 Selected IR spectra of EC (a), the electrolyte (b) and its residue on different substrates: (c) on commercial LiCoO2; (d) on nano-MgO; and (e) on MgO-modified LiCoO2 (note that the 1490 and 1400 bands of EC have been much enlarged). Reproduced from [122] with permission of The Electrochemical Society Inc.

178

Lithium-Ion Batteries: Solid-Electrolyte Interphase

The above results suggest that EC molecules interact strongly with the surfaces of commercial LiCoO2 and nano-MgO particles. Such interactions make EC molecules more reactive to the particle surface and will be more easily decomposed on the electrode surface than DMC. It is thought that the circular structure of EC could contact the surface of the carbon electrode more closely 126 than the acyclic structure of DMC. This may also be true on the cathode surface. In addition, the polarity and dipole moment of EC are greater than that of DMC. These factors make EC molecules preferentially be decomposed on the cathode surface to form a passive film and impede the decomposition of DMC 126 on the surface. Yang et al. reported that only EC is decomposed in a binary solvent consisting of EC/DMC. DMC is not decomposed in the presence of EC. Hence the function of DMC in a binary solvent system is probably to improve the solubility and conductivity rather than to participate in the formation of the passive film. As a result, the formation of the passive film becomes ECdominant. However, the impact of the MgO-modified LiCoO2 on the adsorption configuration of the EC molecules is not as strong as on MgO or LiCoO2 surface. 108, 127 used FTIR spectroscopy to follow the surface chemical Aurbach et al. changes of cathodes including LixCoO2, LixNiO2 and LixMn2O4 spinel. They observed ROCO2Li species in cycled LixNiO2, LixCoO2 and LixMn2O4 cathode in EC/DMC LiAsF6 or LiPF6 electrolyte and concluded that the cathode materials are probably also covered with surface films (initially Li2CO3). In addition, it 128 has been reported that cathode material has a pristine surface film of Li2CO3 as a residue from the synthesis precursors or a product of reactions between CO2 in the atmosphere and the active cathode powder. The vibrational bands of Li2CO3 −1 39-41, 43, 47 Nevertheless no are usually observed around 1505, 1435 and 868 cm . traces of Li2CO3 was detected on commercial or surface-modified LiCoO2 before they were soaked in the electrolyte in this work (Fig. 22). The intense −1 868 cm bands are indeed observed on electrolyte-soaked commercial LiCoO2 and nano-MgO particles (Figure 27). However, the other two bands could not be observed on commercial LiCoO2 in Figure 28. Therefore, it seems that no −1 Li2CO3 exists on commercial or MgO/LiCoO2 particles. The strong 868 cm bands in commercial LiCoO2 in Figure 27 might be due the P-F bonding from 105 LixPFy and/or LixPFyOz. In contrast, it is interesting that traces of Li2CO3 is + detected on the electrolyte-soaked MgO though there is no Li ion source in 2− solid nano-MgO. The CO3 groups can only come from the decomposition of the carbonate electrolyte (the breaking-down of the ring-structure of EC molecules, for example). This is persuasive evidence that strong interaction

Spectroscopic Studies of Solid-Electrolyte Interphase

179

between nano-MgO and the electrolyte takes place and the electrolyte is decomposed upon contacting the nano-MgO particles with high reactivity due to their large surface energies and specific surface areas. These demonstrate that electrolyte decomposition can be a chemical process as well as an −1 −1 electrochemical one. The two intense peaks at 1484cm and 1410cm are from the EC residue on the cathode materials.

Absorbance (a.u.)

140

nano-M gO coated

120 100

electrolyte

80 60

4.8V 4.6V

40

4.4V

20

4.3V 4.2V

0 3100

4.1V

3050

3000

2950

2900

2850

2800

-1

W avenumber (cm ) Figure 29 Typical (MgO-coated LiCoO2) evolution of the IR spectra of the CH stretching modes of the electrolyte and its residue on cathode charged to different voltages. Reproduced from [122] with permission of The Electrochemical Society Inc.

In addition to Li2CO3, it is reported that the SEI surface is composed of more complicated organic compounds like R-CO3Li (R = methyl, ethyl, propyl 32 and butyl, etc.) and (CH2OCO2Li)2. Considering that most of the DMC molecules and quite some of the LiPF6 have been removed from the electrolyte during the washing and drying processes (Figures 23 and 26), the difference of −1 the relative intensity of the hump around 850 cm is surprising. The significant enhancement of this hump on nano-MgO and on commercial LiCoO2 should definitely be assigned to the decomposition of the electrolyte. According to 32 Aurbach et al., this region corresponds to the –CO3 bending in (CH2OCO2Li)2 and RCO3Li. In contrast, the change of this hump in MgO-coated LiCoO2 is rather mild: its height is comparable to that of the electrolyte and only its bandwidth is slightly broadened. Figure 28 demonstrates the IR spectra of the decomposition products of the electrolyte. Based on the assignments by −1 43 Aurbach et al. the intense wide hump at around 1650cm in Figure 28c

180

Lithium-Ion Batteries: Solid-Electrolyte Interphase

(on commercial LiCoO2) is assigned to the –CO2 stretching modes in R-CO3Li while the more intense hump in Figure 28d (on nano-MgO) corresponds to the −1 –CO2 stretching modes in (CH2OCO2Li)2 (approx. 1665 cm ) and in RCO3Li −1 (approx. 1650 cm ). CH and CH3 asymmetric bending modes are also observed −1 between 1540 and1400 cm in these two samples. In comparison, though (CH2OCO2Li)2 and RCO3Li are also detected in MgO/LiCoO2 cathode, their contents should be much lower than on commercial LiCoO2 and nano-MgO, considering that the relative intensities of these new bands are much weaker on MgO/LiCoO2 than on nano-MgO and commercial LiCoO2 in Fig. 28. These results indicate that electrolyte decomposition can take place before any electrochemical treatment is performed to the commercial LiCoO2 electrode. The ROCO2Li species may be formed on the active surface due to possible surface nucleophilic reactions between commercial LiCoO2 and EC, or by + nucleophilic reactions that form Li -free carbonate anion-containing species that can further attack EC molecules. However, coating commercial LiCoO2 with nano-MgO can obviously suppress such chemical reactions on it significantly.

4.3

IR Spectra of EC on Electrodes Charged to Different Voltages

The IR spectra of the electrolyte in the CH stretching region on commercial and MgO/LiCoO2 charged to various voltages have been compared in Figure 29. It is seen that the spectral difference between fresh electrolyte and the electrolyte residue on charged cathodes are striking. However, the IR spectra of the electrolyte residue on charged commercial and MgO/LiCoO2 cathodes are similar to each other, independent of the charge voltage. Therefore, only the IR spectra of MgO/LiCoO2 cathode are shown. Actually the IR spectra of these CH stretching bands on the charged cathodes in Figure 29 are quite similar to those on commercial LiCoO2 and nano-MgO in Figure 24. Two strong bands are −1 observed at 2922 and 2855 cm respectively. These two bands have been attributed to the –CH symmetric and asymmetric stretching of the –CH3 129 130-132 groups, due to the formation of an SEI layer on the LiCoO2 cathode. The composition of the SEI layer on the cathode may be similar to that on the anode 133 and contain functional groups like –CH2 and –CH3. Itoh et al. studied highly polarized LiCoO2 electrode in propylene carbonate (PC) by ex situ FTIR but did + not observe these bands until 5.0 V (vs Li/Li ). This disagreement might be due to the difference of the electrochemical stability of the solvents. These two bands appear at rather low voltages and remain there until high voltages. As their spectral features (relative intensities, band widths and positions) do not

Spectroscopic Studies of Solid-Electrolyte Interphase

181

change with the charge voltage, these two bands should have originated from the same chemical species insoluble to DMC and not volatile, or the same as or similar to the substances observed on uncharged commercial LiCoO2. In −1 contrast, the relative intensity of the 2961 cm band changes randomly depending on the washing condition to the cathodes, implying that this band originates from a substance soluble to DMC. 100 c o a ted 90

A b s o rb a n c e (a .u .)

80 electro lyte

70

4.8V

60 50

4.6V

40 4.5V

30

4.4V

20

4.3V

10

4.2V

0 1 0 00

4.1V

950

900

850

800

750

700

-1

W a ve n u m b e r (c m )

Figure 30 Evolution of the IR spectra of the CH stretching modes of the electrolyte and its residue on MgO-coated LiCoO2 cathode charged to different voltages.

120

160

c o a te d n aked 140

A b s o rb a n c e (a .u .)

100

Absorb ance (a.u .)

120 100

4.8V 4.6V 4.5V

80 4.4V

60

4.3V 4.2V

40

80

4 .80 V 4 .6V

60 4 .5V 4 .3V

40 b e fo re c h a rg e d

4 .2V

4.15V before charged

4.10V

20

EC

20 EC 0 1850

1820

1790

1760

1730 -1

W avenu m ber (cm )

1700

0 1850

1820

1790

1760

1730

1700

-1

W a ve n u m b e r (c m )

Figure 31 Comparison of the evolution of the C=O stretching mode and the overtone of the ring breathing mode of EC adsorbed on commercial LiCoO2 (A) and on MgO-coated LiCoO2 cathode (B) charged to different voltages. Reproduced from [122] with permission of The Electrochemical Society Inc.

182

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Consistent with the spectral features in uncharged MgO/LiCoO2, no traces of Li2CO3 can be detected in charged MgO/LiCoO2 until rather high voltages + (4.4 V vs Li/Li ) though other components can be produced at lower voltages as −1 seen from the shape evolution of the hump around 850 cm (Figure 30). This indicates that Li2CO3 can be formed on MgO/LiCoO2 through electrochemical reactions, but its formation is prevented up to 4.4 V. As the passive film formed on the cathode is electrically insulating, its growth eventually leads to loss of electrical contact between cathode particles, resulting in cell failure. Therefore, suppression of the formation of passive film prolongs the cycle life as well as enhances the reversibility of the cell. Figure 31 compares the C=O stretching bands of the electrolyte residue on charged cathodes with and without surface modification. As the interference of DMC is negligible, all the spectral changes in this region can be attributed to the reaction product of the electrolyte with the electrode and interaction of EC with −1 the electrode material. The overtone of the ring breathing mode at 1774 cm band is narrower and slightly weaker than the C=O stretching mode at −1 1799 cm band in pure EC. However, this relationship becomes the opposite on −1 charged commercial LiCoO2 electrode (Figure 12A). The 1774 cm band “falls” −1 −1 to 1767 cm and becomes much wider and stronger than the 1799 cm upon soaking the electrode in the electrolyte. The impact of the charge voltage on the spectral features of these two bands is not significant. In contrast, the position of the overtone band only slightly shifts downwards with rising charge voltage −1 while the position of the 1800 cm band almost remains unchanged until 4.8 V. Different from that on charged electrode of commercial LiCoO2, the spectral features of the electrolyte residue on MgO/LiCoO2 are rather similar to −1 that of pure EC. With rising charge voltage, the intensity of the 1774 cm band increases slightly and its position remains unchanged until 4.4 V. Only at very high charge voltages (above 4.4 V), do its spectral features become similar to that on commercial LiCoO2. This indicates that formation of an MgO-LiCoO2 interface on the LiCoO2 particle weakens the disturbance of electrode oxidation to the adsorption state of EC at low charge voltages. Only at rather high + voltages (around 4.4 V vs Li/Li ) does the electrode oxidation have a significant disturbance to the EC adsorption comparable to that on charged commercial LiCoO2 electrodes. This means that surface modification to commercial LiCoO2 improves its electrochemical performance by suppressing the interaction between the electrode material and the electrolyte. The complexity of the composition of the SEI film and the lack of data for the exact assignments to the complicated spectra of these absorption bands make it difficult to recognize the chemical species except for Li2CO3 on the particle surface. Tables 4 and 5 list the observed IR absorption bands on

Spectroscopic Studies of Solid-Electrolyte Interphase

183

commercial and MgO/LiCoO2 electrodes charged to different voltages, excluding those from the electrolyte. Table 4 shows that the decomposition products on commercial LiCoO2 are primarily independent of the charge voltage. Most of the chemical groups observed at low voltages (4.1 V, for example) can also be observed at high voltages and vice versa. This may be attributed to the strong interaction between the particle surface of commercial LiCoO2 and the electrolyte. Such strong interaction makes the electrochemical decomposition of the electrolyte less significant. Table 5 demonstrates that the charge voltage influences the chemical species on the surface of the MgOmodified electrodes significantly. Some species can only be observed at rather high charge voltages, Li2CO3, for example. The higher is the charge voltage, the more complex will be the spectra, indicating that the decomposition products become more complicated and the constituent of the SEI film becomes complete.

Table 4 New bands (cm−1) observed on commercial LiCoO2 charged to various voltages (V).

4.1

4.15

4.2

2959 2922 2855

2959 2921 2855 2850

2960 2920 2850

1976

1974

1974 1754 1727 1658 1642

4.3

4.4

2922 2850

2960 2920 2851

1980

1732 1646

1646

1628

1627

1459 1432

1446

1279

1356 1271

835

839

1626 1620 1612 1463 1433 1428 1356

836

1728 1659 1649 1641 1625 1619 1612 1462

1754 1725 1660

1629

4.5

4.6

4.8

2923 2853

2961 2924 2853

1980 1971

1980 1971

2961 2924 2856 2850 1978 1974 1754

1727 1661 1642

1728 1662 1643

1644

1624

1625

1623

1461 1430

1459 1430

1463 1430

1358

1354

1354

865 833

867 838

864 834

Assignment C-H stretching 134 C-H stretching 134 C-H stretching 134 C-H stretching 134 CH2 wagging of aryl 129 CH2 wagging of aryl 129 carbonyl stretching 87 carbonyl stretching 85 CO2 stretching 134 CO2 stretching 134 CO2 stretching 134 CO2 stretching 134 CO2 stretching 134 CO2 stretching 134 CH, CH3 asym. bending 134 CH, CH3 asym. bending 134 CH, CH3 asym. bending 134 CO2 sysm. stretching 134 C=O stretching 134 CO3 bending 134 CO3 bending 134

184

Lithium-Ion Batteries: Solid-Electrolyte Interphase Table 5 New bands observed on MgO-modified LiCoO2 charged to various voltages.

4.1 V

4.2 V

4.3 V

4.4 V

4.6 V

4.8 V Assignment

2960 2919 2851

2959 2923 2852

2958 2923 2852 1838

2958 2920 2850

2962 2919 2848 1833

1724

1728

1727

1725

1564

1565

1562

1725 1640 1617 1567 1527

1462 1432

1463 1426 1354~ 1333 1291

1462 1423 1358~ 1330 1288

1137 1122

1141 1126 1109 1052 1026 921

1281 1144 1126 1101 1050 1027 922

868 852 747 721

868 858 747 719

670

678 667 655 592 560 535 513

C-H stretching 130 C-H stretching 130 C-H stretching 130 CH2 wagging overtone in vinyl and vinylidine groups 125 1729 carbonyl stretching 131 1643 CO2 stretching 130 1621 CO2 stretching 130 C=O stretching 125 1532 C=O stretching 125 1513 C=O stretching 125 1463 CH, CH3 asym. bending 130 1433 CH, CH3 asym. bending 130 1357~ CO2 sysm. stretching 130 1330 1290 CO2 sysm. stretching 130 1280 CO2 sysm. stretching 130 1141 C=O stretching 130 1127 C=O stretching 130 C=O stretching 130 1054 C=O stretching 130 1026 C=O stretching 130 932 CH2 wagging in vinyl and vinylidine groups 125 868 CO3 bending 130 850 CO3 bending 130 747 olefinic CH wagging 125 721 CO2 asym. bending 125 CO2 asym. bending 125 667 CO2 asym. bending 125 655 CO2 asym. bending 125 644 CO2 asym. bending 125 589 Li-O stretching 125 560 Li-O stretching 125 Li-O stretching 125 Li-O stretching 125

1462 1432

1293 1137 1100

1108

915

721

560

560

721

560

585 562

1463 1359~ 1332

2959 2920 2849 1833

Comparison of Table 4 with Table 5 indicates that the composition of the SEI film on commercial LiCoO2 is quite similar to but not the same as that on MgO/LiCoO2 at high charge voltages. The SEI film on the commercial and MgO/LiCoO2 cathodes are mainly composed of ROLi and ROCO2Li but species

Spectroscopic Studies of Solid-Electrolyte Interphase

185

with aromatic structures are present on commercial LiCoO2 while vinyl and vinilidine groups are detected on charged MgO/LiCoO2 electrode. Therefore, it seems that the MgO-LiCoO2 interface does not affect the formation and composition of the SEI film on the cathode at high charge voltages, rather it suppresses the decomposition of the electrolyte and the formation of Li2CO3 at low voltages. There might be two possible reasons for the above spectral variations of electrolyte residue on different substrates and on electrodes charged to different voltages. One is the influence of the chemical adsorption state of the EC molecules on the electrode (substrate). The surface chemistry of the electrode influences their configuration, especially at the uncharged state, and therefore leads to the spectral variation. The other reason is the chemical and electrochemical decomposition of the electrolyte. The decomposition products mix with the electrolyte. Their spectra overlap and change the observed spectral features of some bands. The discussion in the last section has shown that the surface chemistry of the substrate has a critical effect on the absorption spectra of EC molecules.

4.4 XPS Study on Evolution of Electronic Structure of Cathode Materials with Charge Voltages The above phenomena observed by IR absorption spectroscopy imply that the surface properties of the electrode take important part in the interaction between the electrode and electrolyte. In order to interpret the significant impact of the surface on the electrochemical performance of the commercial and MgO/LiCoO2 cathodes, the surfaces of these charged cathodes are studied by 107 XPS spectroscopy. Thomas proposed that the electrolyte decomposition on 4+ charged LiCoO2 surface is induced by the strong oxidizing power of Co 135 cations. However, Montoro et al. reported that the Co ions are in a trivalent 3+ + Co low-spin state in LiCoO2 and remain mostly unaffected by Li extraction in chemically delithiated LiCoO2. On the other hand, the O ions in LiCoO2 are partially reduced upon Li deintercalation. Figure 32 shows the XPS spectra of commercial and MgO/LiCoO2 charged to different voltages. The spectra correspond to Co 2p →Co 3p transitions and are dominated by multiplet effects. The Co 2p spectrum is split by the spin-orbital interaction into Co 2p1/2 and Co 2p3/2 regions. In turn these regions are further split due to Co 2p-3d interaction and crystal field effects. The shapes of the spectra are directly related to the 136 ground state of the Co ions. Only Co 2p bands corresponding to trivalent

186 3+

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Co are observed in commercial and MgO/LiCoO2 before electrochemical treatment, representative of the Co 2p3/2 at 780.5 eV and Co2p1/2 at 795.7 eV. Another two Co 2p bands are observed at 786.6 and 803.2 eV on both the commercial and the MgO/LiCoO2 cathodes after charging the cathodes to a certain voltage. These two new bands might be responsible for the presence of 4+ tetravalent Co ions in the charged electrodes. Interestingly these two bands appear at rather low charge potential (4.1 V, corresponding to the extraction of + approx. 0.3 mole of Li from LiCoO2) and remain there until most of the Li ions + are extracted (4.8 V vs Li/Li ). It is surprising that the detected oxidation states of Co cations in commercial LiCoO2 and MgO/LiCoO2 is independent of the 135 charge voltage, consist with the investigation results of Montoro et al. An explanation is that the XPS spectra are only sensitive to a very small depth (about 20 Å at most) under the surface of a LiCoO2 particle. Further sputtering the electrode surface is expected to remove the surface species and detect the species deeper under the surface. Figure 32C demonstrates that the relative 4+ intensity of the Co band slightly increases with the charge voltage compared 3+ with that of Co . Notice that the LiCoO2 particles are not in the same geometric plane and sputtering the electrode surface can only remove the surface species on some of the LiCoO2 particles. It is understandable that the variation of the 4+ 3+ relative intensity of Co to Co is not as significant as expected, with the charge voltage. The difference of the O 1s spectra of commercial and coated LiCoO2 and their variation with the charge voltage are much more obvious than in the Co 2p spectra. As shown in Figure 32A, the electronic structure of oxygen varies steadily (at 529.4 eV in uncharged commercial LiCoO2 electrode) with the increase of the charge voltage and a new component appears at 532.6 eV, corresponding to oxygen atoms with higher oxidizing power. The content of such oxygen atoms increases with the charge voltage and becomes dominant in 135 the electrode at high voltages. This is opposite to the report of Montoro et al. Oxygen atoms with higher binding energy (at 531.6 eV) in the coated cathode, however, appear at lower charge voltage but their content increases more slowly than in commercial LiCoO2 cathode (Figure 33B). These oxygen atoms cannot dominate the cathode material up to 4.8 V. Moreover as the binding energy of these oxygen atoms is lower in charged MgO/LiCoO2 cathode than in charged commercial LiCoO2, their oxidizing power should also be weaker than those in commercial LiCoO2 cathode. These results indicate that modifying the surface of LiCoO2 particles can help to suppress the formation of oxygen atoms with high oxidizing power and the decomposition of electrolyte. These will both

Spectroscopic Studies of Solid-Electrolyte Interphase

187

elevate the cycling efficiency of the cell and prolong its cycle life especially when it is deeply delithiated. In addition, the suppression of the formation of oxygen atoms with high oxidizing power will improve the safety of the cell, especially at deeply delithiated state.

3500

5000

9000

naked LiCoO2

MgO-coated-2

MgO-coated-1 4500

8000 3000 7000

4.80V

4000

4.8V

4.5V

4.4V

1500 4.3V

4.8V

6000 4.5V

5000

4.6V

4000 4.4V

4.3V

3000

1000 4.2V

2000

4.2V

Intensity (a.u.)

2000

Intensity (a.u.)

Intensity (a.u.)

2500

3500 3000

4.80V

2500

4.5V

2000 4.40V 1500

4.30V

1000

4.1V

500

4.20V

1000

uncharged

0 815

805

795

785

775

Binding Energy (eV)

0 815

4.15V

uncharged

805

795

785

775

Binding Energy (eV)

500 0 815

805

795

785

775

Binding Energy (eV)

Figure 32 Comparison of the Co 2p spectra of commercial (A, left) and MgO-coated (B, middle) LiCoO2 charged to various voltages. C. (right) MgO-coated LiCoO2 after sputtering. Reproduced from [122] with permission of The Electrochemical Society Inc.

Based on the XPS and IR spectroscopic results, the improvement of the electrochemical performances of the modified cathode material can be explained. First, surface modification to LiCoO2 changes its surface chemistry and morphology. These changes make weak the interaction between the adsorbate (the electrolyte) and the adsorbent and stabilize the EC molecules towards disturbance of LiCoO2 in the electrolyte. The surface chemistry of MgO-coated cathode also prevents the decomposition of the electrolyte prior to the electrochemical cycling of the cathode material. Second, the surface layer on the modified LiCoO2 suppresses the electrolyte decomposition upon electrochemical cycling, ensuring that the electrolyte can withstand higher charge voltages. Third, although surface modification cannot prevent the 4+ formation of Co ions at charged state, the coating layer can at least be a physical barrier between the active LiCoO2 and the electrolyte, preventing their direct contact and suppressing the dissolution of the cathode material. At last, the presence of the interface hinders the formation of oxygen atoms with high oxidizing power. The decrease of the content of such oxygen atoms at or near the cathode surface will also hinder the decomposition of the electrolyte at the charged state and enhances the safety of the cell charged to high voltages.

188

Lithium-Ion Batteries: Solid-Electrolyte Interphase

2400

1800 naked LiCoO 2 1600

coated 2000

4.8V

4.8V 4.7V

1600

4.5V

1200 1000

Intensity (a.u.)

Intensity (a.u.)

1400

4.4V

800 4.3V

600 4.2V

400

4.1V

4.6V 4.5V

1200

4.4V

800

4.3V 4.2V

400

4.15V

200 uncharged

0 541

536

uncharged

531

526

0 540

521

535

530

525

520

Binding Energy (eV)

Binding Energy (eV)

Figure 33 Comparison of O 1s spectra of commercial (left) and MgO-coated (right) LiCoO2 charged to various voltages. Reproduced from [122] with permission of The Electrochemical Society Inc.

M gO -coated

1200

Intensity (a.u.)

1000 4.80V 800

600

400

200 4.15V 0 65

60

55

50

45

Binding Energy (eV)

Figure 34 Mg 1s and Co 3p spectra of MgO coated LiCoO2 charged to various voltages (from bottom to top: 4.15 V; 4.2 V; 4.3 V; 4.4 V; 4.5 V; 4.6 V; 4.7 V and 4.8 V).

The Mg 1s spectra of the MgO/LiCoO2 cathode charged to various voltages are also recorded and shown in Figure 34. The Mg 1s peak is observed at about 50.4e V and does not shift with the charge voltage. The other peak at 60.0 eV is 137 assigned to Co 3p. As the Co cations cannot be expected to accumulate at the

Spectroscopic Studies of Solid-Electrolyte Interphase

189

surface of a LiCoO2 particle upon charging, it can be suggested that the content of the Mg cations or the thickness of the MgO coating decreases with increasing charge voltage. It would be interesting to find out whether the electrolyte decomposition on MgO/LiCoO2 at high voltages is related to the decreasing thickness of the coating layer. Indeed it has been observed that the MgO coating 121 is corroded on cycled LiCoO2 cathodes in our previous study. In the current experiment, as DMC is removed during sample washing and drying, the impact of the nano-MgO coating on the interaction of LiCoO2 with DMC is still not clear (not recognizable). Improved experimental design is performed and further investigation results will be published elsewhere.

4.5

Conclusions

Comparative spectroscopic studies on the electrolyte residue on various substrates shows that the solvent molecules are sensitive to the surface chemistry of the substrates. Electrolyte decomposition occurs and Li2CO3, RCO3Li and CH3CH(CO3Li)CH2CO2Li are formed on uncharged commercial LiCoO2 and nano-MgO particles. Surface modification to commercial LiCoO2 alleviates its disturbance to the configuration of the solvent molecules and suppresses the electrolyte decomposition at charged and uncharged states. The substances produced by electrochemical decomposition of the electrolyte show little differences on commercial and MgO/LiCoO2 charged to high voltages but Li2CO3 is not detected below 4.4 V on MgO/LiCoO2 electrode. Due to the hindrance to the formation of oxygen atoms with high oxidizing power, electrolyte in MgO/LiCoO2 cathode based cells can withstand higher charge voltages. In addition, the coating layer prevents the direct contact of the active cathode material with the electrolyte and suppresses the electrolyte decomposition though the surface modification cannot prevent the formation of 4+ Co ions at charged state. These factors allow the MgO/LiCoO2 cathode to be charged to a higher voltage and ensure higher cycling efficiency than commercial LiCoO2 cathodes. Of course, there might be other factors that help to improve the electrochemical performances of the cell such as the morphology, structure and composition of the coating layer and the interface between the coating layer and the active LiCoO2. Further study is necessary and is in progress.

190

5

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Summary and Comments

In the above discussion, various spectroscopic techniques have been applied to study the SEI layers on both the positive and negative electrodes in lithium ion batteries. These techniques provide information to identify the chemical substances by recognizing some of the important functional groups of a molecule. By tracing the variation of the characteristic bands of some chemical species at different charge/discharge states, the evolution of the SEI layer can be proposed. By identifying some critical substances in the SEI layers, the chemical and electrochemical reactions on the electrode surfaces can be suggested and the properties of the SEI layer be predicted. We agree that no single method/technique can answer all the questions concerning the SEI layers on the electrodes. Joint investigation of several methods/techniques is necessary in most cases and each technique is compensation to the others. We expect that in situ techniques at the molecular level and indicating the microstructures of the SEI layers be developed and a further understanding to the SEI layer be realized. So that the performance of the lithium ion batteries be significantly improved.

Acknowledgments  This work was financially supported by the National Science Foundation of China (NSFC) (Contract No. 50272080) and the Special Funds for Major State Basic Research Project of China (Contract No. 2002CB211802).

References 1. 2. 3. 4. 5. 6. 7.

Nagaura T. and Tazawa K., Prog. Batteries Sol. Cells 9 (1990), 20-22. Idota Y., Nishima M., Miyaki Y., Kubota T. and Miyasaki T., EP0651450 A1. Idota Y., Kubota T., Matsufuji A., Maekawa Y. and Miyasaka T., Science 276 (1997), 1395-1397. Courtney I. A. and Dahn J. R., J. Electrochem. Soc. 144 (1997), 2045-2052. Courtney I. A. and Dahn J. R., J. Electrochem. Soc. 144 (1997), 2943-2948. Liu W. F., Huang X. J., Wang Z. X., Li H. and Chen L. Q., J. Electrochem. Soc. 145 (1998), 59-62. Li H., Huang X. J. and Chen L. Q., Solid State Ionics 123 (1999), 189-197.

Spectroscopic Studies of Solid-Electrolyte Interphase

8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

191

Li H., Huang X. J. and Chen L. Q., J. Power Sources 81-82 (1999), 340345. Zhou G. W., Li H., Sun H., Yu D., Wang Y., Huang X. J., Chen L. Q. and Zheng Z., Appl. Phys. Lett. 75 (1999), 2447-2449. Gao B., Sinha S., Fleming L. and Zhou O., Adv. Mater. 13 (2001), 816. Nazar L. F., Goward G., Leroux F, Duncan M., Huang H., Terr T. and Gaubicher J., Int. J. Inorg. Mater. 3 (2001), 191-200. Li H., Huang X. J., Chen L. Q., Wu Z. and Liang Y., Electrochem. Solid State Lett. 2 (1999), 547-549. Winter M. and Besenhard J. O., Electrochim. Acta 45 (1999), 31-50. Yang J., Wachtler M., Winter M. and Besenhard J. O., Electrochem. Solid State Lett. 2 (1999), 161-163. Whitehead A. H., Elliott J. M. and Owen J. R., J. Power Sources 81-82 (1999), 33-38. Li H., Shi L. H., Lu W., Huang X. J. and Chen L. Q., J. Electrochem. Soc. 148 (2001), A915-922. Shi L. H., Li H., Wang Z. X., Huang X. J. and Chen L. Q., J. Mater. Chem. 11 (2001), 1502-1505. Li H., Wang Q., Chen L. Q. and Huang X. J., Chem. Mater. 14 (2002), 103-108. Tamai H., Matsuoka S., Ishihara M. and Yasuda H., Carbon 39 (2001), 1515-1523. Wang G. X., Ahn J. H., Lindsay M. J., Sun L., Bradhurst D. H., Dou S. X. and Liu H. K., J. Power Sources 97-98 (2001), 211-215. Kim I. S., Kumta P. N. and Blomgren G. E., Electrochem. Solid State Lett. 3 (2000), 493-496. Sazhin S. V., Gorodskii A. V. and Khimchnko M. Y., J. Power Sources 47 (1994), 57-62. Shembel E. M., Maksyuta I. M., Nenuzho L. I., Beloshokov A. I., Rmaumenko A. and Rozhkov V. V., J. Power Sources 54 (1995), 416-420. Winter M., Besenhard J. O., Spahr M. E. and Novak P., Adv. Mater. 10 (1998), 725-763. Beaulieu L. Y., Eberman K. W., Turner R. L., Krause L. J. and Dahn J. R., Electrochem. Solid State Lett. 4 (2001), A137-A140. Brousse T., Retoux R., Herterich U. and Schleich D. M., J. Electrochem. Soc. 145 (1998), 1-4. Yang J., Takeda Y., Imanishi N., Ichikawa T. and Yamamoto O., Solid State Ionics 135 (2000), 175-180.

192

Lithium-Ion Batteries: Solid-Electrolyte Interphase

28. Fong R., von Sacken U. and Dahn J. R., J. Electrochem. Soc., 137 (1990), 2009-2013. 29. Dey A. N. and Sullivan B. P., J. Electrochem. Soc. 117 (1970), 222-224. 30. Peled E., J. Electrochem. Soc. 126 (1979), 2047-2051 31. Peled E., Golodnitsky D. and Ardel G., J. Electrochem. Soc. 144 (1997), L208-L210. 32. Aurbach D., Daroux M. L., Faguy P. W. and Yeager E., J. Electrochem. Soc. 134 (1987), 1611-1620. 33. Goren E., Chusid O. and Aurbach D., J. Electrochem. Soc. 138 (1991), L6L9. 34. Aurbach D. and Chusid O., J. Electrochem. Soc. 140 (1993), L1-L4. 35. Chusid O., Ein-Eli Y. and Aurbach D., J. Power Sources 43-44 (1993), 4764. 36. Aurbach D., Weissman I., Zaban A. and Chusid O., Electrochim. Acta 39 (1994), 51-57. 37. Aurbach D., Ein-Eli Y. and Zaban A., J. Electrochem. Soc. 141 ( 1994), L1-L3. 38. Aurbach D., Ein-Eli Y., Chusid O. Cameli Y., Babai M. and Yamin H., J. Electrochem. Soc. 141 (1994), 603-611. 39. Aurbach D., Zaban A., Schechter A., Ein-Eli Y., Zinigrod E. and Markovsky B., J. Electrochem. Soc. 142 (1995), 2873-2882. 40. Aurbach D. Ein-Eli Y., Markovsky B., Zaban A., Luski S., Cameli Y. and Yamin H., J. Electrochem. Soc. 142 (1995), 2883-2890. 41. Naji A., Ghanbaja J., Willmann P., Humbert B. and Billaud D., J. Power Sources 62 (1996), 141-143. 42. Naji A., Ghanbaja J., Willmann P., Humbert B. and Billaud D., J. Power Sources 63 (1996), 33-39. 43. Aurbach D. and Granot E., Electrochim. Acta 42 (1997), 697-718. 44. Kanamura K., Takezawa H., Schiraishi S. and Takehara Z., J. Electrochem. Soc. 144 (1997), 1900-1906. 45. Dudley J. T. and Wilkinson D. P., J. Power Sources 35 (1991), 59-82. 46. Li H., Huang X. J. and Chen L. Q., Electrochem. Solid State Lett. 1 (1998), 241-243. 47. Nyquist R. A. and Kagel R. O., Infrared Spectra of Inorganic Compounds (Academic, 1971), p. 210. 48. Grant W. B., Schulz H., Hufner S. and Pelzl J., Phys. Stat. Sol. B60 (1973), 331.

Spectroscopic Studies of Solid-Electrolyte Interphase

193

49. Le Bellac D. L., Kiat J., Hedoux A., Garnier P., Grebille D., Guinet Y. and Noiret I., Ferroelectric 125 (1992), 215-220. 50. Li J. Z., Li H., Wang Z. X., Huang X. J., Chen L.Q., J. Power Sources 8182 (1999), 346-351. 51. Aurbach D., Daroux M. L., Faguy P. and Yeager E. B., J. Electroannal. Chem. 297 (1991), 225-244. 52. Aurbach D., Zaban A. and Ein-Eli Y., J. Power Sources 68 (1997), 91-98. 53. Kanamura K., Tamura H. and Takehara Z., J. Electroanal. Chem. 333 (1992), 127-142. 54. Kanamura K., Tamura H., Shiraishi S. and Takehara Z., J. Electroanal. Chem. 394 (1995), 49-62. 55. Aurbach D. and Cohen Y., Electrochem. Solid State Lett. 2 (1999), 16-18 56. Nie S. and Emery S. R., Science 275 (1997), 1102-1106. 57. Moskovits M., Rev. Mod. Phys. 57 (1985), 783 58. Otto A., Mrozek I., Grabhorn H. and Akemann W., J. Phys. Condens. Matter 4 (1992), 1143-1212. 59. Inaba M., Yoshida H., Ogumi Z., Abe T., Mizutani Y. and Asano M., J. Electrochem. Soc. 142 (1995), 20-26. 60. Huang W. W. and Frech P., J. Electrochem. Soc. 145 (1998), 765-770. 61. Zhang X. and Frech R., J. Electrochem. Soc. 145 (1998), 847-851. 62. Wang Z. X., Huang B. Y., Xue R. J., Huang X. J. and Chen L. Q., Solid State Ionics 121 (1999), 141-156. 63. Wang Z. X., Huang B. Y., Xue R. J., Chen L. Q. and Huang X. J., J. Electrochem. Soc. 145 (1998), 3346-3350. 64. Wang Z. X., Huang B. Y., Wang S. M., Xue R. J., Huang X. J. and Chen L. Q., Electrochim. Acta 42 (1997), 2611-2617. 65. Wang Z. X. Huang B. Y., Wang S. M., Xue R. J., Huang X. J., Chen L. Q., et al., J. Electrochem. Soc. 144 (1997), 44-48. 66. Wang Z. X. Huang B. Y., Huang H., Chen L. Q., Xue R. J. and Wang F. S., Electrochim. Acta 41 (1996), 1443-1446. 67. Wang Z. X., Huang B. Y., Huang X. J., Chen L. Q. and Xue R. J., J. Raman Spectrosc. 27 (1996), 901-906. 68. Wang Z. X. Huang B.Y., Huang H., Chen L.Q. and Xue R. J., Solid State Ionics 92 (1996), 265-271. 69. Wang Z. X., Huang B. Y., Wang S. M., Chen L. Q., Huang X. J. and Xue R. J., J. Electrochem. Soc. 144 (1997), 778-786. 70. Wang Z. X., Gao W. D., Chen L. Q., Mo Y. J. and Huang X. J., J. Electrochem. Soc. 149 (2002), E148-E154.

194

Lithium-Ion Batteries: Solid-Electrolyte Interphase

71. Wang Z. X., Gao W. D., Huang X. J., Mo Y. J. and Chen L. Q., J. Raman Spectrosc. 32 (2001), 900-905. 72. Wang Z. X., Gao W. D., Huang X. J., Mo Y. J. and Chen L. Q., Electrochem. Solid State Lett. 5 (2001), A148-A150. 73. Wang Z. X., Gao W. D., Huang X. J., Mo Y. J. and Chen L. Q., Electrochem. Solid State Lett. 5 (2001), A132-A135. 74. Wang Z. X., Huang X. J. and Chen L. Q., J. Power Sources 81-82 (1999), 328-334. 75. Wang Z. X., Huang X. J., Xue R. J. and Chen L. Q., Carbon 37 (1999), 685-692. 76. Wang Z. X., Lu Z. H., Huang X. J., Xue R. J. and Chen L. Q., Carbon 36 (1998), 51-59. 77. Fleischmann M., Hendra P. J. and MacQuillan A., Chem. Phys. Lett. 26 (1974), 163-166. 78. Mattei G., Quagliano L. G. and Pagannonne M., Europhys. Lett. 11 (1990), 373. 79. Mo Y. J., Mattei G. and Pagannone M., Appl. Phys. Lett. 66 (1995), 25912593. 80. Kneipp K., Wang Y., Kneipp H., Pereiman L. T., Itzkan I., Kasari R. R. and Feld M. S., Phys. Rev. Lett. 78 (1997), 1667-1670. 81. Pettering B., Wenning U. and Wetzel H., Chem. Phys. Lett. 67 (1979), 192-196. 82. Van Duyne R. P., Chemical and Biological Applications of Lasers, Vol. 4, ed. by Moore C. B. (Academic, New York, 1979), p. 101. 83. Otto A., Surf. Sci. 75 (1978), L392-L396. 84. Honbo H., Takeuchi S., Momose H., NishimuraK., Horiba T., Muranaka Y. and Kozono Y., Denki Kagaku 66 (1998), 939-944. 85. Aurbach, D., J. Power Sources 89 (2000), 206-218. 86. Mo Y. J. and Wachter P., Surf. Sci. 133 (1983), L452-L458. 87. Aurbach D., J. Electrochem. Soc. 136 (1989), 906-913. 88. Morita M., Nishimura N. and Matsuda Y., Electrochem. Acta 138 (1993), 1721-1726. 89. Li H., Mo Y. J., Pei N., Xu X., Huang X. J. and Chen L. Q., J. Phys. Chem. B104 (2000), 8477-8480. 90. Li H. and Mo Y. J., unpublished data. Raman bands for LiPF6, EC-DEC (1:1 v/o) electrolyte are located at: 363(w), 718(m), 743(w), 892(s), −1 906(s), 978(m), 1119(w), 1457(m), 1485(w), 2940(m) and 2985(m) [cm ]. Raman bands for 1 M LiClO4, PC-DMC (1:1 v/o) electrolyte are located

Spectroscopic Studies of Solid-Electrolyte Interphase

91. 92. 93. 94. 95. 96. 97.

98. 99. 100. 101. 102. 103. 104. 105. 106.

107. 108.

195

at: 197(w), 262(W), 380(w), 458(m), 519(s), 580(w), 633(m), 712(m), 851(m), 918(s), 936(s), 960(m), 1070(w), 1123(w), 1153(w), 1222(w), 1347(w), 1459(m), 1486(w), 1755(w), 1792(w), 2849(w), 2885(w), −1 −1 2942(m), 2968(m), 2995(m) and 3038(w) [cm ]. Peak 936 cm is attributed to LiClO4. Farmer V. C., The Infrared Spectra of Minerals (Mineralogical Society, London, 1974), p. 249. Rahner D., Machill S., Schlorb H., Siury K., Klob M. and Pleith W., Solid State Ionics 86-88 (1996), 891-896. Mo Y. J., Wachter P., Martin U. and Reinhart F. K., J. Phys. Chem. Solids 56 (1995), 975-980. Rauh R. D. and Brummer S. B., Electrochim. Acta 22 (1977), 75-83. Aurbach D. Ein-Eli Y., Chusid O., Cameli Y., Babai Y. and Yamin H., J. Electrochem. Soc. 141 (1994), 603-611. Coogan C. K., J. Chem. Phys. 43 (1965), 823-830. Dollish F. R., Fateley W. G. and Bentley F. F., Characteristic Raman Frequencies of Organic Compounds (John-Wiley & Sons, New York, 1974). Behrendt V. W., Gattow G. and Drager M., Anorg. Z. Allg. Chem. 397 (1973), 237. Wickersheim K. A., J. Chem. Phys. 31 (1959), 863-869. White D., Seshadri K. S., Dever D. F., Mann D. E. and Linevsky M. J., J. Chem. Phys. 39 (1963), 2463-2473. Seshadri K. S., White D. and Mann D. E., J. Chem. Phys. 45(1966), 46974700. Hase Y. and Pagotto Y., Chem. Phys. Lett. 65 (1979), 46-49. Andrew L., J. Chem. Phys. 50 (1969), 4288-4299. Phillips B. A. and Busing J. W. R., Phys. Chem. 54 (1971), 4788. Ostrovskii D., Ronci F., Scrosati B. and Jacobsson P., J. Power Sources 94 (2001), 183-188. Aurbach D., Levi M. D., Levi E., Markovsky B., Gomolsky K., Oesten R. and Heider U., in Lithium Batteries, ed. by Surampudi S. and Marsh R., PV 98-16, p. 173, The Electrochemical Society Proceedings Series (Pennington, NJ, 1998). Thomsas M. G. S. R., Bruce P. G. and Goodenough J. B., J. Electrochem. Soc.132 (1985), 1521-1535. Aurbach D., Levi M. D., Levi E., Teller H., Markvsky B., Salitra G., Heider U. and Heider L., J. Electrochem. Soc. 145 (1998), 3024-3034.

196

Lithium-Ion Batteries: Solid-Electrolyte Interphase

109. Uchiyama T., Nishizawa M., Itoh T. and Uchida I., J. Electrochem. Soc.147 (2000), 2057-2060. 110. Eriksson T., Andersson A. M., Bishop A. G., Gejke C., Gustafsson T. and Thomson J. O., J. Electrochem. Soc. 149 (2002), A69-A78. 111. MacNeil D. D. and Dahn J. R., J. Electrochem. Soc. 148 (2001), A1205A1210. 112. MacNeil D. D. and Dahn J. R., J. Electrochem. Soc. 148 (2001), A1211A1215. 113. Zhang X. R., Kostecki R., Richardson T. J., Pugh J. K. and Ross, P. N. Jr., J. Electrochem. Soc. 148 (2001), A1341-A1345. 114. Balasubramanian M., Lee H. S., Sun X., Yang X. Q., Moodenbaugh A. R., McBreen J., Fischer D. A. and Fu Z., Electrochem. Solid State Lett. 5 (2002), A22-A25. 115. Kweon H. J., Kim S. J. and Park D. G., J. Power Sources 88 (2000), 255A261. 116. Cho J., Kim Y. J., Kim T. J. and Park B., Chem. Mater. 13 (2001), 18. 117. Cho J., Kim G. B., Lim H. S., Kim C. S. and Yoo S. I., Electrochem. Solid State Lett. 2 (1999), 607-609. 118. Amatucci G. G., Blyr A., Sigala C., Alfonse P. and Tarascon J. M., Solid State Ionics 104 (1997), 13-25. 119. Endo E., Yasuda T., Kita A., Yamaura K. and Sekai K., J. Electrochem. Soc. 147 (2000), 1291-1294. 120. Cho J., Kim Y. J., Kim T. J. and Park B., J. Electrochem. Soc. 149 (2002), A127-A132. 121. Wang Z. X., Wu C., Liu L. J., Huang X. J. and Chen L. Q., J. Electrochem. Soc. 149 (2002), A466-A471. 122. Wang Z. X., Huang X. J. and Chen L. Q., J. Electrochem. Soc. 150 (2003), A199-A208. 123. Fortunato B., Mirone P. and Fini G., Spectrochim. Acta 27A (1971), 19171927. 124. Silverstein R. M., Basselr F. C. and Morrill T. C., Spectroscopic Identification of Organic Compounds, 4th ed. (Wiley, New York, 1981). 125. Puchert C. J., Aldrich Library of FTIR spectra, 1st ed., Vols. 1 and 2 (Aldrich Chemical Co., Milwaukee, WI, 1980). 126. Yang C. R., Wang Y. Y. and Wan C. C., J. Power Sources 72 (1998), 6670. 127. Aurbach D. Gamolsky K., Markovsky B., Salitra G., Gofer Y., Heider U., Oesten R. and Schmidt M., J. Electrochem. Soc. 147 (2000), 1322-1331.

Spectroscopic Studies of Solid-Electrolyte Interphase

197

128. Matsumoto K., Kuzuo R., Takeya K. and Yamanaka A., J. Power Sources 81-82 (1999), 558-561. 129. Colthup N. B., Daly L. H. and Wiberley S. E., Introduction to Infrared and Raman Spectroscopy, 3rd ed. (Academic, San Diego, 1990). 130. Kanamura K., Toriyama S., Shiraishi S., Ohashi M. and Takehara Z., J. Electroanal. Chem. 419 (1996), 77-84. 131. Vovak P. and Vielstich W., J. Electrochem. Soc. 137 (1990), 1681-1689. 132. Abbate M., Pen H., Czyzyk M. T., de Groot F. M. F., Fuggle J. C., Ma Y. J., Chen C. T., Sette F., Fujimori A., Ueda Y. and Kosuge K., J. Electron Spectrosc. Rel. Phenom. 62 (1993), 185-195. 133. Itoh T., Anzue N., Mohamedi M., Hisamitsu Y., Umeda M. and Uchida I., Electrochem. Commun. 2 (2000), 743-748. 134. Behrendt V. W., Gattow G. and Drager M., Z. Arorg. Allg. Chem. 397 (1973), 237. 135. Montoro L. A., Abbate M. and Rosolen J. M., Electrochem. Solid State Lett. 3 (2000), 410-412. 136. Kavan L., Krtil P. and Gratzel M., J. Electroanal. Chem. 373 (1994), 123131. 137. Handbook of X-Ray Photoelectron Spectroscopy (Perkin-Elmer, Physical Electronics Division).

CHAPTER 4

SCANNING PROBE MICROSCOPY ANALYSIS OF THE SEI FORMATION ON GRAPHITE ANODES MINORU INABA Department of Molecular Science and Technology, Faculty of Engineering, Doshisha University, Kyotanabe-shi, Kyoto 610-0321, Japan E-mail: [email protected] ZEMPACHI OGUMI Department of Energy & Hydrocarbon Chemistry, Graduate School of Engineering, Kyoto University, Sakyo-ku, Kyoto 606-8501, Japan E-mail:[email protected]

1

Introduction

One of the most important breakthroughs for lithium-ion batteries (LIBs) in the early stage of development was no doubt the use of graphite as a lithium 1–4 reservoir at the anode. In 1981, Sanyo first reported in a patent that lithium ion can be electrochemically intercalated into the graphite anode during charging, 5 and deintercalated during discharging. The reversibility of graphite anode depends greatly on the kind of electrolyte solution. Early attempts to electrochemical lithium intercalation into graphite in propylene carbonate (PC)based solutions, which have been used in primary lithium metal cells for more than two decades, were mostly unsuccessful owing to poor compatibility – between graphite and PC.6 9 When a graphite electrode is polarized to negative potentials in a PC-based solution, the solvent decomposes ceaselessly at around 0.9 V vs. Li+/Li and thereby lithium ion is not intercalated. Later this problem was overcome by the use of ethylene carbonate (EC) instead of PC,10 and ECbased mixed solvent systems such as EC + dimethyl carbonate (DMC), EC + diethyl carbonate (DEC), and EC + ethyl methyl carbonate (EMC) are currently used in commercially available LIBs employing graphite as anodes.1 It has been widely believed that a protective surface film, which is conductive for lithium ion but electronically insulating, is formed on graphite anode via reductive decomposition of electrolyte solution in the initial stage of 11, 12 charging. The passivating film, often called the solid electrolyte interface 11 (SEI), suppresses further solvent decomposition and enables lithium ion to be

198

SPM Analysis of the SEI Formation on Graphite Anodes

199

intercalated within graphite. In addition, it plays a beneficial role in improving the safety and cycleability of LIBs. The SEI formation on graphite anode is thus a prerequisite for its stable charging and discharging; however, it is the primary cause for bringing about the irreversible capacity through consumption of a considerable amount of charge. Numerous efforts have been made in search for good solvent systems that give superior SEI with a minimal consumption of charge. For this purpose, understanding of the SEI composition, morphology, stability, formation mechanism, and its influence on battery performance is very important, and has been a focus of much research over the past decade. For example, the composition of SEI formed on graphite anodes has been 13–53 extensively studied with a variety of analytical tools. However, the results in 13–22 these studies are somewhat controversial. Aubach et al. have investigated surface reactions occurring at graphite anodes in various nonaqueous solutions by in situ Fourier-transform infrared spectroscopy (FTIR) and impedance spectroscopy, and suggested that the major constituent of the SEI formed in ECbased solutions is a lithium alkyl carbonate (CH2OCO2Li)2, which is a reduction 22 product of EC. The reaction scheme proposed by them is shown in Figure 1. 23 Yoshida et al. detected carbon monoxide and ethylene by gas chromatography (GC) from LIBs after being cycled, and suggested the presence of a lithium alkoxide (CH2OLi)2 and Li2CO3 in the SEI. X-ray photoelectron microscopy 24 25, 26 (XPS) and transmission electron microscopy (TEM) analysis revealed that the film contains inorganic compounds such as LiOH and Li2CO3. The authors detected oligomers that have repeated oxyethylene units from the SEI formed on graphite anodes by pyrolysis–gas chromatography–mass spectroscopy (Py–GC– MS), and suggested that the SEI consists of polymerized compounds similar to 27 poly(ethylene oxide) (PEO). The presence of polymers in the SEI on graphite 28, 29 30–33 anodes has been also proposed by other researchers. Peled et al. analyzed reaction products formed on the basal and the edge planes of highly oriented pyrolytic graphite (HOPG) that was cycled in EC-based solutions by XPS, time-of-flight secondary ion mass spectrometry (TOF–SIMS), etc. They have reported that the SEI on the edge plane is rich in inorganic compounds, whereas that formed on the basal plane is rich in organic compounds (mainly polymers). The most interesting and important phenomenon to be clarified is the aforementioned difference in SEI-forming ability between EC and PC, and the origin of the difference will be a clue to understanding the mechanism of the SEI formation on graphite anode. These two solvents have very similar chemical structures, and the electron-donating ability of the methyl group of PC

Lithium-Ion Batteries: Solid-Electrolyte Interphase

200

would not be so high as to significantly change its electrochemical properties such as the ease to reduction. It seems that there exist some factors other than the chemical structure of solvent that determine the SEI-forming ability. In this respect, the role of solvent co-intercalation has been underestimated because several research groups attributed the exfoliation of the graphite in PC-based 34, 35 solutions to co-intercalation of PC molecules with lithium ion. Furthermore, 36, 37 Besenhard et al. reported that solvent co-intercalation occurs even in ECbased solutions. They studied the crystal expansion of HOPG during electrochemical reduction in an EC-based solution by dilatometry, and observed a drastic expansion of the graphite matrix (> 150%) at potentials more negative + than 1.0 V vs. Li /Li. They attributed this expansion to solvent co-intercalation, and concluded that the intercalated solvent further decomposes to form an immobile product remaining between the graphene sheets and that this reduction product prevents further solvent co-intercalation and the exfoliation of 36 the graphite sheets as shown schematically in Figure 2.

O

e

+ Li

C O

O

e

O

O

H2C

CH2

CH2 2+ CO3

Low CEC + 2Li

+ Li /O + /+ Li O C /+ O + /Li

CH2

CH2 OCO2 Li+

+ Li

C

graphite

CH2

CH2 (EC)

H2C

further electron transfer e

CH2 OCO2 Li+

O

High CEC EC (Nucleophilic attack)

CH2

or + Li

solution Disproportionation

+

CH2 OCO2

Li O /-

/+C O /-

CH2 OCO2 + 2Li

CH2 OCO2 Li+ CH2 OCO2

+ Li O /-

+

Li

+ Li

(CH2-OCO2Li)2

CH2 CH2

graphite

CH2 CH2

Partial polymerization graphite + passivation Li

O

/CH2 CH2

(CH2CH2)n

+ Li

+ Li

O /-

O /-

C

/+O /-

CH2 CH2

/+

O

/-

C

/+O /-

Li+ solution

CH2CH2

Figure 1 Reduction mechanisms of EC on graphite proposed by Aurbach et al. Reproduced from [22] with permission of Elsevier Science.

SPM Analysis of the SEI Formation on Graphite Anodes

201

(a) Graphene layer Donor solvent

(b)

Decomposition solvent Li Film component

(c)

Figure 2 Solvent cointercalation model for surface film formation on/in graphite proposed by Besenhard et al. (a) before reaction, (b) formation of ternary GIC, Lix(solv)yCn, and (c) film formation by decomposition of Lix(solv)y. Reproduced from [36] with permission of Elsevier Science.

Electrochemical scanning probe microscopy (SPM) is a useful tool that is capable of giving structural and topographical information of interfacial processes in various electrolyte solutions under potential control.38 Knowledge of surface structure could be crucial to the understanding of the SEI formation that is taking place at the electrode surface. Using scanning tunneling – – microscopy (STM) 39 44 and atomic force microscopy (AFM),45 48 the authors have investigated topographical changes of the basal plane of HOPG in several kinds of electrolyte solutions in order to clarify the mechanism of the SEI formation on graphite anode. In this chapter, the authors focus on the role of solvent co-intercalation in SEI formation on graphite anode, and discuss the mechanistic aspects of SEI formation.

2

Charge and Discharge Characteristics of Graphite Anode in EC- and PC-Based Solutions

Figure 3 compares the first charge and discharge curves of composite anodes made of natural graphite powder (The Kansai Coke and Chemicals, NG-7) in 1 – M (M = mol dm 3) LiClO4 dissolved in EC and PC. In 1 M LiClO4/EC, the potential dropped rapidly after subtle retardation at approx. 0.8 V vs. Li+/Li during the first charging. The main intercalation and deintercalation of lithium

Lithium-Ion Batteries: Solid-Electrolyte Interphase

202

ion take place at potentials < 0.25 V, accompanied by successive stage transformations between different stages of lithium–graphite intercalation – compounds (Li–GICs).49 51 The charge consumed during the first charging –1 (approx. 415 mAh g ) was not fully recovered during the following discharging. The capacity that cannot be recovered is called the “irreversible –1 capacity” (Qirr), 65 mAh g in this case, which is generally believed to be consumed by SEI formation as mentioned earlier. The reversible capacity Qrev –1 was 350 mAh g , which is close to the theoretical value of graphite anode –1 (372 mAh g ). On the contrary, the potential was kept nearly constant at about –1 0.9 V up to 2000 mAh g , and then dropped suddenly to 0 V (Figure 3(b)) in 1 M LiClO4/PC. The whole charge was consumed by solvent decomposition and exfoliation of graphite, and the electrode did not have any appreciable discharge capacity during the following discharging.

2.0 +

Potential / V vs. Li /Li

O O

O

PC O O

O

1.5

Propylene carbonate (PC) Ethylene carbonate (EC)

1.0

0.5

0.0

EC

0

100

200

300

Capacity / mAh g

400

-1

Figure 3 First charge and discharge characteristics of natural graphite (NG-7) in 1 M LiClO4 dissolved in PC and EC.

Figure 4 shows cyclic voltammograms (CVs) of freshly cleaved HOPG –1 basal plane obtained at a scan rate of 5 mV s in 1 M/LiClO4/EC + DEC and 41 1 M LiClO4/PC. In each solution, several cathodic peaks appeared in the range of 0.5-1.0 V on the first cathodic sweep, which are related to solvent decomposition and SEI formation processes. The number of cathodic peaks ranges from three to six depending on the kind of solvent, which implies that

203

SPM Analysis of the SEI Formation on Graphite Anodes

solvent decomposition and surface film formation processes are not a simple reaction. The first reduction peak rose at 1.1 and 1.0 V in the EC + DEC and PC solutions, respectively. The cathodic peaks fully disappeared on the second sweep in EC + DEC, which indicates that EC + DEC gives a stable SEI during the first cycle. In the case of PC, however, large reduction currents were observed even on the second sweep, which shows that the protective film was not easily formed in PC. This result is consistent with the charge and discharge characteristics shown in Figure 3.

(a) EC+DEC

(b) PC 2nd

2nd 1st

1st 100 µ A cm

100 µ A cm



 Potential / V

-2

-2

!



 Potential / V

Figure 4 Cyclic voltammograms of HOPG basal plane (0.20 cm2) in 1 M LiClO4 dissolved in (a) EC+DEC (1:1 by volume) and (b) PC. v = 5 mV s–1. Reproduced from [41] with permission of Elsevier Science.

3

Morphology Changes of HOPG Basal Plane in the Initial Stage of Solvent Decomposition

STM images of HOPG basal plane were obtained with a STM unit placed in an argon glove box, in which the dew point was kept < –60°C. Strict inertness of the atmosphere is very important for STM measurements because the surface reactions on graphite anode are very sensitive to moisture and oxygen. A typical STM image of HOPG basal plane is shown in Figure 5(a), which was obtained + 39, 40 A clear step of 3-nm height is at 2.8 V vs. Li /Li in 1 M LiClO4/EC + DEC.

!

204

Lithium-Ion Batteries: Solid-Electrolyte Interphase

seen horizontally in the image, which corresponds to nine layers of graphene sheets. The potential of the HOPG sample was lowered stepwise from 2.8 V, and the morphology change of the surface was observed by STM. When the potential was stepped to 1.1 V for 30 s (Figure 5(b)), at which the first reduction peak rose in Figure 3, part of the basal plane was raised by about 1 nm. The 39–41 authors called such features the “hill-like” structures in their original papers. After the potential was kept at 1.1 V for 4 min, another hill-like structure appeared in the vicinity of the step edge. The height of both hills was 0.8–1 nm, and the hilltop was atomically flat. The shape of the hill at the step edge clearly indicates that it was formed from the step edge and then spread out. When the hilltop was observed with an atomic resolution by STM, typical atomic images of graphite basal plane, every other atom on the hexagonal carbon network of the graphite sheet spaced by approx. 0.25 nm on a two-dimensional triangular 52 lattice, were obtained. This fact indicates that the top surface consisted of graphite sheets of ABAB.... stacking, and thereby the hill was an interior structure raised by something inserted beneath the surface. The observed height of the hill-like structure, 0.8–1 nm, is comparable to the interlayer spacings of stage–1 ternary GICs of alkali metal with organic solvent molecules, such as tetrahydrofuran (THF) and dimethoxyethane (DME), prepared by a solution 53 method. It is thus most probable that solvent co-intercalatio (intercalation of solvated lithium ion) occurred at this potential to form the hill-like structures.

(a) at 2.8 V

(b) after 0.5 min at 1.1 V

(c) after 4 min at 1.1 V

Hill

Hill

 QP  QP

 QP

Hill

Step

0

500 nm

0

500 nm 0

500 nm

Figure 5 STM images (500 × 500 nm) and height profiles of HOPG basal plane obtained (a) 2.9 V, and at (b) 0.5, (c) 4 min after the potential was stepped to 1.1 V in 1 M LiClO4/EC+DEC. The potential of the Pt/Ir tip was 3.0 V. Reproduced from [40] with permission of The American Chemical Society.

SPM Analysis of the SEI Formation on Graphite Anodes

205

Figure 6 shows surface morphology changes obtained for another HOPG 41 sample in 1 M LiClO4/EC + DEC. In this case, the hill-like structure appeared when the potential was stepped at 0.95 V for 1 min (Figure 6(b)). After the potential was kept at 0.75 V for 1 min, a significant change in surface morphology was observed as shown in Figure 6(c). Large swellings in irregular shapes (called “blisters”) were formed on the surface. The maximum height of the blisters was approx. 20 nm, which was much higher than that of the hills (approx. 1 nm). These blisters seem to have been formed by accumulation of decomposition products of the solvated lithium ions that had been intercalated beneath the surface. Similar blisters were observed in 1 M LiClO4 dissolved in 41 EC + DME. These morphology changes suggest that solvent co-intercalation plays an important role in the initial stage of SEI formation on graphite anode in EC-based solutions, which is in agreement with the “solvent co-intercalation 36, 37 model” proposed by Besenhard et al. as mentioned earlier.

(a) 1.1 V (1 min)

(b) 0.95 V (1 min)

(c) 0.75 V (1 min)

Hill 1 nm

0

2 µm 0

2 µm 0

Blister

2 µm

Figure 6 STM images (2 × 2 µm) of HOPG basal plane surface observed at 2.8 V after the potential was kept at (a) 1.1, (b) 0.95, and (c) 0.75 V for 1 min in 1 M LiClO4/EC+DEC. The tip potential was 3.0 V. Reproduced from [41] with permission of Elsevier Science.

In contrast to the above results in the EC-based solutions, morphology 41 changes in a 1 M LiClO4/PC were quite different as shown in Figure 7. Neither hill-like structures nor blisters were formed in 1 M LiClO4/PC, but only rapid exfoliation and rupturing of graphite layers occurred. At 0.75 V, the original step-and-terrace structure of HOPG basal plane was completely lost (Figure 7(c)). Because the exfoliation of graphite sheets leads to regeneration of fresh edge planes, stable SEI should not be formed in the solution.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

206

(a) 1.1 V (30 s)

(b) 0.95 V (30 s) A B

B

(c) 0.70 V (30 s)

A C

C

C

D D

D E

E 0

1 µm 0

E 1 µm 0

1 µm

Figure 7 STM images (1 × 1 µm) of HOPG basal plane surface observed at 2.8 V after the potential was kept at (a) 1.1 V, (b) 0.95 V, and (c) 0.7 V for 30 s in 1 M LiClO4/PC. The tip potential was 3.0 V. Reproduced from [41] from Elsevier Science.

The results obtained by STM revealed that solvent co-intercalation is involved in the initial stage of solvent decomposition (and SEI formation) in both EC- and PC-based solutions. Nevertheless, SEI-forming abilities in these solvent systems are greatly different. Although it is not clear why graphite layers exfoliate only in PC-based solution, it seems that the vigorous exfoliation is partly due to faster kinetics of intercalation of solvated lithium ion or its subsequent decomposition between graphite layers. Other factors may also be + 10, 54 raised, such as gas evolution upon decomposition of Li(PC)n , and a greater interlayer stress caused by the co-intercalation of PC molecules with a larger 41 molecular size.

4

SEI Formation in EC-Based Solutions

In the previous section, morphology changes were observed by STM, and the results indicated that solvent co-intercalation plays an important role in the initial stage of SEI formation in EC-based solutions. Unfortunately, clear images could not be obtained at potentials lower than 0.75 V by STM. This fact implies that an insulating layer is formed on the surface at lower potentials. Hence electrochemical AFM was employed instead of STM to clarify the whole picture of what happens during the SEI formation.

SPM Analysis of the SEI Formation on Graphite Anodes

207

Panel (a) in Figure 8 shows a cyclic voltammogram at a slow scan rate of –1 45 0.5 mV s of HOPG basal plane in 1 M LiClO4/EC + DEC. In the first cycle, three major cathodic peaks appeared at about 1.0, 0.8 and 0.5 V. These cathodic peaks disappeared in the second cycle, and hence are attributed to irreversible decomposition reactions of the electrolyte solution that are closely related to SEI formation as mentioned in the previous section. A large cathodic current rise observed at potentials close to 0.0 V could be assigned to lithium intercalation because of the presence of an anodic lithium deintercalation peak at about 1.0 V. However, the charge consumed for the current rise at around 0 V was much greater than that for the anodic peak, and hence a substantial fraction of the cathodic current at around 0 V was consumed by irreversible processes such as solvent decomposition. AFM images (b)–(f) in Figure 8 shows morphology changes of a 5 × 5 µm area of the HOPG basal plane obtained during the first cycle shown in 8(a). The arrows in parentheses denote the direction of raster of the micro-cantilever; for example, the top and bottom scanning lines of Figure 8(c) were obtained at 1.10 and 0.95 V, respectively. Figure 8(d) shows the morphology in the potential range of 0.95–0.80 V. Many “hill-like” structures, which are formed by solvent co-intercalation, are again seen in the lower part of this image. The height of each structure was either 1 or 2 nm. The hills overlapped with one another so that the pattern made by hill formation was very complicated. At potentials more negative than 0.65 V in Figure 8(e), particle-like precipitates appeared on the HOPG surface. The number of the precipitates increased with lowering the potential down to 0.0 V as shown in Figure 8(f). The precipitates are considered to be decomposition products of solvent molecules, – such as lithium alkoxides,13, 22, 55 lithium alkyl carbonates,13, 22, 55 59 and their polymerized compounds.27, 31, 32 The authors analyzed SEIs formed on natural graphite flakes in EC-based solutions after cycling by Py–GC–MS.27 Figure 9 shows a typical gas chromatogram of the SEI formed on natural graphite flakes in 1 M LiClO4/EC + DEC.27 In addition to solvent molecules (EC, DEC, DME), many oxygen-containing organic compounds were detected by Py–GC–MS. Of these, ethylene glycol, di(ethylene glycol), and tri(ethylene glycol) methyl ester are oligomers that have one, two, and three oxyethylene units, respectively, which were formed by reductive decomposition of EC. The presence of these oligomers suggested that the SEI should contain longer polymerized compounds with repeated oxyethylene units that are similar to poly(ethylene oxide) (PEO). Such polymerized compounds are most probably responsible for the precipitates observed on the basal plane in Figure 8.

208

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Figure 8 Cyclic voltammogram (a) and AFM images (5 × 5 µm) of the HOPG basal plane surface obtained at (b) 2.9 V before CV, (c) 1.10–0.95 V, (d) 0.95–0.80 V, (e) 0.65–0.50 V, and (f) 0.20– –1 0.05 V during the first cycle at 0.5 mV s in 1 M LiClO4/EC+DEC (1:1). Reproduced from [45] with permission of The Electrochemical Society, Inc.

209

SPM Analysis of the SEI Formation on Graphite Anodes

The precipitates formed at lower potentials were scraped off at 2.9 V by repeated AFM scanning. Figure 10(a) shows the AFM image of an expanded 45 area (10 × 10 µm) including the 5 × 5 µm area observed during the first cycle. Many precipitates are clearly seen on the surface outside the 5 × 5 µm area, although they are almost completely scraped off inside the 5 × 5 µm area. Inside the 5 x 5 mm area, many swellings (blisters) formed beneath the surface can be seen. From the height profile, the thickness of the precipitate layer in Figure 10(a) was roughly estimated to be 40 nm. After the expanded image in Figure 10(a) was obtained, the second cycle of CV was performed, while a 2 × 2 µm area nearly at the center of Figure 10(a) was continuously observed. Figure 10(b) shows the AFM image of an expanded area of 10 × 10 µm after the second cycle. Outside a rectangular hole (2 × 2 µm) observed during the second cycle, many precipitates were formed even inside of the 5 × 5 µm area where the precipitates had been completely scraped off after the first cycle. The precipitates in this region were formed in the second cycle, and their thickness was estimated to be about 40 nm from the height profile. Furthermore, the precipitate layer outside of the 5 × 5 µm area grew from 40 to 70 nm during the second cycle. The latter fact indicates that the basal plane was not completely passivated after the first cycle, but the precipitate layer grew during the second cycle. Similar results by AFM observation were reported for HOPG in 1 M 60,61 LiClO4 /EC + DMC by Novak et al., though the thickness of their SEI layer was slightly thinner. OR

DME DECRO

OR

RO

O OR

OR 1,3-butanediol

O RO

OR O

EC RO

O O

R = (H), Li, or COOLi

Retention time (min)

Figure 9 Gas chromatogram of thermally decomposed products at 300°C of the SEI formed on natural graphite flakes after a charge and discharge cycle in 1 M LiClO4/EC+DEC. Reproduced from [27] with permission of Elsevier Science.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

210

The results of AFM observation revealed that SEI formation on graphite anodes involves the following two different steps: (i) the solvent cointercalation and its decomposition beneath the surface at around 1 V and (ii) direct decomposition of solvents on the basal plane to form a precipitate layer at lower potentials. These steps are schematically shown in Figure 11. The intercalation of solvated lithium ions is not so vigorous in EC-based solutions, and its decomposition products do not damage the graphite host. It is reasonable to think that the presence of the decomposition products (blisters) in the interlayer space of graphite prevents further solvent co-intercalation. The cointercalation of solvent molecules and their decomposition are one of causes for bringing about the irreversible capacity of graphite anodes. The degree of solvent co-intercalation is significantly affected by the kind of co-solvents in EC-based solutions, which will be discussed in the following section. It is therefore important to choose a solvent system that allows a minimal amount of solvent co-intercalation. (a) after 1 st cycle

(b) after 2 nd cycle

A

B B

AB A

A

40 nm

B

AB A

70 nm

40 nm

B

Figure 10 AFM images of expanded areas (10 x 10 µm) and height profiles of HOPG basal plane surface obtained at 2.9 V after (a) the first and (b) the second cycle of CV. The dotted square shows the area observed by AFM during the first cycle, and the solid square shows the area observed during the second cycle of CV. Reproduced from [45] with permission of The Electrochemical Society, Inc.

SPM Analysis of the SEI Formation on Graphite Anodes

211

On the other hand, it is clear that the precipitate layer has a role in suppressing further reductive decomposition of solvent molecules on the basal plane. The layer most probably consists of polymerized compounds with oxyethylene units as mentioned above. In EC + DEC, the precipitate layer is fairly porous as shown in Figure 10(a). Hence the surface was not completely passivated during the first cycle, and the precipitate layer grew even during the second cycle, which will lead to an irreversible capacity in practical batteries. In this respect, a solvent system or an additive that gives a dense layer of decomposition products are preferable in order to reduce the irreversible capacity.

In E C -b ase d so lutio ns

Lithium ion Solvent molec ule Solvated lithium ion

C o in te rc a la tio n o f L i(s o lv)n

H ill

P recip itate la yer

B lister

Figure 11 Schematic models of SEI formation in EC-based solutions.

5

Effect of Co-solvent on Solvent Co-intercalation in EC-based Solutions

This SEI-forming ability is a unique property of EC, and thus EC is exclusively used as a solvent in commercially available LIBs that employ graphite anodes. Because EC is highly viscous and hence gives a poor conductivity even at ambient temperature,62 it is mixed with linear alkyl carbonates such as DEC, DMC, and EMC, which have much lower viscosities than EC,63, 64 and the resulting binary or ternary mixed solutions are widely used in commercially available LIBs. The choice of co-solvent is an important issue because it greatly affects not only the conductivity, but also the performance of graphite anodes such as reversible/irreversible capacities and cycleability because the physicochemical properties of SEI formed on graphite anodes depend on the

212

Lithium-Ion Batteries: Solid-Electrolyte Interphase

kind of co-solvent.65 69 To know the correlations between the kind of co-solvents and the properties of the resulting SEI is essential to improve the performance of LIBs. Parts (a), (b), and (c) of Figure 12 show AFM images (5 × 5 µm) of HOPG basal plane obtained in EC+ DEC, EC + DMC, and EC after the first cycle of –1 48 CV at 0.5 mV s . Each image was obtained after the precipitate layer was completely removed. Swellings (blisters) formed by solvent cointercalation are seen in each image. It should be noted that the number of the swellings on the surface was much smaller in EC + DMC (Figure 11(b)) and in EC (Figure 11(c)) than in EC + DEC. This fact indicates that intercalation of solvated lithium ion, i.e. solvent co-intercalation, took place more extensively in EC + DEC than in the other solutions. 68, 69 Smart et al. studied the effect of electrolyte composition on charge and discharge characteristics of a synthetic graphite (KS44) electrode. They reported that the KS44 electrode exhibited a larger irreversible capacity and a smaller reversible capacity in EC + DEC than in EC + DMC up to the fifth cycle. Yang 66 et al. also reported that graphitized meso-carbon micro beads exhibited a smaller reversible capacity in EC + DEC than in EC + DMC in the first cycle. In any of these reports, they did not refer to the reason for the difference in capacity between EC + DEC and EC + DMC. Solvent co-intercalation is most probably responsible for the difference because it consumes charge irreversibly and reduces the intercalation sites for lithium ions in graphite anodes. –

(a) EC + DEC

0

(b) EC + DMC

5 µm

0

(c) EC

5 µm

0

5 µm

Figure 12 AFM images (5 × 5 µm) of HOPG basal plane surface obtained at 2.9 V after the first –1 cycle of CV at 0.5 mV s in 1 M LiClO4 dissolved in EC+DEC (1:1), EC+DMC (1:1), and EC. Reproduced from [48] with permission of Elsevier Science.

213

SPM Analysis of the SEI Formation on Graphite Anodes

To understand the reason why the difference is brought about, the ion– solvent interactions in the three electrolyte solutions were investigated by laser Raman spectroscopy. From the results of Raman measurements, apparent solvation numbers were calculated to analyze quantitatively the interactions between lithium ions and solvent molecules. The apparent solvation number (n) 63 can be calculated using the following equation: n = (CM·IS)/CS(IS + IF)

(1)

where CM is the total molar concentration of solvent molecules in the solution, CS is the salt concentration, IS is the integrated scattering intensity of solvating molecules, and IF is the integrated scattering intensity of free solvent molecules. The calculated values are summarized in Table 1. The total solvation number, n(total), varies roughly from 3 to 5. In each solution, lithium ions are preferentially solvated by EC molecules. DEC participates in solvation and replaces a quarter of EC molecules in the EC + DEC solution, whereas a lithium ion was scarcely solvated by DMC in EC + DMC. The preferential solvation of EC is easily understandable when one compares the donor numbers of the 63 solvents, EC (16.4), DEC (16.0), and DMC (15.1). The other feature is that DEC participates in solvation in EC + DEC, whereas DMC does not in EC + DMC, which is also reasonable because the donor number of DEC is higher than that of DMC.

+ Table 1. Apparent solvation numbers of Li with EC, DEC, and DMC in some EC-based a electrolyte solutions. From [ 48].

Solvent systems

n(EC)

n(DEC)

n(DMC)

n(total)

EC

4.6

–

–

4.6

EC + DEC (1:1)b

3.1

1.1

–

4.2

2.9

–

0.2

3.1

EC + DMC (1:1)b a

b

Lithium salt: 1 M LiClO4; by volume

70, 71

studied the solvation state of lithium ions in various Matsuda et al. EC-based solutions by electrospray ionization–mass spectroscopy (ESI–MS). + + They reported that two species, [Li(EC)2] and [Li(EC)3] , in which lithium ions

214

Lithium-Ion Batteries: Solid-Electrolyte Interphase

were solvated only by EC molecules, were major solvated species in 5 mM LiClO4 dissolved in EC, EC + DEC, and EC + DMC. They also detected a small + + amount of [Li(EC)(DEC)] in EC + DEC, but no [Li(EC)(DMC)] in EC + DMC. Their results are in good agreement with the results obtained by Raman spectroscopy, though the absolute values for the solvation number are somewhat different. It should be noted that solvent co-intercalation took place intensively in EC + DEC, in which lithium ion is solvated not only by EC, but also by DEC. In the other solvent systems, lithium ions are solvated exclusively by EC molecules, and solvent co-intercalation was less vigorous. These results may suggest that the degree of solvent co-intercalation is greatly enhanced by the solvation of linear alkyl carbonates such as DEC and DMC to lithium ions. However, it is not easy at present to explain why the presence of linear alkyl carbonates in the solvation shell enhances solvent co-intercalation. More detailed information about the structures and the ion-solvent interactions of lithium ion in non-aqueous solutions are necessary to answer this question.

6

Additives in PC-Based Solutions

The poor compatibility between graphite electrode and PC results from intensive co-intercalation of PC molecules at about 0.9 V accompanied by vigorous exfoliation of graphite layers, and thereby PC-based solutions cannot be used in LIBs using graphite anode, as mentioned earlier. Nevertheless, PCbased solutions are attractive as electrolyte solutions in LIBs because of their 72 superior ionic conductivity at low temperatures. It has been reported that the addition of certain kinds of organic molecules (typically 5% by volume) to PCbased solutions greatly suppresses solvent decomposition and graphite exfoliation, and enables lithium ion to be intercalated into graphite. 73,74 These include chloroethylene carbonate (Cl–EC), vinylene carbonate 75 76 77 (VC), ethylene sulfite (ES), propylene sulfite (PS), fluoroethylene sulfite 78 79 79 (FEC), methyl chloroformate, α–bromo–γ–butyrolactone, t–butylene 80, 81 58, 82 and 12–crown–4 (12–C–4). In addition to these carbonate (t–BC), additives, co-solvents, such as dimethylsulfoxide (DMSO), diethoxymethane (DEM), dimethoxymethane (DMM), and diethoxyethane (DEE) are also 83 effective for stable SEI formation in PC-based solutions. The molecular structures of these additives and co-solvents are summarized in Figure 13. It seems that all these additives give stable SEI layers on graphite surface;

215

SPM Analysis of the SEI Formation on Graphite Anodes

however, the roles of these additives and co-solvents in SEI formation have not been completely clarified yet.

Main solvents O O

O O

O

O

O

O

TFPCCF3

EC

O

PC

Additives O

O

O O

O

O

c-BC

BC

Co-solvents

O

O O

S O

O

O

O

VC

FEC

O

O

S O

O

O

F

O

O

O

O

DMM

O

O

O

DEM

O O

Cl

DMSO

α-bromoCl-EC Cl γ-butyrolactone

PS

S

O

ES

O Br

O

O

O

O

O O

O

methyl chloroformate t-BC

O

O

O

DEE

12-C-4 Figure 13 SEI-forming ability of various cyclic carbonates, and effective additives and co-solvents in PC based solutions. (O) good SEI-forming ability; (X) poor SEI-forming ability.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

216

6.1

Roles of VC, FEC, and ES as Additives

In this subsection, the authors focus on three additives, VC, FEC, and ES, and summarize their roles in SEI formation on graphite anode. The results of charge and discharge tests of composite graphite anodes (NG-7) in 1 M LiClO4/PC containing the additives (3 wt% each) are shown in Table 2.46 Figure 14 shows the variations of discharge capacity with cycle number in the presence of the additives.46 All the three additives gave good cycleability to graphite anode in PC-based solutions, and the capacity retentions were 96, 88, and 85% at the 50th cycle for PC + VC, PC + FEC, and PC + ES, respectively. In addition, the coulombic efficiencies were high, except for the first cycle, in these electrolyte system, and were comparable with that in EC+ DEC, as shown in Table 2. These results confirm that all the three additives, VC, FEC, and EC, gave effective SEI layers on graphite surfaces in PC solutions, which enables lithium ion to be electrochemically intercalated into and deintercalated from graphite anode. Figure 15 shows CVs of HOPG basal plane in the first cycle between 2.9 –1 46 and 0.0 V at 0.5 mV s in the presence of the additives. In each solution, a distinct reduction peak appeared in the range 1.0–1.5 V in the first cycle. The peak potentials were 1.3, 1.1 and 1.0 V in PC+ VC, PC + FEC, and PC + ES, respectively. These reduction peaks disappeared in the second cycle, and hence are attributed to irreversible decomposition of the electrolyte solutions that are closely related to SEI formation. Table 2. Charge and discharge capacities, and Coulombic efficiencies of graphite anode (NG-7) a during fifty cycles in various electrolyte solutions. From [46]. –1

Charge capacity/mAh g / Discharge capacity/ mAh g Solution

PC + 3 wt%VC PC + 3 wt%FEC PC + 3 wt%ES

EC + DEC (1:1) a

Lithium salt: 1 M LiClO4.

–1

(Coulombic efficiency/%) 1st

10th

30th

50th

439/363

366/363

356/355

350/350

(82.7)

(99.2)

(99.7)

(100)

446/353

361/352

348/440

425/311

(79.1)

(97.2)

(94.8)

(98.7)

554/356

353/349

337/336

320/319

(64.3)

(98.9)

(99.7)

(99.7)

438/365

371/364

368/357

351/351

(83.3)

(98.1)

(99.7)

(100)

217

Discharge Capacity / mAhg

-1

SPM Analysis of the SEI Formation on Graphite Anodes

400 350 300 250 200 VC addition FEC addition ES addition

150 100 50 0 0

10

20 30 Cycle Number

40

50

Figure 14 Variations of the discharge capacity with cycle number for natural graphite powder (NG-7) in 1 M LiClO4/PC containing 3 wt% VC, FEC, and ES. Reproduced from [46] with permission of The American Chemical Society.

3 wt.% VC 3 wt.% FEC 3 wt.% ES

0.0

0.5

1.0

1.5

2.0

2.5

3.0

+

Potential / V vs. Li /Li Figure 15 Cyclic voltammograms of HOPG basal plane in 1 M LiClO4/PC containing 3 wt% VC, FEC, and ES. Sweep rate: 0.5 mV s–1. Reproduced from [46] with permission of The American Chemical Society.

218

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Figures 16 shows morphology changes (5 × 5 µm) of HOPG basal plane in 1 M LiClO4/PC + VC obtained simultaneously with the first voltammogram 46 shown in Figure 15. Figures 16(b) and (c) show AFM images obtained in the potential ranges of 1.40–1.25 and 0.95–0.80 V, respectively. Morphology changes began at a potential around 1.35 V, where particle-like precipitates appeared on the HOPG surface as shown in Figure 16(b). The number of the precipitates increased with lowering the potential down to 0.8 V, and the whole HOPG surface was covered with the precipitates (Figure 16(c)). Very similar 46 results were obtained in PC + FEC and PC + ES. In all the three solutions, the potentials at which the precipitates appeared are well correlated with the reduction peaks centered at 1.3, 1.1 and 1.0 V in PC + VC, PC + FEC, and PC + 46 ES, respectively, in the first cycle shown in Fig. 15. It should be noted that ceaseless solvent decomposition and exfoliation of graphite take place at about 0.9 V in 1 M LiClO4/PC without additives as shown in Figure 4(b). All the additives tested in the present study decompose at potentials more positive than 0.9 V, and the resulting precipitate layers effectively suppress the intercalation of solvated lithium ion.

Figure 16 AFM images (5 × 5 µm) of the HOPG basal plane surface obtained at (a) before and (b, c) during the first cycle of CV in 1 M LiClO4/PC containing 3 wt% VC. Scan rate of the –1 microcantilever: 5 µm s . Reproduced from [46] with permission of The American Chemical Society.

After the potential was scanned back to 2.9 V, an expanded area of 10 × 10 µm including the 5 × 5 µm area was observed in each solution. AFM images and height profiles of the 10 × 10 µm areas in PC + VC, PC + FEC, and PC +

219

SPM Analysis of the SEI Formation on Graphite Anodes

ES are shown in Figures 17(a), (b) and (c), respectively. From the height profiles, the thicknesses of the precipitate layers formed in PC + VC, PC + FEC, and PC + ES were roughly estimated to be 8, 20 and 30 nm, respectively. The thickness of the precipitate layer is the thinnest in PC + VC. This implies that the precipitate layer formed in PC + VC was dense and solid, and hence the most effective as SEI, which is in agreement with the superior cycling characteristics in the presence of VC shown in Table 2 and Figure 14. Another important feature seen in Figure 17 is that no evidence for solvent cointercalation was observed inside the rectangular holes in PC + VC, PC + FEC, and PC + ES. This fact confirms that effective SEI layers were formed by decomposition of the additives on graphite surfaces and that they suppressed the co-intercalation of PC molecules. (b )

(a)

nm 20

nm 40 A

15 B

A

nm 20

0

10 µ m A

8 nm

20 10

0

B

1

30

5

10 0

B

10

2 µm

0 nm 40 A 20

10 µ m 20 n m

B

1

0

0

2 µm

(c) nm 50 40 A

30 20

B

10 10 µ m

0 nm 40 20

30 n m

A

0

1

0

B

2 µm

Figure 17 AFM images of expanded areas (10 × 10 µm) and height profiles of HOPG basal plane surface obtained at 2.9 V after the first cycle of CV in 1 M LiClO4/PC containing 3 wt% (a) VC, (b) FEC, and (c) ES. The square in each image shows the area observed by AFM in the first cycle, where the precipitates were scraped off by the AFM tip. Reproduced from [46] with permission of The American Chemical Society.

220

6.2

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Roles of Other Additives in PC-Based Solutions

The three additives, VC, FEC, and ES, decompose to give effective SEI at potentials more positive than 0.9 V, at which co-intercalation of PC occurs. The vulnerability to reduction of these additives is due to the presence of double bonds or electron-withdrawing atoms in their molecules. From this point of view, PS, Cl–EC, α–bromo–γ–butyrolactone, methyl chloroformate, t–BC in Figure 13, which have been reported to be effective additives in PC-based solutions as well, should have the same role in SEI formation. On the other hand, ether compounds, such as 12–crown–4, DMM, DEM, and DEE, in Figure 13 are less vulnerable to reduction. Figure 18 shows STM images 44 obtained in 1 M LiClO4/PC + 0.5 M 12–crown–4. The addition of 12–crown–4 into PC greatly suppressed the exfoliation. In this solution, however, hill-like structures (Figure 18(b)) and blisters (Figure 18(c)) were observed after the potential was stepped to 0.9 and 0.7 V, respectively. This fact indicates that solvent cointercalation occurs in the presence of 12–crown–4 and that its role in SEI formation is different from that of VC, FEC, and EC discussed in the previous subsection. Ethers generally have higher donor numbers than carbonates, and preferentially solvate lithium ion. The preferential solvation of ethers prevents cointercalation of PC molecules, and thereby suppresses exfoliation of graphite layers even in the presence of PC.

(a) 1.0 V (1 min)

(b) 0.9 V (1 min)

(c) 0.7 V (1 min)

5 nm 1 nm

0

2 µm 0

2 µm 0

2 µm

Figure 18 STM images (2 × 2 µm) of HOPG basal plane surface observed at 2.8 V after the potential was kept at (a) 1.0 V, (b) 0.9 V, and (c) 0.7 V for 1 min in 1 M LiClO4/PC + 0.5 M 12–crown–4. The tip potential was kept at 3.0 V. Reproduced from [44] with permission of Wiley–VCH.

SPM Analysis of the SEI Formation on Graphite Anodes

7

221

Summary and Outlook

In this chapter, the authors focused on the role of solvent co-intercalation in SEI formation on graphite anode, and discussed the mechanistic aspects of SEI formation using the results obtained by STM and AFM. It was revealed that two steps are involved in SEI formation on graphite anode in EC-based solutions: (i) solvent co-intercalation and its decomposition beneath the surface at around 1 V and (ii) direct decomposition of solvents on the basal plane to form a precipitate layer at lower potentials. The presence of the decomposition products (blisters) in the interlayer space prevents further solvent cointercalation. On the other hand, the precipitate layer formed on the surface has a role in suppressing further reductive decomposition of solvent molecules on the basal plane. The precipitate layer most probably consists of polymerized compounds with oxyethylene units. The poor compatibility between PC and graphite anode originates from the fact that exfoliation of graphite during step (i) is too vigorous to form stable SEI. It is not clear why intercalation of PC-solvated lithium ion causes the exfoliation of graphite anode, but it seems that effective SEI formation on graphite anode is realized on a subtle balance of the kinetics between solvent cointercalation and its decomposition in the interlayer space. In this respect, the state of solvation is an important factor that determines the ease of solvent cointercalation. The additives that have been reported so far to be effective to form stable SEI in PC-based solutions are classified into two groups. One is a group of additives that decomposed at potentials higher than 0.9 V before cointercalation of PC takes place. VC, FEC, ES and other halogenated carbonates belong to this group. The other is ethers that preferentially solvate lithium ion and suppress co-intercalation of PC. As emphasized repeatedly in this and other chapters, the SEI plays a vital role in the battery reactions of LIBs, and is also a key material for safety and cycleability. It is recently reported that a continuous growth of SEI and the resulting loss of available lithium ion are the primary reason for long-term 84 degradation of large-scale LIBs. The understanding of SEI on graphite anode is being more and more important in the development of high-performance LIBs. Owing to numerous efforts of many researchers, SEI formation on graphite anode has been clarified to a considerable extent. However, details on the SEI composition, stability, and its influence on the performance of LIBs are still controversial. Careful analysis and discussion are necessary to completely

Lithium-Ion Batteries: Solid-Electrolyte Interphase

222

understand the nature of the SEI formed on graphite anodes. Theoretical considerations for solvent decomposition, which just started to appear in the 85, 86 literature, will be of a great help.

References 1. 2. 3.

4.

5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

Inaba M. and Ogumi Z., IEEE Electr. Insul. M. 17 (2001), 6. Ogumi Z. and Inaba M., Bull. Chem. Soc. Jpn. 71 (1998), 521. Ogumi Z. and Inaba M. in Advances in Lithium-Ion Batteries, ed. van Schalkwijk W. and Scrosati B. (Kluwer Academic/Plenum Publisher, New York, 2002), pp. 79–101. Winter M. and Besenhard J. O., in Handbook of Battery Materials, ed. Besenhard J. O. (Wiley–VCH, Weinheim, 1999), pp. 339–357 and 383– 418. Ikeda H., Narukawa S. and Nakajima H., Japanese Patent 1769661 (1981). Dey A. N. and Sullivan B. P., J. Electrochem. Soc. 117 (1970), 222. Besenhard J. O. and Fritz H. P., J. Electroanal. Chem. 53 (1974), 329. Eichinger G., J. Electroanal. Chem. 74 (1976), 183. Arakawa M. and Yamaki J., J. Electroanal. Chem. 219 (1987), 273. Fong R., von Sacken U. and Dahn J. R., J. Electrochem. Soc. 137 (1990), 2009. Peled E., J. Electrochem. Soc. 126 (1979), 2047. Peled E., in Handbook of Battery Materials, ed. Besenhard J. O. (Wiley– VCH, Weinheim, 1999), 419–458. Ein-Eli Y., Markovsky B., Aurbach D., Carmeli Y., Yamin H. and Luski S., Electrochim. Acta 39 (1994), 2559. Aurbach D., Ein-Eli Y., Chusid (Youngman) O., Carmeli Y., Babai M. and Yamin H., J. Electrochem. Soc. 141 (1994), 603. Aurbach D., Ein-Eli Y., Markovsky B., Zaban A., Luski S., Carmeli Y. and Yamin H., J. Electrochem. Soc. 142 (1995), 2882. Ein-Eli Y., Thomas S. R., Koch V., Aurbach D., Markovsky B. and Schechter A., J. Electrochem. Soc. 143 (1996), l273. Aurbach D., Markovsky B., Schechter A. and Ein-Eli E., J. Electrochem. Soc. 143 (1996), 3809. Ein-Eli Y., McDevitt S. F., Aurbach D., Markovsky B., and Schechter A., J. Electrochem. Soc. 144 (1997), l180.

SPM Analysis of the SEI Formation on Graphite Anodes

223

19. Aurbach D., Zaban A., Ein-Eli Y., Wessman I., Chuid O., Markovsky B., Levi M., Levi E., Schechter A., and Granot E., J. Power Sources 68 (1997), 91. 20. Aurbach D., Levi M. D., Levi E., and Schechter A., J. Phys. Chem. B101, (1997), 2195. 21. Aurbach D., Levi M. D., Levi E., Teller H., Markovsky B. and Salitra G., J. Electrochem. Soc. 145 (1998), 3024. 22. Aurbach D., Markovsky B., Weissman I., Levi E. and Ein-Eli Y., Electrochim. Acta 45 (1999), 67. 23. Yoshida H., Fukunaga T., Hazama T., Terasaki M, Mizutani M. and Yamachi M., J. Power Sources 68 (1997), 311. 24. Kanamura K., Shiraishi S., Takezawa H. and Takehara Z., Chem. Mater. 9 (1997), 1797. 25. Naji A., Ghanbaja J., Humbert B., Willmann P. and Billaud D., J. Power Sources 62 (1996), 141. 26. Dolle M., Grugeon S., Beaudoin B., Dupont L. and Trascon J.-M., J. Power Sources 97–98 (2001), 104. 27. Ogumi Z., Sano A., Inaba M. and Abe T., J. Power Sources 97–98 (2001), 156. 28. Yazami R., Electrochim. Acta 45 (1999), 87. 29. Novak P., Joho F., Lanz M., Rykar B., Panitz J. C., Alliata D., Kotz R. and Haas O., J. Power Sources 97–98 (2001), 39. 30. Menachem C., Peled E., Burstein L. and Rosenberg Y., J. Power Sources 68 (1997), 277. 31. Peled E., Golodnitsky D., Menachem C. and Bar-Tow D., J. Electrochem. Soc. 145 (1998), 3482. 32. Bar-Tow D., Peled E. and Burstein L., J. Electrochem. Soc. 146 (1999), 824. 33. Peled E., Bar-Tow D., Merson A., Gladkich A., Burstein L. and Golodnitsky D., J. Power Sources 97–98 (2001), 52. 34. Winter M., Besenhard J. O., Spahr M. E. and Nobak. P, Advanced Materials 10 (1998), 725. 35. Chung G. C., Kim. H. J. Jun S. H. Choi J. W. and Kim M. H., J. Electrochem. Soc. 147 (2000), 4398. 36. Besenhard J. O., Winter M., Yang J. and Biberacher W., J. Power Sources 54 (1995), 228. 37. Winter M., Wrodingg G. H., Besenhard J. O., Biberacher W. and Novak P., J. Electrochem. Soc. 147 (2000), 2427.

224

Lithium-Ion Batteries: Solid-Electrolyte Interphase

38. Bard A. J., in Scanning Tunneling Microscopy and Spectroscopy, ed. Bonnell D. A. (VCH, New York, 1993), p. 287. 39. Inaba M., Siroma Z., Ogumi Z., Abe T., Mizutani Y. and Asano M., Chem. Lett. 1995 (1995), 661. 40. Inaba M., Siroma Z., Funabiki A., Ogumi Z., Abe T., Mizutani Y. and Asano M., Langmuir 12 (1996), 1535. 41. Inaba M., Siroma Z., Kawatate Y., Funabiki A. and Ogumi Z., J. Power Sources 68 (1997), 221. 42. Inaba M., Kawatate Y., Funabiki A., Jeong S.-K., Abe T. and Ogumi Z., Electrochim. Acta 45 (1999), 99. 43. Inaba M., Kawatate Y., Funabiki A., Jeong S.-K., Abe T. and Ogumi Z., Electrochemistry 67 (1999), 1153. 44. Ogumi Z., Jeong S.-K., Inaba M. and Abe T., Macromol. Symp. 156 (2000), 195. 45. Jeong S.-K., Inaba M., Abe T. and Ogumi Z., J. Electrochem. Soc. 148 (2001), A989. 46. Jeong S.-K., Inaba M., Mogi R., Iriyama Y., Abe T. and Ogumi Z., Langmuir 17 (2001), 8281. 47. Ogumi Z., Inaba M., Abe T. and Jeong S.-K., in Studies in Surface Science and Catalysis, Vol. 132, ed. Iwashita Y., Oyama N. and Kunieda H. (Elsevier Science B.V., Amsterdam, 2001), pp. 929–934. 48. Jeong S.-K., Inaba M., Iriyama Y., Abe T. and Ogumi Z., Electrochim. Acta 47 (2002), 1975. 49. Dahn J. R., Phys. Rev. B44 (1991), 9170. 50. Ohzuku T., Iwakoshi Y. and Sawai K., J. Electrochemical. Soc. 140 (1993), 2490. 51. Inaba M., Yoshida H., Ogumi Z., Abe T., Mizutani Y. and Asano M., J. Electrochem. Soc. 142 (1995), 20. 52. Morita S., Tsukada S. and Mikoshiba J., J. Vac. Sci. Technol. A6 (1988), 354. 53. Setton R., in Graphite Intercalation Compounds I. Structure and Dynamics, ed. Zabel H. S. and Solin A. (Springger-Verlag, Berlin, 1990), 320. 54. Shu Z. X., McMillan R. S. and Murray J. J., J. Electrochem. Soc. 140 (1993), 922. 55. Chusid (Youngman) O., Ein-Eli Y. and Aurbach D., J. Power Sources 43–44 (1993), 47. 56. Ryu Y.-G. and Pyun S.-I., J. Electroanal. Chem. 433 (1997), 97.

SPM Analysis of the SEI Formation on Graphite Anodes

225

57. Menachem C., Peled E., Burstein L. and Rosenberg Y., J. Power Sources 68 (1996), 277. 58. Naji A., Ghanbaja J., Humbert B., Willmann P. and Billaud D., J. Power Sources 63 (1996), 33. 59. Mori S., Asahina H., Suzuki H., Yonei A. and Yokoto K. J. Power Sources 68 (1997), 59. 60. Allita D., Kotz R., Novak P. and Siegenthaler H., Electrochemistry Communications 2 (2000), 436. 61. Novak P., Joho F., Lanz M., Rykart B., Panitz J.-C., Allita D., Kotz R. and Haas O., J. Power Sources 97–98 (2001), 39. 62. Lide D. R. (ed.), Handbook of Organic Solvents (CRC Press, Boca Raton, 1995). 63. Morita M., Asai Y., Yoshimoto N. and Ishikawa M., J. Chem. Soc., Faraday Trans. 94 (1998), 3451. 64. Klassen B., Aroca R., Nazri M. and Nazri G. A., J. Phys. Chem. B102 (1998), 4795. 65. Megahed S. and Ebner W., J. Power Sources 54 (1995), 155. 66. Yang C. R., Wang Y. Y. and Wan C. C., J. Power Sources 72 (1998), 66. 67. Yang C. R., Song J. Y., Wang Y. Y. and Wan C. C., J. Appl. Electrochem. 30 (2000), 29. 68. Smart M. C., Ratnakumar B. V. and Surampudi S., J. Electrochem. Soc. 146 (1999), 486. 69. Smart M. C., Ratnakumar B. V., Surampudi S., Wang Y., Zhang X., Greenbaum S. G., Hightower A., Ahn C. C. and Fultz B., J. Electrochem. Soc. 146 (1999), 3963. 70. Fukushima T., Matsuda Y., Hashimoto H. and Arakawa R., Electrochem. Solid State Lett. 4 (2001), A127. 71. Matsuda Y., Fukushima T., Hashimoto H. and Arakawa R., J. Electrochem. Soc. 149 (2002), A1045. 72. Dudley J. T., Wilkinson D. P., Thomas G., LeVae R., Woo S., Blom H., Horvath C., Juzkow M. W., Denis B., Juric P., Aghakian P. and Dahn J. R., J. Power Sources 35 (1991), 59. 73. Naji A., Willmann P. and Billud D., Carbon 36 (1998), 1347. 74. Shu Z. X., McMillan R. S., Murray J. J. and Davidson I. J., J. Electrochem. Soc. 143 (1996), 2230. 75. Biensan P., Bodet J. M., Perton F., Broussely M., Jehoulet C., Barusseau S., Herreyre S. and Simon B., Extended Abstracts of the 10th International Meeting on Lithium Batteries (Como, Italy, 2000), Abs. No. 286.

226

Lithium-Ion Batteries: Solid-Electrolyte Interphase

76. Wrodnigg G. H., Besenhard J. O. and Winter M., J. Electrochem. Soc. 146 (1999), 470. 77. Wrodnigg G. H., Wrodnigg T. M., Besenhard J. O. and Winter M., Electrochem. Commun. 1 (1999), 148. 78. McMillan R., Slegr H., Shu Z. X. and Wang W., J. Power Sources 81–82 (1999), 20. 79. Naji A., Ghanbaja J., Willmann P. and Billaud D., Electrochim. Acta 45 (2000), 1893. 80. Chung G.-C., Kim H.-J., Jun S.-H. and Kim M.-H., Electrochem. Commun. 1 (1999), 493. 81. Chung G.-C., Kim H.-J., Yu S.-H., Jun S.-H., Choi J.-W. and Kim M.-H., J. Electrochem. Soc. 147 (2000), 4391. 82. Shu Z. X., McMillan R. S. and Murray J. J., J. Electrochem. Soc. 140 (1993), L101. 83. Abe T., Mizutani Y., Kawabata N., Inaba M. and Ogumi Z., Synthetic Metals 125 (2000), 249. 84. Yanagida K., Yanai A. Kida. Y, Funahashi A. Nohma T. and Yonezu I., J. Electrochem. Soc. 149 (2002), A804. 85. Li T. and Balbuena P.B., Chem. Phys. Lett. 317 (2000), 421. 86. Wang Y., Nakamura S., Ue M. and Balbuena P. B., J. Am. Chem. Soc. 123 (2001), 11708.

%*#26'4 



6*'14'6+%#.+05+)*65+0616*'5'+%1/215+6+10 #0&(14/#6+10/'%*#0+5/&'05+6;(70%6+10#. 6*'14;567&+'5 ;+:7#09#0) CPF2'4.#$$#.$7'0#Ŕ &GRCTVOGPVQH%JGOKECN'PIKPGGTKPI5YGCTKPIGP'PIKPGGTKPI%GPVGT 7PKXGTUKV[QH5QWVJ%CTQNKPC /CKP%QNWODKC5QWVJ%CTQNKPC75# 'OCKN YCPI[K"GPITUEGFWŔ DCNDWGPC"GPITUEGFW 



 +PVTQFWEVKQP



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



  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

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ŏU DGJCXKQTU CU 5'+ NC[GT IGPGTCVQTU CTG GZVGPUKXGN[FKUEWUUGF /CP[ GHHQTVU JCXG DGGP FGFKECVGF VQ HKPF RTQRGT CFFKVKXGU VQ 2%DCUGF UQNWVKQPUVJCVYQWNFGPCDNGRTCEVKECNCRRNKECVKQPU QH 2% 5QOG UWEEGUUGU JCXG DGGPTGRQTVGFKPENWFKPIVJGCFFKVKQPQHEJNQTQGVJ[NGPGECTDQPCVG %'%QVJGT

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



JCNQIGPUWDUVKVWVGF ECTDQPCVGU CPF C XCTKGV[ QH WPUCVWTCVGF ECTDQPCVGU UWEJ CU XKP[NRTQR[NGPG ECTDQPCVG XKP[NGPG ECTDQPCVG 8% CPF GVJ[NGPG UWNHKVG

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−JCUQPVJGTGFWEVKXGDGJCXKQTQHUQNXGPVOQNGEWNGU EQQTFKPCVGF YKVJ .K  :−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

) FRDCUKU UGVYJKNGVJGDWNMUQNXGPVGHHGEVKUCEEQWPVGFHQTWUKPIVJG%2%/OGVJQF6JG QXGTCNNRTQEGFWTGKUFGPQVGFCU%2%/$29

) FR(QTVJGECUGU YKVJOQTGVJCPQPGUQNXGPVOQNGEWNGUKPVJGENWUVGTOQFGNU$29YKVJVJG DCUKUUGV) FKUGORNQ[GFVQQRVKOK\GVJGUVTWEVWTGQH.K 5P P CPF

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

VJGP VJG %2%/ OGVJQF KU WUGF VQ EJCTCEVGTK\G VJG DWNM UQNXGPV GHHGEV QP VJG QRVKOK\GF UVTWEVWTG 6JKU RTQEGFWTG KU FGPQVGF CU %2%/$29 

) FR$29) F    

















 





 





 

 

 



 

 






 







 





 





 





 





 







 

 











 









 













  5EJGOG 6JGENWUVGTEQPVKPWWOOQFGNGORNQ[GF VQ KPXGUVKICVG VJG TGFWEVKXG FGEQORQUKVKQP QH UQNXGPV5QNUQNXGPVOQNGEWNGU:−UCNVCPKQPU

 #UUQEKCVKQP QH NKVJKWO CNM[N FKECTDQPCVGU CPF VJG UKOWNCVKQP QH VJGKT +4 YGTG FQPG WUKPI $29) F 6JG HTGSWGPE[ UECNG HCEVQT GORNQ[GF KU  HQT +4 UKOWNCVKQP CPF VJG QVJGT RCTCOGVGTU CTG  )CWUUKCP   .QTGPV\KCPHWPEVKQPUCPFCEO−RGCMYKFVJ6QKPEQTRQTCVGRQUUKDNGYGCM KPVGTCEVKQPUDGVYGGPVJGπDQPFCPFπ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

02# #NN VJG ECNEWNCVKQPU JCXG DGGP RGTHQTOGF WUKPI VJG )CWUUKCP  RTQITCO  

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



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



 





 
















$ 

 





















 







%

 




























#  







 





(









 















  












'


"


&





 





 













 

 

!



(KIWTGC+PKVKCNTGFWEVKQPUQH'%.K 2%.K CPF8%.K YKVJ%2%/$29

) FR IGPGTCVKPITCFKECNCPKQPUVJTQWIJ% ECTDQP[NECTDQP1ENGCXCIG 

5VCTVKPI HTQO VJG RTGEWTUQTU VYQ UQTVU QH JQOQN[VKE TKPI QRGPKPI KP RTKPEKRNG EQWNF JCRRGP XKC JQOQN[VKE ENGCXCIG QH GKVJGT %1 %ECTDQP[N



  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

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

 



 













!

 



 





!

 

"

%



 





 "

#









 























"



$



 





 















"

"





$













 














!




























!

!





#

"

"

'

&

"

 (KIWTGD+PKVKCNTGFWEVKQPUQH'%.K 2%.K CPF8%.K YKVJ%2%/$29

) FR IGPGTCVKPITCFKECNCPKQPUVJTQWIJ% GVJGTGCNECTDQP1ENGCXCIG



%QORCTKUQPQH4GFWEVKXG&GEQORQUKVKQP$GVYGGP'%CPF2%.K '%P CPF.K 2%P P  6QVJGQTGVKECNN[KFGPVKH[VJG5'+NC[GTEQORQPGPVUTGUWNVKPIHTQOVJGTGFWEVKQP QH '% CPF VQ GZRNQTG VJG FGUVTWEVKXG GHHGEV QH 2% QP VJG ITCRJKVG CPQFG KP NKVJKWO KQP DCVVGTKGU VJG TGFWEVKXG FGEQORQUKVKQPU QH '% CPF 2% JCXG DGGP

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



KPVGPUKXGN[KPXGUVKICVGFD[CRRN[KPI&(6OGVJQFUVQVJG.K '%PCPF.K 2%P

PENWUVGTU

 

4GFWEVKQP2QVGPVKCNUCPF4KPI1RGPKPI$CTTKGTUQH'%CPF2%  '% CPF 2% GPGTI[ UWTHCEG RTQHKNGU CTG KNNWUVTCVGF KP (KIWTGU  CPF  TGURGEVKXGN[.KMGVJGCDQXGUOCNNGUVOQFGNCPKQPRCKTRTGEWTUQTKUHQWPFHQT GCEJQH.K '%PCPF.K 2%P P KPYJKEJVJGWPRCKTGFURKPFGPUKV[KU OCKPN[ NQECVGF CV VJG ECTDQP[N ECTDQP % YKVJ C CRRTQZKOCVG EQGHHKEKGPV QH G#EEQTFKPIN[ECTDQP[NECTDQPEJCPIGUKVUJ[DTKFK\CVKQPQTDKVCNUHTQO URVQURCPFVJG%1ITQWRFQGUPQVTGOCKPRNCPCTCP[OQTG6JGUGRTGEWTUQTU

EQWNFWPFGTIQCJQOQN[VKETKPIQRGPKPID[VJGENGCXCIGQH%1DQPFXKC  








" 









!

'

'



 





$

#







" 

  





!



#

$











 

 



  


















 

















 













































 










!

%

&





 





$









! 

 







$

&







$



















!

!









 















 









  (KIWTG2QVGPVKCNGPGTI[RTQHKNGHQTVJGTGFWEVKXGFGEQORQUKVKQPRTQEGUUQH.K 2% WPFGTNKPGF FCVCCPF.K '% RNCKPFCVCECNEWNCVGFYKVJ$29

) FR$29) F6JG UGNGEVGFUVTWEVWTCNFCVCTGHGTVQVJG.K 2%URGEKGU4GRTQFWEGFHTQO=?YKVJRGTOKUUKQPQH #OGT%JGO5QE

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

















#



$






























$

$










 

 

 

!

!










 









 

 

 









 









 











  







 





 

"



 



  

 



 



























"

 

"

( )

"

'





&

*

+

(

( "

'

&

*

 %

"

 (KIWTG2QVGPVKCNGPGTI[RTQHKNGHQTVJGTGFWEVKXGFGEQORQUKVKQPRTQEGUUQH.K 2% WPFGTNKPGF FCVCCPF.K '% RNCKPFCVCECNEWNCVGFYKVJ$29

) FR$29) F6JG UGNGEVGFUVTWEVWTCNFCVCTGHGTVQVJG.K 2%URGEKGU4GRTQFWEGFHTQO=?YKVJRGTOKUUKQPQH #OGT%JGO5QE  CVTCPUKVKQPUVCVG 65TGUWNVKPIKPCTCFKECNCPKQPEQQTFKPCVGFYKVJ.K  RCVJ  (KIWTGUJQYUVJCV VJGCFKCDCVKE GNGEVTQP CHHKPKV[ '# ` −∆' QH .K 2% KU  MECNOQN NQYGT VJCP VJCV QH .K '%  XU  MECNOQN 6JGHQTOCVKQPGPGTI[QHVJGTCFKECNCPKQP TGHGTTGFVQCU'HJGTGCHVGTTGNCVKXG VQ .K 2% KU CNUQ UQOGYJCV NQYGT VJCP VJCV QH .K '%  XU  MECNOQN JQYGXGT VJG TKPIQRGPKPI GPGTI[ DCTTKGT 'C QH  KU DCUKECNN[ KFGPVKECNYKVJVJCVQHVJGKPVGTOGFKCVGQH.K '% XUMECNOQN #NVJQWIJ VJG TKPIQRGPKPI DCTTKGTU 'C CTG CICKP ENQUG  XU  MECNOQNVJG'#CPF'HXCNWGUQH.K 2%CTGNQYGTVJCPVJQUGQH.K '%D[  CPF  MECNOQN TGURGEVKXGN[ 6JG DWNM UQNXGPV GHHGEV KPEQTRQTCVGF D[ %2%/WUKPIVJGICURJCUGQRVKOK\GFIGQOGVTKGUHWTVJGTTGOCTMCDN[FGETGCUGU VJG'#QH.K 2%HTQOMECNOQNKPVJGICURJCUGVQMECNOQN 9  −CPF−MECNOQNHQTCPFTGURGEVKXGN[CPFKVUNKIJVN[TGFWEGU'C D[MECNOQN 9−MECNOQNHQT656JGTGFWEVKQPRQVGPVKCNU 42 RTGFKEVGF D[ VJG ENWUVGT%2%/ OGVJQF HQT VJG TGFQZ TGCEVKQP ∆)UQN  ∆) KP VJGICURJCUG ∆9∆)UQN−P( 42 .K  5 G−→= 5•−?ŧŧ.K ŧŧŧ 5 5'%2% + CTG8CPF8HQT2%CPF'%TGURGEVKXGN[

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



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őYCURTQRQUGFHQT'%DCUGFGNGEVTQN[VGQPVJGITCRJKVKECPQFGUWTHCEGD[0CLK GVCN#WTDCEJGVCNUWIIGUVGFVJGRQUUKDKNKV[QH %*%*%*1%1.KCU C5'+URGEKGUQPNKVJKWOUWTHCEGUHTQOVJGHCEVVJCVVJGRGTEGPVCIGQHVJGCNM[N ECTDQP :25 RGCM CTQWPF  G8 KU OWEJ JKIJGT VJCP VJCV GZRGEVGF HTQO VJGRTQFWEVCNQPG6JGTCFKECNCPKQPEQWNFWPFGTIQCPQVJGTFKOGTK\CVKQP

RCVJ$IGPGTCVKPIGVJ[NGPGRTQR[NGPGICUCPFCPKPUQNWDNGEQORQWPFKPRQNCT QTICPKE GNGEVTQN[VG NKVJKWO GVJ[NGPGRTQR[NGPG FKECTDQPCVG .K1%141%1.K

4%*%*%*%*%*YJKEJJCUWUWCNN[DGGPKFGPVKHKGFCUCOCLQT5'+ EQORQPGPV 6JG TCFKECN CPKQP ECP DG HWTVJGT TGFWEGF VJTQWIJ RCVJ % D[ YJKEJ C YGCM EQORNGZ  QH GVJ[NGPGRTQR[NGPG ICU CPF CP WPRCKTGF PWENGQRJKNKE ECTDQPCVG CPKQP .K%1− CTG IGPGTCVGF 6JG ECTDQPCVG CPKQP

.K%1− OC[ GKVJGT RTGEKRKVCVG CU KPUQNWDNG KPQTICPKE NKVJKWO ECTDQPCVG .K%1D[KQPRCKTKPIYKVJVJGUQNXCVGF.KKQP RCVJ)QTD[PWENGQRJKNKE CVVCEMVQCUQNXGPVOQNGEWNGHQTOKPINKVJKWOGVJ[NGPGRTQR[NGPGFKECTDQPCVG

RCVJ(#FFKVKQPCNN[VJGTCFKECNCPKQPEQWNFCNUQTGCEVYKVJVJGKPVGTOGFKCVGU

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

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

4%*%*%1%*%* %*%*%*%1 %*%*%*  YJKEJ UQ HCT JCU PQVDGGPGZRGTKOGPVCNN[KFGPVKHKGFCUC5'+URGEKGU 6CDNG  CPF (KIWTG  UJQY VJCV RCVJ % JCU XKTVWCNN[ KFGPVKECN )KDDU HTGG GPGTIKGU QH TGCEVKQP ∆) XCNWGU HQT .K '% CPF .K 2% YJGTGCU HQT VJG QVJGTRCVJU.K 2%IGPGTCNN[JCUNGUUPGICVKXG∆)VJCP.K '%#UUJQYPKP (KIWTGVJG∆)UQHRCVJU#$&CPF'CTGCPFMECNOQNNGUU PGICVKXG VJCP VJQUG EQTTGURQPFKPI VQ .K '% (QT RCVJU ( CPF ) VJG[ CTG OWEJNGUUPGICVKXGD[CPFMECNOQNVJCPVJQUGQH.K '%CPFRCVJ( QH .K 2% JCU C PGICVKXG ∆) JKIJGT D[  MECNOQN VJCP RCVJ ) − XU −MECNOQN YJGTGCU VJG[QPN[FKHHGTMECNOQNKPVJGECUGQH.K '%

−XUŌMECNOQN%QPUGSWGPVN[RCVJ(IGPGTCVKPINKVJKWORTQR[NGPG FKECTDQPCVG KU OWEJ OQTG HCXQTCDNG VJCP RCVJ ) NGCFKPI VQ VJG KPQTICPKE RTQFWEV .K%1 VJG FKHHGTGPEG DGKPI NCTIGT KP 2% VJCP KP '% 6JG KPQTICPKE EQORQWPFNKVJKWOECTDQPCVGKUWUWCNN[EQPUKFGTGFCUCDGVVGTRCUUKXCVKPICIGPV VJCPVJGQTICPKEQPGJQYGXGTHTQOVJGCPCN[UKUQHUQNXGPVTGFWEVKQPKVECPDG   6CDNG)KDDUHTGGGPGTIKGU ∆)MECNOQNQHVJGVGTOKPCVKQPTGCEVKQPUHQTVJGTCFKECNCPKQPUHTQO VJGKPKVKCNTGFWEVKQPUQH.K '%CPF.K 2% 

2CVJ

4GCEVCPVU

2TQFWEVU

#

 

41%1.K

−

−

$

 

.K1%141%1.K %*QT%*

−

−

%

 G−

=.K%1? %*QT%*

−

−

&

 

.K41%1.K

−

−

'

 

.K14ŏ1%1.K

−

−

(

 

.K1%141%1.K %*QT%*

−

−

)

 

.K%1 %*QT%*

−

−

4%*%*%*%*%*

∆) .K '%

∆) .K 2%

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



)

)







0



/ .

1











1







1



 

2

 .













,

 )

/ 2

2





,





 3









3



,

,

.



3



,



-



(



)





/ .





1



) 



)







%

'

 &







*



+ #



















#

(











*







+ 



*

+





,





 























 
























 



)



#













#







*

 +











 





 

 





,





"

$







*

+ 

,















,


























)

)



!





















 















 



 





)












 

* *



 

)

) + +

 

  





 



 



,





 



,













 (KIWTG 6JGVGTOKPCVKQPRCVJUHQTVJGTCFKECNCPKQPHTQOVJGTGFWEVKXG FKUUQEKCVKQP KP .K 2% ECNEWNCVGF YKVJ VJG $29

) FR$29) F OGVJQF 6JG )T CV - QH VJGWPFGTNKPGFCPFRNCKPFCVCTGHGTVQ.K 2% 52%4%*CPF.K '% 5'%4* TGURGEVKXGN[6JGUGNGEVGFUVTWEVWTCNFCVCTGHGTVQVJG.K 2%TGFWEVKQPURGEKGU4GRTQFWEGFHTQO =?YKVJRGTOKUUKQPQH#OGT%JGO5QE 

 EQPENWFGFVJCV.K%1RNC[UCOKPQTTQNGKPVJG5'+DWKNFWRGXGPKP'%DCUGF GNGEVTQN[VG CPF KVU TQNG DGEQOGU GXGP NGUU KORQTVCPV KP 2%DCUGF GNGEVTQN[VGU 'ZEGRVHQTVJKUFKHHGTGPEGVJGKTRTQFWEVFKUVTKDWVKQPUCTGXGT[UKOKNCTGIVJG NKVJKWO CNM[N FKECTDQPCVG YKVJ NQPI EJCKP EQPUVKVWVGU VJG OQUV KORQTVCPV EQORQPGPV HQNNQYGF D[ 41.K CPF CPQVJGT NKVJKWO CNM[N FKECTDQPCVG YKVJ UJQTVEJCKPCPFVJGNGCUVNKMGN[KUVJG%.KECTDKFGEQORQWPF6JGTGHQTGCVVJKU

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

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

 

+PKVKCN4GFWEVKQP  6JGKPKVKCNTGFWEVKQPRCVJUCTGUJQYPKP(KIWTGCPF(KIWTG HQT 5.K 8%

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ŧŧŧ.K ŧŧŧ 8%  G−→ 5ŧŧŧ.K ŧŧ= 8%•−?   ++  5ŧŧŧ.K ŧŧŧ 8%  G−→= 2%•−?ŧŧ.K ŧŧŧ 8%    

+++

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO 



(

"

               

$

$

&













!

#



'



 





%

&

&








 !

 







 !

" !



"

#

"

$

(











 +

3 *

 

  )






,

+

+

*



) 

,


(

/ 

1



4







#

(

 

) 

,
















 !

 

 / 1

1

 

  )


,

 

. *

1 / 1

2









"



) )


 ,

,






/ 2

*

*

/ 0

 



)

) 




,

,


 



(

(  



*

. *








)

, 4

0

-








 )

,





*

+

 

 



#



)


,

4

0

0

 




)

,





%





 





# !



 !

$







"



 " 











0

3

.

2



)


,

(







0

3

.

3






)

,





(

 

6 5

(



9

8  





7

:

6

6 (

9

8

    (KIWTG6JGRQVGPVKCNGPGTI[ WPFGTNKPGFFCVCCPF)KDDUHTGGGPGTI[RTQHKNGCV-HQTVJG TGFWEVKXG FGEQORQUKVKQP RCVJU QH 2%.K 8% NGHV CPF '%.K 8% TKIJV KVCNKE FCVC YKVJ $29

) FR$29) F OGVJQF 4GRTQFWEGF HTQO = ? YKVJ RGTOKUUKQP QH#OGT%JGO5QE ;

7

(

 6CDNG4GFWEVKQPRQVGPVKCNU 428QH8%CPF'%2%GUVKOCVGFD[VJG%2%/ENWUVGTOQFGN  %NWUVGTU

'%

C

C

2%

'%.K 8%

 

 



2%.K 8%

D



D

.K 8%

C







E







D

.K '% .K 2% C

8%

D

E

4GH 4GH 4GH



6CDNGUJQYUVJCVVJG42QH8%KUKPFGGFJKIJGTD[CRRTQZKOCVGN[8 VJCP VJQUG QH '% CPF 2% TGICTFNGUU QH VJG ENWUVGT OQFGNU GORNQ[GF HQT VJG ECNEWNCVKQPUCPFVJCVVJG42QH'%KUUQOGYJCVJKIJGTVJCPVJCVQH2%+VKUCNUQ

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG




&



 















 









































δ &











&

&





















 



 















  %

 

  



















. 



)

) 

 .

1







/

'  







*


'    *


 



$ -





.

0



 -

2

 



-

' 













*


' 

 *




5







 

, -

)

' 






*



 .

/

+





 

'






*

 "




.

.

1

'

* 










.

, )

'

*

&

3





2

 



(

)

+

+

-

 ' 4


*

 '




*





 

 

 

,

, )

(

2

1



 '






*

'




*



















 &





























1 /





 



1



  *




 















2







,

'

* 

 

2





 














 





 







!

#

 





&

 (KIWTG6JGRQVGPVKCNGPGTI[ WPFGTNKPGFFCVCCPF)KDDUHTGGGPGTI[RTQHKNGCV-HQTVJG TGFWEVKXG FGEQORQUKVKQP RCVJU QH 2%.K 8% NGHV CPF '%.K 8% TKIJV KVCNKE FCVC YKVJ $29

) FR$29) FOGVJQF6JGHCXQTCDNGRCVJUCTG→ →HQNNQYGF D[VYQEQORGVKVKXGRCVJUCCPFE4GRTQFWEGFHTQO=?YKVJRGTOKUUKQPQH#OGT%JGO 5QE

 PQVGF VJCV VJG 42 QH '% CPF 2% FQ PQV RTQHQWPFN[ EJCPIG YKVJ VJG V[RG QH ENWUVGTOQFGNUWVKNK\GF6JCVOGCPUVJCV8%KURTGHGTGPVKCNN[TGFWEGFCU8%KU RTGUGPVKPVJG'%2%DCUGFUQNWVKQPUCPFVJCV8%FQGUPQVENGCTN[KORTQXGVJG TGFWEVKQPECRCDKNKV[QH'%2% 6JGGPGTI[DCTTKGTQHVJG%1DQPFJQOQN[VKEENGCXCIG RCVJ C→ →QHVJG8%OQNGEWNGKPVJG8%TGFWEVKQPRTGEWTUQT XKCVJG65   KU CDQWV  MECNOQN YJKEJ KU TCVJGT ENQUG VQ VJCV QH .K 8%  MECNOQN 6JG JQOQN[VKE TKPIQRGPKPI DCTTKGTU RCVJ D → → QH VJG '%2%TGFWEVKQP KPVGTOGFKCVGU  CPF  XKC VJG 65  CPF  TGURGEVKXGN[QPN[EJCPIGMECNOQN CPFMECNOQNHQT2% CPFMECNOQNHQT'%#UUJQYPQPVJGRQVGPVKCNGPGTI[UWTHCEGQH(KIWTGU CPF  QVJGT TGCEVKQP RCVJU RCVJ E →→ KP (KIWTG  →→ KP (KIWTG  UVCTVKPI HTQO VJG 8%TGFWEVKQP KPVGTOGFKCVGU  CPF  JCXG CNUQ DGGPNQECVGFKPYJKEJKPVTCOQNGEWNCTGNGEVTQPVTCPUHGTKPFWEGUCJQOQN[VKETKPI

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



QRGPKPIQHVJG'%2%TKPIKPUVGCFQHVJCVQHVJG8%TKPI+PVJGEQTTGURQPFKPI VTCPUKVKQPUVCVGUCPFVJGWPRCKTGFURKPFGPUKV[KUOCKPN[FKUVTKDWVGFCVVJG ECTDQP[N ECTDQP QH 8% CPF VJG NGCXKPI GVJGTGCN ECTDQPU QH 2% QT '% 6JG EJCTCEVGTKUVKEUQH CUCVTCPUKVKQPUVCVGEQPPGEVKPI CPFVJGOQTG UVCDNGTCFKECNCPKQPU JCXGDGGPEQPHKTOGFD[VJGXKDTCVKQPCNOQFGUQH KVUGKIGPXGEVQTUCUYGNNCUD[+4%ECNEWNCVKQPU+P VJG ECUG QH VJG 5.K 8% ENWUVGT VJG GPGTI[ DCTTKGT QH RCVJ E →→ KU EQPUKFGTCDN[ NQYGT VJCP VJCV QH RCVJ C →→ → D[  CPF  MECNOQN HQT '% CPF 2% TGURGEVKXGN[6JCVOGCPUVJCVKVKUFKHHKEWNVVJCVRCVJCEQORGVGUYKVJRCVJEKP VJKU ECUG *QYGXGT RCVJ E →→ HQT VJG ENWUVGT 5.K 8% DGEQOGU NGUUHCXQTCDNGVJCPVJCVQHVJGENWUVGT 5.K 8%6JGDCTTKGTUQHVJGRCVJUCTG QPN[  CPF  MECNOQN NQYGT VJCP VJQUG QH RCVJ C HQT VJG ECUGU QH

'%.K 8%CPF 2%.K 8%TGURGEVKXGN[6JKUVGNNWUVJCVRCVJECPFCCTG XGT[EQORGVKVKXGTGCEVKQPRCVJUHQTVJGOQTGTGNCUVKEENWUVGTU $TKGHN[VJGRTGUGPEGQH8%KP'%2%DCUGFGNGEVTQN[VGUQH.+$URTQDCDN[ EJCPIGUVJGUQNXGPVTGFWEVKXGOGEJCPKUO$GECWUGQH8%ŏ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

 

6GTOKPCVKQP4GCEVKQPQH4CFKECN#PKQPU  6CMKPI  CPF  TGUWNVKPI HTQO VJG ENWUVGT '%.K 8% CU CP GZCORNG VJG VGTOKPCVKQP TGCEVKQPU QH VJG TCFKECN CPKQPU CTG GZVGPUKXGN[ KPXGUVKICVGF CU KNNWUVTCVGFKP(KIWTGUCPFKPENWFKPIVJGKTHWTVJGTTGFWEVKQPUGNHCPFETQUU FKOGTK\CVKQPUCPFTGCEVKQPUYKVJTGNGXCPVKPVGTOGFKCVGUCPFUQNXGPV OQNGEWNGU YJKEJ YKNN IGPGTCVG XCTKQWU QTICPKE CPFQT KPQTICPKE EQORQWPFU UQOG QH YJKEJ ECP RTGEKRKVCVG QP VJG CPQFG UWTHCEG CPF DWKNF WR CP CRRTQRTKCVG 5'+ NC[GT

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

#EEQTFKPIVQ(KIWTGHWTVJGTTGFWEVKQPQH[KGNFU RCVJ%KP(KIWTG  YJGTG VJG ECTDQP[N ECTDQP KP 8% VGPFU VQ EQODKPG YKVJ VJG '% GVJGTGCN QZ[IGPFWG VQ GCUKGT TGFWEVKQP QH 8% EQORCTGF YKVJ '% KG TGEQODKPCVKQP OC[ VCMG RNCEG KP  CHVGT CEEGRVKPI CPQVJGT GNGEVTQP 6JG UWDUGSWGPV KQP RCKTKPITGCEVKQP RCVJ(QHYKVJ '%.K 8%IGPGTCVKPINKVJKWOECTDQPCVG

 JCU C SWKVG EQORCTCDNG ∆) Ō XU Ō MECNOQN VQ  VJCV HTQO

.K%1− KP VJG ECUG QH .K '% KP (KIWTG  4GICTFKPI QVJGT VGTOKPCVKQP TGCEVKQPUQHVJGTCFKECNCPKQPVJGFKHHGTGPEGUQH∆)YKVJVJGEQTTGURQPFKPI TGCEVKQPUQH.K '%CTGCTQWPFMECNOQNGIMECNOQNNGUUPGICVKXGHQT VJG.KECTDKFG IGPGTCVKQPCPFMECNOQNOQTGPGICVKXGHQTNKVJKWO GVJ[NGPGFKECTDQPCVG CPFNKVJKWOFKGVJ[NGPGFKECTDQPCVG TGURGEVKXGN[ CPFPGCTN[KFGPVKECNHQTVJG41.KEQORQWPF 6JKUKORNKGUVJCVVJGOCLQT GHHGEVQH8%FQGUNKGKPVJGKPKVKCNUVGRQHVJGUQNXGPVTGFWEVKQP 4GICTFKPI QVJGT VGTOKPCVKQP TGCEVKQPU QH VJG TCFKECN CPKQP  VJG FKHHGTGPEGUQH∆)YKVJVJGEQTTGURQPFKPITGCEVKQPUQH.K '%CTGCTQWPF MECNOQNGIMECNOQNNGUUPGICVKXGHQTVJG.KECTDKFG IGPGTCVKQP CPF  MECNOQN OQTG PGICVKXG HQT NKVJKWO GVJ[NGPG FKECTDQPCVG  CPF NKVJKWOFKGVJ[NGPGFKECTDQPCVG TGURGEVKXGN[CPFPGCTN[KFGPVKECNHQTVJG4 1.KEQORQWPF 6JKUKORNKGUVJCVVJGOCLQTGHHGEVQH8%FQGUNKGKPVJG KPKVKCNUVGRQHVJGUQNXGPVTGFWEVKQP (KIWTGUJQYUVJCVVJG∆)QHVJGHWTVJGTTGFWEVKQP RCVJ%VQDGEQOGU MECNOQNOQTGPGICVKXGCUEQORCTGFYKVJVJCVQH.K 8%6JGVYQRCVJU

RCVJU#CPF$TGUWNVKPIKPNKVJKWOCNM[NFKECTDQPCVGU CPFGZJKDKV CPFMECNOQNNGUUPGICVKXG∆)TGURGEVKXGN[YJKEJSWCNKVCVKXGN[CITGGUYKVJ VJG %2%/ RTGFKEVKQPU HQT .K 8% #U VQ VJG %.K ECTDKFG IGPGTCVKQP RCVJ

RCVJ & VJGTG KU QPN[ C UNKIJV EJCPIG Ō XU Ō MECNOQN 6JG ∆) QH RCVJ'HQTOKPINKVJKWOCNM[NFKECTDQPCVG VJTQWIJVJGETQUUFKOGTK\CVKQPQH  YKVJVJGTCFKECNCPKQPKUEQPUKFGTCDN[JKIJGTVJCPVJCVQHRCVJ# Ō XUŌMECNOQNYJKEJTGXGCNUVJGETWEKCN TQNG QH VJG WPUCVWTCVGF ITQWR KP UVCDKNK\KPIVJGUGFKECTDQPCVGUVJTQWIJGNGEVTQPFGNQECNK\CVKQP#PQVJGTRCVJ ( VJTQWIJVJGETQUUFKOGTK\CVKQPQHYKVJHQTOKPIJCUC∆)TCVJGTENQUG VQ RCVJ $ Ō XU Ō MECNOQN 1WV QH VJG KUQOGTU CTKUKPI HTQO VJG VGTOKPCVKQPU QH TCFKECN CPKQPU  CPF  VJGKT UVCDKNKVKGU EQWNF DG TQWIJN[ FKXKFGFKPVQVJTGGECVGIQTKGUVJGNKVJKWOCNM[NFKECTDQPCVGUYKVJNQPIEJCKP = %*%*1%1.K UQNXCVGF D[ '%?   .K1%1%*%*%*%* 1%1.K YKVJ VJG VYQ VCKNU UQNXCVGF D[ '% CPF 8% TGURGEVKXGN[ CPF  = %*%*1%1.K UQNXCVGF D[ 8%? JCXG VJG JKIJGUV PGICVKXG TGNCVKXG )KDDU HTGGGPGTIKGU ŌŌCPFŌMECNOQNHQNNQYGFD[VJGNKVJKWO

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO

'



$






 







$

$

 3















2









$

$

'



$


$

































# 

-

'



$



$

%



( '

&

*







+











 -











*





 









 *



 

+

+



 






































 


















2

 



-

$











-



*







+


















1









"

$

*



 

+





'

$













 -



 


 





















 *





+











$

δ

δ
































!
























$











 (KIWTG 6GTOKPCVKQP RCVJU QH VJG TCFKECN CPKQP HTQO VJG '% TKPI QRGPKPI QH '%.K 8% ECNEWNCVGF YKVJ $29

) FR$29) F 4GRTQFWEGF HTQO =? YKVJ RGTOKUUKQPQH#OGT%JGO5QE 

CNM[NFKECTDQPCVGUYKVJUJQTVEJCKPCPFQTNKVJKWOECTDQPCVG= %*1%1.K UQNXCVGFD[8% GVJ[NGPGICUŌMECNOQN?`= %*1%1.KUQNXCVGF D[8%CPF'%KPVJGVYQTGURGEVKXGVCKNU GVJ[NGPGICUŌMECNOQN? =.K%1UQNXCVGFD[VYQ8%CPFQPG'% GVJ[NGPGICUŌMECNOQN? = %*1%1.K UQNXCVGF D[ '%  CEGV[NGPG ICU Ō MECNOQN?   =.K1 %*%1 %*1%1.K UQNXCVGF D[ '% CPF 8% KP VJG VYQ VCKNU Ō MECNOQN?CPFVJG%.KECTDKFGUCPFJCXGVJGNQYGUV∆) ŌCPF ŌMECNOQN

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG 





.

3









.

1











.















*

*















 

! 













!





$







 



























,





"

%



&

 #



 

 



0

4

3

.

/

2 1

5

6

4

0

4

3

3

6

/

.

7

/

8

/








































.

2













.



















 





+





!







 



"











 

 



























(






















)











'















*



















(KIWTG6GTOKPCVKQPRCVJUQHVJGTCFKECNCPKQPHTQO8%TKPIQRGPKPIQH '%.K 8%ECNEWNCVGF YKVJ $29

) FR$29) F 4GRTQFWEGF HTQO =? YKVJ RGTOKUUKQP QH #OGT%JGO5QE 

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

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



.K1%1 %*1 %*1%1.KYJKEJCTGWPFGTKPXGUVKICVKQP%1IGPGTCVKQP KUUVKOWNCVGFD[VJGRTGUGPEGQH8%6JGCOQWPVQH%1VJGTGHQTGKUJKIJGTVJCP VJCV QDVCKPGF KP VJG 8%HTGG ECUG 6JKU EQWNF GZRNCKP VJG TGEGPV GZRGTKOGPVCN HKPFKPIHTQO/%%   

*
































 





 













 *



 

 #



&

%

#











































*







!



* 







"



#

$



 %

&









'

"



(



!

 "

&



 #

#



)



 &









# 

 





 







$













 (KIWTG%1IGPGTCVKQPOGEJCPKUOKP8%EQPVCKPKPI'%QT2%UQNWVKQPUQH.+$U2TKQTVQ%1 IGPGTCVKQPVJTQWIJC EQPEGTVGF RCVJ ++ TGXGTUG &KGNU#NFGT TGCEVKQP C TCFKECN KU HQTOGF D[ VJG HCXQTCDNGRCVJ+D

 6JG ∆)ŏU QH HQTOCVKQP HQT TGCEVKQPU QH VJG OCLQT RTQFWEVU KPXQNXGF KP (KIWTGU  CTG UWOOCTK\GF KP 6CDNG  #EEQTFKPI VQ 6CDNG  VJG VYQ OQUV HCXQTCDNG TGCEVKQPU CTG VJG HQTOCVKQPU QH FQWDNGDQPF EQPVCKPKPI NQPI EJCKP NKVJKWO FKECTDQPCVGU .K1 %*%*%*%*1%1.K %1 4GCEVKQP  CPF

%*%*1%1.K 4GCEVKQP4GCEVKQPUCPFJCXGPGICVKXGHTGGGPGTIKGU CRRTQZKOCVGN[MECNOQNJKIJGTVJCP%*IGPGTCVKQPTGCEVKQPUCPF YJKNGTGCEVKQPKUQPN[MECNOQNJKIJGTPGICVKXGKP)KDDUHTGGGPGTI[ YJKEJKUUKOKNCTVQVJGOQUVHCXQTCDNGTGCEVKQPHQT8%HTGG'%DCUGFUQNWVKQP 6JKU UJQYU VJCV VJG ICU GXQNWVKQP QH %* RTQR[NGPG ICU KP '%DCUGF 2% DCUGFUQNWVKQPKUUWRRTGUUGFD[VJGRTGUGPEGQH8%

246

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Table 3 also shows that two acetylene (CH≡CH) involved reactions, 8 and 9, are rather less favorable than ethylene gas evolution reactions. It is safe to conclude that for the VC-containing EC(PC)-based solutions the amount of ethylene (propylene) gas might be less than that for VC-free solution, and that CO2 could be generated due to the presence of VC. Table 3 Thermodynamics for the major simple reductive reactions of EC/VC in electrolyte solutions (molar Gibbs free energies are calculated with the cluster (EC)Li+(VC)-B3PW91/6 311++G(d,p)//B3PW91/6-31G(d)).

Reactions

–∆Gm kcal/mol

1 2(EC)Li+(VC)+2e–→ EC•••LiO(CH=CHCH=CH)OCO2Li•••(EC)↓+CO2 ↑

285.5

2 2(EC)Li (VC) +2e → EC•••(LiOCO2CH=CHCH=CHOCO2Li) •••EC↓

281.2

3 2(EC)Li (VC) +2e → EC•••LiOCO2CH2CH2CH=CHOCO2Li•••EC ↓

276.2

4 2(EC)Li (VC) +2e → VC•••(LiOCO2CH2CH2)2↓

273.4

5 2(EC)Li (VC) +2e → VC•••(LiOCO2CH)2↓ +CH2=CH2 ↑

256.2

6 2(EC)Li (VC) +2e → EC•••LiOCO2CH=CHOCO2Li•••VC ↓ +CH2=CH2 ↑

255.0

7 2(EC)Li (VC) +2e → VC•••(Li2CO3)•••(VC)(EC) ↓ +CH2=CH2 ↑

250.0

8 2(EC)Li (VC) +2e → EC•••(LiOCO2CHCHOCO2Li) •••EC ↓ +CH≡CH ↑

242.6

9 2(EC)Li (VC) +2e → EC•••(LiOCO2CH2CH2OCO2Li) •••VC↓ +CH≡CH ↑

231.8

10 2(EC)Li (VC) +2e → EC•••LiOCO2CH=CHLi•••(EC)(VC) ↓

212.8

11 2(EC)Li (VC) +2e → VC•••LiOCO2CH2CH2Li•••(EC)(VC) ↓

206.2

+

–

+

–

+

–

+

–

+

–

+

–

+

–

+

–

+

+

5.3

–

–

Summary of Reductive Decomposition of Solvents in the Presence of VC

Based on the above investigation, the role of the additive VC on modifying the SEI film formation can be explained as follows. Since the VC molecule is more easily reduced than EC/PC by approximately 0.2 V in the Li-salt/EC-PC/VC electrolyte, it will be initially reduced to the more stable ion-pair intermediate, which may undergo two competitive homolytic ring openings through paths a and c, as shown in Figures 5 and 6, generating two radical anions, which correspond to ring openings of EC/PC and VC, respectively. The main products through termination reactions of the two radicals are various organic lithium alkyl carbonates, such as LiO(CH=CH-CH=CH)OCO2Li, (CH=CHOCO2Li)2,

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



.K1%1%*%*%*%*1%1.K %*%*1%1.K %*1%1.K CPF

%*1%1.K CPF KPQTICPKE .K%1 7UWCNN[ VJGUG QTICPKE NKVJKWO CNM[N ECTDQPCVGUCTGPQVGCU[VQDGFKUVKPIWKUJGFD[(6+4CPFGXGP:25DGECWUGQH VJGKTUKOKNCTEJCTCEVGTKUVKEU%QORCTGFYKVJUQNXGPVTGFWEVKXGFGEQORQUKVKQPUKP 8%HTGG '%2% UQNWVKQPU VJG GXQNWVKQP QH GVJ[NGPGRTQR[NGPG ICU KU EQORTGUUGFKP8% '%2%UQNWVKQPUYJKEJKUTGHNGEVGFD[VJGNCTIGTFKHHGTGPEG QH∆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Ō•••.K •••1Ō$TKFIGU  6Q IGV UVTWEVWTCN KPHQTOCVKQP CDQWV NKVJKWO CNM[N FKECTDQPCVGU YJGP VJG[ CTG RCTVQHVJG5'+HKNOCPFVQJGNRKPVJGCUUKIPOGPVQHGZRGTKOGPVCN(6+4URGEVTC FGPUKV[ HWPEVKQPCN VJGQT[ KU CRRNKGF KP VJKU UGEVKQP VQ NKVJKWO XKP[NGPG FKECTDQPCVG .8& %*1%1.K NKVJKWO FKXKP[NGPG FKECTDQPCVG .&8&

%*%*1%1.K NKVJKWO GVJ[NGPG FKECTDQPCVG .'& %*1%1.K CPF NKVJKWORTQR[NGPGFKECTDQPCVG .2&%*%* 1%1.K%* 1%1.KHQEWUKPI WRQP K VJGKT CUUQEKCVKQP KG IGQOGVTKECN UVTWEVWTGU CPF GPGTIGVKEU QH CIITGICVGUCPF KKVJGGHHGEVQHVJGKTCUUQEKCVKQPUQPVJGKPHTCTGFURGEVTQUEQR[ QPGQHVJGOCKPVGEJPKSWGUVJCVJCUDGGPWUGFVQKFGPVKH[VJG5'+EQORQPGPVU

 

)GQOGVTKGUCPF'PGTIGVKEU  .KVJKWOCNM[NFKECTDQPCVGUIGPGTCNN[UVTQPIN[CUUQEKCVG VJTQWIJ KPVGTOQNGEWNCT 1•••.K•••1KPVGTCEVKQPUTGUWNVKPIKPXCTKQWUEQPHQTOCVKQPUUWEJCUEJCKPNKMG

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

ECIGNKMGCPFSWCUKRNCPCTIGQOGVTKGU6JGUVCDKNKV[QHVJGCIITGICVGUOCKPN[ FGRGPFUQPVJGPWODGTQHVJG1•••.K•••1KPVGTCEVKQPU .KVJKWOXKP[NGPGFKECTDQPCVG .8& %*1%1.K1RVKOK\GFUVTWEVWTGU HQT EJCKP ECIGNKMG CPF UKFGD[UKFG FKOGTU D E CPF F EJCKP CPF SWCUKRNCPCT VTKOGTU G CPF H CPF SWCUKRNCPCT VGVTCOGTU I CTG KNNWUVTCVGFKP(KIWTGVQIGVJGTYKVJUGNGEVGFUVTWEVWTCNFCVCDQPFQTFGTUCPF ∇ρ T6JGVQVCNGPGTI[QHVJGU[UVGOKUEQPUKFGTCDN[NQYGTGFWRQPHQTOCVKQP QHVJGCIITGICVGUVJTQWIJ1•••.K•••1KPVGTCEVKQPU6JGDKPFKPIGPGTIKGU $'QH XCTKQWU EQPHQTOCVKQPU CTG UWOOCTK\GF KP 6CDNG  VJG $'U CTG FGHKPGF D[ VJG FKHHGTGPEG DGVYGGP VJG GPGTI[ QH VJG CUUQEKCVGU CPF P VKOGU VJG GPGTI[ QH VJG OQPQOGT KPENWFKPI XCTKQWU EQTTGEVKQPU NKMG <2' CPF $55' =$'' POGT P ' OQPQOGT? 1H VJG VJTGG FKOGTU VJG ECIGNKMG E KU C INQDCN OKPKOWO FWG VQ KVU HCXQTCDNG IGQOGVT[ HQTOKPI VJG JKIJGUV PWODGT QH 1•••.K•••1 KPVGTCEVKQPUVJGDKPFKPIGPGTIKGUQHEFCPFDYKVJDQVJ<2'CPF$55' EQTTGEVKQPUCTGŌŌCPFŌMECNOQNTGURGEVKXGN[  





 



α=0.360 










$





#




















$

& #

















 





































 CE 







































 









 











 























!

"


















α=0.362 

D 

#





$





'







#

%

 

) *



α=0.224













'

) *





$







% &



'

# #







 (





α=0.373



 F





$





$





#

%

%





 

(KIWTGEQPVKPWGUQPVJGPGZVRCIG 



6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO 





















 












 



 












 










 



  







 





















 























 













 











 

α=0.325



α=0.369

G  5 %

1

2

 7

8

E



3 4

6 

,

0

.

α /

:

@ D

D

9 ;

* +

-

α=0.405

,

.

,

/

* +

-

α=0.358 5

7

8

5

3 4

6

α  

:

<

=

>

9 ;

 

5

8

?

?

3 4

 6









α



















 

 











:

@

A B

9 ;







 





 



























 

 %









2

2

(













 



#

!

$

&





5

8

5

C

3 4 

!

"

6



#

$

%

α &



% '

"

( )

:

@ B

D

9 ;



%

1



 

#





!

$

&

 















 HI













  (KIWTG5VTWEVWTGUYKVJUGNGEVGFRCTCOGVGTUD[$29) F C.8& DFFKOGTU

GH VTKOGTU I VGVTCOGT 5QNKF DCNNU UVCPF HQT QZ[IGP CVQOU DKI GORV[ QPGU HQT ECTDQP CVQOU UOCNN GORV[ QPGU HQT J[FTQIGP CVQOU CPF VJG DCNNU OCTMGF YKVJ C UVCT UVCPF HQT NKVJKWO CVQOU  6JG PWODGTU KP RCTGPVJGUGU TGHGT VQ QXGTCNN DQPF QTFGT YKVJ 0$1$29) CPF VJQUGKPDTCEMGVUVQ∇ρ TCVVJGDQPFETKVKECNRQKPVUQDVCKPGFYKVJ#+/$29) SCPF α CTG VJG RCTVKCN EJCTIGU CPF VJG VQVCN QXGTCNN DQPF QTFGTU QH .K KQPU ECNEWNCVGF  YKVJ 0$1 $29)TGURGEVKXGN[4GRTQFWEGFHTQO=?YKVJRGTOKUUKQPQH#OGT%JGO5QE

 +PIGPGTCNVJGVTKOGTUQHNKVJKWOXKP[NGPGFKECTDQPCVGEQWNFDGENCUUKHKGF KPVQ QRGP CPF ENQUGF UVTWEVWTGU 6JG QRGP UVTWEVWTGU NKMG EJCKPNKMG G CPF UKFGD[UKFGVTKOGTU PQVUJQYPCTGPQVCUUVCDNGCUVJGENQUGFUVTWEVWTGUECIG NKMG PQV UJQYP CPF SWCUKRNCPCT H +P VJG ECUG QH H VJG VJTGG QWVGT 1•••.K•••1 KPVGTCEVKQPU VGPF VQ DG TCVJGT HNCV ∠1•••.K•••1 ` YKVJ VYQ NQPI1•••.KDQPFU —CPF—YJGTGCUVJGKPPGTKPVGTCEVKQPUCTGXGT[ DGPV ∠1•••.K•••1 ` CPF VYQ 1•••.K DQPFU JCXG XGT[ UKOKNCT NGPIVJU QH —#FFKVKQPCNN[CUKZOGODGTGFTKPIQEEWTUKPVJGEGPVGTQHVJGVTKOGTH YJKEJ KU TGHNGEVGF D[ VJG EQPUKFGTCDNG 0$1 DQPF QTFGTU QH  HQT VJG VJTGG NQPIGT1•••.KDQPFU —CPFHQTVJGQVJGT VJTGG UJQTVGT 1•••.K

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

DQPFU `— +V KUQDUGTXGFVJCVVJGVJTGGKPPGT.KKQPUCTGUJCTGFD[HQWT CFLCEGPVQZ[IGPCVQOUYJKNGVJGQWVGT.KKQPUCTGQPN[UJCTGFD[VJTGGQZ[IGP CVQOU VYQUVTQPICPFQPGYGCM1•••.K #U C TGUWNV QH VJKU VJG QXGTCNN DQPF QTFGTUQHVJGKPPGT.KECVKQPUCTGJKIJGTVJCPVJQUGQHVJGQWVGTQPGU XU HKUOWEJOQTGUVCDNGVJCPG $55' <2'EQTTGEVGF$'ŌXU ŌMECNOQNUGG6CDNG5KOKNCTN[VJGENQUGFSWCUKRNCPCT IKUCICKP VJGOQUV UVCDNG KUQOGT HQT VJG VGVTCOGTU 6JKU KPFKECVGU VJCV .8& YQWNF ITQY HTQOVJGECIGNKMGFKOGTEVQVJGENQUGFSWCUKRNCPCTVTKOGTHVJGP VQVJG ENQUGF VGVTCOGT I YKVJ C ENQUGF UVTWEVWTG 6JGTOQF[PCOKEU HQT VJG ITQYVJ TGCEVKQPU QH NKVJKWO CNM[N FKECTDQPCVGU KU UWOOCTK\GF KP 6CDNG  6JG JKIJ PGICVKXG∆)XCNWGUCNUQUJQYVJCV.8&VGPFUVQCUUQEKCVGHQTOKPIJKIJQTFGT UVTWEVWTGU  6CDNG%CNEWNCVGFTGUWNVUHQTNKVJKWOCNM[NFKECTDQPCVGUCVVJG$29) FNGXGNKPENWFKPI VQVCN DKPFKPI GPGTIKGU 6$' MECNOQN $55' MECNOQN $55'EQTTGEVGF DKPFKPI GPGTIKGU =6$' $55' MECNOQN <2'EQTTGEVGF DKPFKPI GPGTIKGU =6$' <2' MECNOQN XCTKCVKQPU QH VJG VQVCNQXGTNCRYGKIJVGF0#1DQPFQTFGTUHQTVJGKPXQNXGF.KKQPU ∆αCPFXCTKCVKQPUQHVJGVQVCN RCTVKCNEJCTIGUHTQO02#HQT.KKQPU Ō∆SG 6$'C

5VTWEVWTGU

6$' $55'D

6$' <2'E

$'F

∆αG

Ō∆S

D

Ō



Ō

Ō

Ō





E

Ō



Ō

Ō

Ō





F

Ō



Ō

Ō

Ō





G

Ō



Ō

Ō

Ō





H

Ō



Ō

Ō

Ō





I

Ō



Ō

Ō

Ō





















D

Ō



Ō

Ō

Ō





E

Ō



Ō

Ō

Ō





F

Ō



Ō

Ō

Ō





















D

Ō



Ō

Ō

Ō





E

Ō



Ō

Ō

Ō





F

Ō



Ō

Ō

Ō





G

Ō



Ō

Ō

Ō





H

Ō



Ō























Ō

Ō

Ō





 E C

$55'

 Ō

6$'' POGTP ' OQPQOGTD6$' $55'6$' $55'E6$' <2'6$'YKVJ<2'EQTTGEVKQPU $'6$' <2' $55'G∆α∑LαLŌOα OVJGPWODGTQH.KKQPU

F

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



.KVJKWO&KXKP[NGPG&KECTDQPCVG .&8& %*%*1%1.K 'ZEGRVHQT VJGRTGUGPEGQHVJGNQPIGTEQPLWICVGF %*%*ITQWRKVUUKFGD[UKFGFKOGT

PQVUJQYPCUYGNNCUVJGEJCKPFKOGTD (KIWTGUVTWEVWTGUCPFDKPFKPI GPGTIKGU CTG TCVJGT ENQUG VQ VJGKT CPCNQIU QH NKVJKWO XKP[NGPG FKECTDQPCVG *QYGXGTVJGFKOGTENQQMURTQPQWPEGFN[FKHHGTGPVHTQOVJGECIGFKOGTE 6JG VYQ KPPGT .K KQPU CTG OQTG UVTQPIN[ NKPMGF YKVJ VJTGG CFLCEGPV QZ[IGP CVQOU VJCP VJG VYQ QWVGT .K KQPU FQ +P URKVG QH NCTIGT FGXKCVKQPU HTQO RNCPCT UVTWEVWTG EQORCTGF YKVJ VJG VTKOGT H VJG EQTTGURQPFKPI SWCUKRNCPCT VTKOGT JCU CNUQ DGGP HQWPF CU UJQYP KP (KIWTG F 5KOKNCTN[ C UKZOGODGTGF TKPI CNUQ GZKUVU KP VJG VTKOGT EGPVGT YJKEJ EQPVTKDWVGU VQ VJG GZVTC UVCDKNKV[ QH VJG KUQOGTF  










α=0.256



C 









$



$



 















'









α =0 .361













 (





)









)



























 (



D  $



$

#









$



#

'




%











.

 







%








α=0.374 

!

& '



 



%



!

" !




 












!














) *




















2

2



  



















α=0.353



















α=0.411






































 




























 



 

















 





 













 3

























/


















 (

) *



















α=0.401 



$



'

& '

" %









#

%






























$

























%































0












 



EF



α=0.366



(KIWTG 5VTWEVWTGU UJQYKPI UGNGEVGF RCTCOGVGTU ECNEWNCVGF YKVJ $29) 

C.KVJKWO FKXKP[NGPG FKECTDQPCVG DE FKOGT F VTKOGT 6JG U[ODQNU CPF PQVCVKQPU JCXGVJGUCOGOGCPKPICUVJQUGKP(KIWTG

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

.KVJKWO'VJ[NGPG&KECTDQPCVG .'& %*1%1.K6JGECIGCPFEJCKP NKMG FKOGTU D CPF E EJCKP VTKOGT F SWCUKRNCPCT VTKOGT G CPF SWCUKRNCPCTVGVTCOGT HQH.'&CTGKNNWUVTCVGFKP(KIWTG#UGZRGEVGFVJG ECIGNKMG FKOGT JCU OWEJ NQYGT GPGTI[ VJCP VJG EJCKPNKMG FKOGT 6JG SWCUK RNCPCT VTKOGT G TGOCTMCDN[ VYKUVU DWV UVKNN JCU VJG JKIJGUV PGICVKXG DKPFKPI GPGTI[ QH VJG VTKOGTU 6JG 1•••.K•••1 KPVGTCEVKQPU KP VJG EJCKP VGVTCOGT PQV UJQYPKP(KIWTGCTGTCVJGTUKOKNCTVQVJQUGQHKVUEJCKPVTKOGTFCPFVJG EJCKPVGVTCOGTJCUOWEJNQYGTPGICVKXGDKPFKPIGPGTI[VJCPVJGENQUGF SWCUK RNCPCTVTKOGTG ŌXUŌMECNOQN   





















 &

  











α=0.367 









 



& $

 




α=0.264 


















































CD  



































 



 












 

α=0.262 



 






































α=0.371

E











































 

 



























 



 

 

































 











α=0.368

 

F  

(KIWTGEQPVKPWGUQPVJGPGZVRCIG

α=0.370

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO 





















$



$






























α=0.364

$

%



$

 



α=0.419 



 





































G













  

 





α=0.417

α=0.400



















 

 













 























































 









 











 









α=0.374 

 



α=0.369

H



  (KIWTG  5VTWEVWTGU UJQYKPI UGNGEVGF RCTCOGVGTU ECNEWNCVGF YKVJ $29) 

C.KVJKWOGVJ[NGPGFKECTDQPCVG DEFKOGT FGVTKOGT HVGVTCOGT6JGU[ODQNU CPFPQVCVKQPUJCXGVJGUCOGOGCPKPICUVJQUGKP(KIWTG

  .KVJKWO 2TQR[NGPG &KECTDQPCVG .2& %*%* 1%1.K%* 1%1.K  6JG GHHGEVQHVJG.2&OGVJ[NITQWRQPKVUEJCKPFKOGTCUYGNNCUQPKVUEJCKPVTKOGT KU CNOQUV PGINKIKDNG YJKEJ EQWNF DG TGHNGEVGF D[ VJG ECNEWNCVGF  1•••.K•••1 KPVGTCEVKQPUYKVJCNOQUVKFGPVKECNDKPFKPIGPGTIKGUŌMECNOQNHQTVJGEJCKP FKOGTU QH .'& CPF .2& Ō MECNOQN HQT VJGKT EJCKP VTKOGTU *QYGXGT KV FQGUCHHGEVVJGSWCUKRNCPCTVTKOGT6JGDKPFKPIGPGTI[QHVJG.2&SWCUKRNCPCT VTKOGT HQT GZCORNG KU CRRTQZKOCVGN[  MECNOQN NQYGT PGICVKXG VJCP VJCV QH .'&SWCUKRNCPCTVTKOGT#FFKVKQPCNN[CUUJQYPKP(KIWTGVJGSWCUKRNCPCT UVTWEVWTG QH .2& E VYKUVU EQPUKFGTCDN[ OQTG VJCP VJCV QH .'& VJKU HCEV VQIGVJGT YKVJ VJG RTGUGPEG QH VJG OGVJ[N ITQWR OC[ RCTVKCNN[ DNQEM VJG & ITQYVJ QH .2& CPF RTGXGPV HTQO DWKNFKPI WR C VJKEM CPF FGPUG GPQWIJ 5'+ NC[GT 

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG



H 

G





E



 (KIWTG5KFGXKGYUQHVJGSWCUKRNCPCTVTKOGTUQH.8& H.'& GCPF.2& E 



+P UWOOCT[ UGXGTCN NKVJKWO CNM[N FKECTDQPCVGU CTG CDNG VQ UVTQPIN[ CUUQEKCVG VJTQWIJ 1•••.K•••1 KPVGTCEVKQPU YJKEJ EQWNF DG ENCUUKHKGF CU KQPKE DQPFKPI 6JG DKPFKPI GPGTIKGU CTG UQOGYJCV FKHHGTGPV HQT GZCORNG VJG[ CTG ŌŌŌCPFŌMECNOQNHQTVJGENQUGFSWCUKRNCPCTVTKOGTU H F G CPF E QH NKVJKWO CNM[N XKP[NGPG FKXKP[NGPG GVJ[NGPG CPF RTQR[NGPGFKECTDQPCVG6JGGPGTI[ FKHHGTGPEG DGVYGGP KUQOGTU HQT C IKXGP P OGT QH C URGEKHKE FKECTDQPCVG UJQYU VJCV NKVJKWO FKECTDQPCVG YQWNF ITQY CFQRVKPI C ENQUGF UVTWEVWTG EQPHQTOCVKQP VJCV KU VJG[ YQWNF CUUQEKCVG VQ VJG ECIGNKMGFKOGTHKTUVVJGPVQVJGENQUGFRUGWFQRNCPCTVTKOGTVJGPVQVJGENQUGF

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



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νCU%1  UJKHVU HTQO EOŌ QHVJGOQPQOGTVQEOŌQHVJGFKOGTECPFVJGPURNKVUKPVQVYQRGCMU

 CPF EOŌ HQT VJG VTKOGT CPF [KGNFU VJTGG RGCMU   CPF EOŌ HQT VJG VGVTCOGT EQPUKFGTCDN[ EQWRNKPI YKVJ VJG %% UVTGVEJKPI

νEEHQTVJGVYQNCVVGTECUGU KK#PGYYGCMRGCMCUUKIPGFVQ%%UVTGVEJKPI

νEEFWGVQCRRGCTCPEGQHFQWDNGDQPFRQNCTKV[QEEWTUCVEOŌKP(KIWTG DQHVJGFKOGTEDWVKUEQXGTGFD[VJGUVTQPIGTνCU%1 KPVJGECUGUQHVTKOGT

EOŌ OGFKWO CPF VGVTCOGT EOŌ OGFKWO KKK 6JG %1 U[OOGVTKE UVTGVEJKPI RGCM νU%1  CPF VJG %* KPRNCPG DGPFKPI RGCM δ%* CNUQ[KGNFDNWGUJKHVUHTQOOQPQOGTVQFKOGTE

νU%1 XUEOŌ δ%* XU EOŌ *QYGXGT VJG[ CTG PQ NQPIGT YGNN UGRCTCVGF KP (KIWTG EQHVJGVTKOGTHCPFKP(KIWTGFQHVJGVGVTCOGTICPFVJG[CTGGXGP EQPUKFGTCDN[EQWRNGFVQGCEJQVJGTKPI 6JG.K1UVTGVEJKPI ν1.KIQGU VJTQWIJCDNWGUJKHVHTQOVJGOQPQOGTVQVJGFKOGTE ν1.KXUEOŌ CPF VJGP TGF UJKHVU VQ VJG VTKOGT CPF VGVTCOGT  CPF EOŌ YJKEJ KU KP NKPG YKVJ VJG XCTKCVKQPU QH VJG .K1 DQPF NGPIVJU 1VJGT XKDTCVKQPCN OQFGU UWEJ CU %* UVTGVEJKPI ν%*`EOŌ % CNM[N ECTDQP1 UVTGVEJKPI

ν%1`EOŌ %* QWVQHRNCPG DGPFKPI δ%*`EOŌ YGCM CPF %1 DGPFKPI δ%1 `EOŌUJQYQPN[UOCNNEJCPIGUFWGVQCUUQEKCVKQP 









256

Lithium-Ion Batteries: Solid-Electrolyte Interphase 1285 δ CH

3.0

IR Relative Intensity

2.5

2.0

604 ν LiO

δ CH

3

964

725

1280

1140

610 ν

1.5

ν co

LiO

δ co

691

769

δ CH

δ CH

3

973

1682 ν c=c

CH

1340 ν

1069

ν c=o

1708 ν

ν LiO 660

δ co

0.5

ν LiO 626

2

ν co 1166

1056

3

718 763

ν co

δ CH

1689 ν c=c

δ CH

1159

δ co

(d)

ν as,co

1385

δ CH

3

773~754

as,co2

1632

1392

1067

as,co2

1701 ν

c=o

1640 ν

1361

1.0

2

1663 1638

1390 ν c=o

1145

770

ν as,co c=o

1372 δ

1123

δ co

529

1340 ν

ν c-o

ν c=o δ CH

ν as,co

1308

1596

2

(c) as,co2

1681ν

c=c

(b) δ CH

1240

906

0.0

(a)

500

1000

1500

Frequency /cm

2000

-1

Figure 14 Simulated infrared spectra for LVD 10a (a), dimer 10c (b), trimer 10f (c) and tetramer –1 10g (d) with parameters: 50% Gaussian + 50% Lorentzian functions, 20 cm peak width; frequencies and relative intensities based on QC calculations of B3PW91/6-31G(d) with a scaling factor of 0.949. Reproduced from [38] with permission of Amer. Chem. Soc.

Similarly, considerable blue shifts also take place on νas, CO , which are –1 1597, 1662 and 1685-1667 cm for the monomer, dimer, and trimer of LDVD, respectively, as illustrated in Figure 15. A weak C=C stretching peak appears at –1 1628 cm in Figure 15a of the monomer of LDVD, which splits and shifts to –1 –1 higher frequencies, 1622 and 1640 cm for the dimer, 1651 and 1665 cm for trimer, and are covered by a stronger νas, CO peak. Compared with those of the LVD trimer (Figure 14c), the νC=C mode of the LDVD trimer (Figure 15c) has –1 –1 an over 20 cm red shift (1651~1665 vs 1689 cm ). Like LVD, the C=O stretching (νC=O) and C-H in-plane bending peak (δCH) strongly couple in the dimer (Figure 15b) as well as in the trimer (Figure 15c). The rest of the variations, such as Li-O stretching (νOLi), CO2 symmetric stretching (νs, CO ) and –1 CO3 bending (δCO ~770 cm ), are also very similar to the patterns of those displayed in Figure 15c. The effects of association on the infrared spectra of LED and LPD are displayed in Figure 16 and Figure 17, respectively. Associations lead to 2

2

2

3

257

Theoretical Insights into the SEI Composition and Formation Mechanism

ν C=O δ CH 1337

1667 ν

2.0 ν as,CO ν c=c

IR Relative Intensity

δ CH ν c=o

1.5

585

ν LiO

ν CO δ CO

610 686

1100

1071

3

1625

1296 1280

1397 δ

δ CH

1662 ν

CH

ν CO 1340

(c)

ν CO

1643

1358

1.0 δ CO3 633

775

as,CO 2

1204

781 1127

585

as,CO 2

2

δ OP-CH

ν CO ν CO ,δ CH 1302 ν CO 1151 δ CH 1088

1370

ν C=C

δ CH

(1622m,1640m)

1263

920 983

0.5 ν CO

ν LiO

δ IP-CH

ν as,CO

1201

1597

1116

602

δ CO

763

3

(b) 2

δ OP-CH

δ ip-CH

ν C=C

950

1200

1628

0.0

(a)

500

1000

1500

Frequency/cm

2000

-1

Figure 15 Simulated infrared spectra for lithium divinylene dicarbonate 11a (a), dimer 11c (b), and trimer 11d (c) with the same parameters as in Figure 14.

Reproduced from [38] with

permission of Amer. Chem. Soc.

changes similar to those observed for LVD and LDVD. The antisymmetric stretching of CO2 (νas, CO ) also significantly blue shifts and appears in the range –1 of 1620-1680 cm for their quasi-planar trimers. Identifying the SEI layer components, either with in situ or ex situ FTIR measurements in mixtures of various organic lithium dicarbonates, a strong peak is always observed in –1 the range of 1640-1680 cm , which should be assigned to the associates of various lithium alkyl dicarbonates instead of to their monomers since νas, CO of –1 39 the latter is rather lower than 1600 cm . Aurbach et al. measured IR bands of lithium propylene dicarbonate and made the assignments as follows: νC-H 2990m, 2950m, 2870w; νas, CO 1665s 1540s; δCH, CH 1430s; νs, CO 1330m; νC-O 1200w 1150m 1100sh 1070sh 1050s; δCH 920w; δCO 870s 830m; δCO 750w 650w; νLi-O 520s. The simulated results in Figure 17c basically agree with Aurbach’s experimental results, especially with respect to the leading peak 2

2

3

2

2

2

3

2

258

Lithium-Ion Batteries: Solid-Electrolyte Interphase

groups, that is, νC-H, νas, CO (1682-1619 cm ), δCH, CH (~1430), νs, CO (~1330), νC-O (1082-1110), δCO (784); δCO (670-694), and νLi-O (614-642), even though minor differences still exist on details of the spectra. –1

2

3

3

2

2

ν C=O , δ CH 1323

2.5

ν as,CO

2

1628

2

1643 1663

2.0

IR Relative Intensity

δ CH

δ CO Li,ν LiO

ν LiO

626

2

ν CO

685

δ CO

3

1068

δ CH

ν C=O , δ CH

δ LiO

2

ν CO 3

1083 1106

771

1684 (c)

ν C=O ,δ CH

1403 2

1340

δ CO

1622

1433

1130

1.5

1.0

1373

1094

786

656

δ C=O

2

1284

2

1435

ν as,CO

1308

2

1580 (b) 1397

0.5

ν C=O , δ CH

ν LiO δ LiCO

ν CO

620 2

543

δ CO δ CO 2

721

1100

δ CH

2

2

1270

3

792

0.0

(a)

500

1000

Frequency/ cm

1500

2000

-1

Figure 16 Simulated infrared spectra for lithium ethylene dicarbonate 12a (a), dimer 12b (b), and trimer 12e(c) with the same parameters as in Figure 14. Reproduced from [38] with permission of Amer. Chem. Soc.

Comparisons are made in Figure 18 between the IR spectra of the quasiplanar trimers of four lithium alkyl (vinylene, divinylene, ethylene and –1 propylene) dicarbonates in the range of 400-3200 cm . Qualitatively they are so similar that it is not easy to distinguish them with FTIR only. C-H stretching vibrations of LVD and LDVD are weaker than those of LED and LPD, and the –1 frequencies of the former are higher by about 90 cm than those of the latter –1 (3074-3104, 3031-3092 vs 2943-3011, 2924-2997 cm ), which is one of the pronounced features that could be used to determine whether double bond containing species are present in the SEI layer. Bending of CH/CH2/CH3 (δCHn) –1 is another clear difference, the calculated values are 1385 cm for 1j, 1397 for 2f, 1433 for 3f and 1442 for 4e, but this peak is usually covered by the stronger –1 C-O stretching (νCO) with approximately 40-60 cm lower frequency. As

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



FKUEWUUGFCDQXGRGCMUTGNCVGFVQ%%UVTGVEJKPICTGFGHKPKVGN[FGVGTOKPGFYKVJ &(6 ECNEWNCVKQPU CTQWPF EOŌ HQT .8& CPF EOŌ HQT .&8& TGURGEVKXGN[JQYGXGTVJG[CTGPQVYGNNUGRCTCVGFHTQOVJGUVTQPIGTνCU%1 RGCM 5KPEGKVKUEQOOQPN[MPQYPVJCVJKIJQTFGTEQPLWICVKQPOQXGUνEEVQNQYGT HTGSWGPEKGU CPF KPETGCUGU KVU KPVGPUKV[ CNUQ TGHNGEVGF D[ VJG RTGUGPV FKHHGTGPEG HQWPFDGVYGGPXKP[NGPGFKECTDQPCVGCPFFKXKP[NGPGFKECTDQPCVGKVKUGZRGEVGF VJCVVJGWPUCVWTCVGFNKVJKWOFKECTDQPCVGU.K1%1 %*%*P1%1.KYKVJOWEJ JKIJGT QTFGT P  YQWNF NGCF VQ UGRCTCVG VJG %% UVTGVEJKPI RGCMU YKVJ HTGSWGPE[ NQYGT VJCP EOŌ 6JKU CTIWOGPV TQWIJN[ UWRRQTVU #WTDCEJŏU EQPENWUKQPDCUGFQPVJG+4RGCMUHQWPFDGVYGGPCPFEOŌVJCVVJG 5'+ NC[GT EQPVCKPU RQN[OGTKE URGEKGU QH RQN[XKP[NGPG FKECTDQPCVG KP 8% EQPVCKPKPIGNGEVTQN[VGUQNWVKQPU 

 ν







ν



ν 

δ 































1





!

,

δ

-

0








ν 













ν

ν



/











δ





ν 



δ

)





δ 

,























!



























!









ν



















/

'

 













)

ν .







δ



ν 











ν

,

+

δ

*






















!

δ

δ





























!



)

!





 



!

(





 (

ν

'

ν














ν







!







δ 






δ

!

ν







 







 









δ





δ







 









δ 

δ 






































































&

2

% "

3

4

5 6

4

7

8 9

#

$

(KIWTG5KOWNCVGFKPHTCTGFURGEVTCHQT.2&OQPQOGT CECIGNKMGFKOGT DCPFSWCUKRNCPCT VTKOGTE EYKVJVJGUCOGRCTCOGVGTUCUKP(KIWTG4GRTQFWEGFHTQO=?YKVJRGTOKUUKQPQH #OGT%JGO5QE 

 .KVJKWO CNM[N FKECTDQPCVGU TGUWNVKPI HTQO CNM[N ECTDQPCVG TGFWEVKXG FGEQORQUKVKQP UVTQPIN[ CUUQEKCVG VJTQWIJ 1•••.K•••1 KPVGTCEVKQPU CPF YQWNF ITQY CFQRVKPI VJG ENQUGF UVTWEVWTG EQPHQTOCVKQP KG VJG[ YQWNF CUUQEKCVG VQ VJG ECIGNKMG FKOGT HKTUV VJGP VQ VJG ENQUGF RUGWFQRNCPCT VTKOGT VJGP VQ VJG ENQUGF VGVTCOGT CPF UQ QP  #FFKVKQPCNN[ YG HQWPF VJCV HGCVWTGU QP VJG +4 URGEVTC ECWUGF D[ CUUQEKCVKQPU QH NKVJKWO CNM[N FKECTDQPCVGU CTG ENQUGT VQ VJG

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

GZRGTKOGPVCNFCVCVJCPVJGOQPQOGTURGEVTWO$QVJVJGGPGTIGVKETGUWNVUCPF+4 UKOWNCVKQP VJGTGHQTG UWRRQTV VJG EQPENWUKQP VJCV QPEG VJG RTQFWEVU CTG HQTOGF PGCTD[ VJG CPQFG UWTHCEG VJG[ YKNN FGRQUKV QP VJG UWTHCEG DWKNFKPI VJG 5'+ NC[GTCUCUUQEKCVGFEQORQWPFUUWEJCUFKOGTVTKOGTVGVTCOGTCPFGXGPJKIJGT QTFGTPOGTUKPUVGCFQHQHUVC[KPIKPOQPQOGTKEHQTO 








1

,

-

 0






















/







)

,

/






 '

)

. 














-

 ,







+ 

*

)

(





















(







'






















































































&

2

% "

3

4

5 6

4

7

8 9

#

$

(KIWTG%QORCTKUQPUQHVJGUKOWNCVGFURGEVTCDGVYGGPVJGNKVJKWOXKP[NGPGFKECTDQPCVGVTKOGT L C NKVJKWO FKXKP[NGPG FKECTDQPCVG VTKOGT H D NKVJKWO GVJ[NGPG FKECTDQPCVG VTKOGT H E CPF NKVJKWO RTQR[NGPG FKECTDQPCVG G F 4GRTQFWEGF HTQO =? YKVJ RGTOKUUKQP QH #OGT%JGO5QE

  #FUQTRVKQPCPF6YQ&KOGPUKQPCN#UUQEKCVKQPQH.KVJKWO#NM[N &KECTDQPCVGUQP)TCRJKVG5WTHCEGU6JTQWIJ1Ō•••.K ••••π #TGPG +PVGTCEVKQPU  2TQXKFGFVJCVCPGHHGEVKXGCFUQTRVKQPUJQWNFQEEWTQPITCRJKVGUWTHCEGUVQHQTO VJG5'+NC[GTKVKUCNUQPGEGUUCT[VQKPXGUVKICVGVJGCFUQTRVKQPEJCTCEVGTKUVKEUQH NKVJKWO CNM[N FKECTDQPCVGU QP VJG CPQFG UWTHCEG  +P VJG EWTTGPV UGEVKQP &(6 KU CRRNKGF VQ VJG OCKP TGFWEVKQP RTQFWEVU QH 8% '% CPF 2% HQEWUKPI WRQP

KVJGKTCFUQTRVKQPQPVJGGFIGRNCPGCUYGNNCUQPVJGDCUCNRNCPGQHITCRJKVG UWTHCEG KG GSWKNKDTKWO UVTWEVWTGU DKPFKPI GPGTIKGU CPF VJG GHHGEV QH VJG CFUQTDCVGU QP VJG ITCRJKVG UWTHCEG CPF KK VJGKT CUUQEKCVKQPU QP VJG ITCRJKVG UWTHCEG

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



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— CPF VJG[ CTG XGT[ ENQUG VQ VJCV — QH ITCRJKVG YJKNG VJG %% DQPFU EQPPGEVKPIVJGRGTKRJGTCNCPFVJGUGEQPFTKPICTGUNKIJVN[NQPIGTCRRTQZKOCVGN[ — )   +P VJG QWVGTOQUV TKPI VJGTG CTG VJTGG UQTVU QH %% DQPFU CPF— ) VJGHKTUVQHYJKEJTGHGTUVQVJGVYQECTDQP CVQOUFKTGEVN[NKPMGFVQCJ[FTQIGPCVQOCPFDGJCXKPINKMGCFQWDNGDQPF6JG TGUWNVUWUKPIVJGVJTGGDCUKUUGVUUJQYVJCVVJG%%DQPFNGPIVJUCITGGXGT[YGNN YKVJKP  — 6Q MGGR EQORWVCVKQP YKVJKP CP CEEGRVCDNG VKOGHTCOG HWNN QRVKOK\CVKQPU CTG QPN[ RGTHQTOGF CV VJG ) NGXGN HQT VJG TGNGXCPV KPXGUVKICVKQPUQHVJGVYQNCTIGTENWUVGTU5KOKNCTN[VQ%*VJG%%DQPFUCTG KPVJGTCPIGQH—HQTVJGVJTGGKPPGTTKPIUQH%*CPFVJGNGPIVJU QH VJG VJTGG V[RGU QH RGTKRJGTCN %% DQPFU CTG   CPF  — TGURGEVKXGN[ 6JG FKCOGVGT QH %* KU CRRTQZKOCVGN[  — DGKPI UKOKNCT VQ VJCV — QH ET[UVCNNKVGU KP FKUQTFGTGF ECTDQP OCVGTKCNU RTGRCTGF D[ JGCV VTGCVOGPVQHRQN[RJGP[NGPG)GPGTCNN[VJGCXGTCIGEJCTIGUHQTVJGOQUVKPPGT

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Ō KU  G8  MECNOQN %QORCTGF YKVJ VJG PGWVTCN )T QPG EGPVTCN ECTDQP CVQO P   IGVU CRRTQZKOCVGN[GKP)TŌ ŌGXUŌGVJGECTDQPUKPVJGUGEQPF TKPI P   EJCPIG C NKVVNG ŌG XU ŌG CPF EQPUGSWGPVN[ VJG

262

Lithium-Ion Batteries: Solid-Electrolyte Interphase

charged electron mainly delocalizes over the edge carbons (–0.54 e) and on the outside hydrogen atoms (–0.34e).

7.2 Adsorption of Lithium Alkyl Dicarbonates on the Basal Plane of the Neutral Graphite Surface Adsorptions of a single molecule on the basal plane of graphite were simulated with the neutral C54H18. Adsorption energies (AE), are summarized in Table 5, + and distances from Li , carbonyl as well as alkyl carbons to the nearest graphite carbon are listed in Table 6. Two types of adhesion are located with respect to orientations of the dicarbonate molecule either parallel or perpendicular to the graphite surface. Figures 19a and 19b illustrate the top and side views of lithium vinylene dicarbonate (LVD), (CHOCO2Li)2, on the surface of C54H18, respectively, where LVD is located nearly parallel to the graphite surface + (conformation 7.1). Figure 19a clearly shows that the two Li gravitate around 6 the C6 ring center with η -coordination. From Figure 19b, LVD significantly bends toward the graphite surface perhaps due to the strong interactions + – + between Li and the arene π, denoted as O •••Li •••π (arene). Table 5 TotalA dsorption Energies (TAE,kcal/m ol)and BSSE-corrected A dsorption Energies (AE, kcal/m ol) of Lithium A lkylene D icarbonates on N eutraland Charged G raphite Surfaces calculated w ith the PW 91PW 91 m ethod. Conf.

Individual

3-21G a

a

7.1 7.2 7.3 7.4 7.5

LVD+Gr54 LVD+Gr54 LED+Gr54 LED+Gr54 LPD+Gr54

7.6 7.7 7.8 7.9

LVD+Gr54– LED+Gr54– LED+Gr54– LPD+Gr54–

7.10 7.11 7.12

LVD+Gr78– LED+Gr78– LPD+Gr78–

4-31G*

6-31G*

TAE

AE

TAE

AE

TAE

AE

–38.84 –40.34 –41.70 –40.13 –40.97

–7.75 –10.36 –12.74 –9.19 –14.89

–23.27 –21.71 –25.29 –23.68 –25.75

–12.33 –11.30 –14.76 –12.73 –15.35

–21.19

–12.12

–23.46 –21.47 –23.44

–14.93 –12.55 –15.05

–34.67 –37.40 –35.17 –38.05 –40.92 –41.81 –44.57

TAE= E(conformations)-E(lithium alkylene dicarbonate)-E(graphite); AE=TAE+BSSE

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



6CDNG 6JGFKUVCPEG —HTQONKVJKWOECTDQP[NECTDQP %CPF%CPFCNM[NECTDQP %CPF %QHNKVJKWOCNM[NFKECTDQPCVGUVQVJGPGCTGUVITCRJKVGECTDQP  %QPH























+PF

.8&

.'&

.'&

.2&

.8&

.'&

.'&

.2&

.8&

.'&

.2&

)T

)T

)T

)T

)TŌ

)TŌ

)TŌ

)TŌ

)TŌ

)TŌ

)TŌ

.K





















%















%















%















%















.K



























2929) 

2929)  .K

















%

















%

















%

















%

















.K





















2929)  .K









%









%









%









%









.K













(KIWTG D CNUQ UJQYU VJCV VJG FKUVCPEGU HTQO .8& CVQOU VQ VJG PGCTGUV ITCRJKVG % CVQOU RTGFKEVGF D[ 2929)  CTG QPN[ UNKIJVN[ NQPIGT D[ NGUU VJCP — VJCP VJQUG QDVCKPGF WUKPI 2929)  %QPUGSWGPVN[VJGCFUQTRVKQPGPGTIKGUQPN[FKHHGTD[MECNOQN ŌXU ŌMECNOQN  *QYGXGT 2929) UKIPKHKECPVN[ WPFGTGUVKOCVGU VJGUG FKUVCPEGU D[ — 6JG FGHKEKGPE[ QH VJG UOCNNGUV DCUKU UGV ) KU CNUQ TGHNGEVGF D[ VJG CFUQTRVKQP GPGTIKGU YJKEJ CTG EQPUKFGTCDN[ QXGTGUVKOCVGFCVVJG)DCUKUUGVYJGTGVJG$55'KUGXGPVYQVKOGUJKIJGT VJCP VJG CFUQTRVKQP GPGTI[ KVUGNH 5VCTVKPI HTQO C IGQOGVT[ RGTRGPFKEWNCT VQ VJG ITCRJKVG UWTHCEG CPQVJGT UVCVKQPCT[ RQKPV EQPHQTOCVKQP  KU NQECVGF

(KIWTG E KP YJKEJ VJG .8& OQNGEWNG DCUKECNN[ MGGRU KVU RNCPCT UVTWEVWTG YJGTGCUVJGFKJGFTCNDGVYGGPVJG.8&OQNGEWNCTRNCPGCPFVJGITCRJKVGUWTHCEG

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG









































 



" !



 

" !

 



 





 

































 





 



 



+







& "

#

(

, *

)

/ 4

7 5

3











 

 







.

-

0

1

2

-

.



$

   





 





 6







 



















 





 '

& "

#

( )

*

 

 







$



 









 



 







 





 #



& "

%



$





:

<

;

 8



 

9




8









 9



=

> ?

= @

> ?

@


















 8





 9

 8



 9

I

L Q

T R

P


D A

 8





9

 8







C

E

@

F

G

J

K

J

M

N

O

J

K

9










B

S














 













D





C




H

@

B



 (KIWTG#FUQTRVKQPUQH.8&QPVJGDCUCNRNCPGQH%*5QNKFDCNNUUVCPFHQTQZ[IGPCVQOU DKIGORV[QPGUHQTECTDQPCVQOUUOCNNGORV[QPGUHQTJ[FTQIGPCVQOUCPFVJGDCNNUOCTMGFYKVJ C UVCT UVCPF HQT NKVJKWO CVQOU 6JG PWODGTU TGHGT VQ VJG FKUVCPEGU DGVYGGP .8& CVQOU CPF VJG PGCTGUV ITCRJKVG ECTDQPU VJQUG KP DTCEMGVU KP RCTGPVJGUGU CPF QWVUKFG CTG HTQO 2929 )  2929)  CPF 2929) TGURGEVKXGN[ C 6QR XKGY QH   D UKFG XKGYQH EUKFGXKGYQH

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



KU SWKVG NGUU VJCP ° 6JG CFUQTRVKQP GPGTI[ QH  KU  MECNOQN NQYGT PGICVKXG VJCP VJCV QH  YJKEJ OQUV RTQDCDN[ KU CVVTKDWVGF VQ VJG UVTQPIGT KPVGTCEVKQPDGVYGGPVJGπ FQWDNGDQPFQH.8&CPFVJGπ CTGPGQHITCRJKVGKP VJGNCVVGTVJCPVJCVKPVJGHQTOGT 6YQ EQPHQTOCVKQPU  CPF  KP (KIWTGU C CPF D EQTTGURQPFKPI VQ VJGRGTRGPFKEWNCTCPFRCTCNNGNCFUQTRVKQPUCTGNQECVGFCUYGNNHQT.'&JQYGXGT VJGRGTRGPFKEWNCTCFUQTRVKQPEQPHQTOCVKQPKUOQTGUVCDNGD[CRRTQZKOCVGN[  MECNOQN VJCP VJG RCTCNNGN EQPHQTOCVKQP  #' Ō XU Ō CPF Ō XU Ō MECNOQN CV 2929)  CPF )  TGURGEVKXGN[ 6JKUOC[DGFWG VQ NGUU TGRWNUKQP DGVYGGP VJG %* ITQWR CPF ITCRJKVG KP  VJCP VJCV KP  %QORCTKPI VJG RCTCNNGN CFUQTRVKQP EQPHQTOCVKQPU  CPF CNVJQWIJVJGCXGTCIGFFKUVCPEGUHTQO.K VQVJGPGCTGUVPGKIJDQTITCRJKVG% CVQOUCTGCTQWPF—CV) HQTDQVJECUGUVJGUJQTVGUVFKUVCPEGUHTQOVJG ECTDQP[N CU YGNN CU GVJ[NGPG ECTDQPU VQ VJG ITCRJKVG ECTDQPU CTG NQPIGT D[ CRRTQZKOCVGN[CPF—VJCPVJGEQTTGURQPFKPIQPGUQH.8& 4XU —  XU — YJKEJ OGCPU VJCV VJG CFUQTRVKQP QH .'& KU PQV CU EQORCEV CU VJCV QH .8& 5KOKNCT EQPENWUKQP EQWNF CNUQ DG TGCEJGF D[ EQORCTKUQP DGVYGGP VJG RGTRGPFKEWNCT EQPHQTOCVKQPU  CPF  4  —XU—CV2929)  6JGEQORCTKUQPDGVYGGPVJGHCXQTCDNGEQPHQTOCVKQPUKGQH.8%CPF QH.'&UJQYUVJCVUWEJFKHHGTGPEGUDGEQOGEQPUKFGTCDN[NCTIGT  —XU— 1PN[VJGEQPHQTOCVKQPQHRGTRGPFKEWNCTCFUQTRVKQP KP(KIWTGEYCU HQWPFHQT.2&CVVJGVYQJKIJGTNGXGNU+PVJGEQPHQTOCVKQPVJGFKUVCPEGU HTQO VJG ECTDQP[N ECTDQPU VQ VJG PGCTGUV ITCRJKVG ECTDQP CTG NQPIGT D[   CPF  — VJCP VJQUG KP VJG EQPHQTOCVKQP  4   — XU  —CV) VJGFKUVCPEGU4HQTCNM[NECTDQPUCTGCNUQHWTVJGTKPETGCUGF D[CRRTQZKOCVGN[CPF— —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

CUUJQYPKP 6CDNG  VQ VJG ITCRJKVG UWTHCEG YJGTGCU QPG GPF VWTPU WRYCTFU 

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG







" &

!

#

$

) '

%





!

(





























 

 

 



























 






 









 
























 







  

 




















  












  
















 



 

   




















 











 







 ,



" &

#

$

'

%

!



!

(

 8

 8











 9

9






 

 

 






 *

*











+

+




 8





 9

 





 
















 









8





9

 *


8









 +

9  

 8









 



  9









*





+

*







 +





 







*







 +










































" &

' %

.

!

#

$



/

!

(





8



9




 *





+












 

 

 






  



8









 



9



 8

9






8

 8

 8



 



 

9

9

9


 *





 +




 















*

 *











 +


+ 

 *

 *







+

+













 
























































 (KIWTG5KFGXKGYUQHCFUQTRVKQPUQH.'& CDCPF.2& EQPVJGDCUCNRNCPGQH%*6JG PWODGTUJCXGVJGUCOGOGCPKPIUCUKP(KIWTG

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



HTQOVJGUWTHCEGKPVJGECUGUQH.'&CPF.2&GURGEKCNN[VJG NCVVGT QPG  6JG CFUQTRVKQPGPGTIKGUQH.'&CPF.2&CTGXGT[UKOKNCT ŌXUŌ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

 

 #FUQTRVKQP QH .KVJKWO #NM[N &KECTDQPCVGU QP VJG $CUCN 2NCPG QH 0GICVKXGN[%JCTIGF)TŌCPF'FIG2NCPGQH)TŌ



#FUQTRVKQPQH.8&.'&CPF.2&QPVJGDCUCNRNCPGQHVJGPGICVKXGN[EJCTIGF ITCRJKVGUWTHCEG )TŌ EQPHQTOCVKQPUCPFYJKEJEQTTGURQPF VQ EQPHQTOCVKQPU QH    CPF  QH VJGKT CFUQTRVKQP QP VJG PGWVTCN UWTHCEG )T CTG CNUQ KPXGUVKICVGF KP QTFGT VQ KPEQTRQTCVG CFUQTRVKQP QP C PGICVKXGN[ EJCTIGF CPQFG DCUCN RNCPG 5KOKNCTN[ VJGKT CFUQTRVKQP GPGTIKGU FQ PQV FKHHGT OWEJ GKVJGT 6#' .8& Ō .'& Ō CPF Ō .2& Ō  5KOKNCTN[ VQ VJG ECUG QH VJGKT CFUQTRVKQP QP VJG PGWVTCN DCUCN RNCPG )T6CDNGUJQYUVJCVDQVJVJGECTDQP[NECTDQPCPFVJGCNM[NECTDQPCVQOUQH .8& EQPHQTOCVKQP  JCXG UKIPKHKECPVN[ UJQTVGT FKUVCPEGU VQ VJG PGCTGUV ITCRJKVGECTDQPUVJCPVJQUGQH.'& EQPHQTOCVKQPCPFJCXG %%% CPF%EQPHQTOCVKQPCPF—EQPHQTOCVKQP    CPF — EQPHQTOCVKQP     CPF —+PVWTPEQPHQTOCVKQPQH.2&CFUQTRVKQPJCUGXGPNQPIGTFKUVCPEG VJCPVJCVQHEQPHQTOCVKQPQH.'&CFUQTRVKQP6JGCXGTCIGFFKUVCPEGVQVJG PGCTGUV ITCRJKVG ECTDQP KP EQPHQTOCVKQP  KU UJQTVGT D[  —  — CPF —VJCPVJQUGKPEQPHQTOCVKQPCPFTGURGEVKXGN[ 6QR XKGYU HQT CFUQTRVKQP QH VJG VJTGG NKVJKWO CNM[N FKECTDQPCVGU QP VJG GFIGRNCPGQHVJGPGICVKXGN[EJCTIGFITCRJKVG)TŌCTGKNNWUVTCVGFKP(KIWTG +V KU YGNN TGEQIPK\GF VJCV VJG TGCEVKXKV[ QH \KI\CI CPF CTOEJCKT ECTDQPU KP VJG GFIGU KU OWEJ FKHHGTGPV HTQO VJCV QH VJG ECTDQPU KP VJG ITCRJKVG DCUCN RNCPGYJKEJKUTGHNGEVGFD[VJG TCVJGT FKHHGTGPV EJCTIG FKUVTKDWVKQP QH %*Ō 6JKU OC[ CHHGEV VJG TGFWEVKXG FGEQORQUKVKQP QH QTICPKE ECTDQPCVGU CPF VJG 

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

 



 



 





 












 CEQPHQTOCVKQP












































































 DEQPHQTOCVKQP

 EEQPHQTOCVKQP  (KIWTG6QRXKGYUQHCFUQTRVKQPUQPGFIGRNCPGUQHVJGPGICVKXGN[EJCTIGF%*QH.8& C EQPHQTOCVKQP.'& DEQPHQTOCVKQPCPF.2&EQPHQTOCVKQP 

 





























































6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



EQPUGSWGPV CFJGUKQP QH VJG FGEQORQUKVKQP RTQFWEVU 6JG VQVCN CFUQTRVKQP GPGTIKGU 6#'QH.8&.'&CPF.2&QPVJGGFIGQH)TŌCNUQCITGGYKVJKP MECNOQN 6#'.8&Ō.'&Ō.2&ŌMECNOQN+PURKVG QHUOCNNGTFKHHGTGPEGUVJGCXGTCIGFFKUVCPEGVQVJGPGCTGUVITCRJKVGECTDQPJCU VJGUCOGUGSWGPEGCUHQTVJGDCUCNRNCPGQH)TŌGIVJGCXGTCIGFFKUVCPEGUKP EQPHQTOCVKQP  .8& )TŌ EQPHQTOCVKQP  .'& )TŌ CPF EQPHQTOCVKQP .2& )TŌ CTG CPF —TGURGEVKXGN[6JWU KV OC[ DG UCHG VQ EQPENWFG VJCV .8& VJG OCKP RTQFWEV HTQO 8% TGFWEVKQP HQTOUVJGOQUVEQORCEVNC[GTQPVJGDCUCNRNCPGCUYGNNCUQPVJGGFIGRNCPGQH VJGEJCTIGFITCRJKVGCPQFGHQNNQYGFD[.'&VJGRTQFWEVQH'%CPFVJGNGCUV HCXQTCDNGKUVJG.2&RTQFWEVHTQO2%TGFWEVKQP    6YQ&KOGPUKQPCN#UUQEKCVKQPQH.KVJKWO#NM[N&KECTDQPCVGUQPVJG$CUCN 2NCPGQHVJG)TCRJKVG5WTHCEG )T  .KVJKWO CNM[N FKECTDQPCVGU ECP UVTQPIN[ CUUQEKCVG VJTQWIJ 1Ō•••.K •••1Ō KPVGTCEVKQPU CPF ECP ITQY CFQRVKPI VJG ENQUGF UVTWEVWTG EQPHQTOCVKQP VJCV KU VJG[ CUUQEKCVG VQ VJG ECIGNKMG FKOGT HKTUV VJGP VQ VJG ENQUGF RUGWFQ RNCPCT VTKOGT VQ VJG ENQUGF VGVTCOGT CPF UQ QP #UUQEKCVKQPU QH NKVJKWO CNM[NFKECTDQPCVGUTGFWEGVJGKPVGTCEVKQPUQH.K YKVJVJGENWUVGTUWTHCEGYJKEJ KU GXKFGPEGF D[ VJG NQPIGT FKUVCPEGU DGVYGGP .K  CPF KVU PGCTGUV PGKIJDQT ITCRJKVG ECTDQPU 4 KU CRRTQZKOCVGN[ — NCTIGT VJCP VJG OQPQOGT %QPUGSWGPVN[VTKOGTUOC[GZKUVKPRUGWFQRNCPCTIGQOGVTKGUCDQXGVJGITCRJKVG UWTHCEGCUUJQYPKP(KIWTGTCVJGTVJCPKPCPCTEJGFIGQOGVT[NKMGCUKPING OQNGEWNG /QTGQXGT CFUQTRVKQP GPGTIKGU CTG QPN[ MECNOQN JKIJGT PGICVKXG VJCP VJQUG QH VJG OQPQOGTU 6#' Ō Ō Ō MECNOQN XU ŌŌŌMECNOQNCVVJG2929)NGXGN 5KOKNCTN[ VQ UKPING OQNGEWNCT CFUQTRVKQPU VJG VTKOGT QH .8& DCUKECNN[ KU RCTCNNGN VQ VJG ITCRJKVG UWTHCEG (KIWTG C YJKNG QPG QH CNM[N GVJ[NGPG CPF RTQR[NGPG ECTDQPU VWTPU WRYCTF HTQO ITCRJKVG UWTHCEGU (KIWTGU D CPF E 6JG FKUVCPEGU HTQO ECTDQP[N CPF XKP[NGPG ECTDQPU QH NKVJKWO XKP[NGPG FKECTDQPCVG VQ VJG PGCTGUV ITCRJKVG ECTDQPU CTG CICKP UJQTVGT VJCP VJG EQTTGURQPFKPIFKUVCPEGUQHNKVJKWOGVJ[NGPGFKECTDQPCVGCPFNKVJKWORTQR[NGPG FKECTDQPCVG 6JKU TGUWNV CNUQ KPFKECVGU VJCV VJG OQPQNC[GT HTQO .8& KU OQTG EQORCEVVJCPVJQUGHTQOVJGQVJGTVYQUCVWTCVGFFKECTDQPCVGUYJKEJOC[JCXG CRQUKVKXGKORCEVQPVJG5'+NC[GTDWKNFWR 

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

C

D 





(c) Figure 22 Side views of adsorptions on the basal plane of C96H24: (a) lithium vinylene dicarbonate trimer, (b) lithium ethylene dicarbonate trimer, (c) lithium propylene dicarbonate trimer.

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



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

 

 5WOOCT[ #DQWV #FUQTRVKQPU QH .8& .'& CPF .2& QP )TCRJKVG #PQFG5WTHCEGU  6JGCFUQTRVKQPGPGTIKGUQH.8&.'&CPF.2&QPVJGDCUCNRNCPGQHPGWVTCNCU YGNNCUVJGPGICVKXGN[EJCTIGF%*CPFQPVJGGFIGRNCPGQHVJGPGICVKXGN[ EJCTIGF%*CITGGYKVJKPMECNOQNGIVJG[CTGŌŌŌ CPF Ō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

272

Lithium-Ion Batteries: Solid-Electrolyte Interphase

also could be responsible for the graphite destruction in PC solutions, such as 41 propylene gas arising from solvent reduction, are still under investigation.

8

Remarks on the Failure of PC and the Efficiency of VC for the SEI Layer Formation in EC/PC-Based Solutions

This chapter provides a theoretical avenue for exploring the failure mechanism of PC in LIBs employing graphite as anode, and the functioning mechanism of VC in EC/PC-based solutions, by extensively investigating their reductive decompositions, the possible appearance of the reduction products, and their adhesion to the basal plane as well as to the edge plane of the carbon anode. Because of the complexity of these phenomena as well as the limitations from the available theoretical models, the theoretically-derived understanding of this problem is not unambiguous, however it does provide a great deal of interesting information, which will be summarized as follows. We believe that the interplay between experimental and theoretical investigations will advance the insights into the formation and growth of the SEI layer. Not only has PC a very similar structure to EC, it also has similar quantum chemical properties like HOMO and LUMO energies, which control their oxidation and reduction behavior. The reduction potential (RP), for example, of PC is only slightly (<0.1 V) lower than that of EC. Additionally, the thermodynamics as well as the dynamics difference for the reductive decomposition of EC and PC are not enough to account for their opposite performance towards the graphite anode of LIBs, although certain difference does exist. However, it is interesting to note that the appearance of LPD, one of the leading products from PC reduction, is much different from the reduction product (LED) of EC, e.g., the quasi-planar trimer structure of LPD twists more considerably than that of LED, and its methyl group, may partially block the 3-D growth of LPD and prevent it from building up a thick enough SEI layer. Another negative factor is that the adsorption of LPD quasi-planar trimer on the basal plane deforms the graphite structure more considerably than that of the LED trimer. Moreover, the adsorption of LPD on the basal plane as well as on the edge plane is farther from the surface than that of LED, that means that LPD forms a looser mono-layer. Propylene gas resulting from PC reduction probably is also one of the negative factors that act against an efficient PC performance in 41 LIBs.

6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



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∆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

 

4GHGTGPEGU   #TOUVTQPI#4CPF$TWEG2)0CVWTG   2GNGF','NGEVTQEJGO5QE   $GUGPJCTF,1CPF(TKV\*2,'NGEVTQEJGO5QE   $GUGPJCTF,19KPVGT/;CPI,CPF$KDGTCEJGT9,2QYGT5QWTEGU    9KPVGT/CPF$GUGPJCTF,1+P+P.KVJKWO+QP$CVVGTKGU(WPFCOGPVCNU CPF 2GTHQTOCPEGU GF D[ 9CMKJCTC / CPF ;COCOQVQ 1 9KNG[8%* 0GY;QTMR

  

.KVJKWO+QP$CVVGTKGU5QNKF'NGEVTQN[VG+PVGTRJCUG

 ,GQPI 5- +PCDC / #DG 6 CPF 1IWOK < , 'NGEVTQEJGO 5QE 

#  2GNGF ' .KVJKWO UVCDKNKV[ CPF HKNO HQTOCVKQP KP QTICPKE CPF KPQTICPKE GNGEVTQN[VGHQTNKVJKWODCVVGT[U[UVGOU+PNKVJKWODCVVGTKGUGFD[)CDCPQ ,2 #ECFGOKE0GY;QTM  2GNGF')QNQFPKVUM[&/GPCEJGO%CPF$CT6QY& ,'NGEVTQEJGO 5QE   #WTDCEJ&.GXK/&.GXK'CPF5EJGEJVGT#,2J[U%JGO$

  %JWPI)%-KO*,;W5+,WP5*%JQK,9CPF-KO/*, 'NGEVTQEJGO5QE   5JW<:/E/KNNCP45/WTTC[,,CPF&CXKFUQP+,,'NGEVTQEJGO 5QE .  ,GJQWNGV%$KGPUCP2$QFGV,/$TQWUUGN[/CPF6GUUKGT.GUEQWTTGV % $CVVGTKGU HQT 2QTVCDNG #RRNKECVKQPU CPF 'NGEVTKE 8GJKENGU 6JG 'NGEVTQEJGOKECN5QEKGV[2TQEGGFKPIU5GTKGU2GPPKPIVQP0,  9TQFPKII)*$GUGPJCTF,1CPF9KPVGT/,'NGEVTQEJGO5QE

  #WTDCEJ & )COQNUM[ - /CTMQXUM[ $ )QHGT ; 5EJOKFV / CPF *GKFGT7'NGEVTQEJKO#EVC   9CPI;:0CMCOWTC56CUCMK-CPF$CNDWGPC2$,#OGT%JGO 5QE   9CPI;:CPF$CNDWGPC2$WPRWDNKUJGFTGUWNVU  /KGTVWU55ETQEEQ'CPF6QOCUK,%JGO2J[U   /KGTVWU5CPF6QOCUK,%JGO2J[U   %QUUK / $CTQPG 8 %COOK 4 CPF 6QOCUK , %JGO 2J[U .GVV 

  $CTQPG8CPF%QUUK/,2J[U%JGO#   (QTGUOCP,$-GKVJ6#9KDGTI-$5PQQPKCP,CPF(TKUEJ/,, 2J[U%JGO   $GEMG#&,%JGO2J[U   2GTFGY,27PKHKGFVJGQT[QHGZEJCPIGCPFEQTTGNCVKQPDG[QPFVJGNQECN FGPUKV[CRRTQZKOCVKQP+P'NGEVTQPKE5VTWEVWTGQH5QNKFUGFD[
6JGQTGVKECN+PUKIJVUKPVQVJG5'+%QORQUKVKQPCPF(QTOCVKQP/GEJCPKUO



 $TGPGOCP%/CPF9KDGTI-$,%QOR%JGO   4GGF#'%WTVKUUU.#CPF9GKPJQNF(%JGO4GX   (TKUEJ / , 6TWEMU ) 9 5EJNGIGN * $ 5EWUGTKC ) ' 4QDD / # %JGGUGOCP , 4
CHAPTER 6

CONTINUUM AND STATISTICAL MECHANICS-BASED MODELS FOR SOLID-ELECTROLYTE INTERPHASES IN LITHIUM-ION BATTERIES *

HARRY J. PLOEHN, PREMANAND RAMADASS, RALPH E. WHITE, † DIEGO ALTOMARE and PERLA B. BALBUENA Department of Chemical Engineering, Swearingen Engineering Center, University of South Carolina, 301 Main, Columbia, South Carolina 29208, USA E-mail: *[email protected]; †balbuena@ engr.sc.edu

1

Introduction

Molecular theory and simulation build on a foundation of atomic and molecularlevel postulates (e.g. orbital models with varying degrees of complexity and approximations) to describe phenomena at molecular length scales. Descriptions on longer scales of space and time are possible but become limited by computational throughput at scales far less than the macroscopic. Extrapolation from molecular to macroscopic behaviour can yield great insight, albeit qualitative. At the other extreme are “macroscopic” continuum models. Built on the foundations of continuum mechanics, continuum models describe the behaviour of systems on macroscopic scales of space and time — the scales on which we use materials and devices such as lithium-ion batteries. However, the predictive value of any continuum model depends in no small part on the veracity of its constitutive equations and the empirical transport coefficients by which they are parameterized. To bridge the gap between molecular and macroscopic scales, models based on statistical mechanics can be formulated in terms of molecular parameters obtained from rigorous ab initio calculations (in principle) or from experiments as empirical parameters (the usual practice). Using molecular parameters as input, statistical mechanical models yield (1) estimates of constitutive transport coefficients required for continuum models, and (2) important physical insight necessary for formulating realistic, useful continuum models. 276

Continuum and Statistical Mechanics-Based Models for SEI

277

In this chapter we provide illustrations of both continuum and statistical mechanics-based models and discuss the insights obtained through comparisons with experimental data related to SEI layer phenomena in lithium-ion batteries.

2

Continuum Models for SEI Growth

Much work has been done on the lithium SEI problem using molecular simulations because many key issues lie at the molecular level. The apparent physical and chemical complexity of the SEI can be broken down into discrete molecular-level problems amenable to molecular simulation. However, recombining and extrapolating this knowledge to macroscopic scales is problematic. In this case, the physical and chemical complexity of the SEI presents a formidable barrier to the development of continuum models for SEI growth. Nevertheless, the synergistic combination of rigorous continuum models with realistic transport properties from molecular simulations is critical to understanding the role of the SEI in Li-ion cell performance.

2.1

Overview of Previous Macroscopic Models

Other chapters in this volume review the considerable experimental effort that has gone into characterizing SEI formation and its effect on Li-ion cell performance. Much less work has been done in the area of theoretical 1 continuum modelling. A large body of work by Peled and coworkers, known collectively as “the SEI model”, provides both a hypothetical description of SEI morphology and a phenomenological equivalent circuit model for rationalizing data from electrochemical impedance spectroscopy experiments. This approach recognizes the likely physical complexity of the SEI but cannot generate a priori predictions of dynamic behavior or SEI morphology. More sophisticated models formulate ad hoc diffusive transport equations in order to describe the steady-state distributions of potential and charge carriers within the SEI and, ultimately, the SEI current-polarization characteristics. Recent approaches have 2, 3 included the use of the point defect model assuming the diffusion of charge carriers through a defective crystalline lattice, as well as a stochastic transport 4 model postulating a distribution of charge carrier jump distances and energy barrier heights within a structurally disordered SEI. Although these models successfully rationalize current-polarization and impedance data, none can treat

278

Lithium-Ion Batteries: Solid-Electrolyte Interphase

SEI growth dynamics and the related problem of long-term capacity fade in Liion cells. 5 A series of papers by Badiali, Nainville, and coworkers present a more detailed, lattice-based numerical simulation of SEI formation and growth. The simulation incorporates a variety of elementary processes including Li metal + oxidation at the Li/SEI interface, transport of electrons and Li through the SEI, reduction of solvent at the SEI/electrolyte interface, precipitation of reduction products, and the local blocking of layer growth due to the diffusion and reaction of additives in the electrolyte. Moving phase boundaries are an inherent part of the simulation. Under certain conditions the model predicts, in accord with Peled’s SEI model, a “duplex” SEI structure consisting of a thin, compact inner layer and a thick, porous outer layer. The lattice simulation gives a useful, qualitative picture of the influence of various parameters on the dynamics of SEI growth as well as the evolution of SEI morphology. 6 The SAFT group published a dynamic model for SEI growth in Li-ion cells employing graphite electrodes and organic electrolytes. They measured the capacity loss of a variety of Li-ion cells as a function of temperature at an applied voltage held constant by maintaining a small trickle current (“float potential”). Based on their experimental observations, they conclude that electrolyte reduction on the negative graphite electrode is the most important contributor to capacity loss under float potential storage conditions. They also attribute the decreasing rate of capacity loss to the production and deposition of an SEI that reduces the electrolyte reduction rate. In order to rationalize these observations, the SAFT group proposed an aging mechanism for Li-ion cells based on SEI growth on the carbon anode limited by the SEI electronic conductivity. Specifically, their model postulates that the rate of lithium loss (in terms of moles of lithium lost, ` L ) is proportional to SEI electronic conductance (X): d`L kχS B = kX = = dt L L

(1)

where k is a proportionality constant, S is the anode area, and χ is the SEI specific conductivity, dependent only on temperature (for consistency, we use our own notation instead of that in Ref. 6). The last equality combines the latter three quantities into an empirical parameter B that is assumed to be independent of time. The SEI thickness, L, varies with time and can be expressed as

Continuum and Statistical Mechanics-Based Models for SEI

L = L0 + A` L

279

(2)

where L0 is the initial SEI thickness after the first few charge-discharge cycles, and A is another empirical parameter independent of both time and temperature. This expression assumes that lithium, electrons, and electrolyte react to produce an insoluble product I with constant composition and average molar volume. Combining Eqs. (1) and (2), integrating subject to the initial condition of n=0 at time t=0, and rearrangement yield t=

A 2 L0 `L + `L 2B B

(3)

as an implicit expression relating the number of moles of lithium lost to the time the cell is held at the float potential. The SAFT group also assumed, implicitly, that the cell capacity is proportional to the available number of moles of lithium, which we denote as ` 0 − ` L with ` 0 as the initial number of moles of lithium available for cycling. Then the fractional capacity loss can be expressed as x = ` L ` 0 and Eq. (3) becomes t=

A′ 2 L0 x + x 2 B′ B′

(4)

with A′ ≡ A` 0 and B ′ ≡ B ` 0 . Empirical parameters defined in this way 6 differ from those in but do not change the essence of their findings. In effect, the ad hoc model of the SAFT group shows that the capacity loss, lithium loss, and SEI thickness all increase (roughly speaking) with the square-root of time. With increasing SEI thickness, the increasing path length for electronic conduction leads to decreasing rates of lithium consumption, generation of insoluble product, and SEI growth. Experimental data for timedependent capacity loss of a variety of Li-ion cells can be fit by Eq. (4) with good accuracy. Of the three unknown quantities (A, B, L0) grouped into two model parameters, only B presumably depends on temperature. From capacity loss data for cells stored at various temperatures, B was shown to have an Arrhenius dependence on temperature, consistent with what one might expect for electronic conduction through the SEI via a diffusive, site-hopping mechanism. The success of the SAFT model demonstrates that macroscopic models of SEI growth can be valuable for rationalizing experimental observations,

280

Lithium-Ion Batteries: Solid-Electrolyte Interphase

developing a better understanding of underlying mechanisms, and predicting the long-term behaviour of Li-ion cells under certain conditions. This motivates our efforts to develop more rigorous continuum models that may be able to describe Li-ion cell performance under a wide variety of conditions, perhaps including first-cycle irreversible Li loss, potentiostatic charge-discharge, and galvanostatic charge-discharge. We first review the key elements of continuum mechanics in the next section, and then outline the development of specialized models for dynamic growth of the SEI on carbon anodes in Li-ion cells under float potential conditions.

2.2

Elements of Continuum Mechanics

As the name implies, continuum mechanics is predicated on the hypothesis that one may describe the properties and behaviour of physical systems entirely in terms of continuous functions of position and time, at least for a single pure component within a single bulk phase (gas, liquid, or solid). Continuum mechanics makes no reference to the fact that real materials are composed of atoms or molecules. Strictly speaking, rationalization of transport coefficients (like viscosity, thermal conductivity, and diffusivity) in terms of molecular behaviour lies within the realm of statistical mechanics and molecular simulations. Continuum mechanics begins to lose its validity when the characteristic length and time scales in a physical system become comparable to molecular scales. Continuum mechanics thus cannot describe the initial stages of SEI film formation composed of many simultaneous, discrete, discontinuous molecular events. Bearing in mind the limitations imposed by the continuum hypothesis, we may construct, parameterize, and test continuum models of SEI growth and mass transport. The validity of any such continuum model lies wholly in its ability to faithfully describe the behaviour of existing physical systems and to accurately predict the performance of systems yet to be. Three essential elements provide the foundation for continuum mechanics. First, we must have a kinematical framework for mathematically describing the motion of material “particles” — not molecules, but differential portions of a physical entity, or body. Second, conservation principles for mass, charge, linear and angular momentum, and energy serve as fundamental, universal postulates. Various forms of the transport theorem enable translation of conservation principles between recognizable forms for closed and open

Continuum and Statistical Mechanics-Based Models for SEI

281

systems. Finally, we require constitutive equations describing material behaviour — in effect, how material particles interact. By introducing phenomenological transport coefficients, constitutive equations provide a formal connection between macroscopic properties and the underlying molecular reality. In practice, constitutive equations provide the most obvious route by which molecular simulations may inform and guide continuum models. Continuum mechanics also considers multi-component systems without loss of generality. For a single phase containing N chemical species, the overall material body is treated as a superposition of N single-component continua, each with its own variable mass density field. The kinematical framework can be readily generalized to describe the relative motions of the chemically distinct material particles (differential portions, not molecules) in the component bodies. A generalized mass conservation postulate recognizes that chemical reactions may generate and consume various chemical species. This, in turn, necessitates the introduction of new constitutive equations expressing the reaction rates in terms of other variables. These subjects will be familiar to those who have 7 studied mass transfer and related transport phenomena. Systems involving multiple phases can also be treated with great generality in continuum mechanics, although most readers will be less familiar 8 with the required formalism. Complications arise due to the fact that, by definition, field variables (such as mass density) are discontinuous at phase interfaces. The kinematical framework must be extended to account for this discontinuity as well as the motion of phase interfaces. Conservation postulates are presumed to be universally valid, even at phase boundaries. Generalization leads to new, unfamiliar “jump” balances – the rigorous foundation for many types of boundary conditions that are often used (and misused) without due consideration of their genesis. Other interfacial boundary conditions, such as the “no-slip” condition in fluid mechanics, are essentially constitutive in nature. Here, too, molecular simulations are supplanting empiricism to produce increasingly realistic continuum models. Interfacial effects dominate the growth of SEI layers on lithium-based anodes. Chemical reactions of lithium and electrolyte may be localized at phase interfaces, and one or more interfaces may move as a result. The following overview of the key elements of continuum mechanics places special emphasis on rigorous description of kinematics, conservation principles, and constitutive equations associated with phase interfaces. Complete details may be found in 8 the text by Slattery.

282

Lithium-Ion Batteries: Solid-Electrolyte Interphase

2.2.1 Kinematics In continuum mechanics, a body B denotes a specified, closed set of material particles P (Figure 1). In order to identify specific material particles, imagine taking a snapshot of the body at a particular time (taken to be t=0 for convenience). This image defines the reference configuration of the body. By referring to the snapshot, we can uniquely identify each material particle by its vector spatial position R in the reference configuration. At any later time t>0, each material particle may have moved to a new position r(t). In general, the motion of a specific material particle is parameterized by the path r(t; R) where R formally identifies the material particle.

z x

B

y r(t) R P

P

B

t=t1>0

t=0 Figure 1 A material body B in a reference configuration at time t=0 (left), and in an arbitrary configuration at some later time t (right).

The velocity v of a specific material particle, known as the material derivative, can be expressed as the time derivative of position, holding constant the identity of the material particle: Dr  ∂r  v≡  =  ∂t R Dt

(5)

Continuum and Statistical Mechanics-Based Models for SEI

283

The second expression shows the usual notation for a material derivative. For an arbitrary scalar- or vector-valued field variable Q(r, t), the material derivative can be related to partial derivatives with respect to time and position via DQ  ∂Q  = + v ⋅ ∇Q Dt  ∂t r

(6)

Clearly the rate of change of Q associated with a specific material particle may be decomposed into the rate of change of Q at a fixed spatial position, plus the rate of change of Q attributable to the motion of the material particle. Certain quantities are expressed in terms of volume integrals of field variables over the spatial domain of the body (VB). Time derivatives of these quantities are problematic because the spatial domain of the body may itself vary with time (e.g., the body in Figure 1). Through a suitable variable transformation, we may map the domain into that of the reference configuration, interchange time differentiation and volume integration (since the reference configuration does not change), and then transform back into the original domain. The result for an arbitrary field variable Q, d dt

 DQ

∫∫∫ QdV = ∫∫∫  Dt VB

VB

 + Q∇ ⋅ v  dV 

(7)

is known as one form of the transport theorem. Equation (7) can only be used for single bulk phases because the derivatives are undefined at phase boundaries. Using Eq. (6) and the divergence theorem, Eq. (7) becomes d dt

∂Q

∫∫∫ QdV = ∫∫∫ ∂t dV + ∫∫ Q ( v ⋅ n ) dS VB

VB

(8)

SB

where SB is the surface of the body and n is the outward-pointing unit normal vector. The transport theorem may be generalized for bodies containing multiple phases and associated phase interfaces. Consider the two-phase body shown in Figure 2. The phase interface (Σ) moves with a velocity u that varies with position on the surface Σ. Likewise, the unit normal vector ξ describes the

284

Lithium-Ion Batteries: Solid-Electrolyte Interphase

ξB

Σ

u

A

ξA

B

Figure 2 A material body consisting of two phases (A and B) separated by a phase interface Σ.

orientation of the phase interface. We may write an expression of the transport theorem, Eq. (8), for each bulk phase (A and B). Adding these, we find d dt

∂Q

∫∫∫ QdV = ∫∫∫ ∂t dV + ∫∫ Q ( v ⋅ n ) dS − ∫∫[Q (u ⋅ VB

VB

)]dS

(9)

Σ

SB

Application of the divergence theorem (valid within each phase) leads to d dt

 DQ

∫∫∫ QdV = ∫∫∫  Dt VB

VB

 + Q∇ ⋅ v  dV + 

∫∫[Q ( v − u ) ⋅

]dS

(10)

Σ

analogous to Eq. (7). In Eqs. (9) and (10), the boldface brackets are special notation representing the “jump” of the enclosed quantity at the phase interface:

[Q] ≡ (Q A + Q B )

(11)

Σ

wherein the notation implies the limiting values of Q as one approaches the interface from within the superscripted phase. In particular, we have

( − )⋅ + ( − )⋅ = {Q ( v − u ) − Q ( v − u )} ⋅

[Q ( v − u ) ⋅ ] ≡




A

A



A

A



A




B

B

B



B



B

(12)

Continuum and Statistical Mechanics-Based Models for SEI

285

appearing in Eq. (10). The interfacial velocity u is not a property of either bulk phase, but only of the interface. So far, no reference has been made to particular chemical species. Since we represent a body containing N chemical species as the superposition of N single-component bodies, the preceding discussion applies without qualification. For a particular chemical species I (subscript) and associated quantity QI, Eq. (10) becomes d dt

 DQ I  + Q I ∇ ⋅ v I  dV + Dt 

∫∫∫ Q dV = ∫∫∫  I

VB

VB

∫∫[Q

I

( v I − u ) ⋅ ]dS

(13)

Σ

as the appropriate form of the transport theorem.

2.2.2

Conservation of Mass

The generalized form of the transport theorem, Eq. (10), provides a clear route for developing conservation equations valid at phase interfaces as well as in bulk phases. For modelling the growth of SEI layers, we are primarily interested in continuity equations derived from the law of conservation of mass. The total mass of a body, expressed as the volume integral of mass density ρ, does not change over time: d dt

∫∫∫ ρ dV = 0

(14)

VB

For a single-phase body, the transport theorem, Eq. (8) transforms this “closedsystem” statement into the “open-system” statement ∂ρ

∫∫∫ ∂t dV + ∫∫ ρ ( v ⋅ n ) dS = 0 V

(15)

S

relating the mass accumulation in an arbitrary domain to the mass flux across its boundaries. For bodies composed of multiple phases, Eqs. (10) and (14) lead to  Dρ



∫∫∫  Dt + ρ∇ ⋅ v  dV + ∫∫[ρ ( v − u ) ⋅ V

Σ

as a universal law for conservation of mass.

]dS = 0

(16)

286

Lithium-Ion Batteries: Solid-Electrolyte Interphase

The “universality” of this law implies its validity for any arbitrary singlephase region with volume V, and any arbitrary phase interface with surface Σ. Consequently, within any single phase, we must have Dρ + ρ∇ ⋅ v = 0 Dt

(17)

which we recognize as a form of the equation of continuity or the differential mass balance. Likewise, at any phase interface, we must have

[ρ ( v − u ) ⋅ ] = {ρ A ( 

A

− 

)− ρ ( B



B

− 

)}⋅

=0

(18)

which is known as the jump mass balance. For a stationary interface (u ⋅ = 0 ) , the jump mass balance tells us that the total mass flux across the interface is continuous. More generally, it relates the normal components of the total mass fluxes to the motion of the interface in the direction of the unit normal. For individual chemical species, the mass conservation postulate requires modification to account for mass “generation” due to chemical reactions. The mass of component I in a body, expressed as the volume integral of mass density ρΙ, may increase due to the production of I by homogeneous (bulk phase) and heterogeneous (interfacial, σ) chemical reactions: d dt

∫∫∫ ρ dV = ∫∫∫ r dV + ∫∫ r I

σ I dS

I

VB

(19)

Σ

VB

Use of the transport theorem, Eq. (13), and arguments similar to those used to rationalize Eqs. (17), and (18), we can derive the differential mass balance for species I, DρI + ρI ∇ ⋅ v I = rI Dt

(20)

and the jump mass balance for species I,

[ρI ( v I − u ) ⋅ ] = {ρIA ( 

A I

− 

)− ρ ( B I



B I

− 

)}⋅

A

= rIσ

(21)

In this expression, A is the unit normal vector pointing from the interface into phase A. Using the definition of the material derivative, Eq. (6), and the mass flux

Continuum and Statistical Mechanics-Based Models for SEI

287

nI ≡ ρI vI

(22)

∂ρI + ∇ ⋅ n I = rI ∂t

(23)

we can derive

and

(n

A I

)

− n BI ⋅

A

(

= rIσ + ρIA − ρIB

)

⋅ 

A

(24)

as useful alternate forms of Eqs. (20) and (21). For a stationary interface, Eq. (24) tells us that the net mass flux of a chemical species across an interface equals the net rate of production of that species due to interfacial reactions. In the absence of reactions, the motion of the interface can generate a mass flux. Finally, Eq. (24) relates the rates of interfacial reactions to the motion of phase interfaces — perhaps the most significant observation for the problem of modelling SEI growth. For convenience, these equations can be expressed in molar variables by dividing all terms by the component molecular weight, MI: ∂cI +∇⋅ ∂t

I

=

rI MI

(25)

and

(

)

N IA − N BI ⋅

A

=

rIσ + cIA − cIB MI

(

) 

⋅

A

(26)

where cI and NI denote molar concentration and flux, respectively.

2.3

Dynamic Continuum Models for SEI Formation and Growth

The conservation equations presented in the previous section provide the necessary foundation for developing rigorous continuum models of SEI growth in lithium battery cells. Here, we consider the case of SEI growth on carbon anodes in Li-ion cells held at a constant float potential, corresponding to 6 published experiments reviewed in Section 2.1. Two plausible scenarios are treated: growth limited by SEI electronic conductivity, and growth limited by solvent diffusion through the SEI.

288

Lithium-Ion Batteries: Solid-Electrolyte Interphase

2.3.1 Growth Limited by SEI Electronic Conductivity The SAFT group developed an ad hoc model of SEI growth on carbon anodes assuming that the growth rate is limited by the rate of electron transport through the SEI (see Section 2.1). Continuum mechanics can provide a rigorous foundation for this model. Figure 3 illustrates the physical situation of onedimensional conduction-limited SEI growth in rectangular Cartesian coordinates.

Carbon

SEI

e− →

Electrolyte

Li+ ← S + 2e- + 2Li +  →P

z

0

L(t)

Figure 3 SEI growth via electron conduction through the SEI.

Under float potential conditions, a trickle current always passes between the electrodes in order to maintain a specified potential. Assuming that the SEI has a small but finite electronic conductivity, electrons leave the carbon anode, transit the SEI, and reduce one or more components of the electrolyte at the SEI/electrolyte interface at z=L(t). Here, we focus our attention on solvent reduction. In accord with experimental data and theoretical analysis presented elsewhere in this volume, we assume that a solvent component, generically + labelled “S”, undergoes two electron reduction and neutralization by Li to produce an insoluble product “P”. This reaction occurs only at the SEI/electrolyte interface. Assuming the stoichiometry of the pseudo-reaction shown in Figure 3, the molar production rates rIσ M I of the various species are related by

(

)

rσ rσ rPσ rσ = −2 e − = −2 Li + = − S MP M e− M Li + MS

(27)

Continuum and Statistical Mechanics-Based Models for SEI

289

As implied by Figure 3, only electrons are conducted through the SEI; no + solvent or Li pass through the SEI in this scenario. Also assuming no electron conduction through the electrolyte, the jump mass balance for electrons at the SEI/electrolyte interface, Eq. (26), reduces to

z = L (t ) :

− N zSEI ,e − =

reσ− reσ− dL − ceSEI ≈ − M e− dt M e −

(28)

with the z-component of the interface velocity expressed as dL dt . The second term on the right side can be neglected if we assume that either the molar concentration of electrons in the SEI is very small, or if we invoke a pseudosteady-state approximation with respect to the dL dt term. + With regard to Li , the jump mass balance at the SEI/electrolyte interface, Eq. (26), reduces to

z = L (t ) :

rLiσ + rσ El dL + cLi ≈ Li + + M Li + dt M Li +

N zEl,Li + =

(29)

+

Since the molar concentration of Li in the electrolyte is no necessarily small, neglecting the second term on the right side requires that we invoke the pseudo+ steady-state approximation for the dL dt term. Since electrons and Li are consumed at the same rate, Eq. (27), combining Eqs. (28) and (29) yields

z = L (t ) :

N zEl,Li + = − N zSEI ,e −

(30)

+

As expected, the molar fluxes of electrons and Li are equal in magnitude and opposite in direction. The electron flux is usually expressed in terms of a current density i: N zSEI ,e − = iF =

VXF V χ F = S L

(31)

where V is the applied float potential and S is the anode surface area. The second and third forms assume current and potential are related by Ohm’s law + with X and χ as the conductance and specific conductivity of the SEI. The Li + flux in Eq. (30) can be evaluated from a macroscopic mass balance for Li in the system, Eq. (19), d dt

∫∫∫ ρ VB

Li + dV

=

∫∫ r

σ Li + dS

Σ

(32)

290

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Dividing both sides by MLi+, the left side is the rate of change of the total number + + of moles of Li in the system ( ` Li+ ) . Assuming Li consumption only at the SEI/electrolyte interface (area S), and that the rate does not vary over the lateral extent of this interface, Eq. (32) simplifies to d ` Li +  rLiσ + S = M dt  Li +

  

(33) z = L (t )

or, in terms of the rate of lithium loss, d`L d ` Li + ≡− = − S ( N z ,Li + ) z = L(t ) dt dt

(34)

with Eq. (29) used to evaluate the reaction rate. Finally, combining this with Eqs. (30) and (31) leads to d ` L VF χ S B = = dt L L

(35)

which is identical in form to Eq. (1). No product P diffuses through the electrolyte or the SEI; all P precipitates at the SEI/electrolyte interface, leading to SEI growth. A jump mass balance for P at the SEI/electrolyte interface, Eq. (26), reduces to

z = L (t ) :

0=

rPσ dL − cPSEI MP dt

(36)

In view of Eq. (27) relating the reaction rates and Eq. (33), this becomes

z = L (t ) :

cPSEI

dL rPσ 1 rLiσ + 1 d ` Li + = =− =− dt M P 2 M Li + 2S dt

(37)

We assume that the molar concentration of P in the SEI is constant. Integrating from an initial condition t = 0:

L = L0 , ` Li + = ` 0

(38)

to the state at time t, and with the definition of lithium loss as ` L ≡ ` 0 − ` Li + , we obtain

Continuum and Statistical Mechanics-Based Models for SEI

L − L0 = −

1 2 ScPSEI

291

( ` Li+ − ` 0 ) = A` L

(39)

which is identical in form to Eq. (2). Under a certain set of assumptions, continuum mechanics yields governing equations, (35) and (39), that are identical to those proposed by the SAFT 6 group. The predictions of this model were reviewed earlier; detailed comparisons with experimental data for capacity loss will not be reproduced here but may be found in their original publication. In essence, we find that the predictions of the SAFT model are consistent with rigorous continuum mechanics under the assumption of pseudo-steady-state SEI growth limited by SEI electronic conductivity.

2.3.2 Growth Limited by Solvent Diffusion 2-5

Previous studies of the SEI on lithiated carbon, both theoretical and 9 experimental, have recognized that the SEI may have significant porosity. Thus an alternate mechanism for SEI growth controlled by solvent diffusion through the SEI seems plausible. Under float potential conditions, electrons in the carbon anode are readily available to reduce any solvent molecules that may + be present. Since the SEI must be a good conductor of Li to permit the Li-ion cell to function, this reactant is available in excess. Thus the two-electron reduction of solvent takes place at the carbon/SEI interface, growing the SEI at the internal interface. This scenario is consistent with the view that a robust SEI should be able to heal itself as damage occurs due to volume changes of the carbon during charge-discharge cycling. Figure 4 illustrates this physical situation in one-dimensional rectangular Cartesian coordinates. The location of the origin and orientation of the coordinate system differ from those in Figure 3 for mathematical convenience. One way to analyze this problem would be to formulate the jump mass balances for the various components at the carbon/SEI interface, much as was done in the previous section. After invoking the pseudo-steady-state approximation, simplification might result in an expression relating the SEI growth rate to the solvent flux through the SEI. Instead, we recognize the mathematical similarity of this problem to that of the growth of silica layers on 10 silicon surfaces limited by the diffusion of molecular oxygen through the growing silica layer. This leads to a more rigorous mathematical solution involving only one empirical parameter, the solvent diffusivity in the SEI.

292

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Carbon

SEI

Electrolyte

+

e− →

Li ←  S ←

S + 2e- + 2Li +  →P

z

L(t)

0

Figure 4 SEI growth via solvent diffusion through the SEI.

Assuming a constant molar concentration of P and a frame of reference in which the SEI is stationary, the flux of P is zero and the differential mass + balance for P is satisfied identically. We assume an excess of Li in the SEI, so + + the Li mass balance need not be considered, and the Li flux is also zero. Only the solvent differential mass balance is relevant; from Eq. (25), we have ∂cS ∂N z ,S + =0 ∂t ∂z

(40)

(variables represent quantities in the SEI phase unless noted by an appropriate superscript). The constitutive equation for solvent flux, based on the 8 assumption of Fickian diffusion with DS as the effective binary diffusivity of the solvent, can be written formally as N z ,S = xS

∑N

z ,I

− cDS

I

∂xS ∂z

(41)

where c denotes the total molar concentration and xI the mole fraction of component I. Recognizing that c is not constant, and that the only the solvent has a non-zero flux, Eq. (41) reduces to N z ,S = − DS

∂cS ∂z

(42)

after some manipulation. Derivation of Eq. (42) from (41) is non-trivial since c(z,t) is not constant. Substitution into Eq. (40) gives

Continuum and Statistical Mechanics-Based Models for SEI

∂cS ∂2c = DS 2S ∂t ∂z

293

(43)

as the expected diffusion equation for solvent in the SEI. Boundary conditions must be derived from jump mass balances, Eq. (26). The jump mass balances for solvent S and product P at the carbon/SEI interface are

z = L (t ) :

− N z ,S =

rSσ dL − cS MS dt

(44)

and

z = L (t ) :

0=

rPσ dL − cP MP dt

(45)

respectively. Adding these equations and eliminating the reaction rates using Eq. (27) gives

z = L (t ) :

N z ,S N z ,S dL = ≈ dt (cP + cS ) cP

(46)

The second equality follows from the key assumption that the molar concentration of solvent S in the SEI is much less than that of the product P of which the SEI is composed. Substituting this expression into Eq. (45) gives

z = L (t ) :

N z ,S =

rσ rPσ =− S MP MS

(47)

In turn, substituting this expression into Eq. (44) leads to the conclusion that

z = L (t ) :

cS ≈ 0

(48)

At the SEI/electrolyte interface, we assume equilibrium between the solvent in the SEI and that in the electrolyte: z = 0:

cS = ceq

(49)

For convenience, we assume that ceq is simply the concentration of the reactive solvent component in the electrolyte solution.

294

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Equations (43), (48), and (49) are the final set of equations that must be 11 solved. Dimensional analysis shows that these equations can be solved through the similarity transformation u≡

z 4 DSt

(50)

without the need for an initial condition. The details of the solution procedure 11 may be found elsewhere. We find that the solvent concentration in the SEI is  erf (u )  cS ( z , t ) = ceq 1 −   erf (λ ) 

(51)

where λ, a constant, may be found from the solution of

λ=

2 ceq exp ( −λ )

π cP

erf (λ )

(52)

For a given electrolyte composition and solvent reduction product, ceq, cP, and λ are all determined and thus are not adjustable parameters. Only the solvent diffusivity in the SEI may serve as an adjustable empirical parameter in this model. Finally, the model’s analytical solution yields L(t ) = 2λ DSt

(53)

for the SEI thickness as a function of time. SEI thickness increases with the square root of time as would be expected for diffusion-limited film growth. The predictions of this model can be compared with published 6 experimental data from the SAFT group. Several additional assumptions are necessary. First, we assume that all of the “lost” lithium goes into the production of Li2CO3 (i.e., product P) with porosity fixed at ε=0.90. This sets the value of cP in Eq. (52). Given a value of the carbon anode area S, SEI thickness can be related to lithium loss through  M Li 2 CO3  L (t ) =   ` (t )  2ερLi CO S  L 2 3  

(54)

The SAFT group did not report values of S for their cells, so we estimated values of S based on the reported nominal cell capacities, using data for other 12 6 commercial cells as a guide. The SAFT group studied HE prototype cells

Continuum and Statistical Mechanics-Based Models for SEI

295

6 60oC

5 SEI Thickness /µm

22 20 18 16 14 12 10 8 6 4 2 0

4 3 2

30oC

1 0 0

5

10 15 1 ( time/days ) 2

20

% Lithium Loss

employing an electrolyte composed of 1.0 M LiPF6 in a mixture of propylene carbonate, ethylene carbonate, and dimethyl carbonate (PC, EC, and DMC, respectively). We assume that these are mixed in a volume ratio of 1:1:1, from which we may calculate the value ceq in Eq. (52). This fixes the value of λ found from the solution of Eq. (52). Linear regression of SEI thickness plotted as a function of the square root of time determines the value of the one adjustable empirical parameter, the solvent diffusivity DS. 6 Figure 5 shows experimental data (symbols) for lithium loss (right abscissa) and the estimated SEI film thickness (left abscissa) as a function 11 of float time and temperature for HE prototype cells. The one parameter linear regression fits the data with good accuracy. From the slopes, we –19 2 estimate that the solvent diffusivity in the SEI is 5.80 × 10 m /s at 30°C and –18 2 6 2.86 × 10 m /s at 60°C. For MP prototype cells, we find that linear fits of 11 experimental capacity loss data are of similar high accuracy. The solvent diffusion coefficients extracted from these fits are consistently lower (by a factor of about 2.2) than those for the HE prototype cells, most likely due to

25

Figure 5 Experimental lithium loss and estimated SEI thickness (symbols) as functions of time1/2 and temperature for HE prototype cells6 stored at a float potential of 3.9 V at temperatures of 30ºC and 60ºC. Solid lines represent one parameter linear fits of the data in accord with Eq. (53).

296

Lithium-Ion Batteries: Solid-Electrolyte Interphase

differences in the true anode surface area and our estimates. The temperature dependence of the solvent diffusion coefficient follows the Arrhenius law with 11 the same apparent activation energy for both the HE and MP prototype cells. 6 11 Both the electronic conductivity model and the solvent diffusion model do a good job of predicting the time dependence of capacity loss in Li-ion cells. Based on this observation, one cannot discount either model or the underlying physical mechanism. More sophisticated models must be developed that can describe capacity loss behavior under conditions other than float potential. We are currently developing reaction-diffusion continuum models, incorporating realistic kinetic expressions for solvent reduction and lithium intercalation in carbon anodes. We will use these models to predict capacity loss and SEI growth in Li-ion cells under potentiostatic or galvanostatic charge/discharge conditions for comparison with experimental data.

3

Statistical Mechanics-Based Model

As remarked above, a better microscopic understanding of the SEI structural and chemical effects on the open circuit potential (OCP), as well as that of the role of the carbon structure, are needed to complement and provide input to the continuum models used for cell design. Statistical mechanics-based models, usually treated in mean field approximation, provide the nexus between the 13, 14 continuum and the molecular-level models. In this section we present a new lattice-gas model able to predict most of the OCP features that describe the process of lithium intercalation in carbon nanocrystallites. The model, that extends a previous description of the lithium 15 intercalation process in graphite, incorporates deposits representing the SEI layer, which may be either uniformly distributed on the surface or forming patches of varying sizes and chemical strengths. The reactions forming the SEI film consume lithium and are irreversible. We assume that these irreversible reactions have taken place during the first charge/discharge cycle, and that the system can be assumed in thermodynamic equilibrium. Thus, the calculated open circuit potential corresponds and it is compared to lithium intercalation during the second and subsequent cycles. Variable lithium ion-deposit interaction strengths may simulate different chemical compounds present in the SEI layer, such as LiF, the various lithium alkyl carbonates, or Li2CO3. Ordered graphite phases and amorphous carbons are represented by the model.

Continuum and Statistical Mechanics-Based Models for SEI

297

3.1 Description of the Lattice-Gas Model in

out

Let ni be the number of lithium ions inside the carbon layer i and ni be the in out number of lithium ions in the SEI film; then yi = (ni + ni )/N is the total lithium concentration in layer i. It is assumed that a fraction of yi, given by ξi yi, where 0<ξi <1 is intercalated into the SEI film, whereas the difference (1-ξi )yi, is intercalated in the carbon layer. With these definitions, the free energy A of a system of l carbon interlayers modeled by l sublattices each containing N sites in which lithium ions can be inserted, is given by l

l

i =1

i =1

A( y1, y2 ,...., yl ) = N {EoL ∑ (1 − ξi ) yi + EoR ∑ ξi yi +

1 l ∑ J1L (1 − ξi ) 2 yi2 + 2 i =1

 l −1 l l 1 l J 2L  2 1 − ξ ξ − − J1Rξi yi + ∑ ∑ y ( 1 ) y ( 1 ) T Si ( yi )  ∑ ∑ i i j j 2 i =1 2 i =1 j ≠ i d 4  i =1 ij  2

(55)

The model l-layer lattice stack is a composite containing the carbon material and out in a thin film that may cover totally or partially the carbon surface. ξi= ni /(ni + out ni ) is an input to the model, that can be different for each of the l interlayers, and it represents the amount of lithium intercalated in the SEI deposit at interlayer i. The average total lithium concentration y in the l-layer stack is

y=

1 l 1 in yi = (ni + niout ) ∑ l i =1 lN

(56)

Interactions between lithium ions in the carbon interlayer are described by the J1L parameter, whereas those inside the film are given by J1R. Although long-range lithium-lithium interactions may exist within the film, we have neglected them as a first approximation. Thus, long-range interactions among lithium ions from different interlayers are assumed to exist only in the carbon structure and characterized by the J2L parameter. EoL is the host-guest interaction between a single lithium ion and a site on the basal plane of graphite, the same 15, 16 value of EoL is used for each of the carbon sites. EoR is the host-guest interaction between a single lithium ion and a site in the film, and a unique 15 value is used for each of the film sites. We found that attractive in-plane interactions (J1L) and repulsive interlayer interactions (J2L) are needed to reproduce the staging phenomena characteristic of metal-intercalation

298

Lithium-Ion Batteries: Solid-Electrolyte Interphase 17

compounds. The interaction energy between ions present in non-consecutive interlayers is lower than that corresponding to the one between consecutive interlayers; it decays as a function of the inverse of the average separation dij 15, 18 between layers i and j raised to an exponent equal to 4. In this model, dij is parameterized to be equal to 1 for a graphite-like separation distance (3.35 Å), 0 < dij < 1 for blocked (contracted) interlayers, and 1 < dij < 2 for expanded interlayers. For a stack of l interlayers, dij may have different values in each of them, allowing the simulation of amorphous carbons. The last term in Eq. (55) is related to the interlayer entropy Si. It is described by a configurational term for the distribution of lithium ions on the 15 carbon sites, treated in the random approximation, and a parameter q that takes 13 into account vibrational contributions per particle,   N ! q ni Si ( yi ) = k ln    ( yi (1 − ξ i ) N )![(1 − yi )ξ i N ]!

(57)

The chemical potential µi of the lithium ions in interlayer i equals the derivative of the free energy A with respect to the number of particles in the interlayer ni:

µi = EoL (1 − ξi ) + EoR ξi + J1L yi (1 − ξi )2 + J1R yi ξi 2 +

  1 l J2L 1 y j (1 − ξi )(1 − ξ j ) − kT ln  − 1  − kT ln q ∑ dij 4 2 j ≠i  yi (ξi − 1) 

(58)

At equilibrium the lithium chemical potential µ in all the l inter-layers is the same, i.e., µ = µ1 = µ2 = ... =µi = ... = µl. The set of equations (3), with i = 1, 2,…,l is solved iteratively for fixed values of µ, yielding the concentrations yi at each interlayer. 19 The OCP is the potential (V) measured between the two electrodes, zeV = −(

cathode

−

anode

)

(59)

For each value of µcathode, y is obtained using Eqs. (57) and (58). Calculated curves V vs. y are referred to lithium metal anode using equation (59), where µanode is the chemical potential of lithium metal, z is the ionic valence, and e is 20 the electron charge. Other experimental features such as the differential capacity (−dy/dV) are obtained by numerical differentiation of the V−y calculated curves.

Continuum and Statistical Mechanics-Based Models for SEI

299

3.1.1 Implementation of the Model The experimental OCP curve may reflect a combination of various crystallite sizes, carbon structure, and spatial film distribution. Here we focus on these different aspects separately. We have solved the model equations both for a finite system of l interlayers, and for a stack of l interlayers that is subject to periodic boundary conditions. Initially we analyze the effect on the OCP curve of the variation in the fraction of lithium ions accumulated in the film per interlayer, ξI, and that of the strength of the ionic interaction with the film, given by EoR. Then we consider the temperature effect, with all the interaction energy parameters fixed. Finally, at fixed temperature and interaction energy parameters, we investigate the effect of carbon heterogeneity varying the interlayer distances dij in the stack. The parameters accounting for the attractive lithium-lithium intralayer (J1L) interactions and for the repulsive lithium-lithium interlayer (J2L) interaction energies within the carbon structure were set to the 21 values (J1L = −0.11 eV, J2L = 0.046 eV), that reproduce the experimental curve at potentials lower than 0.1 V for calculations without SEI deposits (i.e., ξi = 0 15 for all i = 1,2,…,l). The lithium-carbon interaction energy parameter EoL was 16 set at a value (−1.98 eV) calculated by first-principles methods, and the lithium-lithium interaction in the film, J1R, was set to −2 eV. We tested the sensitivity of the model to the J1R parameter, and found no appreciable differences in the OCP curve when J1R was changed between −2 and −10 eV. In contrast, we found a significant influence of the parameter EoR that represents the lithium-SEI material interaction on the OCP curve, as discussed in the next section.

3.2 Results and Discussions All the results that follow refer to isolated stacks of l interlayers, where no periodic boundary conditions have been imposed. The effect of periodicity is negligible for all the cases considered, because the longest interaction range is within the length of the basic stack of l layers. Figure 6 displays the OCP for a nanocrystallite of l = 70 graphite layers (69 interlayers). This model particle has a uniform film covering its surface, that is represented by a fraction ξi of the total number of ions residing in that interlayer, which is the same for all interlayers. The length of the stack (l = 70)

300

Lithium-Ion Batteries: Solid-Electrolyte Interphase

0.7 0.6

+

V (vs. Li/Li )

y (MCMB)

0.5

0.03 0.02 0.015

0.4

0.001

0.3 0.2 0.1 0.0 0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

fraction of lithium intercalated, y Figure 6 Calculated OCP for model systems containing a uniform SEI layer. Each of the curves represents a film with a lithium content ξ = 0.03, 0.02, 0.015, and 0.001 respectively, which is equal for all interlayers, thus representing a uniform film. ξ is the fraction of the total amount of lithium intercalated per interlayer at each value of V. The rest of the parameters are J1L =−0.11 eV, J2L = 0.046 eV, EoL = −1.98 eV and J1R = −2 eV. The calculated curves are compared with the experimental OCP for a MCMB carbon.

corresponds to carbon nanoparticles of the order of 240Å. This value is in the order of the crystallite sizes detected experimentally in mesocarbon microbeads o 22, 23 (MCMB) treated at temperatures between 700 and 2000 C. Comparing the calculated OCP curves with those corresponding to experimental results for MCMB 1028 (mesocarbon microbeads with 10mm o 21 average particle size, heat-treated at 2800 C), it is observed (Figure 6) that the presence of the model SEI film drastically modifies the OCP curve. The effect of the SEI layer is manifested in the portion of the OCP corresponding to potentials higher than approximately 0.12 V, which clearly shows the intercalation of lithium in the film as the SEI thickness increases (indicated by the value of ξ in Figure 6). In contrast, the region corresponding to lithium intercalation in carbon (potentials < 0.12 V) maintains the same general shape, although the staging phase transitions take place at slightly lower potentials as ξ increases. We observed that when uniform films are modelled (as in Figure 6), is limited to potentials not higher than approximately 0.4-0.5 V. To investigate

Continuum and Statistical Mechanics-Based Models for SEI

301

0.7

0.6

y (MCMB) patches 0.005 0.02

V (vs. Li/Li+)

0.5

0.4

0.3

0.2

0.1

0.0 0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

fraction of lithium intercalated, y Figure 7 Calculated OCP for model systems containing uniform (ξ = 0.02, and 0.005), and nonuniform (see Table 1) SEI layers. The rest of the parameters as in Figure 6. The experimental OCP corresponds to a MCMB carbon.

the effect of a non-uniform film, we assigned to ξ a different value for each interlayer. The calculations shown in Figure 7 illustrate the comparison between the case of uniform film (ξ = 0.005 and 0.02 constant for all interlayers, as in Figure 6), and the case where ξ varies in each interlayer, representing a film with variable thickness, i.e., the presence of “patches”. The effect of varying ξ in the several interlayers clearly can describe more accurately the region of higher potentials. The values of ξ corresponding to the curve identified as “patches”, in Figures 7 and 8, are listed in Table 1, along with the values of the EoR parameter that also varies for each interlayer. The same set of EoR values is used for all curves in Figures 6, 7, and 8. Note also that the curve corresponding to a non-uniform film is in better agreement with the experimental curve in the region of lower potentials (< 0.2 V). This analysis suggests that in isolated nanoparticles, distributed deposits with distinct ionic concentration or distinct chemical nature would be responsible for a protective film of the expected characteristics. This preliminary conclusion provides an illustration of the potential use of this model. More tests model/experiment for different conditions will benefit both interpretation of the experiments and model refinement.

302

Lithium-Ion Batteries: Solid-Electrolyte Interphase

1000 MCMB

-dy/dV x 1/V

800

SEI -patches 600 400 200 0 0.00

0.05

0.10

0.15 0.20

0.25

0.30

0.35

0.40

Potential V

Figure 8 Calculated differential capacity for systems containing a non-uniform SEI layer, as described in Figure 7, and differential capacity data derived from the experimental OCP for a MCMB carbon. Table 1 Lattice-model parameters. The ξι values correspond to the curve labelled “patches”, in Figures 7 and 8. The values of EoR correspond to all calculated curves in Figures 6-8.

Interlayer

ξι

EoR (eV)

1 2 3 4 5 6 7 8 9 10 11 12-69

0.090 0.055 0.037 0.025 0.024 0.022 0.021 0.020 0.010 0.005 0.005 0.005

−9.5 −10.0 −10.5 −11.0 −9.5 −8.0 −7.5 −6.0 −6.0 −6.0 −6.0 −1.0

Another test of the model is obtained calculating the differential capacity values defined as –(dy/dV)/V and comparing them with the corresponding values derived from the experimental OCP curve. The results are displayed in Figure 8 for the curve labelled “patches” in Figure 7, i.e., where the SEI layer is

Continuum and Statistical Mechanics-Based Models for SEI

303

modelled as a non-uniform film. Note that a high value of ξ ι in a given interlayer may imply either a different thickness or a different chemical composition of the SEI layer, since a high fraction of lithium ions are retained at the film portion corresponding to that particular interlayer. Peaks in the differential capacity vs. potential curve represent a phase transition that takes place in the solid structure. Two characteristic peaks are observed in the calculated and experimental curves, at the range of low potentials between 0 and 0.2 V corresponding to lithium intercalation in carbons. The peak at 0.11 V is the transition from Dilute Stage 1 to Stage 2, and the one at 0.08 V denotes the 24 transition from Stage 2 to Stage 1. These two transitions were extensively 15 discussed in previous work. The current model retains such prediction, in good agreement with experimental values (Figure 8). Analyzing the behaviour of the model to the several parameters, we found that the results were quite sensitive to the value of the parameter EoR representing the interactions of the lithium ions inside of the film with the material that constitutes the film. As an illustration, Figure 9 shows the effect of two different sets of EoR values, which are indicated in Table 2. At fixed 1.0 0.9

V (vs. Li/Li+)

0.8 0.7

y (MCMB) basic EoR increased EoR

0.6 0.5 0.4 0.3 0.2 0.1 0.0 0.0

0.2

0.4

0.6

0.8

1.0

fraction of lithium intercalated, y Figure 9 Effect on the OCP of the variation of the interaction energy strength for lithium ions inside the SEI layer, with parameters ξι and EoR shown in Table 2, the rest of the parameters as in Figure 6. They are compared with the experimental OCP for a MCMB carbon.

304

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Table 2 Lattice-model parameters used for the calculated curves in Figure 9. The same set of ξι values were employed for both curves in Figure 9.

Interlayer 1 2 3 4 5 6 7 8 9 10 11 12-69

ξι 0.090 0.055 0.037 0.025 0.024 0.022 0.021 0.020 0.010 0.010 0.010 0.010

Basic EoR (eV)

Increased EoR (eV)

−8.0 −8.0 −8.0 −8.0 −6.5 −6.5 −6.5 −4.0 −1.0 −1.0 −1.0 −7.5

−9.5 −10.0 −10.5 −11.0 −9.5 −8.0 −7.5 −6.0 −6.0 −6.0 −6.0 −1.0

values of ξ ι, the fraction of lithium ions retained in the film, increasing the interaction energy strength causes a shift in the curve, in particular in the region of potentials greater than 0.1 V. The entropic terms in Eqs. (55) and (58) carry all the temperature dependence of this model. Figure 10 shows calculated OCP curves at 30 and o 70 C respectively for the set of parameters in Table 1. The entropic effect is more pronounced at low potentials, affecting the lithium intercalation in the carbon material. The carbon capacity is reduced and the phase transitions o become much less sharp at the higher temperature than those at 30 C; both features are in agreement with experiments. The curve does not change much in the region of potentials higher than 0.2 V, since the model does not have any information about changes due to the chemical nature of the film at high temperatures. Another important variable that influences the OCP curve is the nature of the carbon anode. The model equation 55 takes into account the carbon structure through the dimensionless parameter dij, which represents the interlayer separation relative to that occurring on natural graphite, 3.35 Å.

Continuum and Statistical Mechanics-Based Models for SEI

305

1 0.9

V (vs. Li/Li+)

0.8 0.7

30C 70C

0.6 0.5 0.4 0.3 0.2 0.1 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

fraction of lithium intercalated, y

Figure 10 Temperature effect on the calculated OCP curves for systems covered by a non-uniform SEI layer. The parameters ξι and EoR are displayed in Table 1.

Thus, values of dij lower than 1 imply contracted structures, whereas dij > 1 model regions of expanded structures. Using variable values of dij for the several interlayers, we simulate an amorphous carbon. This effect is illustrated in Figure 11. A value of dij of 0.6 was assigned to a group of 10% of the interlayers forming the crystallite of 70 layers for the curve labelled “contracted”, which is compared to that where all interlayers are separated by 3.35 Å (dij = 1 for all interlayer separations). The main effect of an amorphous structure is reflected in its reduced capacity. Staging is less clearly defined, however, first order phase transitions can still be observed between phases that are not pure stage-k, with k a given number of carbon layers separating filled carbon interlayers.

3.3 Remarks with Respect to the Lattice Model The main features of the OCP for lithium intercalation in graphitic and amorphous carbons are reproduced with a lattice-gas model that describes the SEI layer either as a uniform or as a non-uniform film deposited on the carbon surface. It is found that these deposits are responsible for the OCP shape for + potential values higher than 0.12 V (vs. Li/Li ), approximately, whereas the

306

Lithium-Ion Batteries: Solid-Electrolyte Interphase

0.8 0.7

V (vs. Li/Li+)

0.6

regular separation contracted

0.5 0.4 0.3 0.2 0.1 0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

fraction of lithium intercalated, y

Figure 11 Simulation of an amorphous (curve labelled “contracted”) and a graphitic (regular separation) carbons covered by a non-uniform SEI layer. The parameters ξι and EoR are displayed in Table 1.

carbon heterogeneity reduces the carbon capacity, and modifies staging. This microscopic-based model is a valuable tool that can be used in combination with macroscopic models as discussed in the first part of this chapter for the design of lithium-ion batteries.

References 1. 2. 3. 4. 5. 6.

Peled E., Golodnitsky D., Ardel G. and Eshkenazy V., Electrochim. Acta 40 (1995), 2197-2204. Gao L. and Macdonald D. D., J. Electrochem. Soc. 144 (1997), 1174-1179. Pensado-Rodriguez O., Flores J. R., Urquidi-Macdonald M. and Macdonald D. D., J. Electrochem. Soc. 146 (1999), 1326-1335. Churikov A. V., Electrochim. Acta 46 (2001), 2415-2426. Nainville I., Lemarchand A. and Badiali J. P., Electrochim. Acta 41 (1996), 2855-2863. Broussely M. et al., J. Power Sources 97-98 (2001), 13-21.

Continuum and Statistical Mechanics-Based Models for SEI

7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19.

20. 21. 22. 23. 24.

307 nd

Bird R. B., Stewart W. E. and Lightfoot E. N., Transport Phenomena, 2 ed. (Wiley, New York, 2002). Slattery J. C., Advanced Transport Phenomena (Cambridge University, Cambridge, UK, 2002). Aurbach D. et al., Electrochim. Acta 45 (1999), 67-86. Peng K.-Y., Wang L. C. and Slattery J. C., J. Vac. Sci. Technol. B 14 (1996), 3316-3320. Ramadass P. and Ploehn H. J., unpublished results; manuscript submitted to J. Power Sources (2003). Johnson B. A. and White R. E., J. Power Sources 70-71 (1998), 48. Hill T. L., An Introduction to Statistical Thermodynamics (AddisonWesley, Reading, 1962). Hansen J.-P. and McDonald I. R., Theory of Simple Liquids, 2nd ed. (Academic, San Diego, CA, 1990). Derosa P. A. and Balbuena P. B., J. Electrochem. Soc. 146 (1999), 36303638. Marquez A., Vargas A. and Balbuena P. B., J. Electrochem. Soc. 145 (1998), 3328-3334. Solid State Electrochemistry, ed. by Bruce P. G. (Cambridge University Press, Melbourne, 1997). Safran S. A., Phys. Rev. Lett. 44 (1980), 937-940. McKinnon W. R. and Haering R. R., Physical mechanisms of intercalation. In Modern Aspects of Electrochemistry; ed. by Conway B., White R. E. and Bockris J. M. (Plenum, 1983), Vol. 15, p. 235. Gschneider J. K. A., Sol. State Phys. 16 (1964), 276. Arora P. and White R. E., personal communication. Mabuchi A., Tokumitsu K., Fujimoto H. and Kasuh T., J. Electrochem. Soc. 142 (1995), 1041-1046. Nizhizawa M., Koshika H., Hashitani R., Itoh T., Abe T. and Uchida I., J. Phys. Chem. B 103 (1999), 4933-4936. Dahn J. R., Phys. Rev. B 44 (1991), 9170-9177.

CHAPTER 7

DEVELOPMENT OF NEW ANODES FOR RECHARGEABLE LITHIUM BATTERIES AND THEIR SEI CHARACTERIZATION BY RAMAN AND NEXAFS SPECTROSCOPY GISELLE SANDI Chemistry Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, IL 60439, USA E-mail: [email protected]

1

Introduction

Lithium ion secondary batteries are currently the best portable energy storage device for the consumer electronics market. The recent development of the lithium ion secondary batteries has been achieved by the use of selected carbon and graphite materials as an anode. The performance of lithium ion secondary batteries, such as the charge/discharge capacity, voltage profile and cyclic stability, depend strongly on the microstructure of the anode materials made of carbon and graphite. Due to the contribution of the carbon materials used in the anode in the last five years, the capacity of the typical Li ion battery has been improved 1.7 times. However, there are still active investigations to identify the key parameters of carbons that provide the improved anode properties, as carbon and graphite materials have large varieties in the microstructure, texture, crystallinity, and morphology, depending on their preparation processes and precursor materials, as well as various forms such as powder, fibers, and spherule. There is a strong correlation between the microstructural parameters and electrochemical properties of conventional and novel types of carbon materials for Li ion batteries, namely, graphitizable carbons such as milled mesophase pitch-based carbon fibers, polyparaphenylene-based carbon heat treated at low temperatures, boron doped graphitized materials, and templated carbons. Li-ion cells were commercialized by Sony Corporation in the early 1990’s. This phenomenon led to a major expansion in the application of rechargeable Li batteries for portable electronic devices. Before that, the only Li cells that were commercially available were primary cells. The driven factor of the commercial development of Li-ion cells was the utilization of carbonaceous materials as

308

New Anodes for LIB and Their SEI Characterization

309

host for the Li ions. Prior to the success of Li-ion technology, attempts to produce rechargeable lithium cells were frustrated by technical issues related to safety and the poor cycle life of the lithium electrode. The morphological changes that occur by repetitive cycling of lithium eventually led to cell failure. Lithium metal and lithium metal alloys in nonaqueous electrolytes have shown limited promise in practical rechargeable cells. Recently, a variety of tin oxide based compounds; SnO, SnO3, LiSnO3, and SnSiO3 glass, have been tested as anode materials. These materials demonstrate discharge capacities on the order of 1000 mAh/(g Sn), which is consistent with the alloying capacity limit of 4.4 Li atoms per Sn atom, or 991 mAh/(g Sn). However, the irreversible capacities ranged from 200 mAh/ (g active) to 700 mAh/(g active). By introducing lithium, lithium oxide and tin form first, which is then followed by the formation of various LiSn alloy phases. Other investigations have focused on the copper-tin system at around the composition Cu6Sn5 and have determined the effect on cycling and capacity of electrodes with various ratios of copper to tin. Several techniques have been applied for the characterization of battery 1, 2 materials, but the electronic and structural information has not been properly described yet. NEXAFS can be successfully used to determine the electronic structure of carbonaceous materials that have been synthesized using a unique 3-6 method described in detail elsewhere. NEXAFS measures the excitation of electrons to partially-filled or unfilled molecular orbitals. The signal obtained by electron-yield detection is surface sensitive, while that obtained by fluorescence yield detection is bulk sensitive. While the electron-yield method is sensitive only to the top few atomic layers, the fluorescence yield method can detect species up to a few thousand Å deep into the bulk structure. We employed the electron-yield method in this study. In general for hydrocarbons, the carbon NEXAFS energy range can be divided into three regions characterized by specific resonances: the first π* resonance around 285 ± 1 eV, the C-H* resonances around 288 ± 1 eV, and a broad σ* region between 290 and 315 eV. The existence of the π* and the C-H* resonances can clearly be used to establish respectively, the hybridization of the C-C bonds and the existence of any C-H bonds in the samples. Graphite has 7 unsaturated C-C bonds, as proven by the π* resonance in their spectra, and graphite and diamond do not contain hydrogen, as is evident from the missing C-H* resonance. In particular, the π* resonance may be used to clearly distinguish the two forms of carbon from each other. This characteristic feature can be utilized for determining whether the local bonding in amorphous

310

Lithium-Ion Batteries: Solid-Electrolyte Interphase

hydrogenated carbon films is graphite-like or diamond-like. It is our goal to determine the structural differences among the carbonaceous materials prepared by our unique templating method, and to correlate them with electrochemical performance data. The SEI structure will be also addressed.

2

Carbon as a Host in Lithium Ion Cells 8

At the beginning of the 1970’s, it was discovered that certain intercalation compounds could be used as electrodes for lithium secondary batteries. An intercalation “host” is a solid that can reversibly incorporate “visiting” atoms or molecules inside its crystalline lattice with relatively small changes to the structure. The reactions that occur in the electrodes can be summarized as follows: − δx Li > δx Li+ + δx e in the Li electrode − + δx Li + δx e + LiX (host) > Lix + δx (host) in the cathode δx Li + LiX (host) > Lix + δx (host) total reaction (1) During the 1970’s and 1980’s most of the work on secondary batteries was concentrated on the use of intercalation compounds as the positive 9 electrode and metallic lithium or lithium alloys as anodes. Of these, the rechargeable battery “Molicel2” size AA, developed by Moli Energy Ltd. (British Columbia, Canada) at the end of the 1980’s, was probably the most 10 sophisticated. Apart from a few small secondary cells, the AA did not survive the 1990’s. In certain occasions the cells suffered small explosions, which led 11 to a recall of the products from the market. To understand this behavior, the difficulties associated with the use of metallic lithium must be considered. When lithium is electrodeposited on a metallic lithium anode during an intercalation cell recharge, a porous deposit is formed (more porous than the original metal). This induces electrode expansion with the consequent increase of internal pressure of the cell. The increase of the Li electrode with the cycle number diminishes when 15 atmospheres of pressure are reached. Also, the contact area between metallic lithium and the electrolyte becomes larger as the cell is recycled. Because this interface is not thermodynamically stable, the cell becomes sensitive to thermal, mechanical, and electrical changes. Only if the electrolyte and the lithium do not react at all can this situation be avoided. Also,

New Anodes for LIB and Their SEI Characterization

311

if the surface area of the negative electrode is kept constant, a certain degree of safety can be reached. To improve the safety of secondary lithium batteries, the metallic lithium is replaced by another intercalation compound such as graphite. In addition, the cathode would contain ionic lithium in its structure, which is intercalated in the anode or the cathode depending on the direction of the current. Lithium-ion cells are the most advanced batteries now in the market. These cells supply up to 4 volts, have an energy density close to 120 Wh/kg, and have a long life at room temperature. The technology is based on the use of appropriate lithium intercalation compounds as electrodes. Normally a lithium transition metal oxide is used as the cathode and carbonaceous materials serve as the anode. Selecting the best carbon electrode is a work in progress. This is because hundreds of carbons are commercially available, including cokes, mesocarbons, 12-16 fibers, synthetic and natural graphites, etc. The structure of the carbon influences the lithium intercalation in two aspects: how much can be intercalated and at what voltage. To understand the behavior of different carbons, it is necessary to have a good knowledge of the carbon structure, from the most crystalline to the most disordered (or non-crystalline). Carbons that are able to intercalate lithium reversibly can be grouped into two general groups: graphites and disordered carbons. Graphites are carbonaceous materials with a layered structure, but with a number of structural defects. From a crystallographic point of view, the term graphite is only applied to carbons that have a structure with perfectly ordered layers of either type AB (hexagonal) as shown in Figure 1, or ABC (rhombohedral). The actual structure of most of the carbonaceous materials that are used is crystallographically different from the ideal structure of graphite. Materials consisting of microcrystalline graphite aggregates are also called graphite. In other words, the terms natural, artificial, 17 and synthetic graphite are used synonymously, and all are microcrystalline. Non-graphitic or disordered carbonaceous materials consist of carbon atoms that are mainly distributed in hexagonal planes, but without crystallographic order in the c direction. The structure of these carbons is characterized by amorphous areas combined with more organized or crystalline areas, as shown in Figure 2. The number and size of such areas depend on the organic precursors and the temperatures at which the compounds are synthesized. Most of these materials are prepared by the polymerization of organic hydrocarbons at temperatures lower than 1500°C. If they are heated at temperatures between 1500 and 3000°C, two types of carbons are formed: “soft” and “hard”. “Soft” or “graphitizable” carbons are formed by a particular

312

Lithium-Ion Batteries: Solid-Electrolyte Interphase

mobility that leads to the formation of structures similar to graphite. “Hard” or non-graphitizable carbons do not show an organized structure because the layers are immobile and interlaced. Hard carbons are mechanically less flexible than soft carbons.

B asal S urface P lane

A

B

0.3354 nm A

plane A plane B

Figure 1 Schematic representation of the hexagonal crystalline structure of graphite, showing the AB layers in sequence and the unit cell.

New Anodes for LIB and Their SEI Characterization

313

C rystalline p ha se

a m orp hou s p h ase Figure 2 Schematic representation of a non-graphitic carbonaceous material. 18, 19

The synthesis of intercalation compounds, LixCn began in the 1950's. At room temperature the maximum lithium content is one lithium atom per six carbon atoms. During the intercalation process, the ordering of the graphite layers changes to the configuration AA, and the distance between the layers increases by about 10%. When the structure LiC6 is reached, the lithium atoms are distributed so that they do not occupy nearest neighbor sites. The maximum capacity obtained during the ionic lithium intercalation corresponds to 372 mAh/g. However, in the subsequent lithium de-intercalation in the first cycle, only 85 to 90% of the original lithium is recovered. In additional cycles the performance improves. The charge excess consumed in the first cycle is associated with the formation of the solid electrolyte interface (SEI) and to certain reactions related to the corrosion of LixC6. Like metallic lithium and certain lithium alloys, the lithium/carbon intercalation compounds are

314

Lithium-Ion Batteries: Solid-Electrolyte Interphase

thermodynamically unstable in most of the known electrolytes. This means that the surface exposed to the electrolyte must be kinetically protected by the SEI 20-23 layer. Due to the irreversible lithium consumption, the capacity lost in the first cycle is known as irreversible specific capacity. Figure 3 shows a schematic representation of the voltage-capacity profile for lithium intercalation of carbon electrodes. Typically, the potential of the carbon is >1 V before Li intercalation takes place. For highly graphitized carbon, when current is applied to intercalate Li, the potential initially drops rapidly to near 0.8 V vs. Li where electrolyte decomposition and the formation of the SEI layer occur. When these reactions are taking place, the voltage remains close to a constant value. Following electrolyte decomposition, the potential declines and most of the

Vo ltag e (V )

E lec tro ly te d ec o m p o sitio n

G rap h itic ca rb o n A m o rp h o u s c arb o n

stag in g

C a p a city (m A h /g ) Figure 3 Schematic representation of the voltage profile for lithium insertion in carbon anodes.

New Anodes for LIB and Their SEI Characterization

315

Li intercalation occurs at <0.25 V. With highly graphitized carbons, inflections and plateaus in the voltage-capacity profiles are present and are related to staging phenomena. On the other hand, amorphous carbons do not have the inflections and plateaus, but the present a sloping profile. Electrolyte decomposition is a major concern when intercalating Li ions on a carbonaceous matrix. When graphite is used as the anode, exfoliation of the electrode structure occurs when LiClO4/PC is used as the electrolyte, but the same electrolyte system can be used for disordered carbons such as those derived from petroleum coke. The most common nonaqueous electrolyte is LiPF6 in EC/DEC. A number of products from electrolyte decomposition have 24, 25 been identified by Aurbach et al. Also, inorganic compounds such as LiCO3, Li2O, CO, and H2 have been reported as being produced by reactions with the 26 organic products or trace water. Various approaches have been identified to reduce the extent of electrolyte decomposition and irreversible capacity loss at the carbon negative electrode. By adding additives to PC such as CO2, N2O, CO, the self-discharge and cycling 27 behavior of the lithiated carbon electrodes has improved. These additives affect the film properties by decreasing the low-frequency impedance, thus + permitting a more rapid Li -ion transport. Several research groups, mainly in the United States and Japan, have synthesized carbons with low irreversibility. Novel carbonaceous compounds derived from pillared clays have been synthesized and characterized physically and electrochemically by G. Sandí and her collaborators at Argonne National 28-45 Laboratory, USA. These carbons present a reversible capacity that corresponds to almost twice that of graphite under the same experimental conditions. In these compounds the surface area and porosity are controlled by template design. As a result the irreversible processes are minimized, especially those related to decomposition on the surface and organic electrolyte penetration into the carbon structure. By using state-of-the-art X-ray and neutron scattering techniques, it has been proven that these carbonaceous compounds contain pores with diameters of approximately 4 Å, a size through 33 which lithium ion can easily diffuse. Figure 4 shows a schematic representation of their synthesis, starting from pillared clay templates and using pyrene as the organic precursor. The circles denoted as Al2O3 represent pillars in the clay structure that help to keep the clay layers apart upon heating. Organic precursors are incorporated between the clay layers and the spaces created by the pillars. At the same time, the pillars serve as acid sites for the conversion of organics into carbonized material in the presence of nitrogen at

316

Lithium-Ion Batteries: Solid-Electrolyte Interphase

700°C. Afterwards the inorganic template material is dissolved in acid, leaving behind a disordered carbon with pores where the pillars used to be. pillared clay pyrene m olecule

porous carbon

Figure 4 Synthesis of porous carbonaceous materials using pillared clay templates and pyrene as the organic precursor. Reproduced from [34] by permission from the American Chemical Society, Inc.

Figure 5 shows a scanning tunneling microscopy (STM) picture of a carbon sample prepared from PILC/PILC/benzene and pyrolyzed at 700°C. The black channels (50 nm < r < 100 nm) in the STM micrograph correspond to the pore structure network. Denoted by black arrows are several of the mesopores already characterized by LN2 physisorption. The concentration of these mesopores is much lower in the lower surface area materials (~ 10m2/g). Transmission electron microscopy (TEM) of the carbons (Figure 6) indicated that the carbon formed agglomerates of 1-10 mm, which were composed of a broad range (5-500 nm) of many overlapping particles. The vast majority of these particles were completely amorphous. Several others organic precursors have been incorporated in the pillared clay lattice and the results are promising. For example, styrene, ethylene, and trioxane/pyrene co-polymer were incorporated and the resulting carbonaceous material was electrochemically evaluated. The largest voltage plateau in discharge (around 0 V vs. Li) was obtained for the sample made with

New Anodes for LIB and Their SEI Characterization

317

trioxane/pyrene as the carbon precursor. A plateau near 0.7 V in charge was obtained, and the magnitude increases in the following order: 46 propylene>styrene>trioxane/pyrene. This phenomenon has been observed in 47 materials with high hydrogen content. Most of the capacity is delivered between 1 and 0.1 V, thus avoiding the possibility of safety problems associated with the lithium metal deposition close to 0 V. Carbon electrodes prepared from the trioxane precursor showed the largest hysteresis effect. We believe that oxygen on the surface is the main cause of this undesirable phenomenon in lithium ion batteries.

Figure 5 STM of a carbon sample prepared by using PILC/pyrene/benzene and pyrolyzed at 700°C. Scan size = −500 nm; set point = −1.0 V; scan rate = 5 Hz.

More recently, carbonaceous materials have been derived from ethylene or propylene upon incorporation in the vapor phase in the channels of sepiolite, taking advantage of the Brønsted acidity in the channels to polymerize olefins. Sepiolite is a phyllosilicate clay insofar as it contains a continuous twodimensional tetrahedral silicate sheet. However, it differs from other clays in that it lacks a continuous octahedral sheet structure. Instead, its structure can be considered to contain ribbons of 2:1 phyllosilicate structure, with each ribbon

318

Lithium-Ion Batteries: Solid-Electrolyte Interphase

linked to the next by inversion of SiO4 tetrahedra along a set of Si-O-Si bonds. In this framework, rectangular channels run parallel to the x-axis between opposing 2:1 ribbons, which results in a fibrous morphology with channels running parallel to the fiber length. Channels are 3.7 × 10.6 Å in sepiolite (they are 3.7 × 6.4 Å in palygorskite). Individual fibers generally range from about 100 Å to 4-5 microns in length, 100-300 Å width, and 50-100 Å thickness. Inside the channels are protons, coordinated water, a small number of exchangeable cations, and zeolitic water. Figure 7 shows a bright field TEM of the resulting carbon after the clay has been removed. Carbon fibers (1-1.5 microns long) are obtained whose orientation and shape resemble that of the original clay. The SAED pattern of the carbon fibers shows diffuse rings typical of amorphous carbon; no diffraction spots were observed. The specific discharge capacity of this carbon at the end of the 14th cycle corresponds to 633 mAh/g, which is 1.70 times higher than the capacity delivered by graphitic 48 carbon, assuming 100% efficiency. More research is being conducted to improve the irreversible capacity of the first cycle.

Figure 6 TEM of same sample as in Figure 5.

New Anodes for LIB and Their SEI Characterization

319

Figure 7 TEM of a carbon sample derived from sepiolite/propylene composite. A JEOL 100CXII Transmission Electron Microscope operating at 100 kV was used. Reproduced from [48] by permission from the Electrochemical Society, Inc.

3

Alternative Anode Materials

In addition to carbon, the attention has been focused on alloys and lithiated metal oxides as new materials for anodes in Li-ion cells. The reversible insertion of Li in metal/alloys has been studied for many years because of their application in high-temperature molten salts Li cells. The electrochemical reactions that occur during discharge of a Li alloy electrode is: LixM−LiyM + Li+ + e

−

(2)

where M =Al or Si, and LixM and LiyM represent two solid phases in equilibrium. There are other elements that can alloy with Li, such as Sn, Pb, Bi, Sb, and As. Although the electrochemical capacities of lithium alloys may be very large compared to that of carbon (for example, LiAl and Li4.4Sn = 990 mAh/g), 49 the large volume expansion due to the existence of two phases domains results

320

Lithium-Ion Batteries: Solid-Electrolyte Interphase

in severe particle cracking with loss of electrical contact, giving irreversible capacity losses which prevent the practical use of these materials in Li-ion cells. Recently, metal-based oxides have received much attention as another alternative. They offer higher capacity than carbon and have good cycleability. Nevertheless, large first cycle capacity loss is observed since the metal oxides have to undergo electrochemical reduction before Li can alloy with the metals. Many groups have dedicated time and efforts to minimize this irreversibility by: (i) Using composites of active and inactive materials, such as tin-based composites oxides where a nano-structure active phase is dispersed either in an inert solid electrolyte or in a soft metal matrix formed in the initial charge;50 (ii) Using intermetallic lithium insertion compounds, e.g. Cu6Sn5, where lithium atoms occupy interstitial sites, giving only a small volume expansion;51 (iii) Using mixed active composites (active/active composite) such as SnSb0.14, where stepwise lithium insertion into the different active phases buffers volume expansion.52 53 Zhao et al. found that by adding about 12 wt% graphite additives to ballmilled Zn4Sb3, the reversible capacity in the first cycle reached 580 mAh/g. However, the capacity fade in their material (after only 10 cycles) is close to 54 35%. Crosnier et al. tested samples of small particle size bismuth and electroplated Ni-Sn alloys. They concluded that there is a large volume expansion of the electrodes and that this volume expansion is associated with the particle size of the material, that is, the smaller the particle size, the best 55 cyclability obtained. Nam et al. prepared thin films of pure SnO, of Sn/Li2O layered structure, and of Sn/Li2O by a sputtering method, and assembled a lithium-reacted tin oxide thin film by the evaporation of lithium metal onto a SnO2 thin film. According to their results, the lithium-reacted tin oxide thin film, the Sn/Li2O layered structure, and the Sn/Li2O co-sputtered thin films did not show any irreversible side reactions (formation of Li2O or metallic Sn + near 0.8 V vs. Li/Li ), with an initial charge retention of 50% (20% better charge retention that SnO or tin composite oxide). The cells seemed stable after 500 cycles, but the capacity dropped from 800 mAh/g in the first cycle to about 300 mAh/g. Other materials based on Zn and Sn have been prepared. For 56 example, Belliard et al. prepared ZnO, ZnO:SnO2 ball-milled mixture, and Zn2SnO4. According to their findings, these materials have smaller capacities than SnO2, but show reversible capacities around 500 mAh/g. They also observed, however, a large loss of capacity between the initial and the later cycles, similar to the thin oxides, due to the required reduction of the tin and 57 zinc ions to the bulk material. Limthongkul et al. have performed partial

New Anodes for LIB and Their SEI Characterization

321

reduction of mixed oxides to synthesize composites anodes with reduced irreversible capacity loss during the first discharge and in which phase transformation and volume changes during cycling can be accommodated. One example of a suitable system was Sn-Ti-O. They found lower irreversibility in the first cycle and better cyclability compared to the unreduced SnO2-TiO2 sample. There are as many possibilities for new materials as there are elements in the periodic table. The success of a new material will rely on the decrease of volume expansion and lower irreversible capacities in the first cycle.

4

UV Raman Spectroscopy of Templated-Disordered Carbons

Conventional Raman Spectroscopy using visible laser excitation often suffers from two limitations, inherently low Raman scattering signals and strong fluorescence which often obscure the Raman signal. To avoid interferences from fluorescence, ultraviolet excitation is used because the fluorescence of most molecules and surfaces occurs in the visible region. Raman spectra are very sensitive to changes in the translational symmetry of the solid. In amorphous carbon they provide information about the level of microstructural disorder. In contrast with graphite and diamond with a crystalline structure that is connected with Raman lines, amorphous carbon reveals broad bands caused by the structural disorder with an unsymmetrical Raman band in the wavenumber −1 region between 900 and 1800 cm . Its shape is formed by two more or less −1 significant features, originating from the graphite “G” line at about 1580 cm −1 and the “D” line near 1350 cm . Analyzing the intensity relation ID/IG, the peak position, and their FWHM, one gets information about the diamond-likeness of the films. 4 Graphitic samples which posses D 6h space group symmetry, yield 6 nonzero mode frequencies which can be enumerated as 2B2g + 2E2g + A2n + E1n. Only the two in-plane E2g modes are Raman active and they produce peaks near 1582 and in the presence of disorder, will be accompanied by a disorder−1 induced peak near 1360 cm .

322

Lithium-Ion Batteries: Solid-Electrolyte Interphase

400 360

1595 cm-1

pyrene 1000 °C pyrene 700 °C trioxane 700 °C styrene 700 °C

320

Intensity

280 240 200 160

2330 cm-1 3100 cm-1

120 80 40 0 800

1200

1600

2000

2400

2800

3200

3600

Raman shift, cm-1 Figure 8 UV Raman spectra of disordered carbons prepared by a templated method. The spectra are offset on both axes.

Ultraviolet Raman spectroscopy was conducted in carbon samples 28-45 The 514 nm output of a 18-watt Ar+ ion laser is prepared by Sandí et al. 58 frequency doubled to 257 nm using a temperature-tuned KDP crystal. The 257 nm beam is used to excite Raman scattering. The samples were formed pressed discs. The spectra obtained in these experiments are compared with that of the −1 graphite. At 1575 cm a sharp band called “G” band is found in the spectrum of a graphite single crystal. Figure 8 shows the spectra of the template-derived carbonaceous materials. They exhibit a very wide and asymmetric band with a −1 single maximum near 1600 cm . The first order spectrum is measured from −1 900-1900 and the second order from 2500-3300 cm . The carbon structure present at a temperature of 700°C could be a highly relaxed nature with C-C distances able to accommodate larger amounts of Li than typical GICs.

New Anodes for LIB and Their SEI Characterization

323

Cylindrical Mirror Analyzer (CMA)

Sample Collimators Metal Grid






































































































































Phosphor Screen

Partial Yield Detector

Evaporator Electron Multiplier

Figure 9 Schematic representation of the NEXAFS detector used in these studies.

5

SEI Characterization by NEXAFS

The NEXAFS experiments were performed at the U1 beamline of the National Synchrotron Light Source, Brookhaven National Laboratory. Instrumental details about the optics on the beamline, as well as the UHV chamber with 59 facilities for high pressure reactions have been published elsewhere. A schematic representation of the sample set up is illustrated in Figure 9. The spectra were obtained by measuring the intensity of electron yield, recorded by a channeltron electron multiplier located near the sample holder. All NEXAFS spectra were recorded with the photon beam at the normal incident angle with respect to the sample surface. In order to reduce the intensity contribution from low-energy, secondary electrons from the subsurface regions, the entrance of the channeltron was biased by a negative voltage of 100 eV to repel these electrons. The spectra were measured as a function of the incident X-ray photon energy in the vicinity of the carbon K-

324

Lithium-Ion Batteries: Solid-Electrolyte Interphase

edge (275-325 eV), oxygen K-edge (510-590 eV), and Li K-edge (35-80 eV). The carbonaceous materials were pressed into stainless steel sample holders of about 1.3 cm in diameter and 0.1 cm in depth. The energy transitions are then correlated with the structure of the carbon and the relative hydrogen content. Two sets of samples were analyzed: powders and carbon electrodes. Details about the synthesis of the carbonaceous materials have been published 3-6 elsewhere. In brief, the samples were prepared using pillared clays as templates. Different carbon precursors were loaded into the layered pillared clay and pyrolyzed at 700°C for 4 hours. The structure of the clay remains intact at this temperature as evidenced by X-ray powder diffraction. The clay was then removed by conventional demineralization methods and the resultant carbon was oven-dried at 120°C overnight. Electrodes were prepared using 90% by weight of the carbonaceous materials, 5% by weight of Super S carbon black (Alfa Chemicals), and a binder solution made of polyvinylidene fluoride (PVDF, Aldrich, 99+%) dissolved in N-methyl-pyrrolidinone (NMP, Aldrich, 99+%). The Super S carbon black is used to provide electrical contact between carbon grains. An excess of NMP was added to make a slurry. The slurry was oven-dried at 120°C overnight. This resulting powder is used to make pellets in carbon-steel dies. About 20-30 milligrams of carbon is put into the die and pressed at about 5000 psi. The electrolyte was 1 M LiPF6 dissolved in 50 vol.% ethylene carbonate (EC) & 50 vol.% dimethylcarbonate (DMC) obtained as a solution from FMC Lithium Division (Gastonia, NC). Electrochemical cells were assembled in a heliumfilled recirculating/purification glovebox (Vacuum/ Atmospheres DLX series). Carbon electrode pellets were dried at 80°C in a vacuum oven inside the glovebox prior to assembly. All cell hardware and separator materials were also rigorously dried in like manner. The dual electrode configuration in these cells uses metallic lithium as the anode. The assembly of the cells is described in Ref. 60. The sealed button cells that displayed a good voltage were transported out of the glovebox for electrochemical testing on an Arbin 2400 station cell cycler. After several cycles, the coin cells were opened and stored under argon atmosphere. The carbon electrode was separated from the rest of the components (inside an argon glove box) and placed in a flat sample holder using double-sided tape. The sample holder was transported into the vacuum chamber using a specially designed sealed glass container, previously evacuated and filled with argon. A glove bag continuously flushed with argon, was used to provide an argon blanket while the sample holder was rapidly mounted inside

New Anodes for LIB and Their SEI Characterization

325

the vacuum chamber, thus eliminating the possibility of oxygen adsorption. All samples were corrected against a reference sample (boron nitride). Figure 10 shows a comparison of electron-yield NEXAFS spectra of the carbonaceous materials derived from the templating method before they were used as electrodes. The spectra show several peaks at different energy levels that can be explained as the result of resonance interactions between localized molecular states. For aromatic systems, there is a strong interaction between the localized Β* and the Φ* states, producing a set of delocalized orbitals which are significantly separated in energy. The carbon derived from pyrene shows two Β* antibonding orbitals, the first one at 286 eV, corresponding to the transition C(1s) − Β*1 (e2u) and the second transition at about 289 eV, corresponding to the 61 transition C(1s) − Β*2 (b2g). The peak at about 288 eV represents the C-H* transition and is correlated to the hydrogen content in the carbon sample. The Φ* antibonding orbitals are manifested by the energy transitions higher than 293 eV. Two main transitions were observed at 293 and 296 eV, corresponding to Φ*C-C and Φ*C=C. The spectra of the other carbonaceous materials is similar to that of the carbon from pyrene, except that the relative intensity of the C-H* peak is smaller. The C-H* peak is related to the amount of hydrogen on the surface of the carbonaceous material and does not reflect the total amount present in the sample as determined by CHN elemental analysis. It has been shown that there is a direct correlation between the total amount of hydrogen in 62 a carbon sample and the capacity delivered upon cycling. Computer simulations of Li reactions with disordered carbons containing hydrogen have shown that Li readily bonds to a proton-passivated edge carbon resulting in a configuration similar to the organo-lithium molecule C2H2Li2.63 As a result, it provides a second channel for lithium uptake, which only works if the edge carbons are saturated with protons. Furthermore, theoretical calculations demonstrated that the lithium ion in a Li+-anthracene complex is thermodynamically more stable when the Li+ is in a terminal position than when it is in the center.43 This may explain why a carbonaceous material with some terminal hydrogen delivers higher capacity than other materials with low hydrogen content (terminal H is being replaced by Li ion).

326

Lithium-Ion Batteries: Solid-Electrolyte Interphase

20 18

C-H* B* (C=C) F* C-C 1

F* C=C

16

Intensity

14 12 10 8 6 4

Carbon from pyrene Carbon from styrene Carbon from propylene Carbon from trioxane

B* 2

2 0 280

290

300

Energy, eV

310

320

Figure 10 Comparison of carbon K-edge near-edge spectra of carbonaceous materials synthesized using pillared clays as templates. The NEXAFS spectra were obtained by measuring the intensity of electron-yield with a bias of −100 eV to the entrance of the channeltron multiplier. Reproduced from [33] by permission from Elsevier.

New Anodes for LIB and Their SEI Characterization

327

1.7 Pyrene Styrene Propylene Trioxane

Relative Intensity

1.6

1.5

1.4

1.3

1.2

1.1 520

540

560

580

Energy, eV Figure 11 Comparison of oxygen K-edge near-edge spectra of carbonaceous materials synthesized using pillared clays as templates. Experimental conditions were the same as in Figure 10. Reproduced from [33] by permission from Elsevier.

328

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Figure 11 shows the NEXAFS spectra in the oxygen K-edge region of the carbonaceous materials. The spectra exhibit a Β* and a Φ* resonance approximately 10 eV apart. Of special notice is the relative lower intensity of the peaks compared with the carbon K-edge peaks, indicating that even though oxygen is present in the samples, the amount is low. Morever, the carbons synthesized using trioxane and styrene as the precursor materials contain a larger amount of oxygen on the surface than the other carbons, as evidence by the peaks’ intensity. The presence of oxygen on the surface of the carbon contributes to a higher irreversible capacity and hysteresis in voltage (a higher voltage is necessary to intercalate the lithium ion upon cycling) due to side reactions with the electrolyte. It is not surprising then to find higher irreversible capacities in those coin cells prepared with carbon from troixane/pyrene and styrene than in those prepared using pyrene, ethylene or propylene as the organic precursor. Figure 12 shows the carbon K-edge NEXAFS spectra of the electrodes. The strong carbon near-edge features at 292 eV and 302 eV are identical to 64 NEXAFS spectra of Li2CO3 and these two features can readily be assigned to the electronic excitation to the Β* and the Φ* orbitals of carbonates. This result is also consistent with the formation of a passivating layer formed during the first cycle when lithium metal is used as the anode. Decomposition of electrolyte is the main cause of this layer formation. Another factor that contributes to the formation of this layer is the surface area of the carbon. For disordered carbons, an exfoliation mechanism occurs in which the exposed surface area continues to increase upon cycling. The BET surface area of these carbons increases in the same order as the peak intensities do in the NEXAFS spectra (pyrene<styrene
New Anodes for LIB and Their SEI Characterization

329

25 292 eV Carbon from pyrene Carbon from styrene Carbon from trioxane/pyrene

B*

Relative Intensity

20

15

F* 302 eV

10

5

0 280

290

300

310

320

Energy, eV Figure 12 Carbon K-edge NEXAFS spectra of the carbon electrodes. The strong carbon near-edge features at 292 eV and 302 eV are identical to NEXAFS spectra of Li2CO3. Reproduced from [33] by permission from Elsevier.

330

Lithium-Ion Batteries: Solid-Electrolyte Interphase

1.6 Carbon from trioxane/pyrene Carbon from pyrene Carbon from styrene

Relative Intensity

1.3

1.0

0.7

0.4

0.1 35

40

45

50

55

60

65

70

75

Energy, eV Figure 13 Comparison of Li K-edge near-edge spectra of carbonaceous materials synthesized using pillared clays as templates. Experimental conditions were the same as in Figure 10.

New Anodes for LIB and Their SEI Characterization

6

331

Conclusions

The results of this study demonstrated that NEXAFS and RAMAN techniques are very useful tools in obtaining fundamental information concerning the structural and electronic properties of carbonaceous materials and anodes electrodes derived from them. For highly oriented, crystalline-like materials, the differences in the C K-edge features can be associated to different orientations or different polarization-dependence in these samples. However, the carbon samples analyzed in this study are amorphous materials. For amorphous materials, the relative intensities of all C K-edge features are independent of the orientation of the samples. The relative C-H* intensity derived from the carbon K-edge suggest that the carbon samples derived from pyrene as the organic precursor contain a higher amount of hydrogen in the surface than the samples derived from styrene, propylene, and trioxane/pyrene co-polymer.

Acknowledgements This work was performed under the auspices of the U.S. Department of Energy, Office of Basic Energy Sciences, Division of Chemical Sciences, Geosciences, and Biosciences under contract number W-31-109-ENG-38.

References 1.

2. 3. 4.

5.

Touzain P., Guessan G. K., Bonnin, D., Kaiser P. and Chouteau G., Electrochemically reduced cobalt-graphite intercalation compound. Synthetic Metals 79 (1996), 241-251. Amatucci G. G., Tarascon J. M. and Klein L. C., CoO2, the end member of the LixCoO2 solid solution. J. Electrochem. Soc. 143 (1996), 1114-1123. Winans R. E. and Carrado K. A., Novel forms of carbon as potential anodes for lithium batteries. J. Power Sources 54 (1995), 11-15. Winans R. E., Carrado K. A., Thiyagarajan P., Sandí G., Hunt J. E. and G. W. Zajac, Designer carbons as potential anodes for lithium secondary nd batteries. Proceedings 22 Biennial Conference on Carbon 1 (1995), 812. Sandí G., Winans R. E. and Carrado. K. A., New carbon electrodes for secondary lithium batteries. J. Electrochem. Soc. 143 (1996), L95-98.

332

6.

7. 8. 9.

10. 11. 12.

13.

14.

15. 16. 17. 18. 19.

20.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Sandí G., Carrado K. A., Winans R. E., Brenner J. R. and Zajac G. W., Designer carbons templated by pillared clays: lithium secondary battery anodes. Mater. Res. Soc. Symp. Proc., Macroporous and Microporous Materials 431 (1996), 39-44. Stöhr J., NEXAFS Spectroscopy, Vol. 25 (Springer-Verlag, Berlin, 1992), pp. 199-205. Whittingham M. S., Intercalation Chemistry (Academic, NY, 1982), p. 7. Wainwright D. and Shimizu R., Forces generated by anode growth in cylindrical lithium/molybdenum disulfide cells. J. Power Sources 34 (1991), 31-38. Yoshikawa T., Ogawa A., Okuto T. and Shirakawa S., Battery pack for a portable radiotelegraphic unit. USA Patent 4,910,103 (1990), pp. 1-7. For example see “Cellular Phone Recall May Cause Setback for Moli”, Toronto Globe and Mail, August 15, 1989 (Toronto, Canada). Dahn J. R., Sleigh A. K, Shi H., Reimers J. N., Zhong Q. and Way B. M., Dependence of the electrochemical intercalation of lithium in carbons on the crystal structure of the carbon. Electrochim. Acta 38 (1993), 1179-1191. Tran T. D., Derwin D. J., Zaleski P., Song X. and Kinoshita K., Lithium Intercalation studies of petroleum cokes of different morphologies. J. Power Sources 81-82 (1999), 296-299. Tran T. D., Feikert J. H., Song X. and Kinoshita K., Commercial carbonaceous materials as lithium intercalation anodes. J. Electrochem. Soc. 142 (1995), 3297-3302. Shi H., Barker J., Saiidi M. Y., Koksbang R. and Morris L., Graphite structure and lithium intercalation. J. Power Sources 68 (1997), 291-295. Stocchiero O. and Montorso V., Lid for lead-acid accumulators with elastic and deformable pole seats. USA Patent 4,898,795 (1990). Pierson H. O., Handbook of Carbon, Graphite, Diamond, and Fullerenes (Noyes, Park Ridge, NJ, 1993), p. 43. Hérold A., Insertion compounds of graphite with bromine and the alkali metals. Bull. Soc. Chim. Fr. 187 (1955), 999-1012. Hérold A., Conductivity anisotropy in arsenic pentafluoride-intercalated graphite. In Chemical Physics of Intercalation, Legrand A. P. (ed.), NATO ASI Series, Vol. 172 (1987), p. 3B. Ein-Eli Y., Markovsky D., Aurbach D., Carmeli Y., Yamin H. and Luski S., The dependence of the performance of Li-C intercalation anodes for Li-ion secondary batteries on the electrolyte solution composition. Electrochim. Acta 39 (1994), 2559-2569.

New Anodes for LIB and Their SEI Characterization

333

21. Aurbach D., Ein-Eli Y., Markovsky D., Zaban A., Luski S., Carmeli Y. and Yamin H., The correlation between the surface chemistry and the performance of Li-carbon intercalation anodes for rechargeable "rockingchair" type batteries. J. Electrochem. Soc. 141 (1994), 603-611. 22. Yamaki J., Arakawa M., Tobishima S. and Hirai T., Method to determine lithium cycling efficiency for lithium secondary cells. Proceedings of the Electrochemical Society 87-1 (1987), 266-279. 23. Kominato A., Yasukawa E., Sato N., Ijuuin T., Asahina H. and Mori S., Analysis of surface films on lithium in various organic electrolytes. J. Power Sources 68 (1997), 471-475. 24. Chiang J. C. and MacDiarmid A. G., Polyaniline: protonic acid doping of the emeraldine form to the metallic regime. Synthetic Metals 13(1-3) (1986), 193-205. 25. Aurbach D., Ein-Eli Y., Markovsky B., Zaban A., Luski S., Carmeli Y. and Yamin H., The study of electrolyte solutions based on ethylene and diethyl carbonates for rechargeable Li batteries. II. Graphite electrodes. J. Electrochem. Soc. 142 (1995), 2882-2990. 26. Aurbach D., Markovsky B., Shechter A., Ein-Eli Y. and Cohen H., A comparative study of synthetic graphite and Li electrodes in electrolyte solutions based on ethylene carbonate-dimethyl carbonate mixtures. J. Electrochem. Soc. 143 (1996), 3809-3820. 27. Shu Z. X., McMillan R. S. and Murray J. J., Electrochemical intercalation of lithium into graphite. J. Electrochem. Soc. 140 (1993), 922-927. 28. Besenhard J. O., Wagner M. W., Winter M., Jannakoudakis A. D., Jannakoudakis P. D. and Theodoridou E., Inorganic film-forming electrolyte additives improving the cycling behavior of metallic lithium electrodes and the self-discharge of carbon-lithium electrodes. J. Power Sources 44 (1993), 413-20. 29. Winans, R. E. and Carrado K. A., Novel forms of carbon as potential anodes for lithium batteries. J. Power Sources 54(1) (1995), 11-15. 30. Sandí, G., Winans, R. E. and Carrado K. A., New carbon electrodes for secondary lithium batteries. J. Electrochem. Soc. 143(5) (1996), L95-L98. 31. Sandí, G., Carrado K. A., Winans R. E., Brenner J. R. and Zajac G. W., Designer carbons templated by pillared clays: lithium secondary battery anodes. Materials Research Society Symposium Proceedings 431 (Microporous and Macroporous Materials) (1996), 39-44.

334

Lithium-Ion Batteries: Solid-Electrolyte Interphase

32. Sandí G., Winans R. E., Carrado K. A., Johnson C. S. and Thiyagarajan P., Electrochemical and spectroscopic studies of novel carbonaceous materials used in lithium ion cells. J. New Mater. Electrochem. Syst. 1 (1998), 83-89. 33. Sandí G., Song K., Carrado K. A. and Winans R. E., A NEXAFS determination of the electronic structure of carbons for lithium-ion cells. Carbon 36 (1998), 1755-1758. 34. Sandí G., Thiyagarajan P., Carrado K. A. and Winans R. E., Small angle neutron scattering characterization of the porous structure of carbons prepared using inorganic templates. Chemistry of Materials 11 (1999), 235-240. 35. Sandí, G., Carrado K. A., Winans R. E., Johnson C. S. and Kepler K. D., Carbons for lithium ion cells prepared using sepiolite as inorganic template. Proceedings — Electrochem. Soc. 98-16 (Lithium Batteries) (1999), 11-18. 36. Sandí G., Gerald R. E., Klingler R. J., Rathke J. W., Carrado K. A. and Winans R. E. Studies of electrolyte penetration in carbon anodes by NMR techniques. Proceedings — Electrochem. Soc. 98-16 (Lithium Batteries) (1999), 400-407. 37. Scanlon L. G. and Sandí G., Influence of corannulene's curved carbon lattice (C20H10) on lithium intercalation. Proceedings of the Intersociety Energy Conversion Engineering Conference 223 (1998), 1-6. 38. Scanlon L. G. and Sandí G., Layered carbon lattices and their influence on the nature of lithium bonding in lithium intercalated carbon anodes. J. Power Sources 81-82 (1999), 176-181. 39. Scanlon L. G. and Sandí G., Lithium-endohedral C60 complexes. Proceedings of the Power Sources Conference 38 (1998), 382-385. 40. Gerald R. E., Klingler R. J.; Rathke J. W., Sandí G. and Woelk K., In situ imaging of charge carriers in an electrochemical cell. In Spatially Resolved Magnetic Resonance, ed. by Blümler P., Blümich B., Botto R. and Fukushima E., Chapter 9 (1998), pp. 111-119. 41. Sandí G., Thiyagarajan P., Winans R. E. and Carrado K. A., Small angle neutron and x-ray scattering studies of carbons prepared using inorganic templates. Preprints of Papers — ACS, Division of Fuel Chemistry 42(3) (1997), pp. 854-858. 42. Sandí G., Winans R. E., Carrado K. A. and Thiyagarajan P., Small angle neutron scattering analysis of novel carbons for lithium secondary batteries. In Materials Research Using Cold Neutrons at Pulsed Neutron Sources, ed. by Thiyagarajan P., Trouw F., Marzec B. and Loong C.-K. (World Scientific, Singapore, 1999), pp. 196-200.

New Anodes for LIB and Their SEI Characterization

335

43. Scanlon L. G., Storch D. M., Newton J. H. and Sandí G., A theoretical study of a carbon lattice system for lithium intercalated carbon anodes. Proceedings of the Intersociety Energy Conversion Engineering Conference 32 (1997), 87-92. 44. Winans R. E., Sandí G. and Carrado K. A., Thiyagarajan P. Structure of carbons prepared with pillared clay templates. Proceedings, 23rd Biennial Conference on Carbon 2 (1997), pp. 150-154. 45. Sandí G, Winans R. E., Seifert S. and Carrado K. A., In situ SAXS studies of the structural changes of sepiolite clay and sepiolite-carbon composites with temperature. Chemistry of Materials 14(2) (2002), 739-742. 46. Sandí G., Gerald R. E., Scanlon L. G., Carrado K. A. and Winans R. E., Computational, electrochemical and 7Li NMR studies of lithiated disordered carbons electrodes in lithium ion cells. Materials Research Society Symposium Proceedings (Materials for Electrochemical Energy Storage and Conversion II — Batteries, Capacitors and Fuel Cells) 496 (1998), 95-100. 47. Dahn J. R, Sligh A. K, Shi H, Way, B. W., Weydanz W. J., Reimers J. N., Zhong Q. and von Sacken U., Carbons and graphite as substitute for the lithium anode. In Lithium Batteries-New Materials, Developments and Perspectives, ed. by Pistoia G. (Elsevier, Amsterdam, 1994), pp. 1-30. 48. Sandí G., Carrado K. A., Winans R. E., Johnson C. S. and Csencsits R., Carbons for lithium battery applications prepared using sepiolite as an inorganic template. J. Electrochem. Soc. 146 (1999), 3644-3648. 49. Fauteux D. and Koksbang R., Rechargeable lithium battery anodes: alternatives to metallic lithium. J. Applied Electrochem. 23(1993), 1-10. 50. Idota Y., Kubota T., Matsufuji A., Maekawa Y. and Miyasaka T., Tin-based amorphous oxide: a high-capacity lithium-ion-storage material. Science 276 (1997), 1395-1397. 51. Kepler K. D., Vaughey J. T. and Thackeray M. M., LixCu6Sn5 (0 < x < 13): an intermetallic insertion electrode for rechargeable lithium batteries. Electrochem. Solid State Lett. 2 (1999), 307-309. 52. Yang J., Takeda Y., Imanishi N. and Yamamoto O., Ultrafine Sn and SnSb0.14 powders for lithium storage matrices in lithium-ion batteries. J. Electrochem. Soc. 146 (1999), 4009-4013. 53. Zhao X. B. and Cao G. S., A study of Zn4Sb3 as a negative electrode for secondary lithium cells. Electrochim. Acta 46 (2001), 891-896. 54. Crosnier O., Brousse T., Devaux X., Fragnaud P. and Schleich D. M., New anode systems for lithium ion cells. J. Power Sources 94 (2001), 169-174.

336

Lithium-Ion Batteries: Solid-Electrolyte Interphase

55. Nam S. C., Yoon Y. S., Cho W. I., Cho, B. W., Chun H. S. and Yun K. S., Enhancement of thin film tin oxide negative electrodes for lithium batteries. Electrochem. Communications 3 (2001), 6-10. 56. Belliard F., Connor P. A. and Irvine J. T. S. Novel tin oxide-based anodes for Li-ion batteries. Solid State Ionics 135(2000), 163-167. 57. Limthongkul P., Wang H., Jud E. and Chiang Y-M., Metal oxide composites for lithium-ion battery anodes synthesized by the partial reduction process. J. Electrochem. Soc. 149 (2002), A1237-A1245. 58. Li C. and Stair P. C., Ultraviolet Raman spectroscopy characterization of sulfated zirconia catalysts: fresh, deactivated and regenerated. Catalysis Lett. 36(3, 4) (1996), 119-23. 59. Chen J. G., Fischer D. A., Hardenbergh J. H. and Hall R. B., A fluorescence-yield near-edge spectroscopy (FYNES) investigation of the reaction kinetics of nickel oxide/nickel(100) with hydrogen. Surf. Sci. 279(1-2) (1992), 13-22. 60. Sandí G., Winans R. E., Carrado K. A. and Johnson C. S., Novel carbonaceous materials for lithium secondary batteries. Proceedings, 2nd International Symposium on New Materials for Fuel Cells and Modern Battery System 1(1997), pp. 415-418. 61. Horsley J. A., Stohr J., Hitchcock A. P., Newbury D. C., Johnson A. L. and Sette F., Resonances in the K shell excitation spectra of benzene and pyridine: gas phase, solid, and chemisorbed states. J. Chem. Phys. 83(12) (1985), 6099-7007. 62. Zheng T., Liu Y., Fuller E. W., Tseng S. von Sacken U. and Dahn J. R., Lithium insertion in high capacity carbonaceous materials. J. Electrochem. Soc. 142(8) (1995), 2581-2590. 63. Roethlisberger U. and Klein M. L., Ab initio molecular dynamics investigation of singlet C2H2Li2: determination of the ground state structure and observation of LiH intermediates. J. Am. Chem. Soc. 117(1) (1995), 42-48. 64. Chen J.G., De Vries B. D., Lewandowski J. T. and Hall R., Direct differentiation of surface and bulk compositions of powder catalysts: application of electron-yield and fluorescence-yield NEXAFS to LixNi1xO. Catalysis Lett. 23(1-2) (1993), 25-35.

CHAPTER 8

THE CATHODE-ELECTROLYTE INTERFACE IN A Li-ION BATTERY KRISTINA EDSTRÖM, TORBJÖRN GUSTAFSSON and JOSH THOMAS* Ångström Advanced Battery Centre, Department of Materials Chemistry, Ångström Laboratory, Uppsala University, Box 538, SE-751 21 Uppsala, Sweden *

E-mail: [email protected]

The Solid-Electrolyte Interphase (SEI) layer formed on cycling the graphite anode of a Li-ion battery has been studied extensively in recent years since the first suggestion of its existence by Peled more than twenty years ago; see elsewhere in this book. More recently, the cathode-electrolyte interface has also come to attract considerable attention.

1

Background

It is now realised that a complex surface chemistry is also decisive in determining the long-term viability of the commonly exploited LixMOy 1, 2 cathodes, where M=Mn, Co or Ni, and binary or ternary mixtures thereof. The major difference between the anode and cathode situations is the more significant first-cycle irreversible capacity loss for a graphite anode than that found for any of the more typical cathode materials in current use. Nevertheless, classical Electrochemical Impedance Spectroscopy (EIS) techniques have been able to identify the formation of some type of surface film on LiMOx-based 3-5 cathodes as the source of observed impedance increase on cell cycling, and could thus confirm that cathode-electrolyte interactions can indeed be an important factor in determining long-term cycling and thermal stability in + cathodes. The implication is clear then that Li ions must also travel through an additional (SEI-type) layer between cathode and electrolyte — a process which could even prove rate-limiting if the surface species so formed were poor ion+ conductors and Li -ion diffusion through the electrolyte and bulk electrode material were fast. In a cathode context, electrolyte oxidation has come to be seen as a major cause of film formation. The process could, in turn, be driven by 337

338

Lithium-Ion Batteries: Solid-Electrolyte Interphase 4+

the reduction of unstable M ions in the active electrode material. The case of LiMn2O4 (see below) illustrates this well, where corrosion leads to dissolution of 2+ 6 Mn ions into the electrolyte and subsequent electrolyte oxidation. Highvoltage electrolyte oxidation processes of this type have been shown to result in 7 insoluble reaction products. However, it has not so far been possible to identify the oxidation products satisfactorily, and the related reaction mechanisms therefore remain unclear. In this chapter, we shall attempt to shed some light on this situation by relating surface chemistry to some aspects of battery performance for today’s most commonly used cathode materials; namely, LiMn2O4, LiCoO2, LiNiO2 and LiNi1-xCoxO2, and for a recent newcomer on the scene, LiFePO4.

2

LiMn2O4 3

Many different Li-Mn-O phases exist. We shall here focus solely on LiMn2O4, which is the member of the group most commonly exploited as a Li-ion battery material. The material has a number of advantages: it is cheap (Mn is a relatively abundant metal in earth’s crust), environmentally friendly, and has an acceptably high cycling capacity (148 mAh/g). There are, however, some wellknown problems with this material, the most critical of which is capacity fade observed during cycling, even at ambient temperature. The underlying reason for this has been probed extensively, and it is now believed to be the surface chemistry of the electrode/electrolyte interface. Conventional wisdom holds that Mn is lost from the spinel electrode surface into the electrolyte following 8 3+ disproportionation of trivalent Mn through the Hunter reaction: 2Mn solid  → 4+ 2+ Mn solid+Mn solution. This reaction is dependent on the acidity of the electrolyte, so 9 the choice of salt and solvent is also important in this respect. Jang and Oh report that the amount of spinel dissolved decreases in the order: LiCF3SO3 > LiPF6 > LiClO4 > LiAsF6 > LiBF4. The dissolution of divalent Mn into the electrolyte will not only lead to a decrease in the effective amount of active 2+ cathode material but also influence the anode. Solvated Mn ions are transported through the electrolyte and are ultimately deposited both on and in the graphite anode. The Li content in the anode (and hence the overall cell capacity) is thus also depleted, since the reduction of Mn will oxidize Li from the anode according to: 2+

Mn + 2 LiC6

 → Mn + 2 Li+ + graphite

(1)

The Cathode-Electrolyte Interface in a Li-Ion Battery

339

Other mechanisms have also been proposed which are not intrinsically surface chemical, e.g., bulk-structure instability of the delithiated λ-MnO2 10 phase, self-discharge effects, and loss of oxygen from the host structure as a 11, 12 result of solvent oxidation in the high-voltage region.

Figure 1 XPS spectra for a spinel electrode after 50 cycles at C/3; electrolyte: EC/DMC-LiPF6; solid line: prior to Ar-ion sputtering; dotted line: after 60 s Ar-ion sputtering. Reproduced from [31] with permission from The Electrochemical Society Inc.

Nevertheless, the impression remains that a more detailed understanding of surface-related processes is the key to improving spinel-electrode stability. This is especially true at slightly elevated temperatures, where capacity-fade effects are known to be larger; a major area of application for Li-ion batteries is in electronic circuits, e.g., cell-phones and lap-tops, where considerable heat is generated. Tarascon and co-workers have, in fact, shown that the mechanisms responsible for capacity fade at moderate elevated temperatures are essentially 13-17 the same as those at ambient temperature. Let us proceed then by describing

340

Lithium-Ion Batteries: Solid-Electrolyte Interphase

some of our recent efforts aimed at attaining an improved understanding of surface phenomena in relation to the spinel-electrolyte interface.

2.1

XPS Analysis

It has been found that the most powerful technique for probing the chemical nature of the electrochemical interface in a Li-ion battery context is X-ray Photon Spectroscopy (XPS); surface XRD or Raman spectroscopy are considerably less versatile in this respect. Although a number of XPS studies 18-25 have been performed on different Mn-oxides, there is a noticeable lack of XPS surface investigations for the LiMn2O4 system. For this reason, we shall focus particularly on results from XPS studies in this discussion. Examples of typical XPS spectral bands are given in Figure 1. Some comments follow to acquaint the reader with the types of feature found in such spectra; a more complete summary of relevant XPS peak assignments is given in Table 1: • •

•

• •

•

The F1s peak is mainly from LiF, but there is also a contribution from an LixPFy-type compound. Mn2p gives a somewhat weak signal since the bulk spinel material is effectively coated with an organic/inorganic surface film; it has not been possible to determine the valence state of the surface Mn from these measurements. The O1s spectrum has a sharp feature around 529.5 eV originating from the LiMn2O4 oxygen, and two bands associated with the surface film. A strong band occurs around 533 eV originating from both organic and inorganic material, typically poly(oxyethylene) 26-28 and a phosphorous compound.29, 30 C1s has a major bands at 284.3 eV from the nanocrystalline turbostratic graphite regions in the carbon black used in the cell. P2p has a broad feature with more than one P oxidation state. The higher binding-energy (BE) band is from the LixPFy compound and the lower from some phosphorous oxide compound. Li1s is dominated by the contribution from LiF.

It can also be of interest to illustrate an experimental finesse referred to in the Figure 1 figure text, which has proven to be of great value in post-mortem

The Cathode-Electrolyte Interface in a Li-Ion Battery

341

analyses of Li-ion batteries after different storage and cycling treatments, namely Ar-ion sputtering. Although considerable experience and care are needed to use this technique to its full advantage (mainly because of the risk of incurring spurious sample damage through the ion-irradiation treatment itself and also because of surface charging effects), it is nevertheless clear that it can provide depth-analysis information relating to an electrochemical surface which is difficult to attain by any other method. The method is demonstrated in Figure 2, where depth profiles are shown for cycled and stored spinel electrodes. Table 1 XPS peak assignments in elemental spectra from LiMn2O4 surfaces. Surface component C (graphite) C (amorph.) polyether (PEO) ROCO2Li Li2CO3 LixPFy P2O5 LiF LixBFy LiMn2O4

C1s

O1s

F1s

P2p

B1s

Mn2p3/2

Li1s

(eV)

(eV)

(eV)

(eV)

(eV)

(eV)

(eV)

284.3 284.8 285.5

532.5-533.5

286-287 290-291 290-291

532-533 534-535 531-532.5

55 55 687-688

533.5 531.7 685-686 687-688 529.5 531.5-532

137-138 135.5 56.5-57 195 642

The example serves to show an interesting feature: the relative amounts of the various elements in the two surface films formed (under electrochemical cycling and chemical storage) are essentially identical. A small increase in the amount of Li and F is observed just below the surface, while the C content is seen to decrease. This gives us the first indication of the existence of a layered structure in the surface film. As we move into the sample, the F, O, P and Li concentrations decrease, while the relative amounts of Mn and C increase, as more of the bulk spinel electrode is exposed. The amount of surface material present is also found to increase with cycle number and storage time, suggesting that the layer formed on the LiMn2O4 cathode is not sufficiently dense to serve as a barrier between the electrolyte and the oxidising environment close to the cathode surface. Fresh electrolyte can be transported to the electrode surface,

342

Lithium-Ion Batteries: Solid-Electrolyte Interphase

and the oxidation process is able to continue. We can contrast this with the situation on the graphite anode, where the SEI layer formed during the first discharge covers the electrode surface so as to prevent further reduction of the electrolyte in subsequent cycles. This general observation has motivated us to propose that the cathode surface-layer formed is more appropriately termed a Solid Permeable Interface (SPI) rather than a Solid-Electrolyte Interphase (SEI), which would imply some form of passivating function. The cathode-electrolyte interface can thus limit cell performance by consuming electrolyte continuously as cell cycling proceeds. This can also be seen from cycling curves obtained for LiMn2O4 cells, where the accumulated charge during LiMn2O4 oxidation is often in excess of the charge accumulated in the reduction process, implying some 31 side-reaction occurring in the cell, such as electrolyte oxidation. The formation of organic rather than inorganic compounds closest to the surface would appear 6 to limit the ability of the SPI to passivate the electrode.

Figure 2 XPS depth profiles obtained by Ar-ion sputtering of spinel electrodes in an EC/DMCLiPF6 electrolyte: (A) After 50 cycles at C/3 and ambient temperature – total cycling time 300 h; (B) electrodes stored for 300 h in same electrolyte at same temperature. Reproduced from [31] with permission from The Electrochemical Society Inc.

An interesting observation has also been made concerning the influence of state-of-charge (SOC) on the nature of the surface film formed. The surface of a + spinel-based electrode stored for 300h at SOC 100% (4.3 V vs. Li/Li ) was + compared to that of an electrode stored at SOC 0% (3.4 V vs. Li/Li ). Whereas one might expect the surface film to thicken as the potential increases, in fact, the quantity of surface species was found to decrease at the higher SOC. A possible explanation for this counter intuitive result would be the occurrence of more complete oxidation of the solvents, resulting in gaseous end-products such

The Cathode-Electrolyte Interface in a Li-Ion Battery

343

as CO2. It was also found that, when LiBF4 was used as electrolyte salt, additional small C1s XPS bands appeared at 287-289 eV corresponding to ether- and carbonate-carbonyl groups, and suggesting a surface film containing 6 more organic material than for LiPF6. The question thus arises quite naturally as to the nature of the surface structure of the active spinel particles in the electrode under cycling. XRD 32 measurements on reference samples of Li2Mn2O4 clearly show that LiMn2O4 still remains and that the Li-rich phase is therefore not phase-pure Li2Mn2O4 (see Figure 3). In spite of such clear evidence, an XPS analysis probing Mn content (Figure 4) nevertheless reveals that the surface involves only oxidation 3+ state Mn . Although the electrode was allowed to equilibrate before disassembly, we have also seen that Li remains close to the surface. The situation is further complicated by the fact that approx. 20% of the Li is known 33-35 still to be present in the λ-MnO2 phase, implying that the delithiated λ-MnO2 4+ reference could not be assigned to pure Mn .

Figure 3 Ex situ XRD profiles of spinel-based electrodes after cycling or storage at ambient temperature. Dots: Li2Mn2O4 reflections; diamonds: Al current collector. (A) Discharged to 2.4 V; (B) pristine LiMn2O4; (C) charged to 4.3 V; (D) cycled 50 times at C/3; (E) stored for 300 h at OCV. Reproduced from [32] with permission of The Electrochemical Society Inc.

344

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Analysis of the Mn core-level spectra (Table 2) from the XPS 32 3+ measurements on cycled electrodes reveals an increase in the amount of Mn , thus also supporting the literature results from TEM measurements, where tetragonal Li2Mn2O4 was found on the electrode surface after high-rate 36 charge/discharge of electrodes. An even more reduced average valence-state is found on storage. The structures suggested as being formed on storage all have 13,17 3+ an average valence-state > 3.5. The Mn disproportionation process makes 3+ an Mn -rich phase unstable, which will eventually decompose into a more 4+ 2+ stable Mn compound and an Mn compound soluble in the organic 8, 37 3+ electrolyte. It could be confirmed, however, that an Mn -rich compound is 4+ formed as an intermediate step in the decomposition of LiMn2O4 into an Mn containing compound. This, in turn, promotes oxidation of the electrolyte. The Li-content is found to increase significantly in the cycled and stored electrodes compared to a pristine electrode. It is difficult, however, to ascertain how much of this Li can be related to electrolyte-derived compounds, and how much is present in the Mn-compound. Although the electrode was washed thoroughly before the analysis, the SPI could not be removed completely. Table 2 Binding energies (in eV) derived from Mn core-level spectra. Mn2p3/2 Li2Mn2O4 LiMn2O4 λ-MnO2 50 cycles at C/3 Stored for 300 h

641.4 641.4 642.4 641.6 642.4 641.5 642.5 641.4 642.4

Mn2p1/2 satellite

Mn3p

∆E (Mn3s)

664.2 665.2

4.96 5.63

665.5

48.5 48.4 49.8 48.4 49.7 48.5

664.5

49.9 48.4

666.0

Mn3+/4+ Mn2p3/2

Mn3+/4+ Mn3p

529.5 529.4

− 0.91

− 0.99

4.64

529.6

0.22

0.26

5.24

529.7

1.19

1.16

5.38

529.5

1.19

1.28

O1s

49.9

2.2 Elevated Temperature Effects for LiMn2O4 It is very evident from our systematic studies that the effect of increased temperature on the electrode-electrolyte interface is critical to the functionality and safety of an electrochemical battery system. This has earlier been

The Cathode-Electrolyte Interface in a Li-Ion Battery

345

38

investigated for the graphite anode SEI; we probe this here for the case of the 6 spinel-related cathode surface layer. DSC measurements on electrodes at different SOC’s (Fig. 5) have shown that the temperature for the reactions between the active electrode material and the electrolyte increases as the Licontent of the sample increases: from 120°C for λ-MnO2 to 150°C for LiMn2O4, and as much as 160°C for electrodes cycled or stored at 60°C, indicative of an even more lithiated sample.

A A

B

C

Intensity (arb. units)

Intensity (arb. units)

B

D

C

D

E E

670

665

660

655

650

Binding energy (eV)

645

640 52

50

48

46

Binding energy (eV)

Figure 4 XPS Mn2p(left) and Mn3p(right) peaks from measurements on spinel-based electrodes. (A) Discharged to 2.4 V; (B) pristine LiMn2O4; (C) charged to 4.3 V; (D) cycled 50 times at C/3; (E) stored for 300 h at OCV. Solid lines represent individual fitted peaks and the total fit; experimental data is represented by solid squares. Reproduced from [32] with permission of The Electrochemical Society Inc.

346

Lithium-Ion Batteries: Solid-Electrolyte Interphase

λ -MnO2

1

0 SOD 25% 1

0 SOD 50%

Heat Flow (W/g)

1

0 SOD 75% 1

0 SOD 100% 1

0 SOD 100% (cycled) 1

0 100

150

200

250

300

o

Temperature ( C)

Figure 5 DSC measurements on a series of spinel-based electrodes from cells stored at different SOC’s. Bottom curve: for an electrode cycled 50 times at C/3 using an LiPF6-based electrolyte. Reproduced from [6] with permission of Amer. Chem. Society.

The Cathode-Electrolyte Interface in a Li-Ion Battery

A

A

B

B

C

C

D

D

E

E

F

F

G

G

538 292

347

290

288

286

284

Binding energy (eV)

282

280

536

534

532

530

528

526

Binding energy (eV)

Figure 6 XPS C1s and O1s peaks for spinel-based electrodes. (A) Pristine electrode; (B) cycled 50 times at C/3 and ambient temperature, LiPF6-electrolyte; (C) cycled 50 times at C/3 and 60°C, LiPF6-electrolyte; (D) stored for 300h at 60°C, SOC 0%, LiPF6-electrolyte; (E) stored for 300 h at 60°C, SOC 100%, LiPF6-electrolyte; (F) cycled 50 times at C/3 and 60°C, LiBF4-electrolyte; (G) stored for 300h at 60°C, SOC 0%, LiPF6-electrolyte, washed in DMC. Reproduced from [6] with permission of Amer. Chem. Society.

348

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Additional C1s XPS peaks at 286.8 and 288.8-289.4 eV are observed in stored and cycled elevated-temperature samples (Fig. 6). These originate most likely from distorted carbonate groups involved in polymerized carbonates 13 formed by direct polymerization of EC and initiated by EC oxidation. The C1s peak from poly(oxyethylene) is also visible at 285-285.5 eV. Another possibility is the direct polymerization of EC initiated by strong Lewis acids, 39, 40 e.g., PF5 (and BF3), into a poly(oxyethylene-alt-ethylene carbonate). Such polymers have been shown to exist in the graphite SEI layer, especially in 38, 41 connection with chemical lithiation and storage at elevated temperature. A more exact definition of the types of polymer formed has not so far been possible. Note that considerably more polymeric species are formed using LiBF4 as electrolyte salt than with LiPF6 (Figures X.6F), possibly because BF3 is a stronger initiator of polymerization. The low BE O1s peak around 533 eV dominates for electrodes cycled or stored at 60°C; this can be related to the large amount of P in the surface film and the formation of some phosphorous oxide compound. Polycarbonate-poly(oxyethylene) peaks also appear at 532-533 eV and 534-534.5 eV, respectively. Corresponding comparative XPS P2p spectra for samples cycled at ambient and elevated temperatures (not shown here) reinforce the picture of an increased formation of LixPOyFz rather than of LixPFy compounds with increased temperature.

Figure 7 XPS elemental depth-profile of LiMn2O4 electrodes. (A) cycled 50 times C/3 at 60°C, LiPF6 electrolyte; (B) stored at 300 h and 60°C, SOC 0%, LiPF6 electrolyte. Reproduced from [6] with permission of Amer. Chem. Society.

The Cathode-Electrolyte Interface in a Li-Ion Battery

349

Post-mortem XPS depth-profiling analyses have also been made using Arion sputtering of electrodes extracted from Li-ion batteries cycled under different elevated-temperature storage and cycling conditions. The same striking resemblance is seen between the surface chemistry of electrodes cycled 50 times at C/3 and electrodes stored for 300 h at 60°C (Figure 7); the influence of electrochemical cycling on surface-film formation is clearly as negligible at elevated as it was at ambient temperature. The corresponding elemental analysis is given in Table 3. It is very evident that there is an excess of O, P, F and Li (especially O and P) after cycling or storage at SOC 0%, 60°C compared to other samples measured. Some of the P-O products can be removed through thorough washing with DMC, and a surface containing more of the polymeric carbon species is exposed. After storage at SOC 100% (λ-MnO2), the P-O products are less pronounced and, instead, more carbon products can be observed. The same goes for electrodes cycled in a LiBF4-based electrolyte (B substituted for P). The smaller quantities of B- compared to P-oxides can be explained by the higher instability of the LiPF6 salt, making it more sensitive to impurities in the electrolyte. The observed Li concentration increases after washing, indicative of a Li-rich layer under the outer P-O layer, where LiF is the most common product. Notably, an Mn signal is barely detectable in the 60°C samples, and this does not become stronger with washing; the polymeric/polycarbonate species clearly cover the surface too efficiently. The surface-film thickness thus increases as the temperature is increased. While an anode SEI-layer seems to break down at a slightly elevated temperature and must be reformed in the 38 subsequent reduction cycle, the cathodic surface layer increases in thickness with storage-time, cycle-number and temperature. On the basis of both ambient and elevated temperature observations, a number of general mechanisms emerge relating to the formation of an SPI at elevated temperatures: (A) A layer of polymer/polycarbonate forms closest to the active electrodeparticle surface during cycling or storage at SOC 0%, then comes a layer of LiF, and outermost a region containing LixPOyFz and P-oxides (alternatively, LixBFy and B-oxides for an LiBF4-based electrolyte). (B) Mn dissolution from the spinel material occurs in the lithiated state, where 3+ 8 the disproportionation of Mn proceeds by Hunter’s reaction:

350

Lithium-Ion Batteries: Solid-Electrolyte Interphase

2 LiMn2O4

 → 3 λ-MnO2 + MnO + Li2O

MnO + 2 H

+

(2)

 → Mn2+ + H2O

(3)

2+

(C) Mn then forms complexes with the solvent molecules in the electrolyte and goes into solution; Li2O of Eq. (6) can react directly with the P-compounds 42 formed from the decomposition of LiPF6 to form P2O5 according to the sequence LiPF6

← → LiF + PF5

PF5 + H2O

(4)

 → 2HF + POF3

2 POF3 + 3 Li2O

(5)

 → P2O5 + 6 LiF

(6)

Table 3 Atomic percentage (at. %) of elements on the surface of LiMn2O4 electrodes, EC/DMC 2:1 1 M LiPF6 or LiBF4. The carbon content given in parentheses is the value after subtraction of the graphite contribution, i.e., surface-film carbon only.

Sample

LiMn2O4, 300h storage, 60°C, LiPF6 LiMn2O4, 300h storage, 60°C, LiPF6 λ-MnO2, 300h storage, 60°C, LiPF6 LiMn2O4, 300h storage,

Element C1s

O1s

P2p

29.1

22.6

25.6

46.3

42.9

52.0

RT, LiPF6

Mn2p

7.6

30.5

9.9

0.3

24.1

9.3

31.5

9.3

0.2

18.0

5.1

23.8

6.5

0.3

11.8

3.1

22.8

19.1

0.3

16.4

6.2

0.2

20.9

16.2

0.9

0.6

2.7

(30.6)

(24.6)

LiMn2O4, pristine

Li1s

(15.4)

60°C, LiPF6, washed C/3, 60°C, LiBF4

F1s

(17.8)

LiMn2O4, 50 cycles, LiMn2O4, 50 cycles, C/3,

B1s

19.9

5.3

(37.3) 55.3

5.7

1.0

(18.2) 90.3

6.4

3+

P-oxide formation is thus driven by the disproportionation of Mn , which explains the surplus of P-oxides on SOC 0% compared to SOC 100% samples.

The Cathode-Electrolyte Interface in a Li-Ion Battery

351

This would also explain the increase in P-oxides with temperature, since the 13 dissolution of Mn increases at elevated temperatures. For the LiBF4 case, a smaller amount of B-oxides is formed due to the higher stability of the salt. (D) There is a stronger driving force for the formation of polymerpolycarbonate compounds at SOC 100% than at SOC 0%. Electrolyte oxidation, typically in the form of solvent polymerisation, can clearly be coupled to Liinsertion into the λ-MnO2 host structure, and the driving force for electrolyte 11 (El.) oxidation is given by λ-MnO2 + x Li + x El. +

 → LixMn2O4 + x El.+

(7)

This process may well occur in parallel with electrolyte oxidation, coupled to a loss of oxygen from the spinel structure at the highly delithiated spinel 12 surfaces according to LiyMn2O4 + 2 δ El.

 → LiyMn2O4-δ + δ (oxidized El.)2

(8)

(E) The formation of the lithium-rich surface can be coupled to the Mn disproportionation reaction; the product must be of Li1+xMn2-yO4 type, where y < 17 2x: +

LiMn2O4 + 3x Li + x El.

 → Li1+3xMn2-xO4 + x Mn2+ + x El.+

(9)

However, as we saw from the Mn XPS results, the electrodes have an Mn -rich surface, so the Li-insertion reaction can thus drive the electrolyte oxidation process even during storage in the lithiated state without loss of Mn. 32 These processes are summarized in the form of a schematic model for the formation of the Solid Permeable Interface (SPI) in Figure 8. How then can the formation of an SPI and on the positive electrode be avoided? A logical approach would be to prefabricate an inorganic barrier on the cathode particles which resemble the SEI layer formed on a graphite anode. This barrier would need to be impenetrable for solvent molecules but permit Li diffusion. There have so far been only a few attempts to achieve this; some have been found to work well in the early stages of cycling, but do not improve 16, 43-45 prolonged cycling performance. The introducion of small amounts of inorganic material into the electrolyte has improved capacity retention, probably by reaction of the additives with the HF in the electrolyte, thereby lowering the 43 HF concentration. Coating the cathode particles with Li2CO3 lowers the acidity 3+

352

Lithium-Ion Batteries: Solid-Electrolyte Interphase 44

45

of the system, while coating with Al2O3 has improved stability of Li-Mn-O. More effort is needed to render pure LiMn2O4 a viable commercial Li-ion cathode material.

Figure 8 A proposed model for the SPI surface layer formed on a LiMn2O4 electrode. Reproduced from [6] with permission of Amer. Chem. Society.

Meanwhile much progress has been made by substitution: mono-, di- or trivalent cation substitution of Mn increases the average Mn oxidation state. 3+ This results in the reduction of unstable Mn cations and hence a material

The Cathode-Electrolyte Interface in a Li-Ion Battery

less susceptible to the disproportionation reaction. 49 alternative approach.

3

14, 46-48

353

Anion doping is an

LiCoO2, LiNiO2 and LiNi0.8Co0.2O2

LiCoO2 is the most common cathode material used today in commercial Li-ion batteries by virtue of its high working voltage, structural stability and long 50 cycle-life. However, Co is relatively expensive and the cheaper Mn material suffers from the instability problems described above. Much effort has therefore been made in recent years to find cheaper alternatives. LiNiO2 (isostructural with LiCoO2) is a promising materials in this respect, but has not been commercialised successfully for several reasons: i) difficult synthesis 51 52 conditions, ii) poor structural stability on electrochemical cycling, and iii) 4+ 53 poor thermal stability in its delithiated state as a result of the unstable Ni ion. These problems can be circumvented by partially substitution of Ni by other cations, typically Co. The relative performances of the Li(Ni,Co)O2 family of materials are compared with those of spinel in Table 4: Table 4 Comparison of the performance of various cathode materials.

Property Practical capacity Cycling stability High-T stability

LiCoO2 150 Ah/kg good good

LiNiO2 170 Ah/kg

LiNi0.8Co0.2O2

LiMn2O4

180 Ah/kg

120 Ah/kg

good

a

good

poor

good

a

good

Poor

Power capability

best

good

good

average

Safety

good

poor

unclear

best

Toxicity

poor

poor

poor

best

Material cost

high

acceptable

acceptable

best

a

Under optimised conditions

Co-substituted LiNiO2 (resulting in LiNi1-xCoxO2) has the advantage of combining the favourable properties of LiNiO2 and LiCoO2 and yet also have a higher structural stability than the pure Ni-oxide and a potentially lower cost 54, 55 than LiCoO2. Its crystal structure (Fig. 9) is of the α-NaFeO2-type (spacegroup: R 3 m ), which is a layered, rhombohedral structure in which the lithium

354

Lithium-Ion Batteries: Solid-Electrolyte Interphase

ions can move quite freely in the two-dimensional planes perpendicular to 2the c-axis. The O ions form a close-packed face-centred-cubic (fcc) structure, and the Li, Ni and Co ions occupy the octahedral voids in alternating (111) planes. In this structure, ~0.7 Li can be extracted and inserted during the charge and discharge cycles, corresponding to a capacity of ~190 mAh/g. Further 56, 57 extraction leads to irreversible collapse of the structural framework. The electrochemical charging process is described by the electrode reaction Li(Ni,Co)O2 → Li1–x(Ni,Co)O2 + x Li + x e +

–

(10) 58, 59

LiCoO2 and LiNiO2 both tend to be coated with a layer of Li2CO3, which can be a target for reactions with HF and the subsequent formation of surface LiF in accordance with 2HF + Li2CO3 → 2LiF + H2O + CO2

(11)

c

c

Li Co, Ni O

(3b) (6c)

a

(3a)

a Figure 9 The crystal structure of layered Li(Ni,Co)O2 viewed from two different directions.

Both LiCoO2 and LiNiO2 are also more reductive in the presence of solvent species than LiMn2O4; there is also a higher concentration of LiF on their surfaces as compared to the spinel when exposed to LiPF6 solutions. It is general

The Cathode-Electrolyte Interface in a Li-Ion Battery

355

for this group of materials that the impedance increases considerably on cycling or storage at elevated temperatures; this would seem to be the major cause for capacity-fading.

PVdF

PVdF

14000

1500 10000

C1s

O1s

2000

500

0

294

292

290

288

286

284

282

538

Binding Energy (eV)

8000 6000 4000 2000 0

536

534

532

530

528

-2000 694

526

Binding Energy (eV)

200

1200

50

688

686

684

682

Co2p

1000

Intensity (cps)

Intensity (cps)

3000

100

690

1400

Ni2p

150

692

Binding Energy (eV)

4000

Li1s

F1s

10000

Intensity (cps)

4000

0

Intensity (cps)

12000

1000

Intensity (cps)

Intensity (cps)

8000 6000

PVdF LiF

Li2CO3 LiNi0.8Co0.2O2

graphite

2000 1000

800 600 400 200

0

0 60

55

Binding Energy (eV)

50

890

0 880

870

860

850

Binding Energy (eV)

840

800

790

780

770

Binding Energy (eV)

Figure 10 C1s, O1s, F1s, Li1s, Ni2p and Co2p XPS spectra for a fresh (uncycled) LiNi0.8Co0.2O2 cathode laminate. Reproduced from [70] with permission of The Electrochemical Society Inc.

In this section, we shall focus primarily on the results and implications of recent XPS surface analyses of LiNi0.8Co0.2O2-based cathodes. The studies are here made on LiNi0.8Co0.2O2 powder, fresh laminates (with no previous contact with electrolyte), and cycled or stored electrodes. Measurements on the fresh powder and the laminate reveal that Li2CO3 is present on the LiNi0.8Co0.2O2 particle surface (Figure 10), thus confirming previous investigations on 58, 60 The Li2CO3 is believed to form through the reaction of atmospheric LiNiO2. CO2 with LiNi0.8Co0.2O2 according to Li(Ni,Co)O2 + x/2 CO2 + x/4 O2 → Li1-x(Ni,Co)O2 + x/2 Li2CO3

(12)

Fresh laminate also shows a small impurity of LiF not observed on fresh anode laminates. This is believed to be formed as a consequence of a dehydro-

356

Lithium-Ion Batteries: Solid-Electrolyte Interphase

fuorination reaction in the PVdF binder, generating HF, which then reacts with 60 LiNi0.8Co0.2O2 (or Li2CO3) to form LiF.

polycarb. -CH2CH2-

Li2CO3 LiNi0.8Co0.2O2

3000 6000

C1s

2500

O1s

6000

1500 1000 500

4000 3000 2000 1000

0

0 294

Intensity (cps)

Intensity (cps)

Intensity (cps)

2000

F1s

5000

2000 4000

PVdF LiF 7000

0 292

290

288

286

284

282

-500 538

280

536

Binding Energy (eV)

LixPFy

534

532

530

528

526

694

Binding Energy (eV)

692

690

688

686

684

682

Binding Energy (eV)

LixPFyOz 400

300

P2p

Li1s

6000

Ni2p

200

100

0 144

Intensity (cps)

Intensity (cps)

Intensity (cps)

300 200 100 0 142

140

138

136

134

Binding Energy (eV)

132

62

4000

2000

0 60

58

56

54

52

Binding Energy (eV)

50

48

890

880

870

860

850

Binding Energy (eV)

Figure 11 C1s, O1s, F1s, P2p, Li1s and Ni2p XPS spectra for a typical stored LiNi0.8Co0.2O2 cathode tested at 40°C, SOC 60%. Reproduced from [70] with permission of The Electrochemical Society Inc.

The corresponding XPS spectra for a 40°C, SOC 60% SOC stored electrodes are shown in Figure 11. The new compounds formed on storage are proposed be to organic polycarbonates and polymeric hydrocarbons; salt-based products are also observed. LiF is the main product, but decomposition products of the type LixPFy and LixPFyOz can also be identified. Spectral peaks for electrodes stored at 40, 50, 60 and 70°C are similar, indicating that the same surface compounds are formed; nor are changes observed in the LiNi0.8Co0.2O2 peak positions in the Ni2p and O1s spectra with time, temperature or SOC. The relative intensities of the various components formed are, however, affected by temperature; the organic species increase, while the inorganic salt-based

The Cathode-Electrolyte Interface in a Li-Ion Battery

357

(F-containing) species appear to decrease, probably due to screening by an overlying organic surface layer. The formation of surface species on LiNi0.8Co0.2O2 cathodes can have a number of origins. Purely organic polycarbonates appear to be the major carbonate component. The observed self-discharge (through the transport of lithium into the cathode) can be evidence of electrolyte (El.) oxidation according to the general equation for cathode materials undergoing self61 discharge on storage: LixNi0.8Co0.2O2 + y Li + y El. → Lix+yNi0.8Co0.2O2 + y El. +

+

(13)

+

The implication of “El. ” here is somewhat unclear, however: the removal of an electron from EC could, for example, provoke a polymerisation process and hence explain the presence of polycarbonate on the surface. Polycarbonates have also been observed on LiMn2O4 electrodes, but in relatively larger amounts 6 than for the case of LiNi0.8Co0.2O2. The familiar LiPF6 instability (LiPF6 ↔ LiF + PF5) could also be an important driving force for surface-film formation here. The hydrolysis of the LiPF6 molecule results in LiF and the LixPFyOz compounds seen on all electrodes. PF5 (formed in large amounts from the LiPF6 decomposition) can act as initiator for EC polymerisation. This is supported by Ref. 63, who find that reactions between PF5 and solvent indeed result in the formation of polymeric species. LiF can also result, as seen earlier, from the generation of HF according to Eq. (11). This is most important for the LiNi0.8Co0.2O2 family of materials, with their native surface impurity of Li2CO3 2, 60 which reacts with HF to form LiF. The HF so formed may also react with the oxide according to the reaction: LiNi0.8Co0.2O2 + 2x HF → 2x LiF + Li1-2xNi0.8Co0.2O2-x + H2O

(14)

In both cases, the LiF forms directly on the active LiNi0.8Co0.2O2 surface, which will have a more serious effect on cell performance than if LiF is formed from decomposition reactions throughout the electrolyte. The observed increase in cathode impedance with temperature thus results from the combined effects of a thickening of the polymeric surface layer and LiF formation adjacent to the LiNi0.8Co0.2O2 surface. A schematic model for SPI surface-layer formation in LiNi0.8Co0.2O2 based on accumulated experimental evidence is given in Figure 12; the corresponding situation for LiMn2O4 was summarised in Figure 8.

358

Lithium-Ion Batteries: Solid-Electrolyte Interphase

= LiXPFy, LixPOyFz = LiF = Polymer/polycarbonate 2HF + Li2CO3 → 2LiF + H2O + CO2

= Corrosion product + Li2CO3 = LiNi0.8Co0.2O2

LiPF6 ↔ LiF+ PF5 PF5 + H2O → POF3 + 2HF H2O

LixNi0.8Co0.2O2 + y Li+ + y El → Lix+yNi0.8Co0.2O2 + y El+ LiNi0.8Co0.2O2 + 2x HF → 2x LiF + Li1-2xNi0.8Co0.2O2-x + H2O

Figure 12 A proposed model for the SPI surface layer formed on a LiNi0.8Co0.2O2 electrode.

4

LiFePO4

Interest in LiFePO4 (Figure 13) as a cathode material in Li-ion batteries has 63 increased dramatically since it was first proposed by Padhi et al. It exhibits as high a capacity (170 mAh/g) as that of LiCoO2, which is widely used 64 + commercially today, and a slightly lower voltage (approx. 3.5 V vs. Li°/Li ), resulting in increased safety. This lower voltage results from the inductive effect 3+ 2+ of the phosphate group, which lowers the Fermi level of the Fe /Fe redox couple. The early drawback with LiFePO4 was seen to be its extremely low electronic conductivity, but this has now been effectively combated using nano65, 66 crystalline carbon coating of the particle surface. The problem of low electronic conductivity would also seem to be less serious at elevated 67 temperatures.

The Cathode-Electrolyte Interface in a Li-Ion Battery

359

L FeO6 PO4

a

c Figure 13 The crystal structure of LiFePO4 viewed along its b-axis.

We have here exploited photoelectron spectroscopy using synchrotron radiation (PES-SR) to characterise the interface formed on carbon-coated LiFePO4 particles in the cathode of a lithium-ion battery after storage and electrochemical cycling at 23°C and 40°C in a 1M LiPF6 mixture of ethylene carbonate (EC) and diethyl carbonate (DEC). The PES-SR technique facilitates 68, 69 non-destructive depth-profile analysis of the surface layer. An example is given in Figure 14. Most significantly, products from solvent reactions or decompositions (e.g., polycarbonates, semicarbonates and Li2CO3) could not be detected on the carbon-coated LiFePO4 surface. This indicates that the phosphate group does

360

Lithium-Ion Batteries: Solid-Electrolyte Interphase

not react with the solvents, as do the oxides in LiMn2O4, LiNiO2, LiCoO2 and 2, 6, 31, 61, 70 LiNi0.8Co0.2O2. The surface of uncycled LiFePO4 is also different from that found on pristine LiNiO2 and LiNi0.8Co0.2O2, where Li2CO3 has been 2, 70 detected. This can explain the different surface chemistry of LiFePO4 compared to that of other cathode materials. The surface film formed is here found to consist of salt-based products: LiF, LiPF6, LixFy- and LixPOyFz-type compounds. While larger quantities of the salt-based species and more of the oxygenated species were detected on electrodes cycled at 40°C, the surface chemistry of the electrodes cycled at the two temperatures is similar in terms of the surface species present. Neither electrochemical cycling nor cycling temperature appears to enhance the formation of LiF. Moreover, the fact that the phosphate peak is clearly seen for both electrodes indicates that the thickness of the salt-based surface compounds is less than the detection depth of the experiment (~ 50 Å).

b) Intensity (arb. units)

Intens ity (arb. units )

a)

4 54 e V

1 06 1 e V

145

140

135 130 B inding E nergy (eV )

125

454 eV

1061 eV

145

140

135 Binding Energy (eV)

130

125

Figure 14 P2p photoelectron spectroscopy (PES-SR) spectra for carbon-coated LiFePO4 electrodes cycled at 23°C (a) and 40°C (b). The spectra were obtained at photon energies of 454 and 1061 eV. Spectra are shown together with fitted peaks to clarify the peak assignments.

These results combine to indicate that the surface film formed on a carboncoated LiFePO4 cathode will not serve to limit cell performance as the surface film appears to do for other cathode materials.

5

Summary

In the examples taken here, surface phenomena are clearly the major factors to be considered in selecting cathode materials for practical Li-ion batteries.

The Cathode-Electrolyte Interface in a Li-Ion Battery

361

Evidence for the formation of some type of Solid Permeable Interface (SPI) has been obtained in all cases studied. It can be stated generally that the organic species formed on the different cathode electrodes are more or less the same; varying more in degree than in their precise chemical nature; layer thickness also vary from material to material; they also tend to increase significantly with temperature. However, the inorganic species found are more dependent on electrode material type. Reactions with the lithium-salt anion used are also material dependent. It is especially important to reduce the impact of the PF6 anion and its related contaminants (HF and PF5) on electrode surface chemistry through the implementation of more stable salts. Such a development is currently underway.

References 1. Aurbach D., Zaban A., Ein-Eli Y., Weissman I., Chusid O., Markovsky B., Levi M. D., Levi E., Schechter A., Moshkovich M. and Granot E., J. Power Sources 86 (1997), 91. 2. Aurbach D., Gamolsky K., Markovsky B., Salitra G., Gofer Y., Heider U., Oesten R. and Schmidt M., J. Electrochem. Soc. 147 (2000), 1322. 3. Thomas M. G. S. R., Bruce P. G. and Goodenough J. B., J. Electrochem. Soc. 132 (1985), 1521. 4. Levi M. D., Salitra G., Markovsky B., Teller H., Aurbach D., Heider U. and Heider L., J. Electrochem. Soc 146 (1999), 1279. 5. Chen C. H., Liu J. and Amine K., J. Power Sources 96 (2001), 321. 6. Eriksson, T, Andersson, A. M., Gejke, C., Gustafsson, T. and Thomas, J. O., Langmuir 18 (2002), 3609. 7. Kanamura K., Toriyama S., Shiraishi S., Ohashi M. and Takehara Z., J. Electroanal. Chem. 419 (1996), 77. 8. Hunter J. C., J. Solid State Chem. 39 (1981), 142. 9. Jang D. H. and Oh S. M., J. Electrochem. Soc. 144 (1997), 3342. 10. Jang D. H., Shin Y. J. and Oh S. M., J. Electrochem. Soc. 143 (1996), 2204. 11. Guyomard D. and Tarascon J.-M., Solid State Ionics 69 (1994), 222. 12. Gao Y. and Dahn J. R., Solid State Ionics 84 (1996), 33. 13. Pasquier A. Du, Blyr A., Courjal P., Larcher D., Amatucci G., Gérand B. and Tarascon J.-M., J. Electrochem. Soc. 146 (1999), 428. 14. Amatucci G. G., Pereira N., Zheng T. and Tarascon J.-M., J. Electrochem. Soc. 148 (2001), A171.

362

Lithium-Ion Batteries: Solid-Electrolyte Interphase

15. Amatucci G. G., Schmutz C. N., Blyr A., Sigala C., Gozdz A. S., Larcher D. and Tarascon J.-M., J. Power Sources 69 (1997), 11. 16. Amatucci G. G., Blyr A., Sigala C., Alfonse P. and Tarascon J.-M., Solid State Ionics 104 (1997), 13. 17. Blyr A., Sigala C., Amatucci G., Guyomard D., Chabre Y. and Tarascon J.-M., J. Electrochem. Soc. 145 (1998), 194. 18. Foord J. S. and Jackman R. B., Phil. Mag. A 49 (1984), 657. 19. Castro V. Di and Polzonetti G., J. Electron Spectrosc. Relat. Phenom. 48 (1989), 117. 20. Stranick M. A., Surf. Sci. Spectra 6 (1999), 39. 21. Stranick M. A., Surf. Sci. Spectra 6 (1999), 31. 22. Töpfer J., Feltz A., Gräf D., Hackl B., Raupach L. and Weissbrodt P., Phys. Status Solidi A 134 (1992), 405. 23. Oku M., Hirokawa K. and Ikeda S., J. Electron Spectrosc. Relat. Phenom. 7 (1975), 465. 24. Chigane M. and Ishikawa M., J. Electrochem. Soc. 147 (2000), 2246. 25. Carver J. C., Schweitzer G. K. and Carlson T. A., J. Chem. Phys., 57 (1972), 973. 26. Bar-Tow D., Peled E. and Burstein L., J. Electrochem. Soc. 146 (1999), 824. 27. Aurbach D., Weissman I., Schechter A. and Cohen H., Langmuir 12 (1996), 3991. 28. Shiraishi S., Kanamura K. and Takehara Z., J. Electrochem. Soc. 146 (1999), 1633. 29. Wagner C. D., Riggs W. M., Davis L. E., Moulder J. F. and Muilenberg G. E., Handbook of X-Ray Photoelectron Spectroscopy (Perkin-Elmer, Physical Electronics Division, USA, 1979). 30. Khattak G. D., Salim M. A., Wenger L. E. and Gilani A. H., J. Non-Cryst. Solids 262 (2000), 66. 31. Eriksson T., Andersson A. M., Bishop A. G., Gejke C., Gustafsson T. and Thomas J. O., J. Electrochem. Soc. 149 (2002), A69. 32. Eriksson T., Gustafsson T. and Thomas J. O., Electrochem. Solid State Lett. 5 (2002), A35. 33. Ohzuku T., Kitagawa M. and Hirai T., J. Electrochem. Soc. 137 (1990), 769. 34. Tarascon J. M., Wang E., Shokoohi F. K., McKinnon W. R. and Colson S., J. Electrochem. Soc. 138 (1991), 2859. 35. Berg H. and Thomas J. O., Solid State Ionics 126 (1999), 227.

The Cathode-Electrolyte Interface in a Li-Ion Battery

363

36. Thackeray M. M., Yang S., Kahaian A. J., Kepler K. D., Skinner E., Vaughey J. T. and Hackney S. A., Electrochem. Solid State Lett. 1 (1998), 7. 37. Cho J. and Thackeray M. M., J. Electrochem. Soc. 146 (1999), 3577. 38. Andersson A. M. and Edström K., J. Electrochem. Soc. 148 (2001), A1100. 39. Soga K., Hosada S., Tazuke Y. and Ikeda S., J. Polym. Sci. Poly. Lett. 14 (1976), 161. 40. Vogdanis L., Martens B., Uchtmann H., Hensel F. and Heitz W., Macromol. Chem. 191 (1990), 465. 41. Geniès S., Yazami R., Garden J. and Frison J. C., Synth. Met. 93 (1998), 77. 42. Aurbach D., Markovsky B., Schechter A., Ein-Eli Y. and Cohen H., J. Electrochem. Soc. 143 (1996), 3809. 43. Mao H. and Reimers J. N., U.S. Pat. 5,964,902 (1999). 44. Wang E., Ofer D., Bowden W., Iltchev N., Moses R. and Brandt K., J. Electrochem. Soc. 147 (2000), 4023. 45. Cho J., Kim Y. J., Kim T-J. and Park B., Chem. Mater. 13 (2001), 18. 46. Gummow R. J., Kock A. de and Thackeray M. M., Solid State Ionics 69 (1994), 59. 47. Tarascon J.-M., McKinnon W., Coowar F., Bowmer T., Amatucci G. and Guyomard D., J. Electrochem. Soc. 141 (1994), 1421. 48. Robertsson A. D., Lu S. H. and Howard W., J. Electrochem. Soc. 144 (1997), 3505. 49. Kim J. and Manthiram A., Nature 390 (1997), 265. 50. Winter M., Besenhard J. O., Spahr M. E. and Novák P., Advanced Materials 10 (1998), 725. 51. Arai H., Okada S., Ohtsuka H., Ichimura M. and Yamaki J., Solid State Ionics 80 (1995), 261. 52. Ohzuku T., Ueda A. and Ngayama M., J. Electrochem. Soc. 140 (1993), 1862. 53. Ohzuku T., Yanagawa T., Kougushi M. and Ueda A., J. Power Sources 68 (1997), 131. 54. Zhecheva E. and Stoyanova R., Solid State Ionics 66 (1993), 143. 55. Cho J., Kim G. and Lim H., J. Electrochem. Soc. 146 (1999), 3571. 56. Li W., Reimers J.N. and Dahn J.R., Solid State Ionics 67 (1993), 123. 57. Arai H., Okada S., Sakurai Y. and Yamaki J., Solid State Ionics 95 (1997), 275. 58. Mansour A., Surface Science Spectra 3 (1996) 279. 59. Aurbach D., J. Power Sources 89 (2000), 89.

364

Lithium-Ion Batteries: Solid-Electrolyte Interphase

60. Nagai A., Kureha Chemical Industry Ltd., private communication with K. Amine, ANL (2001). 61. Pistoia G., Antonini A., Rosati R. and Zane D., Electrochim. Acta 41 (1996) 2683. 62. Zhang S. S., Xu K. and Joar T. R., J. Electrochem. Soc. 149 (2002), A1521. 63. Padhi A. K., Nanjundaswamy K. S. and Goodenough J. B., J. Electrochem. Soc. 144, 1188 (1997). 64. Tarascon J.-M. and Armand M., Nature 414 (2001), 359. 65. (a) Ravet N., Goodenough J. B., Besner S., Simoneau M., Hovington P. and Armand M., Abstract 127, The Electrochemical Society and The Electrochemical Society of Japan Meeting Abstracts, Vol. 99-2, Honolulu, HI, 17-22 Oct. 1999; (b) Ravet N., Chouinard Y., Magnan J. F., Besner S., Gautier M. and Armand M., Abstract 166, International Meeting on Lithium Batteries, Como, Italy, 28 May – 2 June 2000; (c) Ravet N., Besner S., Simoneau M., Vallée A. and Armand M., Can. Pat., 2,270,771. 66. Huang H., Yin S. C. and Nazar L. F., Electrochem. Solid State Lett. 4 (2001), A170. 67. Andersson A. S. and Thomas J. O., J. Power Sources 97 (2001), 498. 68. Herstedt M., Stjerndahl M., Nytén A., Rensmo H., Siegbahn H., Gustafsson T., Thomas J. O. and Edström K., submitted to Electrochem. Solid State Lett. 69. Andersson A. M., Henningsson A., Siegbahn H., Jansson U. and Edström K., J. Power Sources. In press 70. Andersson A. M., Abraham D. P., Haasch R., MacLaren S., Liu J. and Amine K., J. Electrochem. Soc. 149 (2002), A1358.

CHAPTER 9

THEORETICAL STUDIES ON THE SOLVENT STRUCTURE AND ASSOCIATION PROPERTIES, AND ON THE Li-ION SOLVATION: IMPLICATIONS FOR SEI LAYER PHENOMENA †

YIXUAN WANG* and PERLA B. BALBUENA Department of Chemical Engineering, Swearingen Engineering Center, University of South Carolina, 301 Main, Columbia, South Carolina 29208, USA † E-mail: *[email protected]; [email protected]

1

Introduction

Cyclic alkyl carbonates, such as ethylene carbonate (EC) and propylene carbonate (PC), their linear analog, dimethyl carbonate (DMC), and especially their mixtures are being widely used as solvents in commercial lithium-ion batteries (LIBs). Because of its higher dielectric constant and lower viscosity, EC is considered a better solvent than PC; however, EC is a solid at ambient temperature (melting point 36.2°C). Therefore, mixtures of EC with liquid solvents such as PC and linear carbonates, i.e., co-solvents of EC, are suggested for practical applications at low temperatures. Whether the mixture is suitable to be used in LIBs depends in part on the interactions among carbonate molecules, i.e., their self- and cross-associations, which for this reason are attracting the attention from both experimentalists and theoreticians. Absolutely, the selection of co-solvents for EC should also take into account another very important factor, i.e., whether their presence in the mixture contributes favorably to forming a proper solid electrolyte interface (SEI), which considerably should control the performance of LIBs through effectively passivating the electrodes and preventing further co-intercalation and decomposition of the supporting solvent, permitting only lithium-ion migration, and having a low electronic 1 conductivity. The linear carbonates, DMC, diethyl carbonate (DEC), and methyl-ethyl-carbonate (EMC) do affect much the SEI build up on the graphite electrode in LIBs where EC+DMC, or EC+DEC are used as solvents, and 2-4 eventually influence the electrode reversible capacity. It is speculated that the co-intercalation of linear carbonates into the electrode structure is most probably 365

366

Lithium-Ion Batteries: Solid-Electrolyte Interphase 5

responsible for their distinct behavior. The ability of the solvent to cointercalate into the electrodes is highly related to their interaction (solvation) with Li ions. Chen et al. in Chapter 3 of this book extensively discuss the adsorption of EC on different substrates, such as LiCoO2, nano-MgO, and nano-MgO-coated 55 LiCoO2. Details about the EC structure are useful to understand some of these results. The most interesting structural property of EC is whether its ring is planar or non-planar. Experimentally, the EC structure has been determined in 6 7-10 crystal phase by Brown and Matias et al., who reported a non-planar ring C2 11 symmetry. Very early experimental work of Angell, however, showed that the EC structure is planar in the liquid and gaseous states. Recently theoretical work 12, 13 of Klassen et al. and Blint predicted a planar structure. In this chapter, attention has been paid to the structure of EC in liquid phase. PC is also a very useful solvent of LIBs because of its superior ionic conductivity over a wide temperature range. However, despite the close structural similarity between EC and PC, PC cannot form as effective SEI films 14 as EC does, for LIBs that employ graphite as negative electrodes. To enable to use PC in these batteries, there have been a lot of efforts focusing on the identification of proper additives and/or co-solvents for PC-based electrolytes, which would help to generate an efficient SEI layer. The typical liquid additives 15 include chloroethylene carbonate (CEC), other halogen-substituted carbonates, a variety of unsaturated carbonates such as vinylpropylene carbonate and 16 56, 57 vinylene carbonate, and ethylene/propylene sulfite (ES/PS). The most common co-solvents are DMC, DEC, EMC, γ-butyrolactone (γ-BL), dimethyl sulfoxide (DMSO), dimethyl formamide (DMF), dimethyl amide (DMA), 1,2dimethoxy-ethane (DME) and 1,2-dimethoxy-methane (DMM). To explore the role of these additives and co-solvents, it is necessary to understand their structures and some properties that may affect the SEI formation on graphite anodes. In this chapter, state-of-the-art first-principles-based theoretical methods are used to investigate the geometries of the afore-mentioned supporting solvents, co-solvents and additives, their self- and/or cross-associations, and the interactions between them and lithium ions. It is intended to explain, at the molecular level and on the basis of these theoretical analyses and calculations, some experimental facts relevant to the SEI phenomena.

Solvent Structure and Association Properties, and Li-Ion Solvation

2

367

Computational Details

Ab initio molecular orbital (MO) theory as well as density functional theory (DFT) are used throughout this chapter. Specific MO methods include HartreeFock (HF) and the second-order Møller-Plesset perturbation (MP2). Since DFT has already provided accuracies similar to higher-order ab initio methods for a large range of compounds, especially organic molecules, with much less computational requirements than MO, it is being widely applied to practical problems in chemistry and chemical engineering. A density functional is composed of exchange and correlation components. The frequently used exchange functionals are: the Becke’s three-parameters hybrid GGA/exact58 exchange functional (B3), the GGA exchange functional of Perdew-Wang 17 1991(PW91), and the Perdew-Wang 1991 exchange functional as modified by 18 Adamo et al. (MPW). The correlation functionals are the Lee-Yang-Parr GGA 19 17 functional (LYP), and the Perdew-Wang 1991 correlation functional (PW91). Four different DFT functionals are yielded by pairing these exchange and correlation functionals:B3LYP, B3PW91, PW91LYP, and PW91PW91. The HF-optimized Gaussian type of orbitals (GTO), such as 6311++G(d,p), and 6-31G(d) basis sets, are used to expand the HF orbitals of MO theory and the DFT Kohn-Sham (KS) orbitals, respectively. The theoretical methods are denoted as B3LYP/6-311++G(d,p), B3LYP/6-31G(d), B3PW91/631G(d), MP2/6-31G(d) and so on. Harmonic vibrational frequencies are calculated at the same theory level as that are used to fully optimize molecular geometries, which enable us to confirm the minima and to evaluate the corresponding zero point energy (ZPE) corrections that were scaled by the empirical factor 0.98 for B3LYP/620 311++G(d,p), proposed by Bauschlicher, and by 0.94 for B3PW91/6-31G(d). Basis set superposition error (BSSE) corrections are estimated using the 21 counterpoise method of Boys and Bernardi for the complex systems, such as + dimers and trimers of carbonates and for the complex of Li and solvents, to accurately get energetic data such as binding energies and Gibbs free energy changes. Solvent effects are taken into account using the super-molecule cluster model, by which the interaction between solute and a few solvent molecules is explicitly included, and bulk solvent effect is added through the polarized 22-24 25 continuum models, such as PCM, conductor-like PCM (CPCM), iso-density 26 PCM (IPCM), and self-consistent isodensity PCM (SCI-PCM), which were developed on the basis of the Onsager reaction field theory and are recognized

368

Lithium-Ion Batteries: Solid-Electrolyte Interphase

to provide reliable results for systems without specific interactions like hydrogen bond. These PCM methods are denoted as CPCM-B3LYP/6311++G(d,p), PCM-B3LYP/6-311++G(d,p) and so on. 27 Finally Car-Parrinello molecular dynamics (CPMD), the unification of + DFT and molecular dynamics, is used to address the solvation number of Li in EC solution. The CPMD method is an implementation of DFT in the KohnSham formulation along with the Car-Parrinello MD scheme. A complete and orthonormal set of plane waves is used as the basis set for the valence electron wavefunctions. Norm-conserving pseudopotentials, describing the interaction between the valence electrons and the ionic cores, replace the cores. The CPMD method is starting to play a very important role in many practical problems. Specifically, the system in this study consists of 12 EC molecules and one Li3 ion, located in a periodic cubic supercell of 11*11*11(Å) . The electronic structure was solved using the KS formulation of DFT within the local density approximation of the exchange-correlation functional augmented by the BLYP generalized gradient corrections. The employed pseudo-potential is the Goedecker type for all the atoms. The KS orbitals are expanded in plane waves up to a kinetic energy cutoff of 70 Ry. The system was equilibrated firstly using the Nose-Hoover approach at 350K, and data was collected at constant energy. Ab initio MO and DFT calculations are performed with the Gaussian-98 28 29 A9-A11 program package, and CPMD is done with the CPMD 3.53 version.

3

Geometric Structures of Various Cyclic/Linear Carbonates, and Effective Additives and Co-solvents to PC-Based Solutions

Molecular geometries (stick and ball model) for various solvents optimized with the DFT method (B3LYP/6-311++G(d,p)) are illustrated in Figure 1. Due to rotation about a C-O single bond, different conformations exist in the linear carbonates such as DMC, DEC and EMC. The eclipsed conformation (C2v symmetry) with respect to two end methyl groups is a minimum for DMC, whereas the staggered one (Cs symmetry) is a maximum (transition state) that is 1 characterized by a unique imaginary frequency (−98 cm− ). This is also true for DEC, where the eclipsed conformation (C2v) with respect to the two ethyl groups is a stable structure. In the cases of two ether co-solvents, the eclipsed conformation (C2v) is a minimum for DMM with approximately 3.0kcal/mol lower energy than the staggered one (Cs). However, besides presenting a more

Solvent Structure and Association Properties, and Li-Ion Solvation

369

complex potential energy surface, the calculated global minimum for DME is the trans staggered conformation (C2h).

DMC PC

DMC

C1

C

C

s

2v

C

VC

ν = -98

2v

S

DEC

C

EMC

2v

C

ES

s

C

1

S S

N

PS

C

γ BL

1

C

DMSO

1

C

C

DMF

s

s

N

DMA

C

DME s

C

2h

DMM

C

2v

Figure 1 Optimized geometries and symmetries of various organic compounds with B3LYP/6311++G(d,p). If not noted otherwise, the black ball stands for an oxygen atom, the big white one for a carbon atom, and the small one for a hydrogen atom. Reproduced from [44] with permission of the Am. Chem. Soc.

370

Lithium-Ion Batteries: Solid-Electrolyte Interphase

In gas phase the stable structure of EC is the non-planar one with C2 30 symmetry, as described elsewhere. Solvent effects are taken into account in two ways to explore its structure in liquid phase. One is a super-molecule cluster model that explicitly includes interactions from surrounding molecules, and the other is the polarized continuum model developed on the basis of Onsager model, specifically the polarized continuum model (PCM) and the conductor25 like polarized continuum model (CPCM). As shown in Figure 2 for the cluster model consisting of 9 molecules, the circled EC molecule interacts with the surrounding solvent molecules via C-H•••O interactions, clearly exhibiting a non-planar ring structure. In the regular arrangement shown in Figure 2, for a given column the highly polar molecules (dipole moment for an isolated molecule =5.64 Debye with B3LYP/6-311++G(d,p)) are arranged with parallel rings, and with alternating dipolar directions. The C=O group of one molecule locates over/below the ring of the adjacent molecule so that the geometry is favorable forming C-H•••O interactions. Although the parallel trimer is less stable than the circle trimer, which will be discussed in the following section, the regular structure in Figure 2 is more stable than the non-regular one originating from the cyclic trimer. The molecular packing shown in Figure 2 is 7 rather similar to the crystal structure of EC.

2.690

2.698

Figure 2 Structure of (EC)9 fully optimized with B3LYP/6-31G(d). The black ball stands for an oxygen atom, the big white for a carbon atom, and the small one for a hydrogen atom.

Solvent Structure and Association Properties, and Li-Ion Solvation

371

The side-view of the planar (C2v symmetry) as well as the non-planar (C2 symmetry) structures, predicted by the CPCM-B3LYP/6-311++G(d,p) method, are given in Figure 3. The planar structure is higher in energy than the nonplanar one, and it is a transition state characterized by a unique imaginary frequency (− −128cm−1). The selected structural data for the non-planar geometry from various methods such as MP2, B3LYP, and CPCM-B3LYP are summarized in Table 1 along with the barrier height (ǻE) from the minimum energy to the planar structure. The B3LYP method for the isolated molecule gives the best agreement for dihedral angles with the information from single crystal neutron diffraction analysis. Despite the not so good agreement from B3LYP for the dihedral angles, the bond lengths, angles, as well as the energy barrier predicted by B3LYP-CPCM in solvent are in quite good agreement with the experimental results. The EC potential energy surface profile obtained by CPCM-B3LYP/6311++G(d,p) is illustrated in Figure 4, which further clearly shows that the planar structure is a transition state connecting two non-planar configurations. Therefore we strongly suggest that the stable structure of EC in gas phase as well as in liquid phase is non-planar.

Figure 3 The EC optimized geometries with CPCM-B3LYP/6-311++G(d,p). Non-planar with C2 symmetry (left) is a stable structure, planar with C2v symmetry (right) is a transition structure.

372

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Table 1 Selected geometrical parameters of EC molecule (C2 symmetry) and energy difference ∆E (kcal/mol) with various methods including MP2/6-311++G(d,p), B3LYP/6-311++G(d,p), and CPCM-B3LYP/6-311++G (d,p) and comparison with experimental results. a Ref.7, b Ref. 31. CPCM-B3LYP

Exp.a

MP2

B3LYP

C1-O2

1.193

1.188

1.202

1.203

C1-O3

1.364

1.61

1.344

1.342

O3-C5

1.433

1.437

1.454

1.457

C5-C6

1.522

1.530

1.526

1.522

O2-C1-O3

124.9

124.9

124.3

124.2

O3-C1-O4

110.2

110.1

111.4

111.7

C1-O3-C5

108.3

109.4

109.3

108.7

C1-O3-C6

108.3

109.4

109.3

108.7

O3-C5-C6

102.1

102.6

102.5

102.2

O3-C5-C6-O4

29.5

23.6

21.7

24.8

C1-O3-C5-C6

−24.8

−20.0

−18.6

−21.3 −21.3

C1-O4-C6-C5

−24.8

−20.0

−18.6

O3-C1-O4-C6

10.3

8.3

7.7

9.0

∆E*

1.04(0.04)

0.33(0.50)

0.37(0.56)

0.67b

*∆E energy difference between C2v and C2; the conformation with C2v is the transition state between two configurations with C2. The values in parenthesis are the energy differences without zero point energy.

0 .0

E n e r g y , k c a l /m o l

-0 .1

-0 .2

-0 .3

-0 .4

-0 .5 -3 0

-2 0

-1 0

0

10

20

d ih e d ra l a n g le (O 3-C 5-C 6-O 4 )

Figure 4 The potential energy surface from B3LYP/6-311++G(d,p). This figure clearly shows that the planar structure is the transition state connecting two non-planar configurations.

30

Solvent Structure and Association Properties, and Li-Ion Solvation

373

On the basis of B3LYP/6-311++G(d,p) calculations, infrared spectra are simulated for nonplanar EC conformation. The non-planar EC with C2 symmetry has 24 fundamental vibrational modes, half of which belong to the respective irreducible representations A and B of the point group C2. Figure 5 shows the simulated infrared spectra for the isolated molecule and that in solvent, together with experimental data. Solvent effect results in the red shift of 1 the stretching vibration for C=O (νC=O, 1858 vs. 1761 cm− ). The relative intensity of the CH2 stretching vibration is reduced. It is noted that the main 12 peaks position agrees quite well with experimental result (underlined data). 1796

1.6

1761

1.4 1.2

ν C=O

Exp.

ν C (H

2

)O

Relative Intensity

1070 ν C O

1.0

1119 δCO

1052 769 969

0.8

3

ν C -C

0.4

EC , solvent

1372 1215 1470

762 950 700

0.6

1392 1419 1472

996

δ CH

2

0.2 EC , gas phase

0.0 500

1000

1500

2000 Frequency/cm

2500

3000

3500

-1

Figure 5 Simulated IR spectra for EC in gas phase (B3LYP/6-311++G(d,p)) and in solvent (CPCM-B3LYP/6-311++G(d,p)). The underlined data are from experiment of EC solvent.12

4

Self and Cross Associations of Cyclic/Linear Carbonates via C-H•••O Interactions

Carbonates probably associate mainly by intermolecular C-H•••O interactions between the carbonyl oxygen and the C-H group of another molecule. Such interactions have been a quite active topic in recent years. One reason is that 32 they are widely found in gas phase, liquid phase, for example in many

374

Lithium-Ion Batteries: Solid-Electrolyte Interphase 33-35

36

biological systems, and even in crystalline environments. Another reason is that these interactions present an abnormal behavior, i.e., a C-H bond 37-39 contraction is accompanied by a blue shift of its vibrational frequency, in many cases in which C atoms of hydrogen donor C-H are saturated 3 (traditionally sp hybridization), which is opposite to the conventional X-H•••Y hydrogen bond (e.g., O-H•••O). Hereby C-H•••O is usually classified as an antihydrogen bond. Despite the fact that they are weaker than conventional OH•••O H-bond interactions, in most of cases because of the lower electronegativity of the carbon atom, there is an agreement that C-H•••O represents specific interactions, which are driven by more than a simple electrostatic force between the partial negative charge on the oxygen and a 38, 40, 41 positive charge that accumulates on C-H. There has also emerged a controversy as to whether the C-H•••O interaction is fundamentally different 38 from the conventional H-bond and about the origin of its unusual behavior. Extensively examining the C-H•••O interactions between FnH3-nCH as proton 38 donors, and several acceptors such as H2O, CH3OH and H2CO, Gu et al. claimed that fundamental distinctions do not exist between the C-H•••O and O-H•••O hydrogen bond interactions, despite the opposite trends shown by the O-H and C-H bonds. The inspection of a set of topological criteria based on 42 atoms-in-molecules (AIM) showed also no essential difference either for the two sorts of C-H•••π as well as C-H•••O interactions with normal (C-H bond stretching) as well as anti-normal (C-H bond contraction) properties for the 43, 44 proton donors of H-bond. Interestingly, for both interactions C-H donors 3 with sp -hybridization carbon usually show a bond contraction accompanied by a blue shift of its vibrational frequency, while a stretching and red shift are 2 observed for the unsaturated (sp and sp hybridization) donor such as CHCH, HCN and VC, analyzed in this chapter.

4. 1

Geometric and Energetic Properties

VC, EC, and PC dimers. Calibrated calculations for the most common conventional H-bond system, H2O dimer, which has comparable binding energy to those of cyclic carbonate dimers (as shown in Table 2), indicate that the values of De (binding energy with BBSE correction, Table 2 gives the definitions in detail) from B3LYP, B3PW91 and MP2 methods are consistent within 0.5 kcal/mol and are also in excellent agreement within the error bars of the two sets of experimental data. However, PW91LYP gives a too high De

Solvent Structure and Association Properties, and Li-Ion Solvation

375

value, and that from PW91PW91 appears within the error bars of the experimental result in Table 2 (5.85 vs. 5.4±0.7 kcal/mol). Therefore, only B3LYP results of energetic data and binding characteristics are shown in Table 2 for alkyl carbonate dimers. VC possesses planar geometry with C2v symmetry. Three dimers have been located, as shown in Figure 6, which are a planar structure D1 (Cs), a bifurcated structure D2 with higher symmetry (C2v), and a T-shape structure D3 (Cs) of two monomer planes perpendicular to each other. The binding energies D0 (binding energy with BSSE and ZPE corrections) of D1, D2, and D3 given by B3LYP are 3.04, 3.02 and 2.78 kcal/mol, respectively. The distance (R) of donor H to acceptor carbonyl O in D2 is about 0.5 Å longer than that in D1, and C-H•••O angle (A) much smaller than that in D1 by 60-70 degrees. Although two CH•••O interactions are present in D2, the closeness of D0 between D1 and D2 shows that the C-H•••O interaction in D2 is much weaker than that in D1, therefore, like conventional H-bonds, C-H•••O also tends to a linear orientation. R in the T type of the VC dimer, D3, is longer than that in D1 and shorter than in D2 by 0.15-0.25 Å, and A in D3 is also between those of D1 and D2. Two sorts of minima (D4, D5) have been found for the EC dimer. As shown in Figure 6, D4 possesses two strong (R ~ 2.45 and 2.55 Å; A~140° and 119° respectively) and one quite weak C-H•••O interactions (R ~ 2.90 Å; A~115°). R of the two C-H•••O interactions in the structure D5 are about 0.3 Å longer than those of the two short such interactions in D4 (2.76 vs. 2.47 and 2.53Å), and A are considerably small (96.8°). Accordingly, the binding energy is also much lower than that of D4 (D0, 3.68 vs. 5.08 kcal/mol). The geometry of the PC dimer (D6) looks rather similar to that of the EC dimer (D4). As shown in Table 2, the distances of H•••O for the two relatively strong C-H•••O interactions are 2.466 and 2.578 Å respectively, and that for the weak interaction is 2.835 Å, which closely agree with those of D4 (2.466, 2.534 and 2.950 Å). The angles in D6 are also very close to those in D4. Due to the electron-donor methyl group, the interaction through C-H•••O in PC is slightly weaker than that in EC. This is reflected by the smaller binding energy at the involved four correction levels (6.34, 5.64, 4.69 and 3.32 kcal/mol vs. 6.85, 6.00, 5.08 and 4.16 kcal/mol), which corresponds to 0.2-0.3 kcal/mol lower per C-H•••O interaction. The binding energy difference between EC and PC dimer (DT~ 4.16 vs.3.32 kcal/mol with B3LYP) may be one of main factors leading to the different states at which the solvents are found at room temperature. Additionally, the methyl group in PC perhaps blocks intermolecular interactions.

376

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Table 2 Binding energies including relevant corrections, ∆E, D, D0 and DT (kcal/mol), the distances between H and carbonyl O (R /Å), and the angles of CH•••O (A/deg), for VC dimers (D1, D2, D3), and water dimer WD, for EC dimers (D4, D5), PC dimer (D6), DMC dimer (D7), and the hetero-molecular dimers of EC+VC (D8, D9), of PC+ VC (D10, D11), of EC+PC (D12), of DMC+EC (D13) and PC+DMC (D14), of EC+DEC (D15), EC+EMC(D16) at the basis set 6311++G (d,p) unless specified. Reproduced from [44] with permission of Am. Chem. Soc.

Method

Dimer

∆Ea

Db

D0c

DTd

R

A

B3LYP B3PW91 PW91LYP PW91PW91 MP2 Exp. /De

WD WD WD WD WD WD

5.83 5.28 7.37 6.79 6.08

2.70 2.15 4.13 3.51 2.07

2.67 2.13 4.10 3.52 2.04

1.933 1.933 1.903 1.893 1.949

175.6 175.0 173.8 174.7 176.9

B3LYP B3LYP B3LYP B3LYP

D1 D2 D3 D4

3.72 3.61 3.98 6.85

5.03 4.46 6.44 5.85 4.45 5.4±0.7e 5.4±0.2f 3.46 3.36 3.26 5.93

3.04 3.02 2.78 5.08

1.54 1.50 1.37 4.16

B3LYP B3LYP

D5 D6

4.47 6.34

4.11 5.38

3.68 4.69

2.16 3.32

B3LYP B3LYP B3LYP B3LYP B3LYP B3LYP

D7 D8 D9 D10 D11 D12

2.44 4.32 5.25 4.49 5.17 6.73

2.09 4.00 4.42 4.13 4.38 5.81

1.66 3.50 3.80 3.66 3.87 5.08

0.23 2.04 2.45 2.20 2.46 3.75

B3LYP B3LYP B3LYP B3LYP

D13 D14 D15 D16

4.11 3.53 4.23 4.14

3.41 2.84 3.50 3.46

2.79 2.31 2.90 2.84

1.44 0.92

2.234 2.692 2.471 2.466, 2.950, 2.534 2.764, 2.761 2.466, 2.578, 2.835 2.479, 2.479 2.209 2.582, 2.950 2.220 2.513, 3.059 2.465, 2.562, 2.891 2.501, 2.523 2.588, 2.541 2.471, 2.557 2.487, 2.533

175.4 105.7 123.4 140.1, 114.3, 119.4 96.8, 96.8 139.1, 118.2, 120.1 149.5, 149.2 174.5 120.4, 149.9 176.0 122.7, 154.6 139.7, 118.9, 118.5 119.6, 159.6 117.4, 163.7 122.4, 157.0 121.5, 160.7

∆E= nE(monomer)-E(n-mer); bD =∆E+BSSE; cD0= =∆E+∆ZPE+BSSE; dDT= ∆E + thermal effect at 298.15K+BSSE; e from Ref. 45; f from Ref. 46. a

Solvent Structure and Association Properties, and Li-Ion Solvation

D1

D7

D6

D9

D13

D8

D11

D10

D14

D5

D4

D3

D2

377

D15

D12

D16

Figure 6 Optimized geometries of carbonate monomers and dimers with B3LYP/6-311++G(d,p). Black ball stands for oxygen atoms, big white ones for carbon atoms and small white ones for hydrogen atoms. Reproduced from [44] with permission of the Am. Chem. Soc.

DMC dimer. The dimer (D7) of DMC has two nearly identical C-H•••O interactions, and they constitute a 10-membered ring that was confirmed by an AIM calculation (electron density at the ring critical point, 0.002 au). The bond distance R and the angle A are 2.479 Å and 149.5°, respectively. The binding

378

Lithium-Ion Batteries: Solid-Electrolyte Interphase

energy of the dimer is about two times lower than those of EC and PC (D0, 1.66 vs. 5.08 and 4.69 kcal/mol), and nearly one time lower than that of VC (D0, 1.66 vs. 3.04 kcal/mol). Compared with the three cyclic carbonates, the inclusion of the temperature effect (298.2K) especially decreases the binding energy of the linear carbonate dimer. EC•••VC, PC•••VC, EC•••PC dimer. Two kinds of cross-association structures of VC with EC and PC have been located. VC acts as a proton donor in the hetero dimer D8. Like D1, C-H•••O interaction in D8 approaches to a linear alignment (A~176.2°), and R is also quite close to that in D1 (2.210 vs. 2.234 Å), but it has 0.6 kcal/mol higher binding energy than D1, indicating that EC is a stronger proton acceptor than VC. One short (R~ 2.582 Å, A~ 120.4°) as well as one relatively long (R~2.950 Å, A~149.9°) C-H•••O interactions are present in D9, where VC acts as a proton donor for the former interaction and EC for the latter one. The binding energy D9 is only 0.4kcal/mol higher than that of D8. PC•••VC complexes behave very similarly to those of EC•••VC with respect to structures and binding energies. For example, R agrees within 0.01Å between D8 and D10, and D0 only differs by 0.16 kcal/mol. The crossassociation structure of EC and PC is rather similar to their respective dimer, i.e., two short (R ~2.47-2.58 Å) and one long (R~ 2.84-2.95 Å) C-H•••O interactions exist in D12 (see Scheme 1b). The binding energy D0 is almost the same as that of the EC dimer D4 and only 0.4 kcal/mol higher than that of the PC dimer D6 (5.08 vs. 5.08 and 4.69 kcal/mol). EC•••DMC/DEC/EMC, PC•••DMC dimers. As shown in Figure 6, the optimized structures for the cross associations of DMC with EC and PC (D13 and D14, respectively) have two different C-H•••O interactions. The characteristics of the C-H•••O interactions in which DMC acts as C-H donator are similar to that in the DMC dimer, i.e., the angles A tending to linear (159.6° and 163.7°). Although these cross associates are more stable than the DMC dimer itself (D0, 2.75 and 2.31 vs. 1.66 kcal/mol), they are much less stable than the EC dimer, PC dimer, as well as cross associates of EC and PC. These results, together with those of EC•••PC could give an interpretation to the experimental fact that a pure EC solution presents intermolecular association that remains unaltered in the presence of PC but it is disrupted by a linear 12 carbonate such as DMC. Most probably PC/DMC could also associate with the EC molecule in their mixtures, however the cyclic carbonate molecules as a whole still behave like associates instead of free molecules in the mixture of EC + PC, due to their close binding energy, while a weak binding between EC and DMC could increase the fluidity of their mixture. In view of this point, both PC

Solvent Structure and Association Properties, and Li-Ion Solvation

379

and VC are not so good as the linear carbonate DMC as co-solvent of EC. The binding energies of EC, and DEC and EMC (D15 and D16) are very similar to that of the dimer of EC•••DMC, and also significantly lower than those of EC, PC and EC•••PC dimers. This means that DEC and EMC are also proper cosolvents of EC, increasing fluidity of their mixtures. VC and EC trimers. Energetic and selected geometrical data are shown in Table 3. Three different local minima have been found in a theoretical survey for the VC trimers. Two of them (T1 and T2, of Cs and C2v symmetry respectively as shown in Figure 7) hold planar geometry, while the two C-H acceptor molecules are perpendicular to the C-H donor molecule in another trimer (T3, C2v symmetry). In line with the conventional hydrogen bond 47, 48 systems, such as H2O and methanol, the global minimum T1 also corresponds to a cyclic structure in which each VC acts as a C-H donor as well as an acceptor. Three C-H•••O interactions are almost identical, which is reflected by the R (2.2350, 2.2301, and 2.2383 Å), A (164.6, 164.1, and 164.7°) and other characteristics such as bond lengths and stretching frequencies of CH, electron density and Laplacian of the electron density at the bond critical point of H•••O. In the case of the open trimer T2, only the middle VC molecule behaves as C-H•••O donor and acceptor simultaneously, while the two end molecules as either donor or acceptor. The four weak C-H•••O interactions slightly differ. T2 lies 2.99 kcal/mol above the global minimum. Another open structure trimer, T3, with the middle VC molecule as a duplicate C-H donor, is predicted to be 4.65 kcal/mol less stable than the global minimum T1. As expected, T3 is completely symmetric, with two identical C-H•••O interactions (R~2.506 Å, A~123.5°). A pseudo-planar geometry (T4) and a sandwich geometry (T5) were found for the EC trimer. T4 looks like that two monomers arrange similarly to the EC dimer (D4) keeping one end apart from each other so that the third monomer is inserted. Consistently five different C-H•••O interactions (R: 2.339, 2.823, 2.418, 2.540, 2.634 Å) are present. Two couples of C-H•••O interactions (R/A: 2.619/123.5; 2.631/121.0) appear in T5, which just has 0.6 kcal/mol lower binding energy than T4. The binding energy of T4 is about 0.9kcal/mol higher than that of the VC trimer T1 (10.59 vs. 9.70 kcal/mol). The binding energy of the sandwich trimer T5 is only 0.6 kcal/mol lower than T4.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

380

Table 3 Binding energies including relevant corrections, ∆E, D, D0 and DT (kcal/mol), distances between H and carbonyl O (R /Å), and angles of C-H…O (A/deg) of VC trimers (T1, T2 and T3) and EC trimer (T4 and T5) with B3LYP/6-311++G(d,p) method. Reproduced from [44] with permission of the Am. Chem. Soc. ∆E

D

D0

DT

R

A

T1

12.16

11.06

9.70

7.06

2.238, 2.235, 2.230

164.7, 164.6, 164.1

T2

8.03

7.43

6.71

3.68

2.662, 2.664

105.4, 105.1

T3

7.57

6.04

5.05

2.23

2.506

123.5

T4

13.84

12.06

10.59

7.86

2.339, 2.823, 2.418 2.540, 2.634

153.3, 120.1, 132.1 131.1, 127.0

T5

13.28

2.619, 2.619, 2.631 2.631

123.5, 123.5, 121.0 121.0

T3

T1

T4

T2

T5

Figure 7 Trimers of VC and EC optimized with B3LYP/6-311++G(d,p). T1, T2 and T3 (VC); T4 and T5 (EC). Reproduced from [44] with permission of the Am. Chem. Soc.

4.2

C-H Bond Lengths and Vibrational Frequencies

The C-H harmonic frequencies of the two vibrational modes for the isolated VC 1 molecule are 3249 (a1, very weak) and 3224 cm− (b2, medium) at B3LYP/6-

Solvent Structure and Association Properties, and Li-Ion Solvation

381

311++G(d,p), respectively. The C-H bond lengths and their harmonic vibrational frequencies in VC dimer and VC trimers are summarized in Table 4. The simulated IR spectra for monomer, the dimer D1 and the trimer T1 are shown in Figure 8. It can be observed from Table 4 that the hydrogen donor CH is stretched by 0.0024 Å in the dimer D1, while the free C-H and those of the hydrogen acceptor remains practically unchanged (not shown in Table 4). Consequently, in the dimer D1 the C H vibrational frequencies of the donor 1 are shifted 7 and 19 cm− respectively, corresponding to a1 and b2 modes in the isolated monomer (termed hereafter as symmetric, νs, and antisymmetric, νas, for n-mers). The frequency shift, mainly for the antisymmetric vibration, can also be clearly seen in Figure 8. The stretching of the donor CH and the red shift of its vibrational frequency are approximately five and six times less than that of 1 methanol dimer (0.015 Å and 148 cm− at B3LYP/6-311++G(d,p)), 48 respectively. In the case of D3, there exists a minor CH stretch (0.0008 Å) 1 and a negligible C-H frequency change (less than 5 cm− ). Nearly linear cases 2 like D1, perform similarly to the cases of a CH donor with sp and sp unsaturated hybridization, such as the interactions of acetylene and HCN with 43 41 benzene, acetylene, and ethylene with H2O. Table 4 Length (L/ Å) and harmonic vibrational frequencies (ν/cm−1) of the CH bond for the monomer (M), and those in the CH•••O interactions of dimer (D1,D2 and D3) and trimer (T1) of VC, calculated with the B3LYP/6-311++G (d,p) method. Reproduced [44] with permission of Am. Chem. Soc. L M

∆L

1.0748

νa

∆ν

νs

νas

3249 (a1)

3224 (b2)

∆νs

∆νas −19

D1

1.0772

0.0024

3242

3206

−7

D3

1.0756

0.0008

3246

3222

−3

−2

T1

1.0785

0.0037

3243

3190

−6

−35

1.0785

0.0037

3243

3190

−6

−36

1.0787

0.0037

3242

3187

−7

−38

1.0754

0.0006

3247

3223

−2

−1

T3

With respect to the trimer T1, the C-H of the donor is further stretched (∆L ~ 0.0037 Å), which results in about one time more red shift for νas but negligible change for νs as compared with D1, also reflected by Figure 8. The

382

Lithium-Ion Batteries: Solid-Electrolyte Interphase

1842

2 .5

R e la tiv e In te n s ity

2 .0 3189

1854

T1

1 .5

1 .0

3206

1875

D1 ν C=O

0 .5

1070 701

874 991

ν CH

1158 1329

3224

0 .0 500

1000

1500

2000 F re q u e n c y/c m

2500

3000

VC

3500

-1

Figure 8 Simulated IR spectra for VC, VC dimer (D1) and VC trimer (T1) with B3LYP/6311++G(d,p).

higher C-H stretching frequency shift could be taken as an evidence of nonnegligible cooperative effects of C-H•••O type of H-bond in VC trimer, as usually occurred in conventional H-bond. The cooperative factor (Ab ~ 2.0 for T1) defined in terms of the relative shift of the C-H stretching frequency shows that cooperativity is considerably important in the three C-H•••O H-bonds of the VC trimer. The bond lengths for free C-H in the EC monomer, L(M), and only those for the C-H involved in a H-bond in the dimer and trimer, L(D) and L(T), together with their corresponding vibrational frequencies are listed in Table 5. The simulated IR for EC, D4 and T4 are given in Figure 9. Opposite effects to those of the VC complexes are observed upon the formation of H-bond. The C-H bonds are considerably contracted, and consistently a blue shift takes place on the C-H vibrational frequencies, (e.g., as shown in Table 6, ∆L=−0.0015Å, 1 the largest ∆ν=+23 cm− for D4; the averaged ∆L=−0.0028Å, the largest −1 ∆ν=+36 cm for T4). The trend is kept for the other dimers in this study, and is extended to the trimers as well, e.g., ∆L=−0.0017 Å for DMC dimer, ∆L=−0.0015 Å for PC dimer, and ∆L=−0.0021 Å for the hetero-molecular dimer of EC and DMC (D13), etc. This shows that the anti-hydrogen bond

Solvent Structure and Association Properties, and Li-Ion Solvation

383

Table 5. Length (L/Å) and harmonic vibrational frequencies (ν/cm−1) of the CH bond for the monomer (M), and those involved in a H-bond of EC Dimer ( D4) and Trimer (T4) calculated with B3LYP/6-311++G (d,p) method. Reproduced from [44] with permission of the Am. Chem. Soc. L(M)

L(D)

∆L(D)

L(T)

∆L(T)

ν(D)c

1.0888 1.0929

1.0873a

−0.0015

1.0871a 1.0885b

−0.0017 −0.0044

3091 3074 3016 2999

1.0906b

−0.0013

1.0869a

−0.0019

3081 3064 3019 3000

1.0899b

−0.0030

∆ν

ν(T)

∆ν

+19 +14 +20 +6

3095 3083 3032 3005

+23 +23 +36 +12

+9 +4 +23 +7

3096 3080 3025 2999

+24 +20 +29 +6

3088 3067 3026 3001

+16 +7 +20 +8

(D)

(T)

a

Associated with the longer C-H bond of VC monomer; bassociated with the shorter bond of VC monomer;c the C-H stretching frequencies of EC monomer are 3072 (1b), 3061(1a), 2997(2b), and 2993(2a), respectively. ν C =O

2.5

νCH

1826

2

2.0 3032 3008 3080

Relative Intensity

1839

T4

1.5

1.0

3017 3000 3072

1858

D4

0.5 2993 3065

0.0 500

1000

1500

2000 Frequency/cm

2500

3000

EC

3500

-1

Figure 9 Simulated IR spectra for EC, its dimer (D4) and its trimer (T4) with B3LYP/6311++G(d,p).

384

Lithium-Ion Batteries: Solid-Electrolyte Interphase

character in terms of C-H bond contraction and blue shift would occur in the 3 cases where a donor H is linked to a saturated carbon, i.e., sp hybridization. The present result for EC dimers and trimer is close to the C-H•••O interactions of methane and fluorinated methane with proton acceptors such as H2O, CH3OH 35, 38 and H2CO.

4.3

C=O Bond Lengths and Vibrational Frequencies

Since the relative intensity of the C=O stretching is very strong, variations of the C=O vibration can also reflect C-H•••O(=C) interactions to some extent. Figure 8 shows that νC=O (C=O vibrational frequency) clearly shifts with 1 1 associations of VC, 1875 cm− for the monomer, 1854 cm− for D1 and 1842 −1 cm for T1. Accordingly, the lengths of C=O bonds (RC=O) are stretched from 1.1863Å in the monomer to 1.1909 Å in D1 and 1.1954 Å in T1. Both ∆νC=O and ∆RC=O for the trimer T1 are absolutely higher than those for the dimer 1 1 (−33 cm− , 0.0091 Å vs. −21 cm− , 0.0043 Å), indicating again that the C-H•••O interaction in T1 is enhanced by cooperative effects. In contrast to C-H pattern, changes similar to VC take place with EC associations, i.e., νC=O (RC=O) is decreased (increased), as shown in Figure 9. The effects of association on IR for the EC+DMC and EC+DEC dimers are illustrated in Figures 10 and 11, respectively. The C=O stretching vibrational 1 frequencies of DMC and DEC are very similar (1756 vs. 1747 cm− ), and they 1 1 are approximately 100cm− less than that of EC (1756/1747 vs. 1858 cm− ). Association through two very defined C-H•••O interactions (see both D14 and D15) stretches the respective C=O bond of EC and DMC/DEC by 0.0049 Å and 1 consequently decreases the stretching frequencies by about 20 cm− . Only very weak coupling of vibrations between EC and DMC/DEC has been found. Except for the shifts of νC=O, the IR of the dimer is almost the additive contributions of the two monomers.

4.4

Characteristics of C-H•••O Interactions Using AIM 42

The theory of “atoms in molecule” (AIM) allows one to identify and characterize bonding interactions between atoms through an analysis of the charge density ρ. Specifically, Koch and Popelier proposed a set of criteria that are indicatives of H-bond, such as positive values of the electron density ρ

Solvent Structure and Association Properties, and Li-Ion Solvation

ν C -O

2.5

385

1268 ν C=O

1836 ν C (H )-O

R elative Intensity

2.0

δ CO

1073 759 848

1.5

1116

778 851

0.5

0.0 500

1747

δCH

1379

1858

2990

3055

D EC

1093 δ CH

ν C -C

757 944

2

2996 3056 EC+DEC

1255

1017

1068

3

1382 1492

3

1012

1.0

ν C H /ν C H

1727

1372

1492

1000

2993 3065

1500

2000

2500

F requency/cm

3000

EC

3500

-1

Figure 10 The simulated IR spectra for the binary system of EC+DMC with B3LYP/6311++G(d,p).

ν C -O

2.5

ν C=O

1 28 7

1 83 7

1 73 6

ν C -O

2.0

1 11 5

R elative Intensity

1 07 3

1.5

EC +D M C

1 27 4 δ CH

9 70 7 85 9 10 1 18 0

1.0

ν CH

1 75 6 3

1 45 6

3

3 06 3 3 08 8 3 09 6

1 85 8

DMC

ν C -O

0.5

1 06 8

1 09 3 ν CH δ CH

0.0 50 0

10 00

2

15 00

2

2 99 3 3 06 5

20 00 F re qu en c y/cm

25 00

30 00

EC

35 00

-1

Figure 11 Simulated IR spectra for the binary system of EC+DEC with B3LYP/6-311++G(d,p).

386

Lithium-Ion Batteries: Solid-Electrolyte Interphase

(typically in the range of 0.002-0.034 au) and its Laplacian ∇2ρ (range of 0.014 to 0.139 au) at the bond critical point of H•••O, and integrated properties of the 49,50 donor H. Bond critical points between the hydrogen atom and the carbonyl oxygen atom are found in all of the C-H•••O interactions, except for those cases in which the R distance is longer than 2.7 Å such as the one in the dimer D4 (R~2.95 Å), and both in D5 (R~2.76 Å), etc. ρ at the bond critical point of H•••O varies from 0.0060au in D2 to 0.0138au in D10, fitting well within the 2 range of values of normal H-bond interactions. The values of ∇ ρ at the bond critical point of H•••O, which vary between 0.025 and 0.050 au, are also within the range of values for typical H-bonded interactions. Except for elongated C-H•••O interactions (such as R~2.835 Å in D6, R~2.823 Å in T4), the net charges of the donor hydrogen atoms are somewhat increased with respect to those with no connection to an acceptor atom, ranging from 0.01 to 0.08e, which also fulfills another important feature for the formation of H bond, i.e., the loss of the charge of the donor hydrogen atom upon formation of the complex. Other necessary conditions of H bond for the donor hydrogen atom are also checked to further confirm the nature of hydrogen bonding in C-H•••O, i.e., energetic destabilization and decrease of dipolar polarization of the donor H atom. The former can be demonstrated by an increased total energy of the donor H upon complexation. As shown in Table 6, the energy increments for the donor H in D1 and T1 agree well with the destabilization observed in water dimer. Although quite low, the quantities (0.0077 and 0.0142 au) for the donor H in D2 and D7 also lie in the range 49 for normal H bonds. Table 6 also shows that the associations result in the Table 6 Integrated atomic energy (E/au) and the first moment (M of the H atom in the isolated momomer (Hiso) and in the complex (Hcom), and their changes (∆) arising from association. Reproduced from [44] with permission of the Am. Chem. Soc. E(Hiso)

E(Hcom)

∆E

M(Hiso)

D1

−0.5758

−0.5541

0.0217

0.1225

0.0897

−0.0328

T1

−0.5708

−0.5433

0.0275

0.1293

0.0977

−0.0316

D2

−0.5585

−0.5508

0.0077

0.1345

0.0906

−0.0439

DMC

−0.6092

−0.5950

0.0142

0.1428

0.1274

−0.0154

WD

−0.3913

−0.3691

0.0222

0.1510

0.1081

−0.0429

M(Hcom)

∆M

Solvent Structure and Association Properties, and Li-Ion Solvation

387

reduction of the first moment of the donor H between 0.015 (D7) and 0.044 (D2), meeting the other necessary condition for H-bond formation, a decrease of the dipolar polarization for the complexed hydrogen atom. The topological properties of electron densities at the hydrogen bond (H•••O) critical points discussed in the preceding sections show no significant difference for the two sorts of carbonate complexes, even though the results in Tables 4 and 5 indicate that there is a fundamental division between the complexes exhibiting features of conventional H-bond or anti H-bond behavior in terms of C-H bond length and its stretching frequency. The reason is still not unambiguous for the two opposite H-bonds, however several authors claim that 37, 44 it may relate to the different dipole moment variation with C-H bond. In the 2 case of the unsaturated C (sp and sp ) of C-H, in which C-H is stretched upon the formation of C-H•••O, the dipole moment, especially the local dipole 44 moment, increases with bond length, which favors the electrostatic interaction recognized as the main contribution to the H-bond energetics; whereas in the 3 case of saturated C (sp ) of C-H, the dipole moment behaves oppositely, therefore C-H is contracted enhancing the dipole moment and eventually favoring the formation of C-H•••O.

5

+

Li Solvation from Alkyl Carbonates

In organic nonaqueous electrolytes of Li-ion batteries, carbonate molecules + solvate Li , and such solvation not only considerably affects salt dissociation, also the solvent reduction potentials and the subsequent decomposition reactions, e.g., the solvent molecules coordinated to lithium ions more actively 30 react with the electrode. For these reasons, the solvation of lithium ions in electrolyte solutions of lithium-ion batteries has been an interesting and still 12, 51 controversial topic due to its complexity.

+

5.1 Interactions Between Li and Organic Solvents +

Strong interactions are found between Li and the electronegative atoms, such as + + + Li •••O=C for cyclic/linear carbonates, Li •••O=S for sulfites and Li •••O-C for DMM and DME. These interactions can be characterized by quite high binding energies as well as by length variations of the bonds that are directly linked with + Li , as summarized in Table 7. Except for the two ethers, DMM and DME,

388

Lithium-Ion Batteries: Solid-Electrolyte Interphase +

where Li is coordinated with two oxygen atoms, a favorable conformation + results when Li is coordinated with the oxygen of either O=C or O=S groups, + + and Li •••O=C or Li •••O=S tends to be in a linear alignment, although the + conformation in which Li coordinates with the two ether oxygen atoms coexists for linear carbonates. + The interaction between Li and a solvent molecule can also be analyzed + + by the infrared spectra, as illustrated in Figure 12 for the EC+Li and DEC+Li systems. In both cases, the stretching frequency of C=O (νc=o) is significantly 1 1 decreased (1858 vs 1748 cm− for EC; 1747 vs 1630 cm− for DEC), whereas that of O-C (νC-O, O: linking with CH2; C: carbonyl group) is increased (1093 vs 1 1 1226 cm− for EC; 1255 vs 1446 cm− ). Such variations are also in line with the corresponding bond changes, i.e., C=O (1.2241 vs 1.1881 Å for EC; 1.2469 vs. 1.2079 Å for DEC) is stretched while O-C (1.3169 vs 1.3613 Å for EC; 1.3028 vs. 1.3380 Å for DEC) is contracted. Table 7 shows that the C=O or S=O stretching are in the range of 0.035 to 0.043 Å.

1323

2 .5

νC=O

ν C -O

ν C -O

1630

1446

1255

2 .0 R e la tiv e In te n s ity

νCH

1 .5

D EC +Li

1748 νC=O

1 .0

DEC

1858 νC=O ν L i- O

514

774

0 .5

898

1103

1068

+

1 2 2 6 ν C -O E C +Li

1093 ν C -O

+

EC

0 .0 500

1000

1500

2000

2500

3000

3500

F re q u e n c y/c m -1

Figure 12 Simulated

IR spectra on the basis of B3LYP/6-311++G(d,p) calculation for EC,

+

EC+Li , DEC, and DEC+Li+ systems. +

The BE of Li (PC) is approximately 2.0 kcal/mol higher than that of Li (EC). This result supports the speculation of Chung et al. that PC solvates + 52 Li more strongly than EC does, and it also agrees with the finding of +

Solvent Structure and Association Properties, and Li-Ion Solvation

389

Fukushima et al. that the lithium ion is solvated preferentially by PC in a 53 LiClO4/EC-PC-MeOH (0.5:1.0/1.0:20 volume) solution. The binding energies (BE) of the two cyclic carbonates (EC and PC) are higher than those of the three linear carbonates (DMC, DEC and EMC) by a few + kcal/mol. This probably could explain why Li is preferentially solvated by the 5 EC molecule in mixtures of EC and the linear carbonates. DEC has the highest BE among the three linear carbonates, followed by EMC, and DMC the lowest. + Therefore DEC may replace one EC molecule and solvates Li , finally forming + the super-molecule cluster (EC)2Li (DEC) in the EC+DEC+Li salt, however DMC has less chances to form such clusters because of its 3.0 kcal/mol lower BE than DEC, this topic is further discussed in the following subsection. Another interesting feature is that VC and ES, two well-defined additives to PC/EC solutions in Li-ion batteries, have approximately 5 kcal/mol lower + binding energy with Li than EC and PC. This indicates that incorporating specific additives to the EC/PC-based solutions, can somewhat suppress cointercalation of solvent molecules. This is consistent with the recent Table 7 Bond lengths (R1 and R2/Å) and vibrational frequencies (Ȟ1 and Ȟ2/cm−1) of C=O or S=O before and after interaction with Li+, bond length (R3, Å) of Li+•••O/S, and binding energy (∆E, kcal/mol) with B3LYP/6-311++G(d,p) method. Solvents

R1

R2

∆R

Ȟ1

Ȟ2

∆Ȟ

R3

∆E a

EC

1.1881

1.2241

0.0360

1858

1748

−110

1.7386

50.1

VC PC DEC DMC

1.1863 1.1889 1.2079 1.2071

1.2213 1.2258 1.2467 1.2447

0.0350 0.0368 0.0388 0.0376

1875 1854 1748 1756

1779 1740 1630 1645

−96 −114 −118 −111

1.7484 1.7348 1.7264 1.7346

45.9 51.7 46.9 43.8

EMC ES PS GBL DMF DMA DMSO DME

1.2073 1.4616 1.4624 1.1972 1.2166 1.2243 1.5143

1.2458 1.5049 1.5002 1.2324 1.2553 1.2648 1.5537

0.0385 0.0433 0.0378 0.0352 0.0387 0.0405 0.0394

1753 1187 1184 1813 1718 1681 1030

1637 1096 1095 1704 1674 1617 1022

−116 −91 −89 −109 −44 −64 −8

1.7304 1.7474 1.7545 1.7444 1.7206 1.7092 1.7069 1.8628 1.8628 1.8942 1.8947

45.4 46.4 48.0 51.4 55.2 57.5 59.6 62.4

DMM a

∆E=E(Li+)+E(S)-E(Li+-S)

57.1

390

Lithium-Ion Batteries: Solid-Electrolyte Interphase 54

observation by Jeong et al. In other words, co-intercalation of the two additives may not take place because their desolvation from Li ions, which migrate from the solution to the intercalation sites of electrodes, is easier than the supporting solvents. It is also of interest to note that ethers cosolvent or addtioves to PC-based + solutions, such as DMM and DME, have quiter higher binding energies with Li + than PC and EC. It tells us that such solvents preferentially solvate Li and solvent co-intercalation may happen as they are present, which may prevent co-intercalation of PC and thereby suppress exfoliation of graphite layer for 59 PC-based solutions.

5.2

Solvation Number of Li

+

+

Supermolecule clusters, such as Li (S1)x(S2)y (S1 and S2: solvent molecules; x and y: coordination numbers), were fully optimized by B3PW91/6-31G(d) method. Table 8 summarizes Gibbs free energies of formation. Although the + + ∆Grs for the formation of coordinated clusters, such as Li (EC) and Li (PC), are very negative (−42.1 and −43.6 kcal/mol), and their subsequent reactions, i.e., + + the formations of the 2-coordinated clusters Li (EC)2 and Li (PC)2, are also very + spontaneous (∆Gr: −27.6 and −28.7 kcal/mol), it can be predicted that Li (S) (S=EC, and PC) will not be the leading species in EC/PC solutions. + Additionally, the positive and small negative ∆Gr for the formations of Li (PC)4 + and Li (EC)4 indicate that they are not either the leading species in the PC and + + EC-based solutions, respectively. It could be concluded that Li (S)2 and Li (S)3 are most probably the two major solvated lithium ion species. The result is in 51 line with the recent electrospray ionization mass spectroscopy (ESI-MS) study + in which a four-coordinated Li species was not found. Dynamic DFT simulations with the CPMD method qualitatively support the above results. As shown in Figure 13, only three EC molecules strongly + + coordinate with Li through Li •••O=C interactions, which is different from + classical molecular dynamics simulations result. The distances between Li and 30 O are approximately 0.2-0.3Å longer than those obtained from static DFT most probably due to the bulk solvent effect incorporated to the CPMD simulations. In the cases of EC+DMC and EC+DEC solutions, since ∆Gr differences of approximately 10 kcal/mol exist between EC and DMC as well as with DEC + (−42.1 vs. −31.8 and −34.9 kcal/mol) for the one-coordinated Li (S), it is + + unlikely that considerable amounts of Li (DMC) and Li (DEC) are present in

Solvent Structure and Association Properties, and Li-Ion Solvation

391

Table 8. Gibbs free energies of reactions (∆Gr, kcal/mol) at 298.2K calculated by B3PW91/631G(d). Reactions 1. Li++PC→Li+(PC) 2. Li+(PC)+PC→Li+(PC)2 3. Li+(PC)2+PC→Li+(PC)3 4. Li+(PC)3+PC→Li+(PC)4 5. Li++EC→Li+(EC) 6. Li+(EC)+EC→Li+(EC)2 7. Li+(EC)2+EC→Li+(EC)3 8. Li+(EC)3+EC→Li+(EC)4 9. Li++DMC→Li+(DMC) 10. Li+(EC)+DMC→(EC)Li+(DMC) 11. Li+(EC)2+DMC→(EC)Li+(DMC)+EC 12. Li+(EC)2+DMC→(EC)2Li+(DMC) 13. Li++DEC→Li+(DEC) 14. Li+(EC)+DEC→(EC)Li+(DEC) 15. Li+(EC)2+DEC→(EC)Li+(DEC)+EC 16. Li+(EC)2+DEC→(EC)2Li+(DEC) 17. Li++VC→Li+(VC) 18. Li+(PC)+VC→(PC)Li+(VC) 19. Li+(PC)2+VC→ (PC)Li+(VC)+PC 20. Li+(PC)2+VC→(PC)2Li+(VC) 21. Li+(PC)3+VC→(PC)2Li+(VC)+PC 22. Li++GBL→Li+(GBL)

∆Gr′

∆Gr

−46.0 −31.9 −14.6 −1.6 −45.0 −30.8 −12.7 −5.9 −34.4 −24.3 6.9 −10.6 −37.6 −26.2 4.9 −10.8 −40.5 −27.6 4.3 −12.0 2.6 −45.5

−43.6 −28.7 −11.2 +2.0 −42.1 −27.6 −9.4 −2.4 −31.8 −21.3 −7.3 −34.9

−7.6 −37.9 −24.6

∆Gr =∆Gr′ +BSSE

the binary solutions. The substitutions (reactions 11 and 15 in Table 8), + + + generating (EC)Li (DMC) and (EC)Li (DEC) from Li (EC)2 respectively, are forbidden as indicated by their positive ∆Gr; however, the additions (reactions 10 and 14) are thermodynamically possible, this means that the two crosscoordinated clusters could be generated through the latter. Again the DEC+ + containing cluster (EC)Li (DEC) has higher negative ∆G than (EC)Li (DMC), + which may partially explain the fact that (EC)Li (DEC) was observed in the + mixture of LiClO4/EC-DEC whereas (EC)Li (DMC) was not found in 53 + LiClO4/EC-DMC using EI MS techniques. Similarly, Li (VC) probably does + not exist in the mixture of PC+VC+Li salt, however (VC)Li (PC) and + (VC)Li (PC)2 can be generated through addition reactions (18, 20) instead of substituent reactions (19, 20).

392

Lithium-Ion Batteries: Solid-Electrolyte Interphase

Figure 13 Snapshot of Li+ solvation in EC solvent obtained by Car-Parrinello molecular dynamics (CPMD) simulation at 350.0K. Li+ and the three strongly coordinated EC molecules are highlighted.

Another interesting case is the mixture of GBL with EC or PC. As indicated + + by the ∆G of Li (GBL) (reaction 22) very similar to those of Li (PC) and + Li (EC)(−45.5 vs −45.0 and −46.0 kcal/mol), various two and three coordinated + Li clusters by EC as well as GBL are present in the mixture of + + + + EC+GBL+LiCLO4, including Li (EC)2, Li (GBL)2, (EC)Li (GBL), [Li (EC)3], + + + [Li (GBL)3], [Li (EC)2(GBL)], and [Li (EC)(GBL)2], as revealed by EI MS 53 measurements. CPMD simulations are currently carried out to incorporate more practical + state conditions, including temperature and concentration, and solvation of Li in mixtures such as EC-DMC, EC-DEC, PC-DMC, PC-DEC, GBL-DEC.

Solvent Structure and Association Properties, and Li-Ion Solvation

6

393

Conclusion: Implications for SEI Layer Phenomena

In this chapter, structures and associations of various commonly used organic solvents (EC, PC, VC, DMC, DEC, EMC, ES, PS, GBL, DMSO, DMA, DME, + and DMM) and their interaction with Li (solvation) were analyzed with static ab initio MO/DFT and dynamic ab initio MD (CPMD). These results could help understand some aspects of SEI layer related phenomena at the molecular level, which are summarized as follows: (1) We strongly suggest that the stable structure of EC in gas phase as well as in liquid phase is non-planar. Its ring twists approximately 20-30° with respect to the dihedral angle of O-C-C-O. (2) Much stronger association exists in cyclic carbonate molecules than in linear carbonate molecules, for example, binding energies of dimers decrease in the order of EC-EC ~ EC-PC > PC-PC > VC > DMC (D0 5.08 ~5.08 > 4.69 > 3.04 >1.66 kcal/mol). Based on the similarities among the hetero-molecular dimer of EC/PC, and the EC and PC respective homo-molecular dimers in terms of both energetic and geometric properties, it can be concluded that PC may not destroy the associated structure of EC in their mixture, whereas the differences between the hetero-molecular dimer of EC/DMC and the homo-molecular dimers of EC and DMC may lead to the conclusion that EC would be less associated in the mixture of EC+DMC. This implies that linear carbonates, like DMC, DEC and EMC, are proper co-solvents of EC that may increase the fluidity of their mixtures. (3) Both static and dynamic DFT analyses indicate that the highest coordination + number of Li in EC solvent most probably is 3 rather than 4. This is opposite to our conventional knowledge based on speculations from experimental results and results from classical molecular dynamics simulations. +

(4) The interaction between Li and various solvents could partially explain some phenomena relevant to SEI layer formation. As a co-solvent of EC, for example, it is suggested that DEC is at least easier to intercalate into the graphite anode than both DMC and EMC because of its stronger interaction + with Li , which increases the irreversible capacity of carbons used in LIBs.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

394

(5) VC and ES, two well-defined additives to PC/EC solutions in Li-ion + batteries, have approximately 5 kcal/mol lower binding energy with Li than EC and PC. This indicates that incorporating specific additives to the EC/PC-based solutions, can somewhat suppress cointercalation of solvent molecules. In other words, cointercalation of the two additives may not take place because of their easier desolvation from Li ions than that provided by the supporting solvents. Definitely VC and ES play their roles as additives to PC-based solutions mainly by their preferential reduction and subsequent generation of better SEI forming species, as discussed in Chapter 5 and elsewhere in this book. (6) It is also of interest to note that either type of cosolvent or additives to PCbased solutions, such as DMM and DME, have quiter higher binding energies + + with Li than PC and EC. It tells us that such solvents preferentially solvate Li and solvent co-intercalation may happen as they are present, which may prevent co-intercalation of PC and thereby suppress exfoliation of graphite layer for PC59, 60 based solutions. DMM and DME work in different way than VC and ES. (7) The interactions of C-H•••O through which alkyl carbonates clearly associate, can be identified as hydrogen bonds and they are fundamentally similar to the conventional hydrogen bonds. However, C-H•••O exhibits two 2 opposite behaviors. The cases of unsaturated C (sp and sp ) in C-H bonds, such as those in VC, show C-H stretching upon the formation of hydrogen bonds; 3 however, in the case of saturated C (sp ) of C-H, the C-H bond is contracted and its stretching frequency blue shifts. Therefore, C-H•••O behaves as an antinormal hydrogen bond for the latter case.

References 1. 2. 3.

4. 5.

Peled E., J. Electrochem. Soc. 126 (1979), 2047. Smart M. C., Ratnakumar B. V. and Surampudi S., J. Electrochem. Soc. 146 (1999), 486. Smart M. C., Ratnakumar B. V., Surampudi, S., Wang Y., Zhang X., Greenbaum S. G., Hightower A., Ahn C. C. and Fultz B., J. Electrochem. Soc. 146 (1999), 3963. Yang C. R., Wang Y. Y. and Wan C. C., J. Power Sources 72 (1998), 66. Jeong S. K., Inaba M., Iriyama Y., Abe T. and Ogumi Z., Electrchim. Acta 47 (2002), 1975.

Solvent Structure and Association Properties, and Li-Ion Solvation

6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.

17.

18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28.

395

Brown C. J., Acta Crystallogr. 7 (1954), 92. Matias P. M., Jeffrey G. A., Wingert L. M. and Ruble J. R., J. Mol. Struct. (THEOCHEM) 184 (1989), 247. Wang J., Britt C. O. and Boggs J. E., J. Am. Chem. Soc. 87 (1965), 4950. Fortunato B., Mirone P. and Fini G., Spectrochim Acta 27A (1971), 1917. Pethrich R. A. and Wilson A. D., Spectrochim. Acta 30A (1974), 1073. Angell C. L., Trans. Faraday Soc. 52 (1956), 1178. Klassen B., Aroca, R., Nazri, M. and Nazri G. A., J. Phys. Chem. B 102 (1998), 4795. Blint, R. J., J. electrochem. Soc. 142 (1995), 696. Besenhard J. O. and Fritz H. P., J. Electrochem. Soc. 3 (1974), 329. Shu Z. X., McMillan R. S., Murray J. J. and Davidson I. J., J. Electrochem. Soc. 142 (1995), L161. Jehoulet C., Biensan P., Bodet J. M., Broussely M. and Tessier-Lescourret C., Batteries for Portable Applications and Electric Vehicles, The Electrochemical Society Proceedings Series, 1997, Pennington, NJ. Perdew J. P., Unified theory of exchange and correlation beyond the local density approximation. In Electronic Structure of Solids, ed. by Ziesche P. and Eschrig H. (Akademie Verlag, Berlin, 1991). Adamo C. and Barone V., Chem. Phys. Lett. 274 (1997), 242. Lee C., Yang W. and Parr R. G., Phys. Rev. B 37 (1988), 785. Bauschlicher C. W., Chem. Phys. Lett. 246 (1995), 40.] Boys S. F. and Bernardi F., Mol. Phys. 19 (1970), 553. Miertus S., Scrocco E. and Tomasi J., Chem. Phys. 55 (1981), 117. Miertus S. and Tomasi J., Chem. Phys. 65 (1982), 239. Cossi M., Barone V., Cammi R. and Tomasi J., Chem. Phys.Lett. 255 (1996), 327. Barone V. and Cossi M., J. Phys. Chem. A 102 (1998), 1995. Foresman J. B., Keith T. A., Wiberg K. B., Snoonian J. and Frisch M. J., J. Phys. Chem. 100 (1996), 16098. Car R. and Parrinello M., Phys. Rev. Lett. 55 (1985), 2471. Frisch M. J., Trucks G. W., Schlegel H. B., Scuseria G. E., Robb M. A., Cheeseman J. R., Zakrzewski V. G., Montgomery J. A., Stratmann R. E., Burant J. C., Dapprich S., Millam J. M., Daniels A. D., Kudin K. N., Strain O. F. M. C., Tomasi J., Barone B., Cossi M., Cammi R., Mennucci B., Pomelli C., Adamo C., Clifford S., Ochterski J., Petersson G. A., Ayala P. Y., Cui, Q., Morokuma K., Malick D. K., Rabuck A. D., Raghavachari K., Foresman J. B., Ciolovski J., Ortiz J. V., Stefanov V. V., Liu G., Liashensko

396

29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50.

Lithium-Ion Batteries: Solid-Electrolyte Interphase

A., Piskorz P., Komaroni I., Gomperts R., Martin R. L., Fox D. J., Keith T., Al-Laham M. A., Peng C. Y., Nanayakkara A., Gonzalez C., Challacombe M., Gill P. M. W., Johnson B., Chen W., Wong M. W., Andres J. L., HeadGordon M., Replogle E. S. and Pople J. A., GAUSSIAN 98, Revision A.11 (Gaussian Inc., Pittsburgh, PA, 1998). Hutter J. et al., copyright IBM Zurich Research Laboratory and MPI fur Festkorperforschung 1995-2001. Wang Y. X., Nakamura S., Ue M. and Balbuena P. B., J. Am. Chem. Soc. 123 (2001), 11708. Alonso J. L., Esposti A. D., Lister D. G. and Palmieri P., J. Chem. Soc., Faraday Trans. 2 82 (1986), 337. Fujii A., Fujimaki E., Ebata T. and Mikami N., J. Am. Chem. Soc. 120 (1998), 13256. Jedlovszky P. and Turi L., J. Phys. Chem. B 101 (1997), 5429. Houk K. N., Menzer S., Newton S. P., Raymo F. M., Stoddart J. F. and Williams D. J., J. Am. Chem. Soc. 121 (1999), 1479. Scheiner S., Gu Y. and Kar T., J. Mol. Struct. (THEOCHEM) 500 (2000), 441. Scheiner S., Hydrogen Bonding: A Theoretical Perspective (Oxford University Press, New York, 1997). Hobza P. and Havlas Z., Chem. Phys. Lett. 303 (1999), 447~452. Gu Y., Kar T. and Scheiner S., J. Am. Chem. Soc. 121 (1999), 9411. Hobza P. and Havlas Z., Chem. Rev. 100 (2000), 4253. Desiraju G. R., Acc. Chem. Res. 29 (1996), 441. Hartmann M., Wetmore S. D. and Radom L., J. Phys. Chem. A 105 (2001), 4470. Bader R. F. W., Atoms in Molecules: A Quantum Theory (Clarendon, Oxford, 1990). Cubero E., Orozco M., Hobza P. and Luque F. J., J. Phys. Chem. A 103 (1999), 6394. Wang Y. X. Balbuena P. P., J. Phys. Chem. A 105 (2001), 9972. Curtiss L. A., Frurip D. J. and Blander M., J. Chem. Phys. 71 (1979), 2703. Reimers J., Watts R. and Klein M., Chem. Phys. 64 (1982), 95. MÓ O., Yáñez M. and Elguero J., J. Chem. Phys. 97 (1992), 6628. MÓ O. and Yáñez M., J. Chem. Phys. 107 (1997), 3592. Koch U. and Popelier P. L. A., J. Phys. Chem. 99 (1995), 9747. Popelier P. L. A. J. Phys. Chem. A 102 (1998), 1873.

Solvent Structure and Association Properties, and Li-Ion Solvation

397

51. Fukushima T., Matsuda Y., Hashimoto H. and Arakawa R., Electrochem. Solid-State Lett. 4 (2001), A127 52. Chung G.-C., Jun S.-H., Lee K.-Y. and Kim M.-H., J. Electrochem. Soc. 146 (1999), 1664. 53. Matsuda Y., Fukushima T., Hashimoto H. and Arakawa R., J. Electrochem. Soc. 149 (2002), A1045. 54. Jeong S.-K., Inaba M., Mogi R., Iriyama Y., Abe T. and Ogumi Z., Langmuir 17 (2001), 8281. 55. Wang Z. X., Huang X. J. and Chen L. Q., Spectroscopic studies of solidelectrolyte interphase on positive and negative electrodes for lithium ion batteries. In Lithium-Ion Batteries: Solid-Electrolyte Interphase, ed. by Balbuena P. B. and Wang Y. X. (Imperial College Press, London, 2004). 56. Wrodnigg G. H., Besenhard J. O. and Winter M., J. Electrochem. Soc. 146 (1999), 470-472. 57. Wrodnigg G. H., Wrodnigg T. M., Besenhard J. O. and Winter M., Electrochem. Commun. 1(1999), 470-472. 58. Becke A. D., J. Chem. Phys. 98(1993), 5648. 59. Ogumi Z., Jeong S.-K., Inaba M. and Abe T., Macromol. Symp 156 (2000), 195. 60. Inaba M. and Ogumi Z., Scanning probe microscopy of the SEI formation on graphite anode. In Lithium-Ion Batteries: Solid-Electrolyte Interphase, ed. by Balbuena P. B. and Wang Y. X. (Imperial College Press, London, 2004).

INDEX 1,2-dimethoxy-ethane, 366 1,2-dimethoxy-methane, 366 12–crown–4, 214, 220 6-311++G(d,p), 229, 231–234, 237, 239–240, 243–244, 367–373, 377, 380–383, 385, 388–389 6-31G(d), 229–230, 233–234, 237, 239–240, 243–244, 246, 249–250, 256, 367, 370, 390–391

anodic polarization, 75, 76 anti-hydrogen bond, 374 armchair, 13, 17, 42, 49, 60, 267 As2O5, 50 atomic force microscopy, 86, 157, 201 atoms-in-molecules, 374 Auger electron spectroscopy, 82 Aurbach, 7, 8, 11, 45, 61–63, 65–66, 132–139, 145, 150, 155, 161, 178– 179, 200, 222–224, 228–229, 235, 257, 259, 274, 275, 307, 315, 332– 333, 362–364 average thickness of the interphase, 11

α−bromo-γ−butyrolactone, 214, 220 α-NaFeO2, 354 ab initio MD, 393 ab initio, 228, 367–368, 393 AC impedance, 46, 59 active metals, 71–72, 75, 88, 90, 93, 107, 129, 131 AES, 77, 82, 87 AFM, 2, 17, 59, 77, 86–87, 124–127, 157, 201, 206–210, 212, 218–219, 221, 228 Ag electrodes, 141, 160, 166 AIM, 249, 374, 377, 384 Al2O3, 12, 168, 315, 353 alkaline, 2, 70, 72, 131 alkaline earth, 70, 72, 131 alkaline earth batteries, 2 alkaline earth metals, 72 alkaline metals, 70, 72 alkanes, 75, 107 alkenes, 75, 107 alkyl ammonium cations, 75 alkyl carbonate, 10, 24–25, 46, 75–76, 88, 107, 109–111, 114–115, 117– 118, 120, 125–126, 128, 130, 142, 149–150, 156, 199, 207, 211, 214, 227, 246, 259, 273, 296, 365, 375, 394 alloying, 54, 142, 151, 155, 309 aluminum, 70, 73, 76,171 anion doping, 354 anode surface, 3, 156, 228, 235, 241, 255, 260, 296 anode-surface layer, 1

B3PW91, 229–234, 237, 239–240, 243–244, 246, 249–251, 253, 256, 367, 374, 376, 390–391 basal planes, 13–14, 16, 21, 23–24, 27, 29, 31–32, 44, 60 basal SEI, 17, 19, 20, 21, 24–27, 30– 31, 60 basis set superposition error, 230 Besenhard, 16, 54, 62, 65–68, 137, 200–201, 205, 222–223, 226, 228, 273–274, 333, 364, 395, 397 BET, 13, 45, 47, 328 bivalent metal, 75 BL, 4, 7, 9–10, 119, 121, 366 Blomgren, 54, 69 Bode, 94, 95 Brønsted acidity, 317 Brookhaven National Laboratory, 323 BSSE, 230, 248, 250, 262–263, 265, 367, 375–376, 391 Butler Volmer, 72 C1s, 17, 30, 34–35, 39, 58, 341–342, 344, 348–349, 351, 356–357 C2H2Li2, 325, 336 Ca, 72–73, 78, 88, 93, 129–131 cage-like, 247–248, 250, 252, 254–255, 259, 269 calcium, 71, 75, 129, 130 calorimetry, 2 398

Index capacity fading, 77, 131,152 carbon anodes, 60, 76, 280, 287–288, 296, 314, 334–335 carbon black, 45, 75, 170, 324, 341 carbon electrodes, 7, 13–14, 31, 42, 45, 47, 73, 90, 97, 100, 116, 118–120, 124, 178, 311, 314–315, 324, 328, 333 carbon precursor, 317 carbonaceous anodes, 16, 60, 140, 227 carbonaceous electrodes, 13, 33 carbonaceous materials, 3, 33, 43, 53,159, 308–311, 317, 322, 324– 325, 328, 330–332, 334, 336 carboxylate groups, 105 cathode-electrolyte interface, 338, 343 cathodic polarization, 73, 75–76, 90, 92, 127 (CHCHOCO2Li)2, 247, 251 CH2Cl2, 73 (CH2OCO2Li)2, 30, 35, 46, 50, 100, 109, 118, 120, 163, 179, 199, 243, 246–247, 252 (CH3)2SO, 73 CH3CH(OCO2Li)CH2(OCO2Li), 247, 253 CH3CH(OCO2Li)CH2-OCO2M, 9 CH3CH2OCO2Li, 155, 163 CH3CH2OLi, 102–103, 163 CH3N, 73 CH3OCH2CH2OLi, 102–103, 108–109, 121 CH3OCO2Li, 30, 35 (CHOCO2Li)2, 120, 243, 247–248, 262 CH3OLi, 108, 109, 121 chain-like, 247, 249, 252 charge transfer, 1, 71, 76, 94, 97, 99, 129, 130, 167 CHELPG, 230 chemical adsorption, 174, 176, 185 chemical vapor deposition, 33 chemically bonded SEI, 51, 52 CHI660 Electrochemical Workstation, 159 chloroethylene carbonate, 214, 228, 366 chronoamperometry, 94 chronopotentiogram, 92

399 chronopotentiometry, 94 coating layer, 168, 169, 187, 189 co-intercalation, 15, 53, 120, 122, 142, 201, 204–207, 210–212, 214, 219– 221, 273, 365 cokes, 33, 47, 311, 332 compact-stratified layer, 6 composite anodes, 55–57, 141, 201 composite electrodes, 75, 85 conductor-like PCM, 229, 367 conservation of mass, 285 continuum mechanics, 280–282, 291 continuum models, 229, 280–281, 287, 296, 367 conventional H-bond, 374, 382, 387 copper electrodes, 9 copper, 9, 55–57, 90, 144, 171 copper-tin, 309 corrosion, 1–2, 4, 70, 313, 339 CPCM, 229–232, 234–235, 238–239, 242, 367, 370–373 CPCM-B3PW91/6-311++G(d,p), 229 CPCM-B3PW91/6311++G(d,p)//B3PW91/6-31G(d), 230 CPMD, 368, 390, 392–393 cross-associations, 365–366 cross-section, 17, 20–21, 23–27, 29–31, 44, 60 cross-sectional SEI, 20–21, 24–25, 27 crown ethers, 121 CV profiles, 155 CVD, 33 cyclic compounds, 10 cyclic voltammetry, 8, 9, 11, 46, 48 cyclic voltammograms, 155, 160, 202 Dahn, 14, 54–55, 57, 63–65, 68–69, 137, 141, 145, 222, 224–225, 307, 332, 335–336, 362, 364 dangling bonds, 52 dangling, 14 de-alloying, 151 DEC, 5, 8–9, 17, 23, 29, 37, 45–46, 49, 55, 114–115, 118, 142–144, 146– 152, 154–156, 159, 162–163, 165, 198, 202–205, 207–209, 211–214,

400

Lithium-Ion Batteries: Solid-Electrolyte Interphase

216, 315, 360, 365–366, 368, 376, 378, 384–385, 388–390, 392–393 DEE, 214, 220 de-intercalation, 313 DEM, 214, 220 DEMS, 48 density functional theory (DFT), 228, 229, 233, 235, 247, 259–260, 367– 368, 390, 393 depth profile, 8, 17, 19, 21, 24, 25, 37, 59 diamond-like carbon, 168 diamond-like, 168, 310 dianions, 89 diethoxyethane, 214 diethoxymethane, 214 diethyl carbonate electrolytes, 34 dimethoxymethane, 214 dimethyl amide, 366 dimethyl formamide, 171, 366 dimethyl sulfoxide, 366, 214 dioxolane solutions, 16 disordered carbon electrodes, 3, 39 disordered carbonaceous materials, 311 disordered carbons, 1, 14, 32, 34, 37, 42, 44, 311, 315, 321–322, 325, 328, 335 disordered-carbon anode, 55 DMA, 366, 389, 393 DMC, 5, 8–10, 12, 22, 44–48, 92, 99, 109, 112, 114–115, 121, 143, 154– 155, 159–161, 170–172, 175–176, 178–179, 181–182, 189, 198, 209, 211–214, 295, 324, 340, 343, 348, 350–351, 365–366, 368, 376–378, 382, 384–386, 389–390, 392–393 DME, 9, 45, 46, 143, 145, 153–156, 204–205, 207, 366, 369, 387, 389, 393 DMF, 73, 171, 366, 389 DMM, 214, 220, 366, 368, 387, 389, 393 DMO, 5 DMSO, 73, 214, 366, 389, 393 DSC, 2, 50, 51, 59, 346–347 dynamic continuum models, 287 (EC)Li+(DEC), 391

(EC)Li+(DMC), 391 (EC)Li+(GBL), 392 EC + DEC, 203, 211–214, 216 EC + DMC, 212–214 EC, 4–5, 8–9, 10, 12, 16–17, 22–23, 29, 44–49, 52, 55, 92, 99, 100, 102, 109, 112, 114–118, 120, 127–128, 142–144, 146–152, 154–156, 159, 162, 165, 170–178, 180–182, 185, 187, 198–213, 216, 220–221, 227– 229, 231–246, 260, 269, 271–273, 295, 315, 324, 340, 343, 349, 351, 358, 360, 365–366, 368, 370–376, 378–380, 382–385, 388–394 EC/DMC, 48, 143, 178, 393 EC-DMC, 108, 112, 391 EC-PC-DMC, 46 ECQCM, 157 ECS, 2, 67 EDAX, 77, 82 edge planes, 14, 28, 42, 45, 60, 122, 124, 127, 199, 205, 268 EDS, 2, 59 EELS, 157 Ein-Eli, 45, 65–67, 134, 135, 136–138, 332–333, 362, 364 EIS, 55, 157, 338 electroadsorption processes, 81 electrochemical impedance spectroscopy, 338 electrochemical quartz crystal microbalance, 9, 46, 83, 157 electrodeposited, 9, 310 electron tunneling, 2, 14, 90, 145 electronic conductivity model, 296 enhancement mechanism, 158 E-NP15, 47 EPR, 2, 59 EQCM, 46, 77, 83–84 ESI-MS, 390 E-SLX2050, 47 ethylene carbonate, 34, 48, 50, 142, 170, 198, 227–228, 295, 324, 333, 349, 360, 365 ethylene glycol, 207 ethylene sulfite, 214, 229 EXAFS, 77, 82 exchange functional, 230, 367

Index exchange-current density, 4, 25, 43, 60 exfoliation, 14, 25, 42, 51, 117, 120, 142, 200, 202, 205–206, 214, 218, 220–221, 315, 328 extended X-ray absorption fine structure, 82 F1s, 31, 35, 341–342, 351, 356–357 face-centred-cubic, 355 Farrington, 16, 63 Fe3+/Fe2+, 359 fibers, 33, 45, 117, 141, 308, 311, 318 film-metal interface, 71 fluorescence, 309, 321, 336 fluorine-containing salts, 10 fluoroethylene sulfite, 214 Fourier transform spectroscopy, 9 Fourier transform infrared spectroscopy (FTIR), 2, 59, 77–80, 87, 100, 102– 105, 107–109, 111, 114, 116, 118– 119, 122–123, 128, 130, 140–141, 143, 145–151, 155–157, 162–163, 165, 167, 171, 173, 178, 180, 199, 247, 255, 257–258 γ-butyrolactone, 7, 48, 105–106, 119, 130, 366 gas evolution, 45, 48, 206, 245–246, 273 Gaussian 98, 230 GBL, 48, 389, 391–393 gel electrolytes, 11–12, 116 GIC, 141, 201, 228 gold electrode, 8 gradient-corrected correlation functional, 230 graphite anodes, 15, 48, 50, 52, 120, 126, 158, 198–199, 201–203, 205, 210–212, 214, 216, 221–222, 229, 232, 261, 267, 269, 272, 338–339, 343, 346, 352, 366, 393 graphite crystallites, 45, 51 graphite electrodes, 46–48, 50, 92, 97, 99, 117–123, 126–128, 150, 198, 214, 228, 247, 271, 273, 278, 365 graphite exfoliation, 214 graphite intercalation compounds, 141, 202

401 graphite, 198 graphite-like, 33, 298, 310 GTO, 367 hard carbon, 36–44, 312 Hartree-Fock, 367 HCON(CH3)2, 73 highly ordered pyrolytic graphite, 3, 17, 199, 228 high-resolution transmission electron microscopy, 140 HOPG, 3, 16–24, 27–28, 32, 40, 43–44, 60, 126, 199–210, 212, 216–220, 228 HPEs, 12 HRTEM, 140, 143–146, 157, 160 Hunter’s reaction, 339, 350 IMLB, 2 impedance spectra, 46, 94–95, 99 impedance spectroscopy, 9, 11, 93–94, 129, 157, 199, 277 impedance spectrum, 94 inert electrodes, 7, 16 inert metal, 7 inert substrate, 7, 10 intercalation compounds, 298, 310, 313 interfacial phenomena, 13, 159, 165 ionic migration, 3 IPCM, 229, 367 iPrOLi, 109 irreversible capacity, 13–15, 44, 47, 49, 52–53, 55, 117, 142, 199, 202, 210– 212, 315–316, 319, 321–322, 329, 338 irreversible capacity loss, 14, 44, 316, 322, 338 isodensity PCM, 229, 367 ISPE, 2 K-edge, 324, 326–331 Kohn-Sham, 367–368 KS6-KS44, 47 Kuribayashi, 52, 67 Lascovich, 33, 65 lattice vibrational modes, 160 lattice-gas model, 296, 305

402

Lithium-Ion Batteries: Solid-Electrolyte Interphase

LDVD, 247, 251, 256–258 LED, 247, 252–256, 258, 262–263, 265–269, 271–272 Li alloy electrodes, 76 Li/Ag cell, 160–161 Li+(DEC), 390–391 Li+(DMC), 390–391 Li+(EC), 231, 391–392 Li+(EC)2, 234–236, 242, 390–392 Li+(EC)3, 234, 391–392 Li+(EC)4, 390–391 Li+(EC)n, 232–233, 235 Li+(GBL)2, 392 Li+(PC)2, 233–237, 390–391 Li+(PC)3, 234, 391 Li+(PC)4, 390–391 Li+(PC)n, 233, 235 Li+(VC), 231, 239–240, 242, 244, 246, 391 Li1s, 341–342, 351, 356–357 Li2CO3, 4, 6–11, 14, 16, 19, 25, 45–46, 50, 59–60, 100, 107–108, 114–115, 119–121, 123, 128, 140–142, 145– 146, 148–152, 156–157, 160–166, 178–179, 182, 185, 189, 199, 235– 236, 243, 246–247, 273, 294, 296, 342, 352, 355–358, 360 Li2O, 8–10, 16, 19–20, 30, 35, 37, 41, 50, 59–60, 92, 113, 115, 120, 315, 320, 351 Li2O2, 115–116 Li2PF6+, 25 Li2S, 46, 113, 120 Li2S2O5, 46 Li2SO2O4, 46 Li2SO3, 46, 113, 120 Li-Ag alloy, 141 LiAsF6, 4, 7, 9–11, 21, 34, 38–42, 44– 46, 76, 92, 94–95, 99–100, 102, 104, 109, 111–114, 116, 118–119, 122–123, 178, 227, 339 LiBF4, 4, 7, 9–10, 45–46, 48, 76, 96– 97, 100, 102, 108–109, 113, 115, 118, 339, 344, 348–352 LIBs, 198–199, 211, 214, 221, 227– 229, 238, 241, 244–245, 261, 272– 273, 365–366, 393 LiC(SO2CF3)3, 100, 102, 104, 112–113

LiC6, 33, 141, 313, 339 LiCF3SO3, 339 LiCl, 4, 7, 16, 49, 60, 113, 129 LiClO4, 6–10, 12, 21, 44–46, 51, 76, 91, 96, 100, 102, 108, 113, 116, 118, 127, 143, 145, 150, 153–156, 159–162, 201–209, 212–214, 216– 220, 315, 339, 389, 391 LiClO4/EC + DEC, 207 LiCN, 12 LiCoO2, 48, 56, 76, 97, 99, 128, 140– 141, 167–189, 227, 339, 354–355, 359, 361, 366 LiF, 4, 7–11, 15, 20–21, 25, 31, 35, 37, 39–40, 42, 44–45, 47, 50, 59–60, 92, 109, 113–116, 128, 142, 162– 165, 167, 296, 341–342, 350–351, 355–358, 361 LiFePO4, 339, 359, 360, 361 Li-imide, 10, 13, 46 Li-ion diffusion, 97 Li-ion migration, 76, 94, 129 LiMn2O4, 76, 97, 99, 128, 158, 167– 168, 227, 339, 341–342, 344–346, 349, 351–353, 355, 358, 361 LiN(SO2CF3)2, 9–10, 100, 102, 104, 113, 116 linear carbonates, 227, 365, 368, 387, 389, 393 LiNi0.8Co0.2O2, 354, 356–359, 361 LiNi1-xCoxO2, 339, 354 LiNiO2, 76, 97, 99, 128, 168, 227, 339, 354–356, 361 LiNiyCo1-yO2, 167 LiO2, 115, 116 LiOCO2R, 9 LiOH, 7–10, 60, 100, 115, 141, 162– 163, 165–166, 199 LiOH·H2O, 141, 162–163, 165–166 LiOR, 9, 236 LiPF6, 7, 9–11, 17, 21–23, 25, 29–31, 34–37, 39–44, 47–49, 55, 76, 109, 111–115, 118, 128, 142–144, 146– 150, 152, 154–156, 159, 162–163, 165, 167, 170, 172, 175–176, 178– 179, 227, 295, 315, 324, 339–340, 343–344, 347–351, 355, 358, 360– 361

Index LiSn alloy, 309 LiSO3CF3, 9–10, 100, 102 lithiated carbon electrodes, 14, 97, lithiated carbon, 14, 46, 52, 73, 75–76, 92–93, 97, 105, 116, 119, 124, 291, 315 lithiated graphite, 50–51, 85, 97, 99, 117, 120 lithiated graphite electrode, 51, 99 lithium alkyl dicarbonates, 230, 242, 247, 250, 254, 257, 259–261, 263, 267, 269 lithium alloys, 53, 55, 310, 313, 319 lithium anode, 11, 310, 335 lithium battery, 1, 2, 7, 287, 335 lithium carbonate, 1, 10, 20, 25, 30, 35, 162, 235–236, 242–243 lithium di-vinylene dicarbonate, 247 lithium electrode, 85, 94, 103, 109, 309 lithium ethylene dicarbonate, 242, 247, 253, 258, 260, 265, 269–270 lithium hexafluorophosphate, 46 lithium intercalation, 14, 16, 25, 33, 46–47, 52–53, 59, 198, 207, 296, 300, 303–305, 311, 313, 332, 334 lithium metal alloys, 309 lithium propylene dicarbonate, 236, 247, 257, 260, 269–271 lithium vinylene dicarbonate, 247, 249, 251, 255, 260, 262, 269–270 lithium-alloy anodes, 53 lithium alloy, 53, 59 lithium-electrode potential, 4 lithium-ion batteries, 5, 13–16, 28, 32– 33, 44–48, 53, 55, 59–60, 72, 76, 120, 128–129, 131, 140–141, 157– 159, 165–169, 190, 198, 227, 229, 232, 236, 267, 306, 317, 335–336, 340, 342, 350, 354, 359–361, 365, 387, 389, 394 lithium-ion cells, 3, 14, 33, 44, 47, 49, 50, 52, 334 lithium-ion migration, 76, 94, 129, 365 lithium-metal anode, 59 lithium metal, 45, 59 lithium-passivation mechanism, 11 lithium-passivation phenomena, 7, 11

403 lithium-tin-based alloys, 53 Li-triflate, 10, 46 LixBFy, 113, 115, 342, 350 LixC6, 4–5, 7, 45, 47, 49–50, 313 LixC6 anode, 4 LixCoO2, 99, 178, 331 LixMOy, 76, 78, 129, 338 LixPF4, 115 LixPFy, 167, 178, 341–342, 349, 357 LixPFyOz, 41, 115, 162–163, 167, 178, 357, 358 LPD, 247–250, 253–259, 262–269, 271–273 LVD, 247– 250, 254, 255–258, 262– 265, 267–269, 271, 273 magnesium, 1, 71–73, 75–76, 78, 88, 93, 95–96, 129–131, 141, 168, 170– 171, 188 mass spectrometry, 48, 83, 111, 199 mass spectroscopy, 87, 213, 390 mean field approximation, 296 mesocarbons, 311 metal oxide anodes, 140, 142 metallic electrode, 1 metallic lithium, 7, 12, 16, 34, 53, 310– 311, 313, 324, 333, 335 methyl chloroformate, 214, 220 methyl formate, 10, 105, 121, 130 MgO/LiCoO2, 141, 169–172, 174–176, 178, 180, 182, 184–186, 189 microcrystalline graphite, 311 microstructural parameters, 308 Mn2p, 341, 346, 351 molecular orbital, 367 molecular vibrational modes, 160 Montoro, 185, 186 Morita, 46, 62, 66, 224–225 morphology, 2–3, 10, 12, 16, 28, 45, 47–48, 51–52, 55, 60, 76, 86, 93, 123, 161, 168, 187, 189, 199, 204– 207, 218, 277–278, 308, 318 Mossbauer spectroscopy, 54 Müller, 6 multilayer surface films, 88, 97 Mylar films, 85

404

Lithium-Ion Batteries: Solid-Electrolyte Interphase

nanocrystallite, 299 nano-MgO, 141, 168, 170–180, 189, 366 nano-SnO, 140, 142–147, 149–157, 161 National Synchrotron Light Source, 323 native film, 10, 88 natural graphites, 44, 47–50, 52, 201– 202, 207, 209, 217, 304, 311 natural population analyses, 230 nutron rdiography, 3 NEXAFS, 308–309, 323, 325–326, 328–329, 331–332, 334, 336 nickel cathode, 1 nickel electrode, 97 nickel substrate, 7, 9 nickel, 1, 7, 8, 33, 54, 90, 102, 336 NMP, 324 NMR, 2, 47, 49, 52, 59, 77, 86, 134, 334–335 noble metals, 70, 75–76, 90, 100, 105– 107, 109, 111, 115, 128, 158 non-aqueous electrochemistry, 73, 131 non-aqueous electrolyte, 45, 73–74, 315 non-aqueous solutions, 71–72, 74–76, 90, 111, 131, 199, 214 non-crystalline, 311 nucleation, 3, 52 Nyquist, 94–97, 99 O1s, 17, 19, 30, 35, 39, 341–342, 345, 348–349, 351, 356–357 OCV, 5, 11, 90, 92, 344, 346 Ogumi, 6, 61, 64, 66, 138, 222–224, 226, 228, 274–275, 394, 397 Ohm’s law, 289 ohmic behavior, 71, 93 oligomers, 18, 199, 207, 273 Onsager reaction field theory, 229 open circuit potential, 296 organic carbonates, 8, 16, 121, 168, 267 Owen, 55, 69 P2O5, 342, 351 P2p, 36, 341–342, 349, 351, 357, 361

PAN, 12, 158 passivating film, 1, 9–12, 14, 18, 21, 45–46, 156, 160, 198 passivating layer, 1, 12, 16, 46, 48, 142, 145, 152, 160, 227, 328 passivation phenomena, 70, 73, 76–77 PC + ES, 216, 218–219 PC + FEC, 216, 218 PC, 1, 5–6, 8–10, 12, 45–46, 49, 51–52, 91, 95–97, 100, 102, 107–108, 116– 117, 121–123, 127–128, 130, 142– 143, 145, 153–156, 159–161, 180, 198–199, 201–203, 205–206, 214– 221, 227–229, 231–241, 244–246, 260, 267, 269, 271–273, 295, 315, 365–366, 368, 374–376, 378, 382, 388–394 PC + VC, 216, 376 PC-based electrolytes, 10, 52, 156, 229, 237, 241, 267, 271, 366 PC-based solutions, 9, 46, 200, 206, 214, 216, 220–221, 228, 238, 240, 272, 389, 394 PCM, 229, 367, 370 PEI, 6 Peled, 1, 55, 61–65, 67, 69, 132, 137, 139, 142, 199, 222–223, 225, 228, 273–275, 277–278, 306, 338, 363, 394 PEO, 12, 13, 55, 116, 199, 207, 342 PES-SR, 360, 361 petroleum coke, 46, 315 phosphate group, 359–360 phyllosilicate clay, 317 pillared clay, 315–316, 324 plateau, 92, 144, 157, 316 platinum, 9 point defect model, 277 polar aprotic organic solvents, 73 polar aprotic solvents, 10, 73, 88, 131 poly(ethylene oxide), 199, 207 polyacrylonitrile, 116, 158 polycrystalline, 5 polyether chain, 167 polymer electrolytes, 1, 7, 11–13, 47 polymer-electrolyte interphase, 6 polymer-in-salt, 158 polymerize olefins, 317

Index polymers, 11, 16, 18, 20–21, 24–25, 33, 37, 40, 42, 50, 58, 60, 142, 199, 349 polyolefins, 17, 18, 26, 30, 34, 36, 60 polyTHF, 60 post-mortem, 350 Pourbaix diagrams, 72 PrOCO2Li, 109 propylene carbonate, 1, 8, 10, 90, 94, 142, 180, 198, 227, 295, 365 propylene glycol carbonate, 5 propylene sulfite, 214, 366 propylene, 1, 5, 8, 10, 12, 90, 94, 122, 142, 180, 198, 214, 227, 232, 235– 236, 244–247, 254, 257–258, 260, 269–273, 295, 317, 319, 328, 331, 365–366 PVDF, 11, 48, 56–58, 122, 143, 324 PW91PW91, 230, 261–265, 269, 367, 375–376 Py–GC–MS, 199, 207 pyrene, 315, 316–317, 325, 328, 331 pyrolysis–gas chromatography–mass spectroscopy, 199 QCMB, 2, 59 quasi-planar geometries, 247 Raman effect, 80 Raman spectroscopy, 59, 80–81, 140, 158, 159, 167, 213, 214, 336, 341 Raman, 2, 9, 59, 77, 80, 81, 133, 134, 140–142, 153–154, 157–167, 321– 322, 336, 341 rate-determining step, 1, 3, 71 rechargeable lithium batteries, 2, 10, 140, 335 reduction potentials, 8, 74, 231, 234, 238, 387 resonance effect, 165 reversible capacity, 14, 15, 49, 53, 55, 117, 142, 202, 212, 238, 315, 320, 365 rhombohedral structure, 354 ring breathing mode, 174, 181, 182 ROCO2Li, 9, 11, 45–46, 50, 90, 100, 102, 107–108, 114–116, 118, 121, 128, 140, 142, 145–146, 148–152,

405 155–157, 161–163, 165–166, 178, 180, 184, 235–236, 342 ROLi, 50, 90, 100, 102, 104, 108, 114– 115, 163, 184 (S)2Li+(VC), 238, 241 (S)Li+(VC), 238, 241 (S)nLi+(VC), 238 SAFT, 278–279, 288, 291, 294 salt anions, 7, 15, 76, 88, 92, 111, 113, 118, 120, 130–131, 230 salt-in-polymer, 158 scanning electron microscopy, 82, 125, 140, 201 scanning tunneling microscopy, 87, 135, 201, 224 SCI-PCM, 229, 367 Scrosati, 13, 53, 63, 68, 135, 222 secondary ion mass spectroscopy, 83 second-order Møller-Plesset perturbation, 367 SEI growth dynamics, 278 SEI layer, 71, 93, 97, 140–146, 150– 151, 154–167, 171, 179–180, 182, 184, 190, 198–199, 201–202, 205– 207, 209–211, 214–216, 219–221, 227–229, 232, 235, 237, 241, 246– 247, 253, 255, 257–260, 269, 271– 273, 277–281, 285, 287–297, 299– 303, 305–306, 308, 310, 313–14, 323, 338, 343, 346, 349, 350, 352, 365–366, 393 selected area electron diffraction, 157 self-consistent isodensity PCM, 229 SEM, 77, 82–83, 140, 169 semicarbonates, 360 Sepiolite, 317, 335 SERRS, 166 SERS, 80, 134, 140–141, 157–158, 160–163, 165–166 silver, 90, 158, 159 SIMS, 77, 83, 199 simulation, 230, 260, 278, 298, 392 SnO anode, 140, 142, 151–157 SnO2-TiO2, 321 soft carbons, 312 solid-electrolyte interphase, 2–3, 9–10, 34, 48, 50, 71, 75, 93, 140, 142, 227

406

Lithium-Ion Batteries: Solid-Electrolyte Interphase

solid permeable interface, 343, 352, 362 solvated lithium ions, 205, 210, 228 solvation, 213–214, 220–221, 229, 366, 368, 387, 392–393 solvent co-intercalation, 200, 205–206, 210–212, 214, 220–221, 228 solvent decomposition, 198, 202–203, 206–207, 214, 218, 222, 238 solvent diffusion model, 296 solvent reduction, 128, 140, 151, 231, 236, 241–242, 272, 288, 294, 296, 387 SPI, 343, 345, 350, 352–353, 358–359, 362 spinel, 168, 178, 339–344, 346–348, 350, 352, 354–355 spinel-electrolyte interface, 341 SPM, 201 spontaneous reactions, 70, 73, 76, 167 spontaneous surface reactions, 88 sputtering, 77, 82, 112, 186, 187, 320, 340, 342–343, 350 standard electrode potential, 3, 4 state-of-charge, 343 statistical mechanics-based models, 296 stimulated surface reactions, 88 STM, 77, 86–87, 201, 203–206, 220– 221, 316–317 stochastic transport model, 277 stripping, 92 styrene, 316–317, 328, 331 superposition, 108, 281, 285, 367 supra-molecular cluster, 229 surface chemistry, 72, 75, 77, 88–89, 104, 108, 111, 113, 115–123, 131, 159, 166–168, 185, 187, 189, 227, 333, 338–339, 350, 361–362 surface enhanced Raman scattering, 140, 158, 166 surface film phenomena, 70, 72, 131 surface film, 70–73, 75–77, 79, 87–90, 92–95, 97, 99, 108, 112–113, 115– 118, 120–121, 123–125, 127–131, 157, 167, 178, 198, 201, 203, 229, 247, 333, 338, 341–343, 349, 361 surface phenomena, 126, 131, 341, 361 synchrotron radiation, 82, 360

synthetic graphites, 47, 49 T15-T44, 47 Tafel slope, 71 Tafel-like behavior, 71–72, 93 Takamura, 52, 67, 132 t–butylene carbonate, 214 Tefzel, 12 TEM, 77, 199, 316, 318, 345 tetra alkyl ammonium, 75 tetravalent Co4+, 186 TGA, 2 theoretical continuum modelling, 277 Thomas, 138, 222, 225, 338, 362–363, 365 time-of-flight secondary-ion mass spectrometry, 23 TIMREX KS6, 48 tin oxide, 142, 157, 160, 309, 320, 336 tin-antimony alloys, 54, 55 TOF SIMS, 17, 23–29, 31–32, 59–60 TPD, 50, 87 transition metals, 72, 128 transport theorem, 280, 283, 285–286 trioxane, 316–317, 328, 331 trivalent Co3+, 185 two-electron reduction, 107, 150–151, 236, 291 ultra-high vacuum (UHV), 77, 80, 83, 87, 104, 111, 323 ultraviolet Raman spectroscopy, 322 ultraviolet, 81, 336 under potential deposition, 92 UPD, 92 UV-Vis, 81 (VC)Li+(PC), 391 vapor-grown carbon fibers, 51 VC, 129, 214, 216–221, 229, 231–232, 238–246, 259–260, 265, 269, 271– 273, 374–376, 378–384, 389, 391, 393–394 vinilidine, 185 vinylene carbonate, 120, 214, 229, 366 visible light, 81 Voigt, 11, 97 voltammetric, 33

Index

Warburg impedance, 7 XPS, 2, 7, 10, 17–18, 20–21, 24–25, 29–31, 33–36, 38–39, 42, 48–49, 58–59, 77, 82, 87, 111–112, 114– 115, 117, 137, 140–141, 157, 171, 185, 187, 199, 235, 247, 340–346, 348–350, 352, 356–357 X-ray absorption, 82 X-ray diffractometry, 85 X-ray photoelectron spectroscopy, 2, 69, 140, 157

407 X-ray powder diffraction, 324 XRD, 77, 85, 142, 341, 344 Yazami, 16, 46, 47, 66, 223, 364 Yudasaka, 33, 65 Zaghib, 44, 64–65 zigzag, 17, 49, 267