Jeremy P. Matthews - QUT ePrintseprints.qut.edu.au/15786/1/Jeremy_Matthews_Thesis.pdf · i Abstract...

KINETIC BEHAVIOUR OF ION

INTERCALATION ELECTRODES AT

ELEVATED TEMPERATURES

PhD Thesis

Jeremy P. Matthews B. App. Sc. (Hons1)

School of Mechanical, Medical and Manufacturing Engineering

Queensland University of Technology

2001

i

Abstract

Electrochromic films undergo a colour change when small ions and electrons are

inserted into them, under the influence of an applied electric field. These films are also

known as ion-intercalation electrodes, and may be incorporated into glazing structures

more commonly known as smart windows. Smart windows are that which may be

used to control the amount of heat and light entering a building and may therefore be

used to minimise the energy consumption associated with heating, cooling and lighting.

The commercial success of smart windows, requires that they operate reproducibly at

temperatures up to approximately 70ºC, for many tens of thousands of colouring and

bleaching cycles. An understanding of the underlying kinetic processes over a wide

temperature range is therefore needed, in order to determine suitable control strategies

and switching conditions capable of fulfilling these requirements.

The research detailed in this thesis has involved an investigation into the kinetic

behaviour of ion-intercalation electrodes, and simulation of the electrical response as a

means of developing a tool for predicting and then optimising electrochromic switching.

More specifically, the electrical and optical properties of electrochromic thin films of

WO3/TiO2 have been studied over a wide temperature range, appropriate for the

operation of electrochromic windows. The magnitude of the voltages required for

coloration and bleaching significantly reduces as temperature increases. Some

irreversibility was observed at high temperature, as well as a reduction in coloration

efficiency. Further investigation revealed that self-bleaching and irreversibility effects

ii

were caused by the presence of water, and this problem was exacerbated at high

temperature. Post-experiment chemical analysis of a film sample revealed that some

trapping of the inserted ions had occurred, however the amount of ions remaining in the

film was much smaller than expected. The results suggested that a large quantity of the

lithium ions injected into the film were lost to the electrolyte after many cycles, possibly

accompanied by some film dissolution.

Experimental work carried out in a dry-box showed that films may be cycled reversibly

in a very dry environment, and the optical properties were independent of temperature

under these conditions. Unfortunately, the conditions which led to reversible cycling

and good electrochromic memory, also resulted in very long response times for film

bleaching. This result implies that a good electrochromic memory and a fast response

are mutually competitive aims.

Data from high temperature experiments was simulated with a mathematical model and

the mobility of lithium ions inside the electrochromic films was estimated in the

process. The estimated diffusion coefficients agreed well with published values, and

exhibited an Arrhenius dependence on temperature. Activation energies for diffusion

were calculated and the results were very reasonable. Some deviation from ideal

Arrhenius behaviour was observed for the estimated diffusion coefficients at high

temperature. It is likely that the rate limiting mechanism changes from diffusive motion

of ions at low temperature, to charge transfer at high temperature.

iii

Keywords: Electrochromic thin films; temperature effects; reversibility; kinetic

behaviour; self-bleaching

v

LIST OF PUBLICATIONS

1. J.P. Matthews, J.M. Bell and I.L. Skryabin, "Effect of temperature on electrochromic

device switching voltages", Electrochimica Acta, 44(18), 3245-3250 (1999).

2. J.P. Matthews, J.M. Bell and I.L. Skryabin, "High temperature behaviour of

electrochromics", Renewables: The Energy for the 21st Century

Proceedings of the World Renewable Energy Congress VI, Brighton, UK (A.A.M.

Sayigh Ed.), 230-235 (2000).

3. J.M. Bell and J.P. Matthews, "Temperature dependence of kinetic properties of sol-

gel deposited electrochromics", Solar Energy Materials and Solar Cells, 68, 249-263 (2001).

4. J.P. Matthews, J.M. Bell and I.L. Skryabin, Simulation of electrochromic switching

voltages at elevated temperatures, Electrochimica Acta, 46, 1957-1961 (2001).

5. J.P. Matthews and J.M. Bell, Self-bleaching, memory effect and reversibility of

electrochromism at elevated temperatures, submitted to Solar Energy Materials and

Solar Cells.

APPENDICES:

1. J.M. Bell and J.P. Matthews, "Glazing materials", Materials Forum, 22, 1-24(1998).

2. J.M. Bell, J.P. Matthews, I.L. Skryabin, J. Wang and B.G. Monsma, "Sol-gel

deposited electrochromic devices", Renewable Energy, 15(1-4), 312-317(1998)

3. J.M. Bell, J.P. Matthews and I.L. Skryabin, "Smart Windows", entry in Encyclopedia

of Smart Materials, accepted for publication.

vii

TABLE OF CONTENTS

ABSTRACT....................................................................................................................... i

KEYWORDS ...................................................................................................................iii

LIST OF PUBLICATIONS .............................................................................................. v

TABLE OF CONTENTS................................................................................................vii

STATEMENT OF ORIGINAL AUTHORSHIP ............................................................ xv

LIST OF ABBREVIATIONS........................................................................................xvi

LIST OF FIGURES ......................................................................................................xvii

ACKNOWLEDGEMENTS .......................................................................................xxviii

Chapter 1 INTRODUCTION ........................................................................................... 1

1.1 DESCRIPTION OF RESEARCH PROBLEM INVESTIGATED............................................. 2

1.2 A BASIC INTRODUCTION TO ELECTROCHROMICS....................................................... 4

1.3 ION INTERCALATION MECHANISM............................................................................. 7

1.4 ION INTERCALATION ELECTROCHEMISTRY ............................................................... 9

1.5 OPTICAL CHARACTERISTICS ................................................................................... 13

1.6 OVERALL OBJECTIVES OF THE STUDY.................................................................... 17

1.7 SPECIFIC AIMS OF THE STUDY ................................................................................ 17

1.8 RESEARCH HISTORY............................................................................................... 18

REFERENCES ................................................................................................................ 24

viii

Chapter 2 LITERATURE REVIEW............................................................................... 27

2.1 ELECTROCHROMISM A BRIEF HISTORY ............................................................... 28

2.2 THE ELECTROCHROMIC REACTION IN WO3 FILMS ................................................. 30

2.3 ELECTROCHROMIC CHARACTERISTICS AT ELEVATED TEMPERATURES .................. 32

2.4 LIFETIME, IRREVERSIBILITY AND SELF-BLEACHING............................................... 36

2.5 MODELS FOR SIMULATION OF ELECTROCHROMIC SWITCHING CHARACTERISTICS. 46

2.6 SUMMARY .............................................................................................................. 58

REFERENCES ................................................................................................................ 59

Chapter 3 EFFECT OF TEMPERATURE ON ELECTROCHROMIC DEVICE

SWITCHING VOLTAGES. ........................................................................................... 63

ABSTRACT.................................................................................................................... 65

3.1 INTRODUCTION....................................................................................................... 66

3.2 EXPERIMENTAL ...................................................................................................... 67

3.2.1 Electrode preparation .................................................................................... 67

3.2.2 Electrochemical Testing................................................................................. 68

3.3 RESULTS AND DISCUSSION...................................................................................... 69

3.3.1 Effect of temperature on applied voltage, Va(t) ............................................. 69

3.3.2 Effect of temperature on colouring efficiency................................................73

3.3.3 Effect of temperature on coloured state electromotive force, emfc, and

maximum colouring voltage, Vc max. ........................................................................ 75

3.4 CONCLUSION .......................................................................................................... 80

ACKNOWLEDGEMENTS................................................................................................. 81

ix

REFERENCES ................................................................................................................ 82

Chapter 4 HIGH TEMPERATURE BEHAVIOUR OF ELECTROCHROMICS ......... 83

ABSTRACT.................................................................................................................... 85

4.1 INTRODUCTION....................................................................................................... 86

4.2 EXPERIMENTAL ...................................................................................................... 88

4.2.1 Electrode preparation .................................................................................... 88

4.2.2 Electrochemical testing.................................................................................. 88

4.2.3 Chemical analysis .......................................................................................... 90

4.3 RESULTS AND DISCUSSION ..................................................................................... 91

4.3.1 Effect of temperature on coloration efficiency...............................................91

4.3.2 Observation of self-bleaching ........................................................................ 95

4.3.3 Determination of trapped lithium in WO3 film by ICP-AES .......................... 98

4.4 CONCLUSIONS ...................................................................................................... 100

ACKNOWLEDGEMENTS............................................................................................... 101

REFERENCES .............................................................................................................. 101

Chapter 5 TEMPERATURE DEPENDENCE OF KINETIC BEHAVIOUR OF SOL-

GEL DEPOSITED ELECTROCHROMICS ................................................................ 103

ABSTRACT.................................................................................................................. 105

5.1 INTRODUCTION..................................................................................................... 106

5.2 THEORY................................................................................................................ 108

5.2.1 Temperature effects on kinetic behaviour ....................................................108

x

5.2.2 Thermodynamics of coloration .................................................................... 111

5.2.3 Modelling of concentration profile .............................................................. 114

5.3 EXPERIMENTAL .................................................................................................... 117

5.3.1 Film preparation .......................................................................................... 117

5.3.2 Electrochemical testing................................................................................ 117

5.4 RESULTS............................................................................................................... 119

5.4.1 Variation in switching voltage with temperature.........................................119

5.4.2 Simulation of Voltage Response of Films ....................................................121

5.5 DISCUSSION.......................................................................................................... 123

5.6 CONCLUSION ........................................................................................................ 127


REFERENCES .............................................................................................................. 128

Chapter 6 SIMULATION OF ELECTROCHROMIC SWITCHING VOLTAGES AT

ELEVATED TEMPERATURES. ................................................................................ 131

ABSTRACT.................................................................................................................. 133

6.1 INTRODUCTION..................................................................................................... 134

6.2 EXPERIMENTAL .................................................................................................... 135


6.2.2 Electrochemical testing................................................................................ 135

6.2.3 Simulation of EC film coloration voltage .................................................... 136

6.3 RESULTS AND DISCUSSION ................................................................................... 138

6.3.1 Voltage characteristics ................................................................................ 138

xi

6.4 CONCLUSION ........................................................................................................ 143


REFERENCES .............................................................................................................. 145

Chapter 7 SELF-BLEACHING, MEMORY EFFECT AND REVERSIBILITY OF

ELECTROCHROMISM AT ELEVATED TEMPERATURES. ................................. 147

ABSTRACT.................................................................................................................. 149

7.1 INTRODUCTION..................................................................................................... 150

7.2 EXPERIMENTAL .................................................................................................... 152


7.2.2 Electrochemical Testing............................................................................... 152

7.3 RESULTS AND DISCUSSION ................................................................................... 154

7.3.1 Observation of self-bleaching at elevated temperatures..............................154

7.3.2 Effects of water on self-bleaching rates at elevated temperatures .............. 157

7.3.3 Effects of water on voltage characteristics ..................................................159

7.4 CONCLUSION ........................................................................................................ 163


REFERENCES .............................................................................................................. 165

Chapter 8 GENERAL DISCUSSION........................................................................... 167

8.1 INTRODUCTION AND IDENTIFICATION OF KNOWLEDGE GAPS............................... 168

8.2 INITIAL CHARACTERISATION IN THE AMBIENT LABORATORY ENVIRONMENT...... 169

8.3 DRY-BOX EXPERIMENTS ...................................................................................... 173

xii

8.4 SELF-BLEACHING EXPERIMENTS.......................................................................... 176

8.5 SIMULATION AND ESTIMATION OF IONIC MOBILITY............................................. 186

8.6 CONCLUSIONS ...................................................................................................... 192

8.7 FUTURE RESEARCH .............................................................................................. 193

REFERENCES .............................................................................................................. 195

Appendix 1 GLAZING MATERIALS. ........................................................................ 197

SYNOPSIS ................................................................................................................... 199

1 INTRODUCTION........................................................................................................ 200

2 REQUIREMENTS FOR GLAZING MATERIALS............................................................. 203

3 COATING SYSTEMS FOR WINDOW GLAZING............................................................ 209

3.1 Spectrally Selective Glazings ..........................................................................209

3.2 Angular Selective Glazings .............................................................................215

3.3 Switchable Glazings........................................................................................ 220

4 CASE STUDIES ......................................................................................................... 227

4.1 Spray pyrolysed fluorine doped tin oxide films............................................... 227

4.2 Angular selectivity of obliquely sputtered chromium films.............................233

4.3 Design and performance of an electrochromic device ...................................238

4.3.1 Film deposition ...................................................................................... 239

4.3.2 Electrode cycling ................................................................................... 240

4.3.3 Polymer electrolyte ................................................................................ 241

4.3.4 Device testing......................................................................................... 242

4.3.5 Device modelling ................................................................................... 244

xiii

5 SUMMARY ............................................................................................................... 246

REFERENCES .............................................................................................................. 247

Appendix 2 SOL-GEL DEPOSITED ELECTROCHROMIC DEVICES. ................... 257

ABSTRACT.................................................................................................................. 259

1 INTRODUCTION........................................................................................................ 260

2 ELECTROCHROMIC FILM DEPOSITION...................................................................... 261

3 ELECTROCHROMIC FILM PERFORMANCE................................................................. 262

4 ELECTROCHROMIC DEVICE FABRICATION AND PERFORMANCE............................... 264

5 ENERGY PERFORMANCE OF ELECTROCHROMIC DEVICES ........................................ 267

6 CONCLUSIONS ......................................................................................................... 269

REFERENCES .............................................................................................................. 270

Appendix 3 SMART WINDOWS. ............................................................................... 273

1 OUTLINE.................................................................................................................. 275

2 INTRODUCTION........................................................................................................ 276

3 ARCHITECTURALGLAZING APPLICATIONS FOR SMART WINDOWS .......................... 277

3.1 Physics of Windows......................................................................................... 278

4 SURVEY OF SMART WINDOWS................................................................................. 283

4.1 Electrochromic Smart Windows...................................................................... 286

4.1.1 Inorganic Electrochromic Smart Windows............................................ 286

4.1.2 Organic Electrochromic Smart Windows .............................................. 290

4.2 Thermochromic Devices.................................................................................. 291

xiv

4.3 Thermotropic Devices ..................................................................................... 293

4.3.1 Hydrogels............................................................................................... 294

4.3.2 Polymer Blends...................................................................................... 295

4.3.3 Applications ........................................................................................... 296

4.4 Polymer Dispersed Liquid Crystal Devices ....................................................296

4.5 Suspended Particle Devices ............................................................................299

4.6 Gasochromic Devices ..................................................................................... 300

4.6.1 Gasochromic Technology ............................................................................ 302

4.6.2 Applications ................................................................................................. 303

5 ELECTROCHROMIC SMART WINDOWS ..................................................................... 304

5.1 Electrochromic Smart Window Structures......................................................305

5.1.1 Type 1 - Ion Conducting Layer and Passive Counterelectrode ............. 305

f5.1.2 Type 2 Combined Ion Conducting Layer and Counterelectrode....... 307

5.1.3 Type 3 Ion Transport Layer and Complimentary Counter-electrode . 307

5.2 Materials used inelectrochromic devices........................................................308

5.2.1 Electrochromic Materials....................................................................... 308

5.2.2 Counter-electrode materials ................................................................... 310

5.2.3 Ion Transport Layer ............................................................................... 310

5.2.4 Transparent Electronic Conductors.............................................................. 311

5.3 Control of Electrochromic Smart Window......................................................312

5.4 Future Directions ............................................................................................ 316

REFERENCES .............................................................................................................. 317

xv

STATEMENT OF ORIGINAL AUTHORSHIP

The work contained in this thesis has not been previously submitted for a degree or

diploma at any other higher education institution. To the best of my knowledge and

belief, the thesis contains no material previously published or written by another person

except where due reference is made.

Signed:...................................

Date: ...................................

xvi

LIST OF ABBREVIATIONS

CE Counter electrode

EC Electrochromic

emf electromotive force

ICP-MS Inductively-Coupled-Plasma-Mass Spectrometry

ITO Indium-tin oxide

PC Propylene carbonate

PECVD Plasma enhanced chemical vapour deposition

PET Polyethylene terephthalate

RE Reference electrode

STA Sustainable Technologies Australia LTD

SIMS Secondary-Ion Masss Spectrometry

TCO Transparent conducting oxide

WE Working electrode

η Coloration efficiency

XRD X-Ray Diffraction

xvii

LIST OF FIGURES

Chapter 1 - INTRODUCTION

Figure 1.1 Transmittance spectra for coloured and bleached states of an STA

electrochromic device. .............................................................................................. 5

Figure 1.2 Typical structure of an electrochromic device............................................... 6

Figure 1.3 Simulated concentration profile for t=200s (Qinj=20mC/cm2). ................... 11

Figure 1.4 Determination of coloration efficiency from a plot of change in optical

density versus injected charge density. Coloration efficiency = 45.5cm2/C.......... 16

Chapter 3 - EFFECT OF TEMPERATURE ON ELECTROCHROMIC DEVICE

SWITCHING VOLTAGES

Figure 3.1 Curves of applied voltage versus time, measured during coloration and

bleaching of WO3 thin film electrode and plotted for four temperatures, for an

injected charge density of 15mC/cm2. .................................................................... 70

Figure 3.2 Maximum voltages required to colour WO3 thin film electrode to 15mC/cm2

at elevated temperatures.......................................................................................... 71

Figure 3.3 Photocell voltage versus injected charge for WO3 thin film electrode at four

temperatures. ........................................................................................................... 73

Figure 3.4 Change in optical density versus injected charge for WO3 thin film electrode

at four temperatures................................................................................................. 74

xviii

Figure 3.5 Maximum voltages required for coloration of WO3 thin film electrode to

5mC/cm2, and corresponding emf values measured between 20°C and 50°C........ 76

Figure 3.6 WO3 film maximum coloration voltage versus temperature, for an injected

charge density of 5mC/cm2. .................................................................................... 78

Figure 3.7 Log of coloration voltage versus reciprocal temperature for WO3 film with

an injected charge density of 5mC/cm2................................................................... 79

Chapter 4 - HIGH TEMPERATURE BEHAVIOUR OF ELECTROCHROMICS

Figure 4.1 Change in optical density versus injected charge for WO3 films cycled to

15mC/cm2 at elevated temperatures. The results shown in (a) are for an experiment

carried out in the ambient environment, while the results shown for (b) are for an

experiment carried out in a dry-box. ....................................................................... 91

Figure 4.2 Reversibility of cycling at elevated temperatures, represented as the

percentage of the injected charge density trapped per cycle................................... 93

Figure 4.3 Change in (a) photocell voltage and (b) emf of WO3 electrode during self-

bleaching experiment. ............................................................................................. 96

Figure 4.4 Correlation of estimated and measured quantities of charge lost during self-

bleaching. ................................................................................................................ 97

xix

Chapter 5 - TEMPERATURE DEPENDENCE OF KINETIC BEHAVIOUR OF

SOL-GEL DEPOSITED ELECTROCHROMICS

Figure 5.1 Simulated concentration profile for t=1s (Qinj=0.1mC/cm2). .................... 115

Figure 5.2 Simulated concentration profile for t=200s (Qinj=20mC/cm2). ................. 116

Figure 5.3 Applied voltage for colouration and bleaching of a sol-gel WO3 film to

15mC/cm2.............................................................................................................. 119

Figure 5.4 Dependence of emf on temperature and surface lithium concentration

predicted using equation 5.8. ................................................................................ 120

Figure 5.5 Experimental and simulated voltages during charge injection of a sol-gel

electrochromic film. (a) Temperature = 20.1ºC, D = 2.07x10-12cm2,

(b) temperature = 30.3ºC, D = 6.33x10-12cm2, (c) temperature = 40.3ºC,

D = 1.33x10-11cm2 and (d) temperature = 50.0ºC, D = 1.71x10-11cm2. ................ 122

Figure 5.6 Variation in estimated diffusion coefficients with temperature plotted for

range (a) 20.1 < T < 50.0ºC and (b) 20.1 < T < 40.3ºC. ....................................... 124

Chapter 6 - SIMULATION OF ELECTROCHROMIC SWITCHING

VOLTAGES AT ELEVATED TEMPERATURES.

Figure 6.1 Applied voltage during colouration and bleaching of a sol-gel WO3/TiO2

film to 15mC/cm2/s. .............................................................................................. 138

xx


electrochromic film. (a) T = 35.8ºC, D = 8.68x10-13cm2/s, (b) T = 46.5ºC,

D = 1.54x10-12cm2/s, (c) T = 56.2ºC, D = 4.02x10-12cm2/s, (d) T = 65.3ºC,

D = 1.48x10-11cm2/s and (e) T = 76.4ºC, D = 6.00x10-11cm2/s............................. 139

Figure 6.3 Arrhenius plot showing the variation in estimated diffusion coefficients with

temperature............................................................................................................ 141

Chapter 7 - SELF-BLEACHING, MEMORY EFFECT AND REVERSIBILITY

OF ELECTROCHROMISM AT ELEVATED TEMPERATURES.

Figure 7.1 Change in optical density (a) and measured voltage (b) during self-bleaching

of a WO3/TiO2 film in dry electrolyte................................................................... 154


of a WO3/TiO2 film, after firing at 250ºC for hours. ............................................ 155

Figure 7.3 Rate of self-bleaching for WO3/TiO2 films under various conditions....... 158

Figure 7.4 Voltage characteristics during cycling of WO3/TiO2 films under various

conditions. ............................................................................................................. 159

Figure 7.5 Dependence of electrochromic cycling characteristics on electrolyte water

concentration. ........................................................................................................ 161

xxi

Chapter 8 - GENERAL DISCUSSION


bleaching of WO3 thin film electrode and plotted for four temperatures, for an

injected charge density of 15mC/cm2. .................................................................. 170

Figure 8.2 Log of absolute value of maximum coloration voltage versus reciprocal

temperature for WO3 film, for an injected charge density of 5mC/cm2................ 171


at four temperatures............................................................................................... 172


percentage of the injected charge density trapped per cycle................................. 174

Figure 8.5 Change in optical density with time for a film coloured to 15mC/cm2 at

various temperatures. ............................................................................................ 177

Figure 8.6 Change in optical density during self-bleaching of a WO3/TiO2 film in dry

electrolyte.............................................................................................................. 179


of a WO3/TiO2 film, after firing at 250ºC for hours. ............................................ 180

Figure 8.8 Rate of self-bleaching for WO3/TiO2 films under various conditions and

temperatures. ......................................................................................................... 181

Figure 8.9 Voltage characteristics during cycling of WO3/TiO2 films to 15mC/cm2

under various conditions. ...................................................................................... 182


concentration. ........................................................................................................ 184

xxii


electrochromic film. (a) Temperature = 20.1ºC, D = 2.07x10-12cm2 and

(b) temperature = 50.0ºC, D = 1.71x10-11cm2....................................................... 188

Figure 8.12 Arrhenius plot showing the variation in estimated diffusion coefficients

with temperature between 20.1ºC and 50.0ºC....................................................... 189


with temperature between 35.6ºC and 76.4ºC....................................................... 190

Appendix 1 - GLAZING MATERIALS

Figure 1 The three components related to the ambient radiation environment which

need to be considered for design of windows. The three distinct spectral regions

correspond to the wavelengths 0.3<λ<2.5 µm (solar radiation), λ>3µm (thermal

IR), and 0.37<λ<0.77µm corresponding to the visible response of the human eye.

............................................................................................................................... 205

Figure 2 The instantaneous energy balance in a double pane window. The radiative

heat transfer between the panes is dependent on the absorption of solar radiation in

the outer pane (inward) and by the temperature of the inside pane relative to the

outside pane (outward). Both are reduced by using low emittance coatings on

surfaces 2 and/or 3. ............................................................................................... 208

Figure 3 Transmittance spectra for a solar control coating based on ZrN. Note the high

reflectivity in the IR and narrow band transmittance in the visible. From Roos and

Karlsson [18]. The dotted line shows a theoretical calculation of the spectrum

xxiii

based on material optical properties...................................................................... 212

Figure 4 One of the principles of angular selectivity, with different levels of

transmittance for radiation incident on either side of the normal. ........................ 216

Figure 5 The relationship between deposition geometry (atom flux direction with

respect to substrate) and the column orientation. Adapted from Smith [75] ....... 217

Figure 6 Typical structure of an electrochromic device.............................................. 221

Figure 7 Development of an absorption band in WO3 deposited by sputtering onto

substrates held at various temperatures. The graph shows the coloration efficiency

as a function of wavelength, which is directly proportional to absorption. Note the

shift in the absorption band peak position as the substrate temperature increases and

the film becomes more crystalline. From Wang and Bell [88]. ........................... 222

Figure 8 Schematic of the 'in-line' spray pyrolysis process. ....................................... 228

Figure 9 Transmittance (solid lines) and Reflectance (dashed lines) spectra for four

TCO samples: - LOF TEC 8 SnO2:F; - LOF TEC10 SnO2:F ; and -

sputtered ITO (Donnelly Applied Films, 15Ω/square); and - LOF TEC20

SnO2:F. Note the very sharp spectral selectivity and the very low visible

absorption of the sputtered ITO film compared to the spray pyrolysed FTO. This is

characteristic of the smoother and finer grain structure of the sputtered films (see

figure 10)............................................................................................................... 230

Figure 10 Micrographs of three different TCO films. (a) LOF TEC8 (b) LOF TEC10

and (c) Donnelly Applied Films, sputtered ITO, nominally 15Ω/square. ............ 231

Figure 11 Thickness of the layers in LOF TEC15 and TEC20 glass products according

the ellipsometric analysis by von Rottkay and Rubin [153]. ................................ 232

xxiv

Figure 12 Spectral transmittance of a 90nm Cr film, for s-polarised and p-polarised

light at various angles of incidence....................................................................... 234

Figure 13 The voltage non-uniformity ratio along a device for two different electrolyte

resistivities. ........................................................................................................... 242

Figure 14 Transmittance spectra for coloured and bleached states of an STA

electrochromic device. The ratio of Tvis/Tsol in both states are shown on the graph,

indicating the increase in spectral selectivity in the coloured state. The dynamic

range for the device is Tsol=51.5% to Tsol=20.6%, with injected charge of

10mC/cm2. From Bright [151]. ............................................................................ 243

Appendix 2 - SOL-GEL DEPOSITED ELECTROCHROMIC DEVICES

Figure 1 (a) Transmittance spectra of sputtered (solid lines) and sol-gel (dashed lines)

in the coloured and bleached states, and (b) the electrical characteristics of the sol-

gel deposited film, obtained using constant current charge injection. .................. 262

Figure 2 The variation of the injected and extracted charge for a sol-gel deposited film

cycled in 1 M LiClO4 electrolyte, and the difference between the maximum

injected (extracted) charge and the quantity Qvmax (Qvmin), which is the value of

injected (extracted) charge when the preset voltage limit is reached. The change in

characteristics after cycle 1000 is attributed to a long break in cycling (nearly 1

month), which disappears after 300 cycles. .......................................................... 263

Figure 3 The transmittance spectra for a sol-gel deposited electrochromic device in

both coloured and bleached states......................................................................... 266

xxv

Figure 4 The relative coloration measured as the ratio of a photocell voltage at time t to

the photocell voltage in the bleached state. The inserts show the initial coloration

and bleaching of the device................................................................................... 267

Figure 5 The energy savings calculated from a simulation (DOE2.1E) of a prototypical

office building in three locations in Australia resulting from using an

electrochromic glazing instead of a bronze tint glazing........................................ 268

Appendix 3 - SMART WINDOWS

Figure 1 The three components related to the ambient radiation environment which

need to be considered for design of windows. The three distinct spectral regions

correspond to the wavelengths 0.3<λ<2.5 µm (solar radiation - ϕsol(λ)), λ>3µm

(thermal IR), shown here as ϕth(λ) for T=300K, and 0.37<λ<0.77µm corresponding

to the visible response of the human eye, ϕvis(λ). ................................................. 279

Figure 2 Schematic illustration of the different types of switching of transmitted

radiation which are possible with different switchable window systems. ............ 280

Figure 3 The variation in optical density for four differently electrically activated smart

window devices. • - a sol-gel deposited device manufactured by Sustainable Technologies

Australia Ltd, and based on WO3; ×- a sputtered device manufactured by Asahi Glass Co,

and based on a WO3-NiO complementary device (see Section 4.2); +- an organic

electrochromic from Gentex Corporation, based on viologens; - a polymer dispersed

liquid crystal device manufactured by 3M Corporation. The different curves represent

different spectra used in calculating the optical density: a broad band transmittance;

a− − − the visible transmittance Tvis; and the solar transmittance Tsol. The

xxvi

divergence in the optical density for visible and solar transmittance for the Gentex

viologen-based device is a reflection of the type of spectral change (Type D in figure 2).

............................................................................................................................... 282

Figure 4 A Schematic illustration of the structure of a typical inorganic electrochromic

device, showing the five layer structure................................................................ 287

Figure 5 The spectral coloration efficiency, which directly proportional to change in

optical density in sputtered WO3. This shows the development of an absorption

band in WO3 during ion injection. Note the shift in the absorption band peak

position as the substrate temperature increases and the film becomes more

crystalline, representing the change from absorption to reflection modulation.... 289

Figure 6 The transmittance of (a) VO2 and (b) W-doped VO2 at temperatures above

(80°C) and below (20°C) the thermochormic transition. The spectra are of type C

in both cases (see Figure 2). Reprinted from Solar Energy Materials and Solar

Cells, Volume 44, M.A. Sobhan, R.T. Kivaisi, B. Stjerna and C-G. Granqvist,

Thermochromism of Sputter Deposited WxV1-xO2 films, 1996, pages 451-455, with

permission from Elsevier Science......................................................................... 292

Figure 7 The structure of a themortropic laminate. In the low temperature state (a), the

device is fully transparent, but in the high temperature state the thermotropic state

separates into discreet particles and the layer becomes scattering........................ 294

Figure 8 The normal-hemispherical reflectance of a thermotropic hydrogel laminate at

a range of temperatures below (30°C) and above (>35°C) the thermotropic

transition temperature. The spectra are of type B, demonstrating a high degree of

selectivity. ............................................................................................................. 295

Figure 9 The structure of a PDLC device. Unlike the electrochromic device, it is not a

xxvii

conducting device, with the switching dependent on the electric field across the

PDLC layer. .......................................................................................................... 297

Figure 10 Normal-hemispherical transmittance spectra for the PDLC device for applied

voltages of 0, 10, 20, 30 and 100 V. The device is opaque (in the sense of being

unable to perceive images through the device in the off state (0V), but still

transmits significant energy. Reproduced with permission of Dr Arne Roos. .... 298

Figure 11 A typical structure of a gasochromic device............................................... 301


electrochromic device. The selectivity in both states are shown on the graph,

indicating the increase in selectivity in the coloured state. The dynamic range for

the device is Tsol=51.5% to Tsol=20.6%, with injected charge of 10mC/cm2. ...... 306

Figure 13 Transmittance spectra of an Asahi electrochromic device in the bleached

state and for 3 colored states, with 5mC/cm2, 10mC/cm2 and 15mC/cm2 injected

into the WO3 layer at a constant potential of 1.5V. .............................................. 308

Figure 14 Electrochromic device performance at 50°C (+) and 18°C (×). The solid

lines represent the device voltage and the dashed lines the relative transmittance of

the device. In both cycles the operational parameters are: current

density=0.1mA/cm2 , Qin=13mC/cm2; Qout=-13mC/cm2, preset switching voltages

are: Vmax= 1.8V, Vmin=-0.7V. Note that at 50°C voltage does not reach Vmax. The

high temperature cycle period is shorter and lower voltages are required, however,

transmittance change in the two cycles is almost identical................................... 313

xxviii

ACKNOWLEDGEMENTS

This work was supported by an Australian Postgraduate Award (Industry) scholarship

from the Australian Research Council, and Sustainable Technologies Australia Limited

(STA). The work described in this thesis has been supported by the Australian

Cooperative Research Centre for Renewable Energy (ACRE). ACREs activities are

funded by the Commonwealths Cooperative Research Centres Program.

The author would also like to thank Mr Graeme Evans for technical assistance and for

provision of WO3/TiO2 films for use in this project, Mr Mark Hayne for technical

assistance and Mr Pat Stevens for his assistance with chemical analyses.

CHAPTER 1

INTRODUCTION

2

INTRODUCTION

1.1 Description of Research Problem Investigated

This thesis is concerned with the operation of electrochromic (EC) windows under

conditions expected during their normal operation. Owing to the absorption of tungsten

oxide in the coloured state, the windows are expected to heat up to between 50 and

70ºC. Smart windows operate by the movement of charge between the electrodes and

the coloration mechanism involves a redox process in the tungsten oxide electrode (See

section 1.3), therefore it is expected that the kinetics of coloration and bleaching will be

affected as the temperature increases. The research problem described in this thesis

therefore involves the electrochemical and optical properties of electrochromic films at

elevated temperatures. The goal of the work is to develop a detailed understanding of

the switching voltage and current, and reversibility of the electrochromic reaction as a

function of temperature. This should enable control strategies to be developed for smart

windows to increase device lifetime and permit successful commercialisation of

electrochromic systems. The work was performed using sol-gel deposited films of

WO3-TiO2 produced by STA (Sustainable Technologies Australia) as a part of their

development of smart window systems.

The kinetics of the individual reaction mechanisms will change with temperature, and

the switching regime should reflect these changes. The application of high voltages or

currents leads to fast coloration but also increases the possibility of side reactions

occurring, thereby degrading device performance and shortening the useful lifetime.

3

Low voltages and currents make switching safer thereby increasing lifetime, but the

slower response may be perceived as a loss in product value by the end user.

It is therefore important to optimise the device-switching regime to achieve maximum

lifetime, while still achieving reasonable response times. Such an optimisation requires

detailed knowledge of the kinetic properties of electrochromic films over a range of

conditions applicable to real electrochromic device application. Simulation of the

device response with a mathematical model may then be used as a valuable tool for

predicting how a device will function under specific conditions. The ability to predict

how the kinetics of an electrochromic device will change in response to real conditions

is extremely useful in terms of device design and optimisation.

The research problem has therefore involved determining the effects of temperature on

the kinetic properties of the electrochromic reaction in mixed oxide WO3/TiO2 films.

Specifically, this has involved measurement of the optical and electrical responses of EC

films during cycling over a wide temperature range from room temperature to

approximately 70ºC. A large portion of this work has involved specific investigation

into the causes of detrimental effects such as self-bleaching and charge insertion

irreversibility. This research work has also involved the use of a mathematic model to

simulate device-switching characteristics during the coloration process, and to estimate

the dependence of ionic mobility on temperature.

4

1.2 A basic introduction to electrochromics

An ion intercalation electrode is a solid substrate, into which ions can be inserted by the

application of an electric field. Considerable research has been conducted on the

properties of tungsten oxide (WO3) as an ion intercalation electrode since the early

1970s. In 1969 Deb published results showing that colour centers could be formed in

thin films of tungsten oxide by the application of an electric field [1]. Deb called the

tungsten oxide film electrochromic, a term previously only associated with some

organic compounds. This work revealed the potential of inorganic electrochromic thin

films, and sparked much interest in electrochromic research.

Inorganic electrochromic materials such as tungsten oxide (WO3) change their

transmittance when ions and electrons are injected into the material under the influence

of an electric field. This is schematically illustrated in the reaction scheme below, using

WO3 as the electrochromic material, V2O5 as a counter-electrode material and Li+ ions

as the mobile ionic species:

WO xLi xe Li WO working electrode

transparent deep blue

Li V O Li V O xLi xe counter electrode

V

Vx

y

V

Vy x

3 3

2 5 2 5

+ + →←

→← + +

+ −−

+

+

−−

+ −

(1.1)

WO3 is transparent while LixWO3 is a deep blue colour, therefore the colour of the WO3

film changes when the above reaction occurs. Figure 1 shows the transmittance

properties of an electrochromic device in the coloured and bleached states. The

transmittance of the infrared radiation component (λ>1000nm) is inherently lower than

5

the visible part of the spectrum (370nm<λ<780nm) which means that these films may

be used to reduce solar gain entering a building through the windows, while still

allowing significant daylighting to improve comfort and reduce the requirement for

lighting energy.

0

10

20

30

40

50

60

70

500 1000 1500 2000

trans

mitt

ance

(%)

λ (nm)

coloured T

bleached


electrochromic device.

A major application of electrochromic materials is in the fabrication of smart

windows. Smart windows are active glazing devices that can be used to control the

amount of visible and solar radiation entering a building, in order to minimise the

buildings energy load normally associated with heating, cooling and lighting. Thin

films of tungsten and vanadium oxide can be incorporated into a smart window structure

with transparent conducting oxides (TCOs) and a polymer electrolyte, as shown

6

schematically in Figure 2. The TCO layers allow electron current to flow between the

working and counter electrodes while maintaining transparency, a property mandatory

for window applications. The polymer electrolyte serves as an electrical insulator and

ionic conductor, as well as adding mechanical strength and physically holding the

device together.

The charge that is used to colour the WO3 film (ie, Li+, H+ etc) must be stored outside

the working electrode to achieve bleaching, and this is accomplished using a suitable

counter electrode. Vanadium oxide is a good counter electrode material because a

relatively large amount of lithium ions may be intercalated into it, without any

significant colour change.

Figure 2 Typical structure of an electrochromic device.

7

The voltages associated with the redox reactions outlined in equation (1.1) depend on

several factors including:

o value of x in LixWO3 (ie. film emf)

o Li+ diffusion coefficients in the films

o charge transfer at WO3/electrolyte interface

o series resistance of the cell

o temperature

As well as the electrochromic reactions, there are a number of undesirable side reactions

that ultimately lead to degradation of a device. EC devices absorb a proportion of the

incident radiation, and so will heat up considerably under normal operating conditions.

At elevated temperatures the voltages required to colour and bleach the smart windows

are considerably reduced, due primarily to increased diffusion coefficients and normal

Nernstian behaviour. This means that excessive overvoltages may be applied at high

temperatures if the room temperature voltage limits are applied. Thus maximum

lifetime of electrochromic devices can only be obtained if the kinetic behaviour of the

electrodes is understood and the control strategy is optimised over a wide temperature

range.

1.3 Ion intercalation mechanism

The response of electrochromic systems to an applied electrical signal is very complex,

but the process may be broken down into a number of simpler stages. When a current or

voltage is applied between the working and counter electrodes, an electric field is

8

established across the electrolyte. During coloration the mobile ions in the electrolyte

move towards the working electrode under the influence of this electric field, and

electrons enter the electrode from the TCO contact. The ions in solution must lose their

solvation sheaths and then combine with an electron at a tungsten site on the film

surface. The WO3 lattice expands slightly as the metal ion is intercalated into an

interstitial position. The electron is localised on a tungsten site, according to equation

(1.1) and an optical transition occurs. There is some controversy regarding the exact

nature of the initial species formed when the ion migrates from the electrolyte to the

electrode surface [2], but it is commonly accepted that the process involves the double

injection of the metal ion and an electron [3,4].

As coloration proceeds, the concentration of metal ions just below the surface of the

WO3 electrode will increase, saturating the surface sites available for intercalation.

Ion/electron pairs must migrate deeper into the film if more ions are to be intercalated.

The conductivity of the cycled WO3 film is quite high, and it is generally assumed that

there is no drop in electrical potential internal to the electrode. This means that there is

no electric field present to drive the ions deeper into the film, and so the ions must

diffuse into the film solely under the influence of the concentration gradient. Upon

bleaching the ions must migrate from within the film to the electrode/electrolyte

interface, before they can be extracted. The research presented in this thesis focuses on

the coloration process, so the mechanics of bleaching will not be discussed further here.

9

1.4 Ion intercalation electrochemistry

Experimental measurements made on EC films usually involve a three-electrode cell

consisting of a working electrode (WO3 film), a counter electrode (in this case an inert

platinum electrode) and a reference electrode, all immersed in some electrolyte solution.

A current or voltage is applied across the working and counter electrodes to colour or

bleach the film. The potential of the working electrode is measured relative to the

reference electrode, as current flow between these electrodes is negligible and the

reference electrode potential is constant.

The measured voltage during coloration of an EC film in an electrolyte solution is the

sum of all of the potential drops between the TCO (of the working electrode) and the

reference electrode. The conductivity of the transparent conductors and the electrolyte

is relatively high, which means that the electrical potential drop across these layers will

be small. Assuming that there is no internal resistance for the WO3 film, the large

majority of the potential will be applied across the electrolyte/electrode interface.

The voltage measured during an experiment depends on several factors, one being the

chemical composition of the film. If no voltage or current is applied to the film and

significant time has passed allowing diffusion to occur (ie. the system is at equilibrium)

then the measured voltage is the difference between the electromotive force (emf) of the

film and the electrical potential of the reference electrode. The emf of a film of

composition MyWO3 is related to the free energy change associated with introducing y

moles of the atom M into the solid lattice of WO3. This emf is given by equation (1.2)

10

−

−+=y

yFRTbyaemf

1lnν (1.2)

where a, b and ν are constants which are discussed in detail in section 2.5, and y is the

stoichiometric lithium coefficient (ie. y in MyWO3). This equation is similar to the

Nernst equation for redox reactions in solution, except that in this case we are

considering the concentration of a species in a solid solution. Cell potentials are related

to the free energy change (∆G) of reaction by equation (1.3)

nEFG −=∆ (1.3)

where n is the number of electrons in the process, E is the cell potential and F is

Faradays constant. Spontaneous reactions occur when the free energy change is

negative (ie. the reaction will tend to move in the direction of the energetically more

stable state) and therefore when the sign of the emf is positive. As y in MyWO3

increases and the emf of the working electrode (relative to the reference electrode)

becomes more negative, indicating that the film is being reduced and is therefore

moving cathodically further from equilibrium.

During switching of an EC film, the system is displaced from equilibrium and the

voltage response is considerably more complex. Unless ionic diffusion is extremely

fast, the surface concentration of ions will be greater than the average concentration.

Assuming no internal potential drop in the electrode, the voltage during switching is

related to the concentration of injected ions at the electrode surface so we need to

investigate the distribution of ions within the film. If the diffusion coefficient is high,

the surface concentration of ions will be low because the atoms move very quickly into

11

the film. This will mean that the emf and consequently the measured voltage are

expected to decrease as ionic mobility increases. Increases in temperature should result

in faster diffusion and hence lower voltages are expected at higher temperatures.

Figure 3 shows a graph of stoichiometric lithium concentration (ie. y in MyWO3) versus

film depth (x), for different values of the chemical diffusion coefficient. The data on

this graph was calculated using a model reported by Wang [5] which is detailed in

section 2.5. It is evident from this graph that the surface concentration will be

considerably higher then the average concentration inside the film, or alternately, the

greater the diffusion coefficient, the flatter the concentration profile will be.

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 50 100 150 200

D = 1x10-12 cm2/s

D = 2x10-12 cm2/s

D = 5x10-12 cm2/s

D = 1x10-11 cm2/s

Con

cent

ratio

n (

)

Distance x (nm)

y

Figure 3 Simulated concentration profile for t=200s (Qinj=20mC/cm2)

Various models have been proposed to describe the electrical response of EC films and

devices according to different rate limiting mechanisms. Many rate-limiting

12

mechanisms have been proposed including the diffusion limited motion of ions in the

film [6], the presence of a barrier to charge transfer at the electrolyte/electrode interface

[7] and the series resistance of the complete cell. Several of these rate-limiting

mechanisms are discussed in more detail in section 2.2.

Regardless of the limiting mechanism for a particular EC system, processes such as

diffusion and charge transfer will occur more readily at high temperatures. It is

therefore likely that the electrical potential required to achieve a specific charge density

will depend heavily on temperature. EC films and devices for use in smart window

applications are known to heat up considerably in the coloured state [8], because they

change their optical constants by a principle known as absorption modulation (See

section 1.5), and therefore absorb a significant proportion of the incoming radiation.

The way in which temperature affects the kinetic behaviour of electrochromic systems is

therefore of great interest to anyone wishing to control devices in a manner which will

reduce the possibility of degradation and maximise device lifetime. The effects of

temperature in EC cycling are discussed in more detail in sections 2.3 and 2.4, and

various models describing the electrical response are discussed in section 2.5.

13

1.5 Optical characteristics

Electrochromic films undergo a colour change by either reflectance modulation or

absorption modulation [9]. Reflectance modulation results from an increase in the free

electron density in the material [10] and is observed in crystalline electrochromic

materials. Amorphous electrochromic materials exhibit absorption modulation, caused

by the development of an absorption band [11]. Crystalline electrochromic films have

significantly lower ionic mobility than amorphous films, which limits the maximum

switching speed, and these films are also difficult to produce in combination with other

layers. Amorphous electrochromic layers are therefore preferable for smart window

components. The sol-gel deposited films used in this work are all amorphous and

therefore colour by absorption modulation. The following discussion of optical

characteristics is then limited to amorphous films, and the optical properties of

crystalline films are not considered further in this thesis.

There is some controversy as to the exact nature of the colour centre formed in EC

materials, but it is commonly accepted that the coloration process involves the double

injection or extraction of ions and electrons [4]. Several models have been proposed to

describe the coloration process including the formation of colour centers by electron

trapping at oxygen vacancies [12] (analogous to the formation of F centers in alkali

halides), an interband transition model and a small polaron model [13,14].

14

Perhaps the most widely accepted model describes the colouration process as arising

from an intervalence charge transfer between a tungsten(VI) atom and a neighbouring

tungsten(V) atom according to equation (1.4) [3]

W5+(A) + W6+(B) + hν ⇔ W6+(A) + W5+(B) (1.4)

According to this explanation, an electron trapped at a tungsten(V) site is transferred to

a neighbouring tungsten(VI) site when incident radiation is absorbed. It is this

absorption which is thought to give rise to the colour change of the EC material.

Several parameters are used to discuss the optical properties of EC materials. The

optical density (OD) is a parameter that describes the level of coloration, and is defined

for a particular wavelength (λ) by

( )λλλ xTTOD /log 0= (1.5)

where T0λ and Txλ are the transmittances of a reference sample and measured sample

respectively, at wavelength λ. The optical density is therefore a comparison between a

measured transmittance and some reference transmittance. The change in optical

density (∆ODλ) is frequently used to describe the optical state of EC systems and it is

defined by equation (1.6).

( )λλλ cb TTOD /log=∆ (1.6)

where Tcλ and Tbλ are the sample transmittances in the bleached and coloured states

respectively, at wavelength λ. The coloration efficiency (ηλ) is a measure of the

intensity of the colour change per injected ion concentration, and is defined as

15

( )QOD /λλη ∆= (1.7)

where Q is the injected charge density per unit area. In reality the optical density is only

proportional to the injected charge density over a limited region. The optical transition

as denoted by equation 1.4 requires the presence of both W5+ and W6+ sites. If charge is

injected until the stoichiometric lithium coefficient (y) is 0.5, there will be equal

numbers of W5+ and W6+ sites. Further charge injection will decrease the number of

transitions possible because there will be too few W5+ sites available. This site

saturation model predicts that the coloration efficiency will be zero at y=0.5 and then

become negative for higher charge densities, and this behaviour has been observed

experimentally [15].

In practise the coloration efficiency may be determined experimentally as the slope of

the linear region of a ∆ODλ versus Q plot. Figure 4 shows such a plot for a tungsten

oxide film coloured with lithium ions. Linear regression of the data shown in Figure 4

gives a coloration efficiency of 45.5cm2/C.

16

0.00

0.20

0.40

0.60

0.80

0.0 5.0 10.0 15.0 20.0 25.0

OD

Charge injected (mC/cm2)

∆

Figure 4 Determination of coloration efficiency from a plot of change in optical density

versus injected charge density. Coloration efficiency = 45.5cm2/C.

Some electrochromic devices have the ability to maintain a given level of coloration for

several hours, when disconnected from the external circuit. This ability of a film to

maintain coloration is called electrochromic memory and is a very desirable attribute for

a real EC system. If an EC device does not have a good electrochromic memory, the

optical density will decrease with time and the device will consume more power if it is

forced to maintain constant optical density. In the past, some coloured EC films have

been observed to self-bleach when exposed to various environmental conditions, and

self-bleaching has been observed to be heavily dependent on humidity.

ηλ=45.5cm2/C

17

In this work, self-bleaching was observed to be associated with irreversible charge

injection. The amount of charge available for transfer between the electrodes of a real

EC device is determined during manufacture. If self-bleaching occurs in a real device,

the slow reduction in the amount of mobile charge available for switching, may

eventually render it useless. The ability to produce an electrochromic device with an

excellent memory, very reversible charge injection and fast response times is the holy

grail for EC researchers.

1.6 Overall Objectives of the Study

The overall objectives of this study were to provide detailed information about the

kinetic processes occurring in sol-gel electrochromic films and the ways that these

processes are affected by increasing temperature. This information is required in order

to improve device design, and to make progress towards successful electrochromic

device commercialisation. The effects of temperature on the optical and electrical

properties of electrochromic films were determined, and the results modelled over a

wide temperature range. The effects of water and temperature on the reversibility and

response time of the EC process were also investigated.

1.7 Specific Aims of the Study

The specific aims of the study are:

o To investigate the effect of temperature on the voltage response of

electrochromic films

o To investigate the effects of temperature on the optical response of

electrochromic films

18

o To determine the effect of temperature on reversibility of the electrochromic

reaction

o To determine the effect of water on the reversibility of the electrochromic

reaction

o To validate and extend an existing model [5] by simulating experimental data

collected at elevated temperatures, thereby assessing the validity of this model.

o To demonstrate that information about the ionic mobility may be extracted from

the process of simulating experimental data obtained at elevated temperatures

o To gain information regarding the causes of self-bleaching, and the effect of

water and temperature on this phenomenon.

1.8 Research History

The research reported in this thesis took place in several stages, each culminating in a

published journal article or the presentation of a conference paper. This section is a

brief historical account of the research progress in my PhD program, to give the reader a

sense of the context and linkage of each paper. The experimental component of the

research is discussed in much more detail in Chapter 8.

The initial experimental work in this research project was focussed on simply

determining the voltage response of an EC film cycled in a liquid electrolyte over a

range of temperatures. A WO3 film was cycled in the ambient laboratory environment

at temperatures between 20 and 50ºC and the electrical and optical properties were

observed. The films required significantly smaller voltages for coloration and bleaching

when cycled at high temperature. The reversibility of the electrochromic reaction was

19

observed to decrease, with almost 4% of the injected charge unable to be extracted per

cycle at 50ºC. The coloration efficiency was also seen to decrease slightly at the higher

temperatures and this research was published in Electrochimica Acta with the title

Effect of temperature on electrochromic device switching voltages (Thesis Chapter 3)

in 1999. The decrease in coloration efficiency at high temperatures was anomalous and

experiments were planned to investigate the effect further.

It seemed plausible that the charge trapping observed in initial experiments was due to

the presence of water, and so the experimental apparatus was moved into a nitrogen-

filled dry-box of approximately 1ppm absolute humidity. Experiments in a dry-box

showed the same voltage-temperature behaviour as seen previously, however the

coloration efficiency and cycling reversibility were now independent of temperature. It

was then evident that both temperature and moisture play a significant role in the

reversibility of the EC process.

A set of self-bleaching experiments was carried out in the ambient laboratory

environment in order to better determine the effects of temperature and water on cycling

reversibility. A film was coloured to a specific charge density and then the counter

electrode was disconnected. The electrodes remained in the electrolyte solution for a

half-hour period, and the films were observed to slowly bleach. The emf shifted towards

more anodic potentials and the optical density decreased with time. The self-bleaching

rate was observed to increase with temperature, as did the amount of irreversibly

inserted charge (Qin-Qout). At 50ºC approximately 4.5mC/cm2 of the injected charge

20

(Qin=20mC/cm2) could not be extracted. Optical and electrical measurements were used

to estimate the rate of self-bleaching, and the estimates correlated well with the

measured values. This work showed that charge trapping is a real phenomenon that will

rapidly reduce the useful lifetime of an EC device and was published in Renewable

Energy, in a paper titled High temperature behaviour of electrochromics (Thesis

Chapter 4).

A major objective of this work was to simulate high temperature experimental data

using a mathematical model, and extend the work of Wang [5]. The data from

irreversible cycling was not suitable for modelling because it was very difficult to

estimate the amount of charge in the film at a given time. Film cycling in the dry-box

was very reversible and hence provided data to use in the simulation process. The

modelling work involved the simulation of experimental data obtained during coloration

at temperatures ranging from 20ºC to 50ºC. Estimates were made for the chemical

diffusion coefficient of lithium, and the variation in surface lithium concentration with

time was used to model the voltage during the charge injection process. Good fits were

obtained for data collected between 20ºC and 50ºC, and the estimated diffusion

coefficients increased with temperature, obeying Arrhenius type activation behaviour.

This allowed for the extrapolation of the activation energy for diffusion of lithium ions,

of 0.73eV. This work was published in Solar Materials and Solar Cells in a paper titled

Temperature Dependence of kinetic behaviour of sol-gel deposited electrochromics

(Thesis Chapter 5) and was the first publication to report the estimation of ion mobility

by simulation of voltage-time data. This was very significant because the direct

21

measurement of diffusion coefficients is a process that usually involves very time

consuming experimental work, often requiring costly equipment.

It was desirable to increase the temperature range at which the films were cycled, so an

experiment was carried out from 36ºC to 76ºC. The data was again simulated and

reasonable fits to the data were obtained over the wider temperature range. An

activation energy of 0.99eV was calculated, and although this was significantly larger

than the activation energy determined previously, it was in accord with the voltage

characteristics and indicated lower ionic mobility within the film. The results of this

modelling work suggested that the EC process was limited by diffusional motion of

atoms inside the EC film at low temperatures, and by the charge transfer step at high

temperatures. This work has been accepted for publication in Electrochimica Acta in a

paper titled Simulation of electrochromic switching voltages at elevated temperatures

(Thesis Chapter 6).

The earlier work on self-bleaching showed that films would lose their coloration when

exposed to an electrolyte in a moist environment for an extended period of time,

however no results were collected in a very dry environment. The same experiments

were then repeated in very dry conditions, in order to support the hypothesis that water

causes irreversibility and self-bleaching. Films in the dry-box were observed to slowly

self-bleach but at a rate much slower than in the ambient environment. A fresh film was

fired at 250ºC for several hours in an attempt to drive off any adsorbed water, and then

used in a similar self-bleaching experiment. This time the film did not self-bleach

22

significantly over a half-hour period. The cause of the self-bleaching effect was then

ascribed to the presence of water in different states. The presence of water in the

electrolyte led to relatively high irreversibility, and when the electrolyte was dry,

reversibility improved considerably. Removal of adsorbed water from the film itself

resulted in very reversible cycling of the film however the response time was very long.

Some cycles required in excess of 15 minutes in order to extract all of the injected

charge, thereby bleaching the film. This final experimental work on self-bleaching and

reversibility represented the end of the experimental work in this project, and

culminated in the paper Self-bleaching, memory effect and reversibility of

electrochromism at elevated temperatures (Thesis Chapter 7), which has been

submitted to Solar Energy Materials and Solar Cells.

Three other papers have also been published as a result of this PhD project, and these

form the Appendices of this thesis. These papers are more general than those described

above, and give the reader a more fundamental understanding of not only

electrochromic materials and systems, but advanced optical systems in general.

Appendix 1, Glazing Materials, was published in Materials Forum and is a review of

glazing materials, including but not limited to electrochromic systems. Other systems

such as liquid crystal, thermotropic, and angular selective devices are discussed here,

giving the reader an overview of the state-of-the-art in advanced glazing systems.

Appendix 2, Sol-Gel Deposited Electrochromic Devices has been published in

Renewable Energy and describes results of devices produced from sol-gel deposited

electrochromic films. The results presented include electrical and optical characteristics,

23

and also an analysis of the potential energy saving benefits of these devices. Appendix

3, entitled Smart Windows is an encyclopedia entry in the Encyclopedia of Smart

Materials, and discusses several types of glazing technologies. This paper also gives a

discussion of the physics appropriate to window glazings, and defines several of the

parameters used to characterise advanced glazing systems. If the reader is not familiar

with the technology described in the heavily research oriented papers discussed above,

then reading the appendices will provide much of the background necessary for a more

complete understanding of this research work.

24

REFERENCES

[1] S.K. Deb, Appl. Optics, Suppl. 3 on Electrophotography, 192-195 (1969).

[2] S.K. Deb, Solar Energy Materials and Solar Cells, 25, 327-338 (1992).

[3] B.W. Faughnan, R.S. Crandall and P.M. Heyman, R. C. A. Rev., 36, 177-197

(1975).

[4] B.W. Faughnan and R.S. Crandall. Electrochromic Display Based on WO3 in J.I.

Pankove, ed., Topics in Applied Physics V40, Display Devices, Spring-Verlag Berlag,

New York, Ch. 5. (1980)

[5] J. Wang, PhD Thesis, University of Technology, Sydney (1998).

[6] C. Ho, I.D. Raistrick and R.A. Huggins, J. Electrochem. Soc., 127(2), 343-350

(1980).

[7] S.K. Mohapatra, J. Electrochem. Soc., 125(2), 284-288 (1978).

[8] C.M. Lampert, A. Agrawal, C. Baertlien and J. Nagai, Solar energy Materials and

Solar Cells, 56, 449-463 (1999).

[9] C.G. Granqvist. Introduction to Materials Science for Solar Energy Conversion

Systems in A. A. M. Sayigh, ed., Materials Science for Solar Energy Conversion

Systems, Pergamon Press, New York, 1 (1991).

[10] J.S.E.M. Svensson and C.G. Granqvist, Appl. Phys.Lett., 45(8), 828-830 (1984).

[11] C. Wang and J.M. Bell, Solar Energy Materials and Solar Cells, 43(4), 377-391

(1996).

[12] S.K. Deb, Phil. Mag., 27, 801(1973).

[13] O.F. Schirmer, V. Wittwer and G. Bauer, J. Electrochem. Soc., 124, 749(1977).

[14] V. Wittwer, O.F. Schirmer and P. Schlotter, Solid State Commun., 25, 977(1978).

25

[15] M. Denesuk and D.R. Uhlmann, J. Electrochem. Soc., 143(9), L168-188 (1996).

CHAPTER 2

LITERATURE REVIEW

28

LITERATURE REVIEW

2.1 Electrochromism A Brief History

Between the late fifties and mid-nineteen sixties there was considerable interest in the

properties of tungsten bronzes [1-6]. It was observed that metal ions could be

preparatively incorporated into the lattice structure of tungsten oxide (WO3), and the

resulting non-stoichiometric compounds exhibited strong colour centres and metallic

properties.

In 1969 Deb reported that an electric field could be used to form colour centres in thin

films of tungsten oxide, producing colours similar to those of the tungsten bronzes [7].

Deb reported his results as A Novel Electrophotographic System, where a WO3 film

was incorporated with a photoconductive (CdS) layer. A picture could be reversibly

recorded on the WO3 film by the application of an electric potential across the film

combined with the projection of light (through a photographic negative) onto the film.

The light projected through the negative increased the conductivity of the CdS layer

allowing colouration of the tungsten oxide under the applied potential.

In the following ten years, the potential of electrochromic films in devices such as

smart windows was realised, and interest in the field grew rapidly [8-21]. Initial

research into electrochromics concentrated primarily on the optical and electrical

properties of HxWO3 [12,14], while other research investigated a number of other

possible electrochromic and counter electrode materials [15-18]. Metal oxides which

29

have been shown to exhibit electrochromism include NiO, TiO2, CeO2-TiO2 and IrO2.

NiO and V2O5 have been studied extensively as counter electrode materials. NiO is a

particularly attractive counter electrode material as is colours anodically (upon extraction

of ions), as opposed to WO3 which colours cathodically (upon ion insertion). Devices

which use both anodically and cathodically coloured layers exhibit a greater

transmittance range between the coloured and bleached states and are known as

complementary devices.

In recent years the developments in processing and production of transparent conducting

oxides (TCO) have increased the potential for application of electrochromic films in large

area smart windows. The commercialisation of on-line spray pyrolysis coating of

fluorine doped tin oxide (SnO2:F) [19] has led to cheaper TCOs and hence more

economically viable large-area devices.

In these devices it is undesirable to have liquid components and much research has been

carried out searching for suitable all solid state devices. The electrolyte layers used today

generally either consist of an organic polymer (doped with the ion inserted in the EC

reaction), or an inorganic fast-ion conductor such as M-β-alumina, (where M = Na, Li,

etc) [20,21]. Numerous reviews of electrochromic systems have been published recently,

and examples are given [22,23].

Some recent work closely related to this project involves the modelling of the

current/voltage characteristics of electrochromic films and devices. Bell et al [24] have

modelled the electrical characteristics of small and large area electrochromics based on

30

an equivalent network of distributed electrical devices, in order to predict the behaviour

of devices under given conditions. The small device model is applicable to the samples

used in this work, while the large device model is applicable to devices of size suitable

for commercial window application. Zhang et al [25,26] reported a model which

predicted WO3 film behaviour under constant voltage cycling, given the cell series

resistance, the lithium diffusion impedance and the emf (electromotive force) of the film.

This model has been modified recently [27] for constant current cycling of WO3 films

and also has the advantage of not assuming the semi-infinite approximation. This means

that this model can be applied to films of low thickness (typical of the films dealt with in

this work), and should describe the electrical characteristics of devices at high

temperatures. One of the aims of this work is to validate this model over a broad

temperature range, and future work is aimed at modification of this model to further

describe the behaviour of V2O5 counter electrodes and eventually complete devices.

2.2 The Electrochromic Reaction in WO3 Films

Electrochromic reactions occur when ions and electrons are simultaneously inserted into

the host lattice of an ion-intercalation electrode under the influence of an electric field.

The process occurs in several steps [28] as outlined below:

(i) ions in the electrolyte migrate towards the surface of the WO3 film under the

influences of diffusion and the applied electric field, and electrons from the

external circuit are inserted into the WO3 film via the ITO conductor

31

(ii) charge transfer occurs between the ion and electron at the WO3/electrolyte

interface. The cation is inserted into an interstitial position in the host lattice

between WO6 octahedra [29, 30]

(iii) the ion/electron couple diffuses into the oxide film (ie. towards the ITO)

under the influence of the concentration gradient

(iv) an intervalence charge transition takes place between a pair of neighbouring

tungsten atoms and an adjacent ion/electron pair resulting in a change in the

materials optical properties according to equation (2.1)

W5+(A) + W6+(B) + hν ⇔ W6+(A) + W5+(B) (1.1)

where (A) and (B) are neighbouring tungsten sites, the W6+ site is reduced

when it receives an intercalated electron from a W5+ site and light of energy

hν is absorbed [12].

The EC reaction therefore combines charge transfer (step (iii)) above and diffusion (steps

(iii) and (i) above) processes and the rates of these will depend on the way in which the

system is switched. There has been much discussion regarding the particular steps which

limit the rate of the EC reaction, and these will be discussed more fully below.

Failure of electrochromic systems may occur in a variety of ways, resulting in loss of the

ability to make the EC material undergo a significant optical transition. Several

degradation mechanisms have been proposed for EC systems, but there is no single

generally accepted mechanism that applies to EC WO3 films. Faughnan and Crandall

illustrated that lifetime will be greatest for switching conditions involving smaller

changes in optical density, and hence the application of lower electrical potentials.

32

Conversely, the higher the change in optical density achieved during cycling, the lower

the lifetime [12].

It is commonly believed that water plays an important role in the lifetime of EC systems.

One of the first investigations into the effects of water on EC cycling was published by

Arnoldussen [31] in 1981. Arnoldussen reported that WO3 films will dissolve in water by

forming tungstate species. The small size of the water molecule allows it to penetrate the

WO3 lattice and hydrolyse W-O-W bonds to form two W-OH bonds. It has been

suggested that these W-OH bonds are potential sites for irreversible lithium intercalation

[32], which supports the notion of water being a major cause of EC film and device

degradation. Svensson and Granqvist [33] concluded in 1984 that a long electrochromic

life for an EC device could only be achieved if water was carefully excluded from the

system. The degradation issues due to the presence of water in an electrochromic system

have resulted in a search for more aprotic solvents. The effects of water on the cycling

characteristics are discussed in more detail in section 2.4.

2.3 Electrochromic Characteristics at Elevated Temperatures

Electrochromic devices reduce the transmission of light by absorbing a significant

proportion of the incoming radiation. The absorption of this energy is associated with an

increase in the temperature of the device and hence EC devices may typically reach

temperatures exceeding 65ºC [34,35] during their normal operation. It is therefore

essential to understand the ways in which the kinetic properties of EC devices are

affected by temperature, so that the control regime for switching these devices may be

optimised for a wide temperature range. The response time for EC devices slows down

33

as temperature decreases [22], and conversely the mobility of ions within EC systems

increases with increasing temperature. It should then be possible to switch EC devices

with the application of smaller voltages at high temperatures, as predicted by Nernstian

behaviour (see section 1.4)

, or alternately higher switching currents (and faster switching) may be possible without

device degradation. There is little published information in the literature regarding the

effects of temperature on electrochromic device switching properties and this is the

primary knowledge gap that this work aims to address.

Hackwood et al [17] reported a study in 1980 that illustrated the dependence of switching

speed of electrochromic iridium oxide films on temperature. The EC iridium oxide films

were cycled by insertion and extraction of protons (in aqueous media), and colouration

and bleaching speeds were measured at temperatures ranging from +20 to 43ºC. In this

work it was observed that the switching speed of these films increased with increasing

temperature and the response was explained by an activation-controlled scheme, with an

activation energy of approximately 0.25eV. The colouration efficiency at +20ºC was not

significantly different from that at 30ºC and was therefore observed to be independent of

temperature. The iridium oxide films discussed in this paper have application in fast

switching (switching speeds <0.5s) display devices, where it is unlikely that their

operational temperature range will approach that of ECs for smart window applications.

Although the temperature range, film material and inserted species are dissimilar to the

tungsten oxide EC films common to smart window research, the paper is important as the

first reported experimental work on temperature effects on EC systems and the reported

34

trends in kinetic behaviour may reasonably be expected to be observed in other EC

systems.

In 1996 Badding et al [34] reported results from durability testing of a monolithic EC

device over a wide temperature range. The multilayer device utilised a solid-state ionic

conductor as the electrolyte and used a constant voltage step to colour and bleach the

device, applying the same voltage limits for all temperatures. Switching speed was

defined as the time taken to reach 90% of the full range of visible coloration and this was

measured from 27 to +70ºC. At 27ºC the device was observed to switch very slowly,

taking almost 25 minutes to reach 90% of its full coloration. Switching speed increased

rapidly as temperature was increased, taking less than five minutes to colour the device

for all temperatures above 0ºC. At 25ºC and 70ºC the switching times were

approximately 2 minutes and no significant increases in speed were observed over this

temperature range. The paper does not report any attempt to calculate an activation

energy for diffusion from the results of the switching speeds, or the effect of faster

switching speeds on device degradation. As switching currents were not measured during

cycling, the amount of charge injected and extracted per cycle was not calculated and no

conclusions may be made regarding the cycling reversibility at high temperature. The

increases in switching speed indicate that kinetic processes such as ion transport are

occurring faster at high temperatures, which is in agreement with Hackwood et als work

on iridium oxide films.

Another study by Tulloch et al [36] reported the cycling voltages of a sol-gel EC film at

18ºC and at 50ºC, under constant current operation. In this work a complete device using

sol-gel deposited electrodes was cycled in an oil bath and electrical and optical

35

measurements made. The device was cycled using a constant current technique in order

to control the quantity of charge injected and hence the colouration level achieved. The

switching algorithm involved cycling of the device under a constant current step until a

specific injected charge density was reached. If the voltage reached a preset safe limit

during coloration or bleaching, the mode of operation changed to constant voltage at the

preset voltage limit in order to protect the device from high overpotentials. The voltage

required to colour the device to 10mC/cm2 at 18ºC was approximately 1.6V while at 50ºC

only 1.0V was required to achieve the same injected charge density. A total of

13mC/cm2 was injected in the cycles reported and the maximum voltage limit on

coloration was set to +1.8V. The device cycled at 18ºC reached this limit after

approximately 13.1mC/cm2 was injected, but never reached this limit when cycled at

50ºC. This work demonstrated that significantly lower voltages are required to colour

and bleach EC devices at elevated temperatures and hence the way in which we switch

them is very important. The paper also showed that a voltage-limited constant current

technique has the advantage (over constant voltage techniques) that the amount of charge

injected and hence the level of coloration is the same each cycle regardless of the

temperature. Tulloch et al did not however elaborate on the kinetic mechanisms

responsible for the lower voltages except to comment that the result was expected

because diffusion coefficients and polymer conductance increase with temperature.

It is evident from the papers discussed above that the general effect of temperature is to

increase the ease by which EC reactions occur. Identification of the mechanism or

mechanisms responsible for this increased reaction rate (constant voltage cycling) or

reduction in switching voltage (constant current cycling) is not possible from the previous

36

work reported in the literature, and more work is needed to clarify this. Nernstian

behaviour (see section 1.3) dictates that the magnitude of the electromotive force (emf)

for a given state of charge injection (or film composition) will decrease with increasing

temperature. The result is that lower switching voltages are required at high

temperatures. A concern however, is that side reactions leading to device degradation

will also proceed more easily and so the application of the same switching voltages over a

wide temperature range may adversely affect device lifetime. A better understanding of

the kinetic properties and rate limiting mechanisms of ECs at elevated temperatures will

allow us control EC devices in a fashion that will use knowledge of material properties to

maximise device lifetime.

2.4 Lifetime, Irreversibility and Self-Bleaching

Ideally EC films and devices may be cycled reversibly between coloured and bleached

states, so that the charge injected into the working electrode is always fully recovered

upon bleaching. In practice this is not always possible and the reversibility is dependent

upon various combinations of deposition conditions, microstructure and cycling

conditions. Irreversibility has been observed many times over the last three decades and

several papers outline the anomalous effects observed.

In 1978 Randin [37] reported work whereby the stability of EC WO3 films was studied in

10:1 glycerin/sulfuric acid (H2SO4) mixtures and in various non-aqueous electrolytes.

The WO3 films were observed to slowly dissolve in glycerin/H2SO4 regardless of whether

the films were actively cycled or simply stored in the electrolyte solution, however the

37

dissolution rate was significantly increased by film cycling. WO3 films cycled in non-

aqueous electrolytes were observed to dissolve much more slowly, presumably due to the

lack of available water. Randin proposed that the tungsten oxide dissolved in water from

the electrolyte to form a soluble polytungstic acid [38] species. Randin also suggested

that films cycled in lithium salt/organic electrolytes with small amounts of water present

would undergo dual insertion of protons and lithium ions, because proton migration

occurs much faster than for lithium. Randin observed one anodic peak in the cyclic

voltammograms of films cycled in aqueous media, but two peaks for films cycled in non-

aqueous electrolytes. These two peaks were attributed the to solid-state diffusion of

protons and lithium ions as described by the space-charge limited bleaching model of

Faughnan and Crandal[14]. Randins paper was the first to investigate the stability of

tungsten oxide electrodes in different solvents and to report the effects of various solvents

on the reversibility of electrochromic cycling. This paper also showed that cycling

lifetime is dependent on switching frequency and to discuss the implications of this in

terms of validation of accelerated tests.

In 1990 Zhong et al [39] reported work where WO3 films were coloured and bleached at

room temperature and significant amounts of lithium were observed to remain inside the

films, even after the bleaching process. The crystallographic structure of the tungsten

oxide changed upon insertion of either H+ or Li+ ions, and Zhong et al monitored these

changes using x-ray diffraction techniques. The x-ray diffraction studies showed

characteristic peaks attributable to the formation of the tungsten bronzes HxWO3 and

LixWO3 when the WO3 films were coloured with H+ and Li+ ions respectively. When the

films were bleached, only the peaks from tungsten oxide were present. The crystal

38

structure was therefore observed to revert back to its original form after bleaching, even

though a significant proportion of the injected ions still remained inside the films.

Secondary ion mass spectroscopy (SIMS) and inductively coupled plasma-mass

spectrometry (ICP-MS) experiments revealed that on average, approximately 50% of the

ions injected during the coloration process were still inside the films after bleaching.

Using this combination of XRD, SIMS and ICP-MS techniques, Zhong was able to show

that a proportion of injected ions remained inside the WO3 EC films after bleaching,

however these ions were not held in interstitial positions in the tungsten oxide lattice and

they did not contribute to coloration.

Zhong et al hypothesised that during the bleaching process, some of the ions were trapped

in optically inactive sites outside the WO3 lattice such as on grain boundaries. An ion

trapped at a grain boundary would not participate in the EC reaction because its electron

would also be trapped and therefore would not be available for donation to a tungsten(VI)

atom. Zhong et al observed that the conductivity of bleached samples was the same as

that for as-deposited samples, which showed that the electrons (associated with the

remaining injected cations) were not available for electrical conduction. This supports

the hypothesis that the cations that remain inside the films are not in interstitial lattice

positions and so cannot contribute to coloration.

In 1991 Hashimoto and Matsuoka [32] reported a study on the electrochromic lifetime of

mixed oxide WO3-TiO2 films. The films were prepared by electron beam deposition with

TiO2 concentrations ranging from zero to 30 mole percent. The longest lifetime was

observed for a film with 15.6 mol % TiO2 and the lifetime was five times greater than for

39

a pure WO3 film. Hashimoto and Matsuoka used SIMS to measure the amount of lithium

in as-deposited, bleached and coloured films and found that there was a significant

amount of lithium in all of the bleached samples. The bleached WO3 sample contained

approximately 44% of the injected lithium ions, while the bleached 15.6 mol% TiO2

sample only contained about 20% of the injected lithium ions. This was the first report of

evidence relating cycling lifetime to the reversibility of individual cycles. Hashimoto and

Matsuoka used X-ray photoelectron spectroscopy (XPS) to obtain information on the

binding of lithium ions in coloured and bleached films and showed that the lithium in the

bleached films is not located in the body centred position in the WO3 lattice, as it is for a

tungsten bronze. They suggested that the lithium is present in the bleached film as O-Li

after exchange between a proton and lithium ion of an O-H group. The XPS experiments

also showed that the number of lithium ions present as O-Li in the bleached samples was

smallest for those doped with TiO2. Hashimoto and Matsuoka investigated the cause for

increased lifetime of the mixed oxide films by using low loss region EELS and showed

that the electrons in WO3-TiO2 are more tightly bound than in pure WO3, and used

Raman spectroscopy to show that the W-O bond length is decreased by TiO2 doping.

Hashimoto and Matsuoka proposed that there are several defects present in pure WO3,

including W-O-H and W=O sites, which may act as lithium trapping sites. These bonds

may be broken by adding small amounts of TiO2 to the structure, the Ti atoms bonding to

oxygens (W-O-Ti-O-W) to remove the trapping sites. Hashimoto and Matsuoka

proposed in earlier work [40] that crystalline lithium tungstate was irreversibly formed

when relatively large amounts of lithium were intercalated into films of amorphous WO3

40

thereby degrading the amorphous film. The increase in lifetime upon addition of TiO2 to

WO3 films was therefore explained as a result reducing the number of defect bonds in the

lattice, and thereby preventing the accumulation of lithium and the subsequent formation

of lithium tungstate.

In 1992 Duffy et al [41] reported on a series of experiments in which several

electrochromic devices were fabricated with dry or damp polymer electrolytes. A

polymer electrolyte was prepared by adding solid orthophosphoric oxide to polyethylene

oxide, and plasticising with pre-dried acetonitrile. A portion of the resulting polymer was

used to fabricate dry devices, while the remainder was exposed to the ambient

laboratory environment for several hours before being used to fabricate wet devices.

Duffy et al found that after coloration the dry EC devices retained their colour almost

indefinitely (when disconnected from the external circuit) and therefore had an excellent

electrochromic memory, while the damp devices lost all colour within one week under

the same conditions. When a damp device had lost its colour it could no longer be cycled

unless the WO3 was first discharged (cycled with WO3 as the anode) or if the counter

electrode was replaced with a fresh piece of protonated ITO. Duffy concluded that the

loss of colour in devices with damp electrolyte was not caused by loss of the working

electrode or by deactivation of the WO3.

Duffy et al investigated the mechanism of self-bleaching by using impedance

spectroscopy and electron microscopy on devices with dry and damp electrolytes. The

impedance spectra for the dry and damp devices were very different, the latter exhibiting

41

blocking electrode behaviour and had no charge transfer semicircle in its spectrum. The

high frequency resistance was the same for coloured and self-bleached samples, which

rules out the possible explanation for self-bleaching of protons migrating into the ITO

layer. The electron micrograph of a film that was cycled in a damp device for one day

showed signs of swelling and buckling of the film. Another film was cycled for one

month under the same conditions and the electron micrograph clearly showed the film

cracking and breaking away from the substrate.

Duffy et al believed that the mechanism for self-bleaching involved selective dissolution

of tungsten(V) atoms so that the intervalence charge transfer which gives rise to

coloration (see section 2.2) is unable to occur. Duffy et al suggested that protons and

electrons migrate to internal surfaces within the electrode where WV species dissolve into

the electrolyte with the hydrogen and oxygen atoms. They also reasoned that this

dissolution would lead to an increase in film porosity which in turn would lead to water

uptake and swelling of the film.

In 1993 Zhang et al[42] reported what is currently the most complete and systematic

study into self-bleaching of electrochromic tungsten oxide films. Zhang et al studied the

effects of preparation conditions, the injected ion, environmental conditions, substrate

and film surface roughness. In this work WO3 films were coloured either 1M

LiClO4/propylene carbonate (PC) or dilute H2SO4 electrolyte, then washed and dried and

transferred to a sealed chamber. Into this chamber they introduced several different gas

mixtures to study their effects. The coloured films were exposed to pure argon, nitrogen

42

or oxygen, and the ambient laboratory environment of ~21%O2, ~79%N2 and 33%

relative humidity at 23ºC.

Sputtered WO3 films coloured with lithium ions were observed to bleach almost

completely (transmittance increased from 30% to 81%) over an 8 hour period, when

exposed to the ambient environment, whereas the same films only increased in

transmittance by 10% and 3% in dry oxygen and nitrogen atmospheres respectively.

When exposed to pure argon atmosphere or a 10-6 torr vacuum, negligible self-bleaching

was observed and Zhang et al concluded that the self-bleaching of EC films coloured

with lithium ions is dominated by the reaction of Li+ ions with water.

The same sputtered films coloured by proton injection were also exposed to environments

of ambient, oxygen and argon and generally exhibited much less self-bleaching than the

films coloured with Li+ ions. These films did not self-bleach significantly in argon, but

increased in transmittance from 30% to approximately 50% when exposed to either

oxygen or the ambient environment for an 8 hour period, so Zhang et al concluded that

self-bleaching in EC films coloured with protons is dominated by the reaction between

the inserted protons and oxygen gas.

Electrochromic WO3 films coloured with lithium ions and protons were exposed to

ambient and argon environments in order to examine the effect of deposition conditions

on the self-bleaching process. The films coloured by proton injection exhibited slower

self-bleaching than those coloured with lithium ions, and the evaporated WO3 films self-

bleached much more slowly than the sputtered films under the same exposure conditions.

43

The authors also mentioned that they have films deposited by plasma enhanced chemical

vapour deposition (PECVD) which were coloured and then stored in air for four years,

and have retained much of their original coloration. The differences in the self-bleaching

rates of films deposited by different techniques illustrates that self-bleaching behaviour of

a film is strongly dependent on the morphology.

Zhang et al also studied the effect of surface thickness and observed that self-bleaching

rate decreased monotonically with increasing film thickness. They found that the self-

bleaching rate was related more specifically to the ratio of the surface roughness to film

thickness, which is a measure of the amount of surface area relative to the thickness of

the film. Zhang et al used the results from all of these experiments to hypothesise a

model for the mechanism of self-bleaching in WO3 EC films. The model describes the

process occurring in two stages. Initially gases from the environment react with injected

protons or lithium ions on the surfaces of internal pores according to equations (2.2) and

(2.3) respectively.

2H+ + 2e- + 1/ 2O2 ⇒ H2O (1.2)

2Li+ + 2e- + 2H2O ⇒ 2LiOH + H2 (1.3)

Due to the ready availability of these surface ions, this step will be reaction limited and

will result in an initial fast decrease in the colouration centers formed by the protons and

lithium ions. Films with large surface roughness will therefore self-bleach faster and

films with high thickness will bleach more slowlyslower because a smaller portion of the

injected ions is at the surface. The second process involves diffusion of protons/lithium

44

ions to the surface or pores and diffusion of oxygen/water into the film. This process is

therefore diffusion limited and explains why the self-bleaching rate at long time periods

(>2 hours) increases with film thickness.

In 1993 Burdis and Siddle [43] published their work on the effect of sputtering conditions

on the reversibility of ionic insertion into EC WO3 films. Burdis and Siddle produced

films with a wide range of sputtering conditions including various pressures, reactive gas

mixtures and choice of metal or metal oxide targets. They grouped the films produced

into the broad classifications of polycrystalline and amorphous materials. They observed

that the polycrystalline materials obeyed Beer-Lambert law [44] upon ionic insertion, so

the optical density increased linearly with injected charge density. The amorphous

materials exhibited different behaviour in which large amounts of charge were inserted

before there was a linear increase in the optical density of the film. In one case

75mC/cm2 of lithium ions was injected into a film in order to achieve a solar optical

density of approximately 0.27 and this optical density returned to zero when only

15mC/cm2 was extracted from the film. The remaining 60mC/cm2 was injected

irreversibly, as seen by the excessive voltage (6V vs. Li) required to remove further

charge. The amount of charge irreversibly intercalated per cycle was observed to

decrease rapidly with progressive cycling but the total amount of irreversible charge was

as large as 100mC/cm2. Burdis and Siddle dissolved the WO3 film off the substrate and

carried out atomic absorption spectroscopy (AAS) on the solution to check whether the

lithium was actually still inside the film. Approximately 97% of the expected lithium

concentration was recovered from the film, proving that irreversible charge intercalation

45

is a real phenomenon and that lithium may be intercalated into sites which are either

optically active or inactive.

Burdis and Siddle also carried out coulometric titrations on both polycrystalline and

amorphous samples, and showed that the amorphous film started to precipitate out a

second phase upon intercalation of lithium ions. They postulated that the second phase

(possibly Li2O) was a result of reaction between inserted lithium ions and oxygen gas

trapped within the structure from the sputtering process. Evidence supporting this

includes the fact that sputtering conditions that increase the amount of trapped gas also

increase the irreversibility, and the irreversible effect is much smaller for evaporated

films.

In 1996 Michalak and Owen [45] reported a study regarding the observation of parasitic

currents during cycling of electrochromic WO3 films in non-aqueous electrolytes.

Michalak and Owen defined the parasitic charge as the difference between the inserted

and extracted charge for a cycle that begins and ends at the same transmission. The

parasitic current is an electrical current due to a reaction other than the reversible

electrochromic reaction, and Michalak and Owen outlined four possible effects of these

parasitic currents:

(1) Gas evolution

(2) Formation of passivating layers of reaction product

(3) Corrosion of electrode, losing host ions into the electrolyte

(4) Charge imbalance between the colouring and bleaching cycles

46

Michalak and Owen integrated the applied current while using cyclic voltammetry to

colour and bleach a WO3 electrode, then calculated the total amount of charge inserted

and extracted. Approximately 15mC/cm2 of lithium ions was irreversibly inserted on the

first cycle. The initial insertion of this 15mC/cm2 during the first cycle was not

accompanied by an optical change of the film, and so the authors concluded that the

injected lithium ions were reacting with some impurities in the film, possibly peroxide or

O- species present in the virgin electrode. After the first cycle, a constant coloration

efficiency was then observed for any further cycles. Michalak and Owen reported that

the anodic parasitic current was negligible up to 4.5V vs Li+/Li and that the cathodic

parasitic current was approximately 0.1µA/cm2 at 2V vs Li+/Li.

These papers address the issues of lifetime, reversibility and self-bleaching. It is evident

that WO3 films coloured with lithium ions often deviate from ideal reversible behaviour.

When cycling EC films (as opposed to complete devices) in a liquid electrolyte this

problem has a minor effect on lifetime because there is an almost infinite ion source

available. In a complete electrochromic device however, there is a finite ion source, and

the lifetime of the device may be dictated by the total amount of charge incorporated into

the device during fabrication. When devices are made it is common to incorporate more

charge than will be injected and extracted from the working electrode per cycle, and the

excess charge is stored in the counter electrode. Any irreversibility or inability to extract

charge from the working electrode upon bleaching will deplete this excess charge. Once

the sum of the irreversible charge (over all cycles) exceeds the initial excess charge that

was incorporated into the device, there will be a reduction in the amount of charge able to

47

be transferred between the working and counter electrodes. This effect will be observed

as a reduction in the transmittance range of the device and this change may well signal

the end of the life of the device. Minimising irreversible charge injection is therefore

very important in maximising device lifetime.

The papers discussed here tell us that reversibility/self-bleaching and hence lifetime are

heavily dependent on factors such as deposition conditions, water content, the injected

ion and electrolyte composition. These papers do not address these issues in sol-gel films

nor do they address the issue of the high temperatures that devices will attain in real

operation. A major aim of this work is therefore to start to fill the knowledge gap in

terms of lifetime, reversibility and self-bleaching at high temperatures, specifically in sol-

gel deposited electrochromic WO3-TiO2 films.

2.5 Models for Simulation of Electrochromic Switching Characteristics

Models are mathematical tools that may be used to simulate or interpret an observed

response, or to predict a response under a certain set of conditions. Modelling of

electrochromic phenomena allow us to extract useful information about the kinetic

processes occurring on a molecular scale, but several assumptions must be made in order

to simplify the modelling process. The rate of the EC reaction will be limited by the

slowest step in the overall EC process, so we must consider all steps including diffusion

and charge transfer processes as well as electrical processes such as resistances across

interfaces between components.

48

In 1975 Faughnan et al[12] published a review paper which included a qualitative model

for the colouration and bleaching of electrochromic WO3 films and for the optical

transition which results. According to the coloration model, electrons are injected from

the TCO and ions are injected from the electrolyte, under the influence of the applied

electric field. The electrons and ions will migrate into the film, where they combine to

form HxWO3. The ion is accommodated inside the lattice of the WO3, and the electron is

localised on a nearby tungsten site thereby reducing W6+ to W5+. Faughnan et al

explained the optical transition as an intervalence charge transfer occurring according to

equation (2.1). An electron localised on a W5+ atom reacts with an adjacent W6+ atom,

after absorbing sufficient energy to cross a potential barrier. The simultaneous oxidation

and reduction reactions that occur are accompanied by a radiationless transition of

energy, sometimes called a phonon emission.

The electrical conductivity of a HxWO3 film was observed to change from insulating in

the bleached, and approach metallic conductivity when highly coloured and the authors

estimated proton mobility to be in the range of 10-10-10-6cm2/s. Faughnan et al described

the bleaching process as being dominated by a space-charge limited current flow of

electrons and ions. The surface of the WO3 film will bleach first, and remaining ions

must move across a region of decreasing charge density in order to reach the electrolyte

interface.

Later in 1975 Faughnan et al [13] presented a quantitative model for the current response

during bleaching of WO3 films, when extracting protons under a constant voltage step.

The model elaborated on the notion of a space-charge limited current being the rate-

49

limiting step and the authors developed mathematical equations which described the

experimental data very well. When a voltage is applied to bleach a WO3 film electrons

leave the film via the TCO and ions migrate back into the electrolyte. This results in a

region of film adjacent to the electrolyte being depleted of electrons and so any current in

this region is purely a proton current. There is also a similar region at the TCO interface

that is depleted of protons, hence only an electron current flows. The third region is the

coloured part inside the film and it has both proton and electron currents flowing through

it during bleaching.

Faughnan et al reasoned that the voltage drop across the film is equal to the sum of the

voltage drops across each of the three regions described above. The potential drop across

the neutral plasma region (the coloured part) will be negligible because the conductivity

of this region is very high. Faughnan et al measured that the electron mobility was

several orders of magnitude higher than proton mobility, which means that the voltage

drop across the region of electron current will be very small. The entire switching

voltage is therefore applied to the neutral plasma region, which will shrink with time.

The mathematical description of this process defines the bleaching current density as a

function of time under constant voltage, as follows:

( )( )4

3

21

41

03

4)(

t

VtJ pµκερ= (1.4)

where )(tJ is current density, ρ is the volume charge density of protons in the neutral

plasma region, κ is the relative dielectric constant, 0ε is the permittivity of free space,

50

pµ is the mobility of the protons, V is the voltage and t is time. The 43

−t dependence of

the current was confirmed experimentally and the model simulated the data very well.

Crandall and Faughnan [14] also published a quantitative model for the coloration

process when protons are inserted into WO3 films. This model assumed that a barrier to

charge transfer at the electrolyte/electrode interface limited the current flow. The barrier

is associated with charge passing the electrolyte double layer at the electrode surface,

losing a solvation sheath and attaching to the electrode surface. The equation for

coloration current density (as a function of time) described the experimental data very

well for short time periods, except for at the highest voltage (0.5V). The authors

concluded that it was likely that other mechanisms limited current flow at voltages in

excess of 0.5V. The inherent limitation of this model is that the process of diffusion is

not considered, and so it is only applicable for small voltages or short time periods and

hence low colouration densities.

In 1976 Crandall et al [46] derived an expression for the emf of a tungsten film after the

electrochemical insertion of protons and electrons. Crandall et al derived the model

using a first principles approach to the thermodynamic properties of the species present in

the EC reaction and the model was used to simulate experimental measurements of film

emf for various injected charge densities. Crandall et al used the general equation

E=∆G/F to relate the emf (E) to the free energy change (∆G) that occurs when x moles

of protons are transferred from the electrolyte to the solid lattice of composition HxWO3

(F is Faradays constant). Free energy is a function of the chemical potential of the

51

reactants and products, so Crandall et al used electrical measurements to obtain

information about the free energy changes upon ion insertion. Information about the free

energy change was then used to ascertain the dependence of the chemical potential on the

composition of the HxWO3 film.

Crandall et al constructed a simplified expression for the free energy of HxWO3, by

grouping the possible contributions by their dependence on the compositional parameter,

x. The free energy of formation of pure WO3 (G0) was assumed to be independent of

x. Terms with a linear dependence on x arise from the free energy changes of reducing

x moles of W6+ to W5+, and the associated changes in interactions between W5+-O and

H-O pairs. Interactions between pairs such as W5+- W5+, W5+- OH-, OH-OH- were

assumed to have a quadratic dependence on x, and contributions to free energy of a

higher order in x were neglected. Finally the energy of distribution (Gd) was included

to describe the entropy changes which occur when x moles of hydrogen is distributed in

one mole of WO3, and all the terms were combined to give the the theoretical expression

for the molar free energy WO3 as

dGBxAxGG +++= 20 (1.5)

where A and B are constants. The chemical potential of hydrogen in HxWO3 (µH) was

then determined by partially differentiating equation (3.5) with respect to x to give

−

++=x

xnRTBxAH 1ln2µ (1.6)

where A and B are constants and n is a factor which describes the spatial correlation of

the injected proton and its associated electron. An expression for emf was also given as

52

−

−+=x

xF

nRTbxaemf1

ln (1.7)

where a combines all constant terms and F

Bb 2−= .

Crandall et al measured the variation in emf for compositions ranging from x=0.002 to

x=0.5 and were able to use equation (2.7) to successfully simulate the experimental data.

In 1980 Faughnan and Crandall [30] outlined the possible mechanisms that may limit the

dynamics of the coloration process as:

1) transport of the electrons and cations through the bulk of the WO3 film

2) a barrier at the TCO/WO3 interface

3) a barrier at the electrolyte/WO3 interface

4) a barrier at the counter electrode

5) charge transport in the electrolyte

Faughnan and Crandall assumed that mechanism 2) is not significant in real systems

because the TCO and WO3 are in electrical contact, and mechanism 5) may also be

neglected if we choose an electrolyte of high conductivity. Mechanism 5) may also be

neglected if we choose a suitable counter electrode material so mechanisms 1) and 3)

may be considered the most likely rate limiting mechanisms for electrochromic reactions.

Mechanism 1) describes the diffusional motion of ions and electrons within the EC film.

Ions are inserted at the WO3/electrolyte interface at vacant interstitial positions, and

hence migrate deeper into the film if more ions are to be intercalated. The rate of this

process depends on how fast the ions may diffuse through the film and this will also

53

depend on temperature. Increases in the thermal energy of mobile ions as well as the host

lattice expansion will allow ionic diffusion to occur more readily at higher temperature.

Diffusion is often considered to be one of the most important rate limiting mechanism

because this behaviour is frequently observed in experimental results [26,28,47,48,49].

Mechanism 3) describes the charge transfer process across the electrode/electrolyte

interface. The applied electrical potential provides energy to assist the ions in the

electrolyte to lose their solvation sheaths and combine with an electron inside the host

lattice, so the greater the electrical potential, the faster this step will be [14].

In 1980 Reichman et al [47] proposed a digital simulation model for the electrical current

during electrochromic insertion of protons in WO3 films. Reichman et al assumed that

the current was limited by charge transfer across the electrolyte/electrode interface,

combined with the diffusive motion of protons within the film and accumulation of

protons inside the film approaching some saturation level. The model was used to

simulate the current-voltage and current-time relationships observed experimentally for

anodically deposited and evaporated WO3 films. Good fits to the experimental data were

obtained for longer time periods of up to 60 seconds, but fits were poor for very short

times(t~5s). The model was also used to successfully predict the dependence of the i-V

characteristics of cyclic voltammetry on the scan rate. The modelling procedure involved

estimating a charge transfer rate constant, the diffusion coefficient of hydrogen atoms

inside the film and it was also necessary to experimentally measure the electrochemical

isotherm in the modelling process. Hydrogen diffusion coefficients estimated from the

54

modelling process ranged from 1x10-9 - 2x10-10cm2/s for evaporated films to

5x10-8cm2/s for anodic films.

Nagai and Kamimori [48] reported an improved kinetic model for WO3 electrochromism

in 1983, which explained the results of emf , chronoamperometry, voltammetry and AC

impedance measurements. The model assumed that the overall impedance of the cell and

the diffusion of atoms inside the film limit the current flowing during coloration or

bleaching. Nagai and Kamimori used the following formula for emf as a function of x (in

LixWO3)

−

−+=x

xFRTbaxemf

1ln)( ν (1.8)

where a, b, and ν are all constants which are used to fit experimental data to the model.

Equation (2.8) simulated the experimental data for emf (x) very well using a = -0.66V, b

= -0.87V and ν = 5.76, when the film was cycled in 1M LiClO4/PC electrolyte.

Nagai and Kamimori carried out chronoamperometric experiments and found that the

current was nearly constant for short time periods (t<0.5s) immediately after the start of

coloration and bleaching. This current was described by the equation

RxemfEi app /))(( −= (1.9)

The resistance in equation (2.9) was determined from impedance measurements to be the

sum of the resistances of the solution, electrode, charge transfer and mass transfer. The

authors assumed that charge transfer occurred at the electrolyte/ WO3 interface and

estimated a diffusion coefficient of 5x10-10cm2/s. Nagai and Kamimori concluded that

55

the cell impedance was a combination of the series cell resistance (~100Ωcm2) and a

capacitance (~104-105µF/cm2), and therefore viewed the electrical characteristics of the

EC system as being similar to the charging and discharging of a capacitor in series with a

resistor. This analogy with electrical components explained the initial dependence of

current on R and the dependence of rate on the cells time constant and diffusion

coefficient.

Raistrick and Huggins [50] reported a model in 1982 which described the transient

electrical response of solid-solution electrodes such as EC WO3 films. The authors

theoretically calculated an electrode impedance by using Laplace transforms to solve

diffusion equations for solid solution electrodes and then used this impedance to

determine the response of the system to an applied voltage or current. The model is very

useful because it can be used to model electrical characteristics for either constant current

or voltage switching, but it does not consider the effect of electrode emf on switching

dynamics.

Bohnke et al [49] reported a model in 1992 that described the way in which the

colouration current for WO3 films is limited by the chemical potential of the HxWO3 film,

the applied overpotential and a series resistance. They observed that current was

proportional to voltage for low levels of charge injection and hypothesised that the

current was limited by a resistance, rather than being limited purely by the charge transfer

process. Vuillemin and Bohnke attributed this resistance to the conductivities of the

electrolyte, the HxWO3 bronze and substrate, and a charge transfer resistance. Impedance

56

spectroscopy was used to measure this resistance experimentally for various WO3 films.

The charge transfer resistance was observed to decrease at the beginning of insertion for

very small time periods (<100ms). Vuillemin and Bohnke [28] later reported the use of

this model to simulate experimental measurements of current versus time, and estimated a

diffusion coefficient of 5x10-10cm2/s for lithium ions in WO3 film.

In 1993 Zhang et al [25] improved on the work of Raistrick and Huggins [50] by

extending the previous model to describe the effects of the cell series resistance, the

lithium diffusion impedance and the emf of the film. The authors reported that each of

these factors may limit the current under certain experimental conditions, such as the film

thickness, applied potential, etc. The model described by Zhang et al therefore

encompassed the previous models and under certain conditions, may be mathematically

reduced to the forms reported previously. A limitation of this model is that it assumed

the semi-infinite approximation (ie for t << L2/D where t is time, L is film thickness and

D is the average diffusion coefficient), which means that it is only applicable for

relatively thick films (L > ~100nm). Zhang et als model describes the switching current

during coloration by a constant voltage step and hence is of limited use in describing EC

processes under constant current switching.

Wang [27] improved on the modelling work of Raistrick and Huggins [50] and Zhang et

al [25,26] in 1998 by adapting it to describe the voltage response to constant current

charge injection and extraction. The model reported by Wang also did not assume the

semi-infinite approximation, and hence is applicable for relatively thin films (~100 nm).

57

Wangs model for the voltage during the coloration process is described by equation

(2.10)

−

−++=),0(1

),0(ln),0('.)(ty

tyFRTtybaRitV ccaν (1.10)

where y(0,t) is the stoichiometric coefficient of lithium (in LiyWO3) at the

electrode/electrolyte boundary (ie y(0,t) = Vm.c(0,t) where Vm is the molar volume of the

film and c(0,t) is the surface lithium concentration per unit volume).

−=

dydEbb'

where dE/dy is the slope of the coulometric titration curve, and a, b, and ν are the same

constants as in Nagai and Kamimoris model (see above). Rc is the series resistance

during coloration and ic is the colouring current. This model requires the determination

of the surface concentration of lithium ions, and Wang calculated this using equation

(2.11)

Γ=DnF

jtc 2),0( (0.11(a))

and ππt

Dtlkt

Dtklerfc

Dkl

kk

−−−=Γ ∑∑∞

=

∞

= 1

22

1

)][exp(2]2[ (2.11(b))

where n is the number of electrons in the process, l is film thickness, j is current density,

F is Faradays constant and D is the chemical diffusion coefficient of lithium ions. The

diffusion coefficient must be estimated in order to determine the surface concentration of

lithium ions, and then calculate the voltage. Information regarding the mobility of

lithium ions inside the WO3 film may then be gained from the simulation process. Wang

used this model to simulate the voltage response of sol-gel deposited electrochromic WO3

58

films, coloured under various current densities. Although the model described the data

very well, it was only tested at room temperature.

All of the modelling work on EC systems has involved making at least one assumption

about the identity of the rate limiting mechanisms and attempting to validate the model

by demonstrating its ability to simulate experimental data. Analysis of the literature

presented above reveals that there is still much ambiguity regarding the rate limiting

process. If a model is to be of some use, it must accurately describe the cycling

characteristics over some useful range of conditions such as injected charge density,

voltage and temperature. To date no models have been used to describe the

electrochemical and optical properties of electrochromics over a wide temperature range.

The aim of this PhD research is to validate the modelling work of Wang [27] by

simulating voltage responses at elevated temperatures. This should allow the estimation

of diffusion coefficients at various temperatures, and therefore provide an insight into the

underlying kinetic processes that are occurring at high temperature.

59

2.6 Summary

It is evident from the literature reviewed above, that the electrical characteristics of

electrochromic films and devices are very complex and depend on many factors.

Faughnan et al [13] found in 1977 that evaporated WO3 films prepared by similar

techniques had different colouring and bleaching response times. Other papers have

reported that films prepared by different techniques (eg. evaporation, sputtering,

electrolytic) have large differences in response times and electrochemical characteristics

[51,52]. Key issues affecting the EC characteristics of various systems have been

identified as the presence of water and film porosity [37-52]. It is also evident that the

rate limiting mechanisms will depend on factors including film composition, applied

voltage or current, film thickness, etc[26].

At elevated temperatures the switching characteristics of electrochromics change

considerably. Although there is little published material regarding the high temperature

behaviour of electrochromics, it is evident that switching occurs more readily at high

temperature. An in-depth knowledge of the dependence of switching characteristics on

temperature is required in order to choose a suitable control strategy which will ensure

maximum device lifetime.

REFERENCES

[1] M.E. Straumanis, J. Am. Chem. Soc., 71, 679-683 (1949).

[2] M.E. Straumanis and A. Dravnieks, J. Am. Chem. Soc., 71, 683-687 (1949).

[3] A.S. Ribnick, B. Post and E. Banks, Advances in Chemistry Series, No. 139, American Chemical Society (1963).

60

[4] H.R. Shanks, P.H. Sidles and G.C. Danielson, in Electrical Properties of the Tungsten Bronzes, Advances in Chemistry Series, No. 139, American Chemical Society, (1963)

[5] M.J. Sienko, in Non-Stoichiometric Compounds, Advances in Chemistry, Series No. 139, American Chemical Society, (1963).

[6] B.O. Loopstra and P. Boldrini, Acta Cryst., 21, 158-162 (1966).

[7] S.K. Deb, Appl. Optics, Suppl. 3 on Electrophotography, 192-195 (1969).

[8] J.M. Berak and M.J. Sienko, J. Solid State Chem., 2, 109-133 (1970).

[9] R.S. McEwen, J. Phys. Chem., 75, 1782-1789 (1971).

[10] S.K. Deb, Phil. Mag., 27, 801-822 (1973).

[11] D.W. Lynch, R. Rosei, J.H. Weaver and C.G. Olson, J. Solid State Chem., 8, 242-252 (1973).

[12] B.W. Faughnan, R.S. Crandall and P.M. Heyman, R. C. A. Rev., 36, 177-197 (1975).

[13] B.W. Faughnan, R.S. Crandall and M.A. Lampert, Appl. Phys. Lett., 27, 275-277 (1975).

[14] R.S. Crandall and B.W. Faughnan, Appl. Phys. Lett., 28, 95-97 (1976).

[15] S.K. Deb, Proc. Roy. Soc. A, 304, 211-231 (1968).

[16] B.W. Faughnan and R.S. Crandall, Appl. Phys. Lett., 31, 834-836 (1977).

[17] S. Hackwood, G. Beni, W.C. Dautremont-Smith, L.M. Schiavone and J.L. Shay, Appl. Phys. Lett., 37, 965-967 (1980).

[18] H.-T. Zhang, P. Subramanian, O. Fussa-Rydel, J.C. Bebel and J.T. Hupp, Solar Energy Materials and Solar Cells, 25, 315-325 (1992).

[19] P.F. Gerhardinger and R.J. McCurdy, MRS, 426, 503 (1996).

[20] M. Green and K.S. Kang, Thin Solid Films, 40, L19-L21 (1977).

[21] W.C. Dautremont-Smith, M. Green and K.S. Kang, Electrochim. Acta, 22, 751-759 (1977).

[22] C.M. Lampert, Solar Energy Materials, 11, 1-27 (1984).

[23] S.K. Deb, Solar Energy Materials and Solar Cells, 25, 327-338 (1992).

[24] J.M. Bell, I.L. Skryabin and G. Vogelmann, Proceedings of the 3rd Symposium on Electrochromic Materials, International Electrochemical Society, 96-24, 396 (1996).

61

[25] J.-G. Zhang, C. Edwin Tracy, D.K. Benson and S.K. Deb, J. Mater. Res., 8, 2649-2656 (1993).

[26] J.-G. Zhang, D.K. Benson, C. Edwin Tracy and S.K. Deb, J. Mater. Res., 8, 2657-2667 (1993).


[28]. B. Vuillemin and O. Bohnke, Solid State Ionics, 68, 257-267 (1994).

[29] C.G.Granqvist, Solar Energy Materials and Solar Cells, 32, 369 (1994).

[30] B.W. Faughnan and R.S. Crandall, Electrochromic Displays Based on WO3in Topics in Applied Physics, Display Devices (Edited by J. I. Pankove), Vol. 40, Springer-Verlag, New York (1980).

[31] T.C. Arnoldussen, J. Electrochem. Soc., 128, 119 (1981).

[32] S. Hashimoto and H. Matsuoka, J. Electrochem. Soc., 138(8), 2403-2408 (1991).

[33] J.S.E.M. Svensson and C.G. Granqvist, Solar Energy Materials, 11, 29 (1984).

[34] M.E. Badding, S.C. Schulz, L.A. Michalski and R. Budziak, Electrochemical Society Proceedings, 96-24, 369-384 (1996).

[35] C.M. Lampert, A. Agrawal, C. Baertlien and J. Nagai, Solar energy Materials and Solar Cells, 56, 449-463 (1999).

[36] G. Tulloch, I. Skryabin, G. Evans and J. Bell, Proceedings of the SPIE, Vol 3136, 426-432 (1997).

[37] J.P. Randin, J. Electron. Mater., 7, 47-63 (1978).

[38] D.L. Kepert, in Progress in Inorganic Chemistry (Edited by F.A. Cotton), Vol. 4, Interscience, New York, 199-274(1962).

[39] Q. Zhong, S.A. Wessel, B. Heinrich and K. Colbow, Solar Energy Materials, 20, 289-296 (1990).

[40] S. Hashimoto, H. Matsuoka, H. Kagechika, M. Susa and K.S. Goto, ibid., 137, 1300 (1990).

[41] J.A. Duffy, M.D. Ingram and P.M.S. Monk, Solid State Ionics, 58, 109-114(1992).

[42] J.-G. Zhang, D.K. Benson, C.Edwin Tracy, J. Webb and S. Deb, Proceedings of the SPIE, Vol 2017, 104-112 (1993).

[43] M. Burdis and J.R.Siddle, Thin Solid Films, 237, 320-325 (1993).

[44] P.W. Atkins, Physical Chemistry, Chapter 18, Oxford University Press, London (1978).

62

[45] F.M. Michalak and J.R. Owen, Solid State Ionics, 86-88, 965-970 (1996).

[46] R.S. Crandall, P.J. Wojtowicz and B.W. Faughnan, Solid State Comm., 18, 1409-1411 (1976).

[47] B. Reichman, A.J. Bard and D. Laser, J. Electrochem.Soc., 127(3), 647-654 (1980).

[48] J. Nagai and T. Kamimori, Jap. J. Appl. Phys., 22, 681-687 (1983).

[49] O. Bohnke, M. Rezrazi, B. Vuillemin, C. Bohnke and P.A. Gillet, Solar Energy Materials and Solar Cells, 25, 361-374 (1992).

[50 I.D. Raistrick and R.A. Huggins, Solid State Ionics, 7, 213-218 (1982).

[51] H.R. Zeller and H.U. Beyeler, Appl. Phys., 13, 231-237 (1977).

[52] B. Reichman and A.J. Bard, J. Electrochem. Soc., 126, 583- (1979).

CHAPTER 3

EFFECT OF TEMPERATURE ON ELECTROCHROMIC DEVICE

SWITCHING VOLTAGES.

J.P. Matthews, J.M. Bell and I.L.Skryabin

Published: Electrochimica Acta, 44, 3245-3250 (1999).

64

Contributions of Authors

This paper presents the results of experimental work carried out by J.P. Matthews, under

the supervision of J.M. Bell and I.L. Skryabin. The paper was written by J.P. Matthews

and revised by J.M. Bell before final submission of the manuscript.

65

EFFECT OF TEMPERATURE ON ELECTROCHROMIC DEVICE

SWITCHING VOLTAGES.

J.P. Matthews1, *, J.M. Bell1 and I.L.Skryabin2

1 School of Mechanical, Manufacturing and Medical Engineering,

Queensland University of Technology, Australia

2 Sustainable Technologies Australia,

11 Aurora Ave, Queanbeyan, NSW, Australia

* Author to whom correspondence should be addressed.

Abstract

Excessive switching voltages in electrochromic devices cause rapid degradation in

performance. Optimisation of switching voltages is therefore critical in order to realise

the maximum possible device lifetime, and to produce a commercially reliable product.

The magnitude of the voltages required to colour and bleach a device are temperature

dependent, with lower voltages being required at higher temperatures. In real

applications, electrochromic devices may attain temperatures as high as 70°C. Use of

the room temperature switching regime at elevated temperatures may impose an

overvoltage on the device, which can significantly reduce both lifetime and optical

performance.

A voltage limited constant current charge injection technique was used to cycle sol-gel

66

deposited WO3 films at elevated temperatures. The voltages required for colouring and

bleaching at these temperatures were determined and correlated with the level of

coloration achieved. The results show that the variation in switching voltages is

significant, and therefore inclusion of temperature in the switching algorithms is

necessary to achieve maximum lifetime for electrochromic devices.

Keywords: Electrochromic thin films; switchable glazing; temperature effects; lifetime;

switching voltages

3.1 Introduction

The lifetime of electrochromic (EC) devices is an issue of critical importance, and is a

parameter that must be maximised in order for smart windows to become

commercially viable [1]. The useful lifetime of EC devices depends on many factors

including device composition, switching algorithm, current density during switching, the

level of charge injected and extracted and operating temperature. The use of excessive

switching voltages provides an overpotential which enables potentially damaging side

reactions to occur in EC devices [2]. The choice of safe switching voltages is

therefore paramount for successful device operation over a timescale of several decades.

Electrochromic devices based on the LixWO3 system absorb a significant proportion of

incoming radiation in the coloured state, so it is foreseeable that these devices will attain

temperatures in excess of 60°C. The effect of temperature on EC device switching

voltages is complex, and published literature on this topic is scarce.

67

This paper reports initial findings of a study aimed at determining the effects of

temperature on electrochromic device switching voltages, with the long term goal of

optimising the switching regime to achieve maximum device lifetime. The films and

devices reported in this study have been cycled under a voltage limited constant-current

charge injection method [3]. The measured or applied voltage during coloration, Va(t),

can be simply described by the equation [4]

Va(t) = emf(t) + Vc(t) (3.1)

where emf is the electromotive force of the cell, and Vc(t) is the overvoltage associated

with the ion insertion. The effect of temperature on the applied voltage Va, therefore

depends on the individual responses of both the electrochromic film emf and colouring

voltage Vc to temperature. Equation 3.1 somewhat simplifies the behaviour of real cells

because it does not consider other possible parameters such as the iR drop of the

electrolyte and the conducting glass substrate. The effects of temperature on these

parameters in experimental situations is however small enough for the model to be of

use, in terms of describing the change in switching potentials with temperature.

Experiments were carried out in order to evaluate these responses. The maximum

colouring voltages required for a specific level of coloration were measured over an

extended temperature range, and the emf of the film was also determined for the fully

coloured state. This has enabled the observation of trends in the emf and colouring

voltage Vc with temperature, and the implications of the results are discussed in the

context of a switching method which enables maximum lifetime to be achieved.

68

3.2 Experimental

3.2.1 Electrode preparation

The WO3 electrochromic films were deposited onto 10cm x 10cm substrates of LOF

TEC8/3 glass using the sol-gel dip coating method [5]. The alkoxide precursor solutions

used in the sol-gel dipping have been described previously [6].

3.2.2 Electrochemical Testing

Electrochemical measurements were made using a three electrode cell. The counter

electrode was a platinum foil (Area=64cm2) and the reference electrode was a Ag/AgCl

cell filled with an ethanolic solution of KCL, saturated with AgCl. The electrolyte used

was 1M LiClO4 in propylene carbonate. The electrolyte solution was dried over

molecular sieves and maintained under a nitrogen atmosphere. Experiments were

carried out in a glass tank (filled with electrolyte solution), which was partly submersed

in a larger glass tank of heating oil. An electrical heater/stirrer unit was used to control

the temperature.

The WO3 films were cycled using a voltage-limited constant current technique,

described previously [3,7]. The experiments were carried out using the convention that

coloration was due to positive currents, hence the corresponding electrical potentials

were also positive. The currents and potentials measured during bleaching are therefore

negative. This convention has been used throughout this paper. Under this convention,

the emf values reported for coloured films are therefore positive, and the emf increases

69

with increasing coloration, and hence with increasing x.

The results reported in sections 1 and 2 are for a film cycled to 15mC/cm2 (Current

density = 0.05mC/cm2, film area = 100cm2). The temperature was ramped from 20° to

50°C at approximately 1°C per cycle. Measurements were performed as the temperature

increased, but were not made as the system cooled.

The results reported in section 3 are for a film cycled to 5mC/cm2 (Current

density = 0.05mC/cm2, film area = 100cm2). The temperature was ramped from 20° to

50°C at approximately 1°C per cycle. Measurements were performed during both

increasing and decreasing temperature.

3.3 Results and discussion

3.3.1 Effect of temperature on applied voltage, Va(t)

Figure 3.1 shows a graph of applied voltage, Va(t), versus time for a sol-gel deposited

WO3 film, cycled at four different temperatures, with a charge injection level of Qin/out =

15mC/cm2. The coloration and bleaching were carried out using a voltage limited

constant-current technique as described previously [3, 7].

70

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

0 100 200 300 400 500

20.6oC

28.4oC32.3oC

50.0oC

E/V

t/s


bleaching of WO3 thin film electrode and plotted for four temperatures, for an injected

charge density of 15mC/cm2.

During coloration the voltage limit was not reached, hence the maximum applied

colouring voltage is readily observed. During the bleaching process the voltage limit (-

0.1V) was reached every cycle, and therefore the true minimum voltage reached during

the bleaching process was not observed. The bleaching data must therefore be discussed

somewhat qualitatively, based on the relative curves of Va(t) for each temperature.

It is evident from Figure 3.1 that the magnitude of the applied voltage decreases as

temperature is increased, and this behaviour is observed throughout both colouring and

bleaching stages. This behaviour is expected, considering the effects of temperature on

diffusion and charge transfer processes. The diffusion coefficients will increase with

71

increasing temperature so that charge is transported more readily within the WO3

electrode, aiding the colouring and bleaching processes. The extra thermal energy at

elevated temperatures also helps overcome the charge transfer activation energy, and

thus reduces the voltages required for coloration and bleaching [8]. Given these results,

it is possible that higher current densities may be used to switch these devices at higher

temperatures without adverse effects, thereby enabling faster response times.

Figure 3.2 shows the variation in applied voltage, Va(t), with temperature, and shows

that variation in switching voltage is greater for lower temperatures. The consequence

of this is that small deviations from room temperature will significantly alter the

electrical characteristics of electrochromic films and devices.

0.95

1

1.05

1.1

1.15

1.2

20 25 30 35 40 45 50 55

E/V

T/oC

Figure 3.2 Maximum voltages required to colour WO3 thin film electrode to 15mC/cm2

at elevated temperatures.

72

The implications of the effect of elevated temperatures on film and device cycling are

dependent on the switching methods used, and constant voltage cycling is likely to be

affected most by these results. Constant voltage coloration of films at various

temperatures while using the same voltage limits, would result in higher injected charge

densities at higher temperatures (and hence progressively darker films), with the

possibility of exceeding the reversible limit of x=0.4 in LixWO3 [4]. The consistent

application of potentials exceeding those necessary to colour the film will also promote

gas generation and film decomposition reactions and undoubtedly decrease the useful

lifetime of the film.

Constant current cycling on the other hand is less sensitive to the effects of temperature,

as the charge injected is measured (by integration of current with respect to time),

during the cycle and hence the same amount of charge is injected and extracted each

cycle. The same level of coloration should therefore be attained at all temperatures (see

section 3.3.2).

The applied or measured voltage, Va(t), can be written as the sum of the electromotive

force (emf) and the voltage associated with the ion insertion (see equation 3.1 above).

Temperature is a variable in Nernstian expressions for the emf of LixWO3 films so this

is one source of the voltage variation. The effects of temperature on both the colouring

voltage and the emf are considered below.

73

3.3.2 Effect of temperature on colouring efficiency

Figure 3.3 shows the relative transmittance (as measured by a photocell) versus injected

charge for cycles at four different temperatures. The response is essentially the same at

all four temperatures, indicating that the same amount of charge has been injected and

extracted each cycle, as we anticipate when using the constant current charge injection

technique. (Note that the curves have each been offset by 0.5V for visual clarity).

0

1

2

3

4

5

6

0 2 4 6 8 10 12 14 16

20.6oC

28.4oC32.3oC

50.0oC

E/V

Q/mCcm-2

Figure 3.3 Photocell voltage versus injected charge for WO3 thin film electrode at four

temperatures.

74

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0 2 4 6 8 10 12 14 16

20.6oC

28.4oC

32.3oC

50.0oC

OD

∆

Q/mCcm-2


at four temperatures.

The change in optical density, ∆OD, has also been plotted against injected charge, in

Figure 3.4. The slope of these lines defines a useful characterisation parameter called

the coloration efficiency [9] for each of the curves shown. The coloration efficiency

appears to be approximately constant at the lower temperatures, (CE=43.9cm2/C,

R2=0.93) but decreases slightly to CE=38.1C/cm2, (R2=0.98) at 50°C (These values

were determined from simple linear regresion of the data for both the coloration and

bleaching processes). It is unclear at this time, whether this difference is due to a

change in optical response of the tungsten oxide film, or is due to errors in the optical

measurements introduced at high temperatures.

A particular problem associated with high temperature experiments is the changing

75

refractive indices of components between the laser and the optical detector, as this is a

single beam measurement. As the temperature rises, the refractive indices of the heating

oil, glass tanks, and LiClO4/PC electrolyte all increase. If the light transmission path is

even slightly off normal incidence, the laser beam will deviate as the temperature rises,

which would introduce a systematic change in photocell response with temperature.

This problem can be corrected by ensuring that the laser beam is normally incident on

all phase boundaries (heating tank walls, WO3 and counter electrode film surfaces, etc).

However this is experimentally difficult and was not ensured at the time of this work.

A further series of experiments is being carried out to determine firstly the normal

photocell response with respect to temperature, and then to establish whether indeed the

coloration efficiency does change with temperature (See section 4.3.1). If the coloration

efficiency is found to decrease with temperature, it may be necessary to modify the

switching to vary the charge injection level to compensate for changing coloration

efficiency at different temperatures.

3.3.3 Effect of temperature on coloured state electromotive force, emfc, and maximum

colouring voltage, Vc max.

Figure 3.5 shows the temperature dependence of the maximum applied voltage

(Va max) and coloured film emf (emfc). The emf was measured by disconnecting the film

and measuring the potential after sixty seconds had elapsed, thereby allowing time for

the system to equilibrate. The maximum colouring voltage was also calculated using

equation (3.1). The experiment was carried out by cycling a film to 5mC/cm2 while

76

slowly ramping the temperature from 22 to 50°C, and then allowing the system to cool.

The lower part of the curves correspond to increasing temperature, and the upper parts

were recorded as the system cooled.

0

0.1

0.2

0.3

0.4

0.5

0.6

20 25 30 35 40 45 50 55

emfc

Va max

E/V

T/oC

Increasing T

Decreasing T

Figure 3.5 Maximum voltages required for coloration of WO3 thin film electrode to

5mC/cm2, and corresponding emf values measured between 20°C and 50°C.

The hysteresis associated with the data in Figure 3.5 is believed to be due to charge

trapping (incomplete charge extraction) during cycling. If the amount of charge

extracted is less than the amount injected per cycle, lithium will accumulate in the film.

This will have the effect of gradually increasing the value of x corresponding to the

coloured state, and hence increasing the films emf. The emf of a LixWO3 film can be

described by the equation [10]

−

++=x

xFRTbxaxemf

1ln)( ν (3.2)

77

where a, b and ν are parameters which can be determined by fitting to experimental

data, and x (in LixWO3) is calculated from the amount of charge injected, the molar

volume and film thickness. If the intercalated charge of the coloured film was the same

each cycle (ie. constant x, no charge trapping), we would expect a graph of emf versus

temperature to yield a straight line with negative slope, proportional to (νR/F), with no

hysteresis. If we assume some fraction of the injected charge to remain in the films after

bleaching, we expect the emf to steadily increase. The difference between the emf

values for increasing and decreasing temperature is therefore attributed to the amount of

charge trapped in the film during the intervening cycles. The hypothesis of charge

trapping is supported by the optical data, which shows a slightly decreasing

transmittance for both coloured and bleached states as the experiment progresses. The

level of charge trapped between the first and last cycle was determined (from the optical

and electrical results) to be 1.25mC/cm2.

The relationship between applied voltage and emf is described by equation (3.1), where

the applied or measured voltage is the sum of the film emf and colouring voltage at any

given time. The applied voltage therefore incorporates the emf, so it is not surprising to

observe the same hysteresis in both graphs of Figure 3.5. When the maximum colouring

voltage, Vc max, is calculated from equation 3.1 and plotted against temperature (Figure

3.6) the hysteresis effect is removed, and the maximum colouring voltages are

essentially the same for both increasing and decreasing temperature, within the error of

the experimental noise. This fact supports the hypothesis that the hysteresis seen in

Figure 3.5 is purely associated with the film emf.

78

0.1

0.12

0.14

0.16

0.18

0.2

0.22

0.24

20 25 30 35 40 45 50 55

Vc max

E/V

T/oC

Figure 3.6 WO3 film maximum coloration voltage versus temperature, for an injected


If charge is indeed being trapped in the films, the emf will steadily increase each cycle

and so the maximum applied voltage Va will soon approach the safe colouring voltage

limit. A reduction in the amount of trapped charge is therefore critical for a long

lifetime, and must be minimised by adjusting cycling conditions (eg. injected charge

density or current density), or other film characteristics.

The data for the maximum colouring voltage Vc max as a function of temperature can be

linearised by plotting log Vc max versus 1/T. This graph is shown in Figure 3.7, together

with the equation for the line of best fit. It is foreseeable that this linear relationship

may be used to determine kinetic information regarding the ion injection process (eg.

diffusion coefficients, charge transfer resistance). This information is also useful in

79

determining the required voltage to achieve a particular state of coloration, given the

emf of the film. If the effect of temperature on the emf of the film can be successfully

modelled, this can be incorporated with the temperature dependence of the coloration

voltage in order to predict safe switching voltages over a wide range of operating

temperatures.

-0.95

-0.9

-0.85

-0.8

-0.75

-0.7

-0.65

-0.6

0.003 0.0031 0.0032 0.0033 0.0034 0.0035

log Vc

E/V

T-1/oC-1

y = 771.59 x - 3.3194R2 = 0.8916

Figure 3.7 Log of coloration voltage versus reciprocal temperature for WO3 film with

an injected charge density of 5mC/cm2.

80

3.4 Conclusion

The magnitudes of the voltages required to colour and bleach electrochromic thin films

of tungsten oxide decrease with increasing temperature. The response of the applied

voltage to temperature was dependent on the individual responses of electromotive force

(emf) and colouring voltage. The emf of the coloured state of these films decreased

with increasing temperature, however the observation of this effect was hindered by

charge trapping during repeated cycling. This charge trapping resulted in a hysteresis

effect for the emf data recorded during increasing and then decreasing temperature. The

maximum colouring voltage was found to decrease with increasing temperature, and this

relationship was described by a linear plot of logVc vs 1/T.

The voltage limited constant current charge injection method used, resulted in the same

level of charge injection and extraction into the WO3 film at high temperatures. This

allowed the WO3 films to be coloured to approximately the same optical density each

cycle over a wide temperature range, even though the voltages required for coloration

decreased with increasing temperature.

Knowledge of the emfc and Vc response to temperature enables one to predict the

maximum applied voltages required to colour electrochromic films under specific sets of

conditions. If these voltages are exceeded, the extra potential provided will promote

side reactions and reduce device lifetime. An understanding of the temperature

dependence of electrochromic device switching voltages can therefore be used to

determine safe switching conditions.

81

Planned future research into this behaviour focuses largely on the modelling of the

electrical characteristics of tungsten oxide EC films and devices as a function of

temperature, and the eventual utilisation of the model as a tool for optimising the

switching algorithm to achieve the maximum possible device lifetime.

Acknowledgements

This work is supported by an Australian Postgraduate Award (Industry) scholarship

from the Australian Research Council, and Sustainable Technologies Australia (STA).

82

REFERENCES [1] C. M. Lampert, IEEE Circuits and Devices, 8, 19-26 (1992).

[2] B.W. Faughnan and R.S. Crandall, Electrochromic Displays Based on WO3in

Topics in Applied Physics, Display Devices (Edited by J. I. Pankove), Vol. 40, Springer-

Verlag, New York (1980).

[3] J. M. Bell and I. L. Skryabin, Solar Energy Materials and Solar Cells, 56 (1999)

437.

[4] J.-G. Zhang, D.K. Benson, C. Edwin Tracy and S.K. Deb, J. Mater. Res., 8, 2657-

2667 (1993).

[5] J. M. Bell, G. B. Smith, I. L. Skryabin, B. G. Monsma, N. C. Ruck and T. Dinh,

Sol-gel Deposited Electrochromic Devices, in Proceedings of Windows Innovations

Conference WIC 95, 383-391, Minister of Supply and Services, Canada.

[6] A. Koplik, Australian Patent Application, PP0274 (1997).

[7] I. L. Skryabin and J. M. Bell, Control of Electrochromic Devices, International

Patent Application PCT/AU97/00697.

[8] A. J. Bard and L. R. Faulkner, in Electrochemical Methods: Fundamentals and

Applications, Wiley, New York (1980).

[9] C.M. Lampert, V-V. Truong, J. Nagai and M.G. Hutchins, in Characterization

Parameters and Test Methods for Electrochromic Devices in Glazing Applications,

International Energy Agency Task X-C Final Report, University of California (1994).


CHAPTER 4

HIGH TEMPERATURE BEHAVIOUR OF ELECTROCHROMICS

J.P. Matthews, J.M. Bell and I.L.Skryabin

Published: Renewables: The Energy for the 21st Century


Sayigh Ed.), 230-235 (2000).

84




and revised by J.M. Bell before final submission of the manuscript. This paper was

presented by John Bell as an invited paper at the World Renewable Energy Congress VI,

Brighton, UK, 2000.

85

HIGH TEMPERATURE BEHAVIOUR OF ELECTROCHROMICS

J.P. Matthews1, J.M. Bell1 * and I.L.Skryabin2

1 Research Concentration in Materials Technology,

School of Mechanical, Manufacturing and Medical Engineering,


2 Sustainable Technologies, Australia,


* Author to whom correspondence should be addressed.

Abstract

Sol-gel deposited electrochromic films have been cycled at elevated temperatures under

various environmental conditions. Significant irreversibility was observed during

cycling of films when moisture was present in the electrolyte, especially at high

temperatures. A proportion of the injected charge did not cause colouration under these

conditions, which caused an apparent decrease in colouration efficiency at high

temperatures. An experiment was carried out which enabled the observation of the slow

bleaching of these films in an electrolyte solution, even though the working electrode

was electrically isolated from the external circuit. This self-bleaching was associated

with irreversible charge injection under conditions where moisture was present. Films

cycled under very dry conditions exhibited very reversible behaviour, and the

colouration efficiency was found to be independent of temperature.

86

Keywords: Electrochromic thin films; switchable glazing; colouration efficiency; self-

bleaching; temperature effects

4.1 Introduction

Previous experiments investigating the effects of temperature on electrochromic device

switching voltages [1] have shown that the magnitude of the voltages that are required to

colour and bleach electrochromic (EC) films decreases with increasing temperature.

When the films were cycled at temperatures exceeding approximately 30ºC some

irreversibility was observed in the charge injection/extraction process. Some of the

charge injected during colouration was unable to be extracted during the bleaching

process however the optical density of the bleached film was consistent with cycles that

were totally reversible. The amount of charge unable to be extracted each cycle

increased with temperature and this trapped charge apparently did not contribute to the

colouration of the film. Although the amount of charge trapped per cycle was relatively

small, the cumulative effect over many cycles is very significant. The amount of charge

available for transfer between the working and counter electrodes of an EC device is

limited by the amount of charge incorporated during device fabrication. After device

fabrication no more charge can be introduced so a reduction in the reversibility of the

EC process limits the maximum possible change in optical density and ultimately the

device lifetime.

In order to better understand the effect of this charge trapping phenomenon, a series of

self-bleaching experiments were carried out. These experiments involved colouring of

87

EC films to a specific charge density, and disconnecting the counter electrode thereby

electrically isolating the working electrode from the external circuit. These coloured

films were observed to undergo a slow self-bleaching process and this change was

monitored by continual measurement of the electromotive force (emf) and optical

density of the working electrode.

After the self-bleaching experiment the WO3 substrate was dissolved off the glass/FTO

substrate with an alkaline solution. Chemical analysis of this solution revealed that

there was a large amount of lithium still present in the film, confirming that there

actually was lithium still inside the film, and that some of this lithium did not contribute

to colouration.

The reversibility problems encountered at high temperatures hindered simulation of the

experimental data because the amount of charge extracted varied for each cycle. In

order to try and establish reversible cycling at elevated temperatures, experiments were

repeated in a nitrogen filled dry-box, in which very low levels of humidity were

stringently maintained. It was found that the EC reaction was reversible over the entire

experimental temperature range and that the colouration efficiency was linear and

independent of temperature.

88

4.2 Experimental

4.2.1 Electrode preparation

Mixed tungsten-titanium oxide electrochromic films were deposited using sol-gel

processing from organic precursors onto 10cm × 10cm substrates of LOF TEC8/3 glass

using the sol-gel dip coating method [2]. The alkoxide precursor solutions used in the

sol-gel dipping have been described previously [3].

4.2.2 Electrochemical testing



cell filled with an ethanolic solution of KCl, saturated with AgCl. The electrolyte used

was 1M LiClO4 in propylene carbonate, which was stored over molecular sieves after

preparation. Experiments were carried out in a glass tank (filled with electrolyte

solution) partly submersed in a larger heating tank filled with mineral oil. An electrical

heater/stirrer unit was used to control the temperature of the oil bath, and hence the

electrolyte solution. The electrolyte solution was also stirred during the experiments to

minimise the temperature difference between the heating oil and the electrolyte solution.


described previously [4,5]. Films were cycled to 15mC/cm2 (Current density =

0.1mA/cm2, film area = 100cm2). Optical measurements were made by directing a

1mW, 670nm laser beam through the electrodes, and onto a silicon photodiode. The

89

photocell voltages reported in the results are the output voltages of the silicon

photodiode. As the film is coloured, the intensity of the laser beam reaching the

photodiode is reduced, hence photocell voltages decrease with increasing optical density

of the film.

Experiments were carried out either in the ambient environment of the laboratory, or

inside a dry-box. The experiments carried out in the ambient laboratory environment

were carried out while slowly bubbling dried nitrogen through the electrolyte solution

before and during testing to maintain a slight positive pressure and minimise the mount

of moisture in the electrolyte. During the dry-box experiments the complete

experimental apparatus including electrolyte tank, oil bath, heater unit and optical bench

remained in a nitrogen-filled dry glovebox. The atmosphere inside the glovebox was

kept dry by exposing it to P2O5 desiccant, and recirculating the nitrogen through a

column of dried molecular sieves. The humidity level inside the glovebox was

monitored with a HMP235 humidity and temperature transmitter, manufactured by

Vaisala. The humidity during the dry-box experiment was maintained at 1.05ppm

absolute humidity (Relative humidity = 5.3% and temperature = 22.4ºC immediately

prior to experiment).

The self-bleaching experiment was performed under the ambient laboratory conditions

described above. Films were coloured to 20mC/cm2 (Current density = 0.1mA /cm2,

film area = 100cm2), and the counter electrode was disconnected from the electrical

circuit immediately after completion of the coloration. The counter electrode was

90

disconnected for 30 minutes and the electrolyte solution was maintained at the

appropriate temperature during this time. At the end of the 30 minute self-bleaching

period, the counter electrode was reconnected, the film was bleached and the

temperature was increased for the next data set.

4.2.3 Chemical analysis

The amount of lithium and tungsten in a bleached film was determined, after the self-

bleaching experiment, by inductively coupled plasma-atomic emission spectroscopy

(ICP-AES). The film was washed off the glass/FTO substrate with aqueous sodium

hydroxide, and the solution was diluted to 50mL. Tungsten and lithium stock standard

solutions were used prepared a series dilution of calibration standards. Calibration

graphs were prepared and used to determine the concentrations of lithium and tungsten

in the sample solution. The ICP-AES measurements were performed on a Spectroflame

spectrometer, manufactured by Spectro Analytical Instruments, West Germany.

91

4.3 Results and Discussion

4.3.1 Effect of temperature on coloration efficiency

Figure 4.1 shows the changing optical properties of two sol-gel deposited WO3 films

during colouration and bleaching at various temperatures. The experiment shown in

Figure 4.1(a) was carried out in ambient laboratory conditions, and nitrogen was

bubbled through the electrolyte solution in an attempt to keep water (from the air) out of

the system. The results shown in Figure 4.1(b) are for an experiment carried out in an

extremely dry environment, inside a nitrogen filled glove box. Plots of change in optical

density versus injected charge are expected to be linear for EC films and devices, and

the slope is defined as the colouration efficiency (CE) [6]. It is evident that this

behaviour is observed only in the very dry case (Figure 4.1(b)) where the plots at

different temperatures are virtually the same with an average CE of 38.6cm2/C

(R2=0.99).

Figure 4.1 Change in optical density versus injected charge for WO3 films cycled to

15mC/cm2 at elevated temperatures. The results shown in (a) are for an experiment

carried out in the ambient environment, while the results shown for (b) are for an

92

experiment carried out in a dry-box.

The colouration efficiency for the first experiment (Figure 4.1(a)) is approximately

linear for the room temperature cycle (20.6ºC), and the colouration efficiency is

determined from a linear regression to be 43.9cm2/C (R2=0.93). As temperature

increases the optical density, for each level of injected charge density, decreases

indicating that some of the injected charge is not contributing to colouration. The

reversibility of the charge injection process in this experiment was also affected by

temperature, with significant irreversibility noticeable for temperatures above 30ºC.

The film was cycled using a constant current charge injection/extraction technique (as

described in the experimental section, above), and the bleaching process was terminated

when the voltage reached some safe limit, predetermined to prevent damage to the film.

Any charge remaining in the film after bleaching was therefore unable to be extracted

without applying larger voltages which would have damaged the film. Figure 4.2 shows

the relative amount of charge unable to be extracted from the films each cycle, during

cycling at elevated temperatures, for both of the experiments discussed above.

93


percentage of the injected charge density trapped per cycle.

It is evident that the ion injection process for the film cycled in ambient conditions was

much less reversible than for the film cycled in the dry-box. The relative amount of

charge not extracted each cycle in the ambient case increases with temperature, and at

50ºC approximately 3.5% of the charge injected was unable to be extracted during the

bleaching process. The reversible limit of x = 0.4 (in LixWO3) [7] was not exceeded

during these cycles so the irreversibility must be accounted to some other reaction

involving the lithium ions.

At room temperature the EC reaction of the film cycled in the dry-box was very

reversible, and the percentage of charge trapped is very close to zero, within the limits

of experimental error. As temperature increases, the amount of charge extracted

actually exceeds the level of charge injected for that cycle, which would suggest some

94

experimental errors. This may be due to the combination of the switching regime used,

and the reduction in switching voltages which occurs at elevated temperatures. Before

the experiment was carried out, the film was pre-loaded with charge. The films do not

cycle reversibly for the first few cycles, so 20 cycles were performed where it is

common for a significant proportion of the injected charge to remain in the films after

bleaching, even though the films appear to be bleached as normal.

As temperature rises, the magnitude of the voltage required to achieve a given charge

density decreases and so applying a set voltage limit for the bleaching cycle, we are

driving the bleaching process further at high temperature. In this experiment the same

voltage limit was applied to the bleaching process at all temperatures, so it is possible

that some of the pre-loaded lithium was removed at higher temperatures. The fact that

the amount of charge extracted increases with temperature for the dry-box experiment

supports this proposition. It is also possible that there is a small experimental error

associated with the measurement of the currents, such as a bias towards the

measurement of the bleaching current. An electrical calibration error of this kind would

be expected to be independent of temperature, and this would also mean that the amount

of charge trapped in the ambient conditions was even greater than that shown in Figure

4.2. The fact that there are some negative results for the charge trapped during the

drybox experiments therefore does not affect the conclusions made about the experiment

carried out in the ambient environment.

The large differences between the results observed from films cycled in ambient and

95

very dry conditions suggests that the problems associated with irreversibility and charge

not causing colouration may be ascribed to water present in the system. Although an

attempt was made to keep the ambient experiment dry by bubbling dry nitrogen

through the electrolyte, the high humidity of the experimental location (Brisbane,

Ausralia) combined with the highly hygroscopic nature of the propylene carbonate

electrolyte means that it is unlikely that there was no water present in the electrolyte.

Inside the drybox, it is relatively easy to ensure that there is very little water present and

so the presence of water is thought to be the major difference in the conditions of the

two experiments described above. In order to further investigate the cycling

irreversibility and the proportion of injected lithium not causing colouration at high

temperature, a self bleaching experiment was carried out using the ambient environment

conditions described in the experimental.

4.3.2 Observation of self-bleaching

Figure 4.3(a) shows the change in photocell voltage of the WO3 film during self-

bleaching at elevated temperatures and Figure 4.3(b) shows emf measurements taken

over the same time period. The WO3 electrode was electrically isolated at 150seconds

(after the end of colouration) and then reconnected after a 30 minute period, just prior to

the bleaching half-cycle. The increase in photocell voltage observed in Figure 4.3(a)

indicates a reduction in the optical density and suggests that the concentration of lithium

in the film is decreasing or that some of the lithium is being converted to an optically

non-active form. The drift in emf observed in Figure 4.3 indicates a changing chemical

96

potential of the film, and the drift towards more positive potentials is consistent with a

decreasing lithium concentration over time and at higher temperature. These results

suggest that the film is bleaching as per the normal EC reaction but with the counter

electrode disconnected there is no path for electron flow from the back of the working

electrode. Any lithium reaction must therefore be with some species already present in

the system which also supports the theory that water in the system is responsible for

some of the injected ions not causing colouration and for the irreversibility observed.

Figure 4.3 Change in (a) photocell voltage and (b) emf of WO3 electrode during self-

bleaching experiment.

The currents measured during colouration and bleaching were integrated with respect to

time to determine the amount of charge injected and extracted respectively. These

values were used to calculate the measured amount of charge which was trapped per

cycle. Calibration curves of photocell voltage and emf versus injected charge density

were constructed for each temperature, by interpolation of measurements made at the

highest and lowest temperatures. These calibration curves were used in conjunction

with photocell and emf values at the start and end of the self-bleaching period (ie. from

97

(a) and (b)), in order to estimate the amount of charge apparently lost during the 30

minute self-bleaching period.

These estimated values of charge loss were then correlated with the measured values

obtained from the difference between the integration of the colouration and bleaching

currents.

Figure 4.4 shows a plot of the estimates of charge lost versus the measured charge loss,

during the self-bleaching period. If the lithium ion concentration in the film was

decreasing (eg. lithium was reacting at the electrode surface to form a new species

outside the film) we would expect the plots of estimated versus measured charge loss to

be linear with a slope of one and intercept of zero.

Figure 4.4 Correlation of estimated and measured quantities of charge lost during self-

bleaching.

98

These plots are indeed linear however the slope is not one and the intercept is not zero.

The photocell measurements are an indication of the number of lithium ions contributing

to colouration, and the charge loss estimated from photocell measurements is therefore

an indicator of the amount of lithium no longer causing colouration at the end of the

self-bleaching period. The measured amount of charge remaining in the film after

bleaching at each temperature is smaller than the estimates, which suggests that some of

the charge that was extracted was not contributing to colouration of the film. For

example, at the end of the self-bleaching period at 50.5ºC, the amount of injected charge

no longer causing colouration is estimated from the photocell voltage (immediately prior

to bleaching) to be 4.7mC/cm2. The amount of charge remaining in the film after the

subsequent bleaching cycle was 4.3mC/cm2. Approximately 4.7mC/cm2 of lithium ions

therefore were not causing colouration after the half hour self bleaching period, and

0.4mC/cm2 of this was later electrochemically extracted from the film. The remaining

4.3mC/cm2 of lithium ions either stayed in the film, was lost into the electrolyte solution

to a side reaction or a combination of both.

4.3.3 Determination of trapped lithium in WO3 film by ICP-AES

In order to answer some of the questions regarding the location of lithium ions which

could not be extracted from the WO3 film by the bleaching process, a portion of a film

used in another self-bleaching experiment was subjected to further chemical analysis.

The WO3 film was washed off the glass/FTO substrate with a sodium hydroxide

solution, and then diluted to 50mL. The film area used was 55.6cm2, and inductively

99

coupled plasma-atomic emission spectroscopy (ICP-AES) was used to determine the

lithium and tungsten concentrations of this solution. The ICP-AES analysis revealed

that there was approximately 100µg of lithium and 8mg of tungsten in the 50mL

solution, which corresponds to x=0.33 in LixWO3 or an injected charge density of

approximately 15mC/cm2 (for a 200nm thick film, with molar volume of 42cm2/mol).

The total measured charge lost during this experiment was approximately 130mC/cm2, a

value clearly very much larger than the amount of ions recovered from the film.

The fact that a large proportion of the measured injected charge was not recovered

suggests that a large amount of the injected ions were either lost to side reactions or that

the ion injection process was not 100% efficient.

Considering that the charge measurement was made by integration of the electron

current with respect to time, any side reactions occurring simultaneously along with the

normal ion intercalation would contribute to the measured current and hence the

measured charge. If the measured current resulted solely from ion injection, the

remaining charge which was not recovered was presumably lost to side-reaction(s) to

form a new species, which then dissolved into the electrolyte. Another possibility is that

side reactions such as gas evolution occurred during ion injection and made a significant

contribution to the measured current, however no evidence of gas evolution was

observed during the experiment.

100

4.4 Conclusions

Sol-gel deposited EC films were cycled under various conditions and a range of

temperatures. Significant irreversibility was observed for films cycled with moisture

present, especially at high temperature. This irreversibility was associated with a

proportion of the injected charge not causing colouration, and consequently there was an

apparent reduction in colouration efficiency at high temperatures. In very dry

conditions, films cycled very reversibly and colouration efficiency was independent of

temperature.

Coloured films were observed to slowly self-bleach in an electrolyte which was not

completely dry, even though the counter electrode was disconnected. After leaving the

film in this electrolyte for 30 minutes, some of the charge remained in the film even

after the bleaching process, and this charge did not give rise to colouration.

Measurements of the photocell voltage and emf of the film during the 30 minute period

were used to estimate the amount of charge trapped during the self-bleaching period.

These estimates were compared to the measured values of charge remaining in the film

after bleaching, and a reasonable correlation was attained.

ICP-AES was used to confirm that there was actually lithium trapped in the film after

self-bleaching, but only a small proportion of the expected amount of lithium was found.

This implied that some charge was trapped in the film while a much larger proportion

was lost to a side reaction, probably reaction of lithium ions with water present in the

electrolyte.

101

Acknowledgements



The work described in this paper has been supported by the Australian Cooperative

Research Centre for Renewable Energy (ACRE). ACREs activities are funded by the

Commonwealths Cooperative Research Centres Program. We would also like to thank

Pat Stevens for his advice and assistance with the ICP-AES measurements.

REFERENCES [1] J.P. Matthews, J.M. Bell and I.L. Skryabin, Electrochimica Acta 44 (1999) 3245.






437.



[6] C.M. Lampert, V-V. Truong, J. Nagai and M.G. Hutchins, in Characterization

Parameters and Test Methods for Electrochromic Devices in Glazing Applications,

International Energy Agency Task X-C Final Report, University of California (1994).


2667 (1993).

CHAPTER 5

TEMPERATURE DEPENDENCE OF KINETIC BEHAVIOUR OF

SOL-GEL DEPOSITED ELECTROCHROMICS

J.M. Bell and J.P. Matthews

Published: Solar Energy Materials and Solar Cells, 68, 249-263 (2001).

104




and revised by J.M. Bell before final submission of the manuscript. This journal article

was an invited paper for a special edition of Solar Energy Materials and Solar Cells, on

sol-gel technologies.

105

TEMPERATURE DEPENDENCE

KINETIC BEHAVIOUR OF SOL-GEL DEPOSITED ELECTROCHROMICS

J.M. Bell* and J.P. Matthews

Research Concentration in Materials Technology

School of Mechanical, Manufacturing and Medical Engineering,

Queensland University of Technology, GPO Box 2434, Brisbane, Qld, 4001, Australia

* Author to whom correspondence should be addressed. E-mail [email protected]

Abstract

The kinetic behaviour of sol-gel deposited electrochromic films is affected by

temperature in a complex manner and may be modelled by considering the reaction

mechanism, and in particular the rate limiting steps. If assumptions are made about the

rate limiting steps in a reaction, a model may be formed which can be used to provide

information about the kinetic parameters such as diffusion coefficient and charge

transfer resistance. Changes in the free energy of a reaction are observed as changes in

the electrical potential associated with the cells and electrodes. We have measured

changes in the switching characteristics of a sol-gel deposited electrochromic film and

modelled these results in order to extract information about the change in lithium

chemical diffusion coefficient (D) with temperature. Values of D estimated using the

model described in this paper are in close agreement with those determined by other

means, however there are some anomalies at high temperatures.

Keywords: Electrochromic film; kinetic behaviour; temperature dependence; diffusion

coefficient;

106

5.1 Introduction

Electrochromic materials undergo a change in their transmittance of heat and visible

light when a small voltage or current is passed through them [1,2]. The cycling of

electrochromic films and devices between coloured and bleached states involves the

injection and extraction of small cations and electrons into the EC material. The guest

cations move through the electrolyte under the influence of an applied electric field, to

the surface of the electrode. The cations must then combine with an electron (provided

by the external circuit) at the electrode surface and lose their solvation sheaths, (thereby

overcoming an associated charge transfer resistance) in order to intercalate into the

crystal structure of the host substrate. After intercalation the ions and electrons diffuse

into the film under the influence of the concentration gradient which results from the

injection of charge at the electrode surface. The bleaching of EC films involves the

extraction of these ions from the electrode (back into the electrolyte) and they must

therefore diffuse from a position within the electrode to the electrode/electrolyte

interface. The ease with which ions can cross the electrode/electrolyte interface and

diffuse within the host substrate is strongly dependent on the microstructure of the film

and also on temperature.

Electrochromic Smart Windows, are glazings that enable the amount of heat and light

entering a building to be controlled to optimise energy efficiency [3]. The operation of

these devices, and also electrochromic display devices, depends critically on

understanding the kinetics of ion injection in these materials. The expense of film

deposition in smart window manufacture is a drawback to commercial viability

107

especially when considering the large area of some glazings (from 1m2 upwards).

Several techniques exist for the deposition of electrochromic films onto glass substrates,

however the sol-gel route has significant cost advantages for large area coatings over

other more energy intensive and high capital-cost processes such as sputtering,

electrolytic deposition and vacuum evaporation [4]. The microstructures of the films

produced by these routes differ significantly, and hence the kinetic behaviour of the

various electrochromic films is also different [5,6,7]. In order to be able to optimise

processes such as manufacturing conditions, device switching and lifetime it is

necessary to understand the kinetics of the electrochromic reaction mechanism.

The description of kinetic behaviour includes information about the reaction mechanism

itself, rates of reaction, rate limiting steps and measurement of parameters such as

diffusion coefficients and charge transfer resistances. Smart windows will be required

to operate at temperatures up to 60ºC (and potentially higher) owing to absorption of

solar radiation in the electrochromic (EC) film in the coloured state [8], and hence it is

useful to understand how changes in temperature affect the kinetic behaviour of

electrochromics. Various models have been proposed to describe the kinetic behaviour

of sol-gel deposited ECs, and several well-defined experiments exist for the

determination of kinetic parameters.

This paper will outline one current kinetic model for sol-gel ECs, and the experimental

techniques used for the determination of relevant parameters. Measurements of the

diffusion coefficient of sol-gel deposited films have been made at various temperatures

from 20°C to 50°C, and they show behaviour close to the activated behaviour expected

108

for diffusion, although it appears that there may be some deviation from this at the

higher temperatures.

The theoretical basis of temperature dependence of ion intercalation in electrochromic

films is discussed in section 5.2, and the experimental work is described in section 5.3.

The analysis of the results is outlined in section 5.4, and the final section is a discussion

of the significance of these results and how these experiments and their analysis can be

improved.

5.2 Theory

5.2.1 Temperature effects on kinetic behaviour

Temperature affects the kinetic behaviour of electrochromic films in several ways,

depending on the mechanism of the reaction and in particular the rate limiting step for a

given process. Generally accepted rate limiting mechanisms include charge transfer at

the electrode/electrolyte interface [9,10], diffusion limited mass transfer inside the

electrode [11,12] and the series resistance of the cell [13,14]. The mobility of the ions

in the electrolyte is very high compared to the mobility within the electrode, so it is

unlikely that ionic diffusion in the electrolyte will limit the reaction rate.

Ions in the electrolyte will move towards the electrode under the effect of the applied

electric field. In order for an ion in solution to intercalate into the host lattice of the

electrode, it must lose its solvation sheath and also overcome the activation energy

109

associated with crossing the electrolyte/electrode boundary, and combine with an

electron at a host site. The energy required for this can come from the applied electric

field and also has a temperature contribution. The larger the potential difference across

the electrode/electrolyte interface, the larger the electrostatic force drawing ions across

the boundary. At high temperatures the ions will have more thermal energy which will

contribute to the ease with which they can be intercalated. Temperature therefore

provides extra driving force for the charge transfer step and lowers the electrical

potential difference required for intercalation. At elevated temperatures we therefore

expect the magnitude of the voltages required for colouration and bleaching of an

electrochromic film to be lower, if the charge transfer process is the rate limiting step.

At high voltages, the charge transfer process will be fast and hence this process is often

considered the rate-limiting step when small switching voltages are applied [15].

Once ions are intercalated into the film, they combine with an electron at a host atom

site. For example, consider the reaction

yLi+ + WO3 + ye- ⇔ LiyWO3 (5.1)

The tungsten atom is reduced from a +6 oxidation state to a +5 state, accompanied by an

optical transition (colour change) of the film. As ion intercalation proceeds the tungsten

sites at the electrode surface will soon become saturated with lithium ions, and

ion/electron couples will need to migrate further into the film if we are to continue to

inject ions into vacant tungsten sites. The concentration of ions at the electrode surface

is much higher than the concentration of ions within the film, and so the ions will move

according to Ficks laws of diffusion [16]. The ions will experience a force to move

110

under the concentration gradient formed upon intercalation. The greater the

concentration gradient and the greater the diffusion coefficient, the faster the ions will

move within the film. If the ions cannot move easily from the surface of the film, the

electrical potential required to intercalate further ions will increase. Large diffusion

coefficients will therefore lead to lower voltages for continued ion intercalation.

Diffusion is a thermally activated process which may be described by the equation [17]

)(

0RTQd

eDD−

= (5.2)

where Qd is the activation energy for diffusion, R is the ideal gas constant, T is absolute

temperature and D0 is a pre-exponential term. The exponential relationship between

diffusion coefficient and temperature means that relatively small increases in

temperature are accompanied by significant increases in the diffusion coefficient and

hence reductions in the switching voltages required for diffusion controlled processes at

elevated temperatures. It has been proposed that diffusion of ions within the

electrochromic electrode is the rate limiting step when switching voltages are relatively

large (ie. charge transfer is fast) [15].

The series resistance of a cell is the sum of the resistances of all components in the

circuit. This includes the external circuit, the transparent conducting oxide (TCO)

electrodes, the films themselves and the electrolyte. Temperature only has a small effect

on the series resistance of the cell, so we expect the series resistance to be fairly

constant. There will be a slight increase in the electrical conductivity of the electrolyte

with increasing temperature, and therefore a slight reduction in the ohmic drop across

the electrolyte, but this is small when compared to the voltages applied for colouration

111

and bleaching and can be minimised by reducing the electrode separation.

The voltage required to switch an electrochromic film is also dependent on the emf

(electromotive force) of the film, which in turn is related to the chemical composition of

the film at a given time by the Nernst equation [16] and is discussed below. The effect

of temperature on the kinetic behaviour of electrochromics is therefore a complex

combination of the temperature dependencies of diffusion, charge transfer and emf.

Generally however, we expect reactions to proceed more easily at elevated temperatures

and hence switching voltages should be lower, regardless of the rate limiting mechanism

for a particular system. This qualitative understanding helps explain the general trends

we observe when switching electrochromic films at elevated temperatures, but if we are

to optimise factors such as electrical and optical efficiency and lifetime hence improving

device design, we require more quantitative information.

5.2.2 Thermodynamics of coloration

The basic quantity governing these processes and reactions is free energy. The electrical

potential (E) associated with an electrochemical reaction is related to the free energy

change of the reaction (∆G) by

nEFG −=∆ (5.3)

where n is the number of electrons involved in the reaction, F is Faradays constant and

E is the potential in volts [16]. A cell emf (electromotive force) is therefore a way of

describing the free energy change of the reaction and a positive emf indicates a negative

free energy change, and hence a spontaneous reaction. The sign of the emf tells us

112

which direction a reaction will spontaneously proceed in. The standard free energy

change is given by ∆Go = -nEoF. The standard free energy change is also related to the

equilibrium constant (K) by

)ln(KRTG −=∆ (5.4)

where K = products

reactants

activityactivity when the system is at equilibrium.

Combining these two equations gives an expression for the standard cell potential of a

system in equilibrium,

)ln(º KnFRTE = (5.5)

The standard electrode potential is the potential of the electrode when the activities of all

species are defined by the equilibrium constant (ie. activityproducts =

K×activityreactants when expressed in some standard unit (eg Molality or mol/Kg). The

potential of a system not at equilibrium (ie E ≠ E0) is described by the Nernst equation

0

0

ln( )

ln( )

red

ox

ox

red

aRTnF a

aRTnF a

E E

E E

= −

= + (5.6)

The entropy change of a reaction describes the way in which temperature affects the free

energy change, and can be described mathematically by [16]

( )ET P

S nF δδ∆ = (5.7)

The change in cell emf with temperature is therefore directly related to the entropy

change of reaction and by measuring the change in emf with temperature, we can

113

determine this entropy change. When considering electrochromic reactions occurring in

thin film electrodes, the emf is related to the concentration (or more strictly, the activity)

of the injected ion at the electrode/electrolyte interface (c(0,t)). The emf associated with

the WO3 electrode (relative to some reference electrode) in the electrochromic reaction

described by equation 5.1 is given by the equation [18,19]

−

−+=),0(.1

),0(.ln),0(..

tcVtcV

FRTtcVbaemf

m

mm

ν (5.8)

where Vm is the molar volume of the WO3 and c(0,t) is the charge density in moles per

unit volume. a and b are constants related to the free energy of formation of WO3 and

free energy changes upon reduction of W6+ to W5+ and the resultant changes in

interactions between the atomic centres within the film, and ν is related to the entropy

change associated with the reaction [18]. In practise these parameters are used to fit

experimental data to equation 5.8. As this reaction proceeds to the right, the electrode

potential decreases (becomes more negative) indicating that the system is being

cathodically pushed further from equilibrium.

When colouring an electrochromic cell, the measured or applied voltage is related to the

emf and the colouration voltage (Vc(t)) by [6]

)()()( tVtemftV ca += (5.9)

and Vc(t) is given by [20]

),0(..)( tcVdydERitV mccc

−= (5.10)

114

where dydE is the slope of the coulometric titration curve, which gives the cell emf versus

stoichiometric coefficient for lithium (y). The total applied voltage observed during

colouration in an experiment is then [21]

−

−++=),0(1

),0(ln),0('.)(ty

tyFRTtybaRitV ccaν (5.11)

where y(0,t) is the stoichiometric coefficient of lithium (in LiyWO3) at the

electrode/electrolyte boundary (ie y(0,t) = Vm.c(0,t)) and

−=

dydEbb' . Equation 5.11

therefore provides a quantitative measure of the free energy change for a reaction in

terms of the measured cell voltage. We can extract information about the underlying

processes and reaction mechanisms occurring by measuring the cell voltages under

various conditions, such as varying temperature. Changes in the free energy of reaction

and hence changes in the measured potential can then be used to gather information

about the rate limiting mechanisms by modelling the voltage behaviour.

5.2.3 Modelling of concentration profile

The way in which the ions are distributed within the film can be described by the

equation [21]

Ψ=DnF

jtxc ),( (5.12(a))

and

( ) ( )

)4

)1(2)1(24

224

²)1(2exp24

²2exp2(0

Dtxlkerfc

Dxlk

Dtxklerfc

Dxkl

Dtxlkt

Dtxklt

k

−+

−+−

+

+−

−+−+

+−=Ψ ∑

∞

= ππ (5.12(b))

115


F is Faradays constant and D is the chemical diffusion coefficient of lithium ions. We

can use equation 5.12 to determine the concentration profile during ion injection for

various values of diffusion coefficient and time.

Figure 5.1 and Figure 5.2 show concentration profiles for t=1s and t=200s respectively,

simulated using equation (5.12). These results are for a 200nm thick film, coloured at

0.1mA/cm2. The concentration (y) is expressed as the stoichiometric coefficient of

lithium (in LiyWO3) and is therefore equal to c(x,t).Vm.

-0.01

0

0.01

0.02

0.03

0.04

0.05

0 50 100 150 200

D = 1x10-12 cm2/s

D = 2x10-12 cm2/s

D = 5x10-12 cm2/s

D = 1x10-11 cm2/s

Con

cent

ratio

n (y

)

Distance x (nm)

Figure 5.1 Simulated concentration profile for t=1s (Qinj=0.1mC/cm2)

116

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 50 100 150 200

D = 1x10-12 cm2/s

D = 2x10-12 cm2/s

D = 5x10-12 cm2/s

D = 1x10-11 cm2/s

Con

cent

ratio

n (y

)

Distance x (nm)

Figure 5.2 Simulated concentration profile for t=200s (Qinj=20mC/cm2)

It is evident from Figure 5.1 and Figure 5.2 that the surface lithium concentration is

strongly dependent on the diffusion coefficient, and hence on temperature. As the

diffusion coefficient increases, the concentration profile becomes flatter and hence the

concentration of ions at the electrode surface is decreased. The emf of a cell at high

temperature will therefore be lower (Equation 5.8) than that for low temperatures, for a

specific injected charge density, since the surface charge density will be lower owing to

an increase in the rate of diffusion away from the surface of the film.

The surface concentration of lithium ions for a diffusion coefficient of 5 x 10-12cm2/s

(Qin = 20mC/cm2) is approximately y = 0.49 (for y in LiyWO3). If the diffusion

coefficient is doubled to 1 x 10-11cm2/s for the same injected charge density, the surface

lithium concentration is approximately y = 0.46. This difference in surface

concentrations of lithium is the major cause of the observed differences in switching

characteristics of electrochromic films at elevated temperatures.

117

The voltage calculated with Equation 5.8 depends on the surface concentration of

lithium ions, because the potential drop occurs solely across the electrolyte/electrode

interface (assuming there is no internal potential drop in the electrode). The surface

lithium concentration is then calculated by solving equation 5.12 for the special case

where x = 0. Solving these, we obtain [21]

Γ=DnF

jtc 2),0( (5.13 (a))

and ππt

Dtlkt

Dtklerfc

Dkl

kk

−−−=Γ ∑∑∞

=

∞

= 1

22

1)][exp(2]2[ (5.13(b))

5.3 Experimental

5.3.1 Film preparation

Mixed tungsten-titanium oxide electrochromic films were deposited using sol-gel

processing from organic precursors onto 10cm × 10cm substrates of LOF TEC8/3 glass

using the sol-gel dip coating method [22]. The alkoxide precursor solutions used in the

sol-gel dipping have been described previously [23].




cell filled with an ethanolic solution of KCl, saturated with AgCl. The electrolyte used

was a solution of 1M LiClO4 in propylene carbonate. The electrolyte solution was dried

118

over molecular sieves prior to use and stored and used inside a dry glovebox, which was

maintained with a nitrogen atmosphere under a slight positive pressure. Experiments

were carried out in a glass tank (filled with electrolyte solution) partly submersed in a

larger heating tank filled with mineral oil. An electrical heater/stirrer unit was used to

control the temperature of the oil bath, and hence the electrolyte solution. The

electrolyte solution was also stirred during the experiments to minimise the temperature

difference between the heating oil and electrolyte solution. The complete apparatus

including electrolyte tank, oil bath and heater unit was kept in the dry-box during all

experiments.

The atmosphere inside the glovebox was kept dry by exposing it to P2O5 dessicant, and

recirculating the nitrogen through a column of dried molecular sieves. The humidity

level inside the glovebox was monitored with a HMP235 humidity and temperature

transmitter, manufactured by Vaisala. The humidity during the experiment was

maintained at 1.05ppm absolute humidity (Relative humidity = 5.3% and temperature =

22.4ºC immediately prior to experiment).


described previously [24,25]. The results reported are for a film cycled to 15mC/cm2

(Current density = 0.1mA/cm2, film area = 100cm2) hence the first 150s correspond to

the colouration cycle. The film was bleached after a 2s delay, with a current density of

0.1mA/cm2. During the experiment the temperature was ramped from 20° to 50°C at

approximately 1°C per cycle. Measurements were performed as the temperature

119

increased, but were not made as the system cooled.

5.4 Results

5.4.1 Variation in switching voltage with temperature

Figure 5.3 shows the results obtained from measurement of a sol-gel deposited

electrochromic film maintained under rigorously dry conditions before and during

cycling, as described above.

-1

-0.5

0

0.5

1

0 50 100 150 200 250 300 350

20.1oC

30.3oC

40.3oC

50.0oC

Volta

ge (V

)

Time (s)

Figure 5.3 Applied voltage for colouration and bleaching of a sol-gel WO3 film to

15mC/cm2

It is observed that the magnitude of the applied voltages decreases with increasing

temperature as predicted from considerations discussed in section 5.2.1. The decrease in

applied voltage with increasing temperature is greatest at low temperatures, which

indicates that small increases from room temperature result in significant decreases in

the voltages required for colouration and bleaching as predicted by equation 5.2.

120

Equation 5.11 describes this variation in voltage with temperature, in terms of the

surface concentration of lithium ions (y(0,t)). The effect of temperature on the cell emf

(for a given lithium concentration) is relatively small, and arises purely from the

Nernstian contribution to equation 5.11.

Figure 5.4 predicts the cell emf versus injected charge density for the same temperatures

as the experiment in Figure 5.3, using equation 5.8 and a=-0.66V, b=-0.87V, ν=5.76

[19]. The variation in emf with temperature for a given surface lithium concentration is

very small, and does not explain the large variation in switching voltages seen in Figure

5.3.

Figure 5.4 Dependence of emf on temperature and surface lithium concentration

predicted using equation 5.8.

A more marked effect of temperature arises from the change in diffusion coefficient, and

hence the way in which the ions are distributed within the films. Ions are injected into

121

the electrochromic film at the interface with the electrolyte (x = 0) and then diffuse into

the film under the influence of the concentration gradient. The concentration of ions in

the film is therefore greatest at x = 0, and decreases with increasing film depth. At

elevated temperatures the diffusion coefficient of lithium ions will be greater, and hence

the surface concentration of lithium ions (for a specific injected charge density) will be

lower. We must therefore take the concentration profile into account, in order to

determine the surface concentration of lithium ions (y(0,t)) and hence simulate the

experimental data using equation 5.11.

5.4.2 Simulation of Voltage Response of Films

We can now use equation 5.13 to simulate the variation of c(0,t) with time during charge

injection, and therefore indirectly determine the diffusion coefficient for each

temperature. The 20.1ºC voltage/time data for colouration of the film in Figure 5.5 was

simulated using equation 5.11, using a least squares method to obtain the best fit by

adjusting the parameters a, b, ν, Rc and D. The constants a, b, ν and Rc were assumed

to be intrinsic to the electrochromic film being studied, and hence only the diffusion

coefficient varied with temperature. The best fit for the 20.1ºC data was obtained with

the following constants:

a = -1.06V,

b' = -0.65V,

ν= 4.2,

Rc = 40Ω and

D = 2.1 x 10-12cm2/s

122


electrochromic film. (a) Temperature = 20.1ºC, D = 2.07x10-12cm2,

(b) temperature = 30.3ºC, D = 6.33x10-12cm2, (c) temperature = 40.3ºC,

D = 1.33x10-11cm2 and (d) temperature = 50.0ºC, D = 1.71x10-11cm2.

The voltage/time data for the colouration at the other temperatures was then simulated

using the above values for a, b, ν and Rc, and only changing the diffusion coefficient.

The experimental and simulated data for four temperatures are shown together in and

the values estimated for the diffusion coefficients and the least squares sum in the

simulations are shown in Table 6.1.

123

T (ºC) T (K) 1/T(K-1) D (cm2/s) ln D Σ(Vsim-Vexp)2

20.1 293.3 0.00341 2.07E-12 -26.90 0.0061

30.3 303.5 0.00330 6.33E-12 -25.79 0.0077

40.3 313.5 0.00319 1.33E-11 -25.04 0.0058

50.0 323.2 0.00309 1.71E-11 -24.79 0.0062

Table 5.1 Diffusion coefficients used to simulate experimental voltage/time data for

sol-gel WO3 film coloured to 15mC/cm2, and least squares sum from data fitting.

5.5 Discussion

It is evident from the theoretical fits (See Figure 5.5(a)-(d)), using equation 5.13 to

estimate the surface lithium concentration during ion injection, and equation 5.11 to

calculate the corresponding voltages, to the experimental data is very good. We observe

that the diffusion coefficients estimated for each data set increase with temperature, so

the trend is as we predicted in section 5.2.1. There is a small deviation between

experimental and simulated data for long time periods, but in general the model

describes the data well considering that the diffusion coefficient was the only parameter

changed for the simulations. The chemical diffusion coefficients estimated from the

simulations are within the range of those reported in the literature for lithium in WO3

[5,19,26]. The room temperature (20.1ºC) diffusion coefficient of 2.01x10-12 cm2/s

seems a little low and would imply that the surface lithium concentration exceeded the

124

reversible limit during colouration, however this may be due to a number of limitations

and assumptions made in the model which are discussed below.

We would expect the values of diffusion coefficients shown in Table 5.1 to follow the

temperature dependence of equation 5.2. An Arrhenius plot of ln D versus 1/T should

then yield a straight line of slope R

Qd− and intercept equal to ln(D0). Figure 5.6(a)

shows one such plot for the diffusion coefficient values extrapolated from the

experimental data simulations. We see that equation 5.2 is satisfied for low temperature

data, but the plot deviates from linearity as temperature increases past approximately

40ºC. Figure 5.6(b) shows the same information plotted for temperature values between

20.1 and 40.3ºC and the data very closely approximates a linear relationship, with a

Pearson moment correlation coefficient, R2 = 0.98.

Figure 5.6 Variation in estimated diffusion coefficients with temperature plotted for

range (a) 20.1 < T < 50.0ºC and (b) 20.1 < T < 40.3ºC.

The calculated diffusion activation energy using the slope of the regression line from

125

Figure 5.6(a) is 0.58eV and the pre-exponential term (D0) is 2.8x10-2 cm2/s. We have

been unable to find published values in the literature for activation energies of diffusion

of Li+ ions in electrochromic films. We may however reasonably expect a small ion

moving within a lattice of relatively large atoms (like lithium in WO3) to have a

diffusion activation energy of approximately the same order of magnitude as, for

example, carbon atoms in α-iron. The activation energy for carbon atoms in α-iron is

0.83eV/atom, and so compares reasonably well with the result determined from this

experimental work. The pre-exponential (D0) for carbon atoms in α-iron is 1x10-1

cm2/s and so the value determined in this work is also in an appropriate range.

The diffusion activation energy determined from the regression line of Figure 5.6(b) is

0.73eV and the pre-exponential (D0) is 7.3x10 cm2/s. The activation energy therefore

seems reasonable, but the pre-exponential term is unusually high and would seem

erroneous. The fact that the data in Figure 5.6 appears more linear at low temperatures

may suggest that the low temperature data is more accurate. A closer analysis of the

assumptions and limitations used in this modelling provides some possible explanations

of this observation.

The model assumes that there is no voltage drop internal to the electrode (WO3 film), so

the measured potential drop occurs solely across the electrode/electrolyte interface and

in the external circuit. In reality the conductivity of the WO3 layer increases as charge is

injected and approaches metallic conductivity as the free electron density approaches the

Mott critical density [11,27,28]. This means that there is a real potential drop internal to

126

the electrode during the early stages of colouration that will provide an electrostatic

force to drive ions in the same direction as their diffusive motion (ie. away from the

electrode surface). The effect of this is to increase the speed at which ions move for low

values of the stoichiometric lithium coefficient (y) and the diffusion coefficient will

appear larger at the start of colouration.

The model also assumes that the slope of the coulometric titration curve dydE is constant

throughout the charge injection, and also for changing temperature. In reality, this slope

is larger for low values of y and decreases with charge injection and will decrease

significantly as temperature increases. The effect of this assumption is to underestimate

the values of Va(t) for low injected charge density and for high temperatures and so

introduces another error in this method of analysis of the data.

The series resistance during coloration, Rc, is also assumed to be independent of

temperature, but in reality will decrease with increasing temperature due to the presence

of more electrons in the conduction bands of the various components in the circuit. The

error introduced by this assumption is however small, as changes in resistance of the

circuit components is very small in relation to the voltages being measured. The

experimental voltages were not corrected for iR drop of the electrolyte, but this effect

will also be small as the working-reference electrode separation was minimal (approx 5-

10mm) and the electrolyte conductivity is relatively high.

The assumptions made in order to simulate the data with this model make complete

interpretation of the results more difficult, however the trend of the results is in

127

accordance with kinetic theory. Future experiments are planned to take into account

these assumptions and produce a better model capable of accurately predicting the

measured voltages during ion insertion and extraction, and explaining the observed

changes in kinetic behaviour with temperature.

5.6 Conclusion

A simulation model based on a combination of diffusion-limited motion and series

colouration resistance has been used to simulate the voltage-time characteristics of

lithium injection into sol-gel deposited tungsten-titanium mixed oxide films. The model

was successfully used to simulate experimental data over an extended temperature

range, by changing only the chemical diffusion coefficient. The diffusion coefficients

estimated from the simulations increased with temperature in accordance with

Arrhenius-like activation behaviour, however there was some deviation from this

behaviour at temperatures above 40ºC. It has been proposed that deviations from ideal

diffusion-temperature behaviour are due to some of the assumptions made in the

modelling process, and future experiments have been planned to minimise these

limitations. Diffusion coefficients in the range of 10-11-10-12cm2/s were estimated with

this model, which are in good agreement with values published in the literature.

Acknowledgements




128


Commonwealths Cooperative Research Centres Program.

REFERENCES [1] B.W. Faughnan, R.S. Crandall and P.M. Heyman, R. C. A. Rev., 36, 177-197 (1975).

[2] C.M. Lampert, Introduction to Chromogenics in: C.M. Lampert and C.G. Granqvist

(Eds.), Large-Area Chromogenics: Materials and Devices for Transmittance Control,

Optical Engineering Press-SPIE, Bellingham, WA, 1990, 378.

[3] C. M. Lampert, IEEE Circuits and Devices, 8, 19-26 (1992).

[4] N. Ozer and C.M. Lampert, Solar Energy Materials and Solar Cells 54 (1998) 147.

[5] J.-G. Zhang, C. Edwin Tracy, D.K. Benson and S.K. Deb, J. Mater. Res., 8, 2649-

2656 (1993).


2667 (1993).

[7] C.G. Granqvist, Appl. Phys. A 57 (1993) 3.


Solar Cells, 56, 449-463 (1999).


[10] B.W. Faughnan and R.S. Crandall, Electrochromic Displays Based on WO3in

Topics in Applied Physics, Display Devices (Edited by J. I. Pankove), Vol. 40, Springer-

Verlag, New York (1980).


(1980).

[12] B. Reichman, A.J. Bard and D. Laser, J. Electrochem.Soc., 127(3), 647-654

129

(1980).

[13] B. Vuillemin and O. Bohnke, Solid State Ionics, 68, 257-267 (1994).

[14] O. Bohnke, M. Rezrazi, B. Vuillemin, C. Bohnke and P.A. Gillet, Solar Energy

Materials and Solar Cells, 25, 361-374 (1992).


[16] A. J. Bard and L. R. Faulkner, in Electrochemical Methods: Fundamentals and

Applications, Wiley, New York (1980).

[17] W.D. Callister, Materials Science and Engineering: An Introduction, John Wiley

and Sons, Inc., New York (1985).

[18] R.S. Crandall, P.J. Wojtowicz and B.W. Faughnan, Solid State Comm., 18, 1409-

1411 (1976).


[20] I.D. Raistrick and R.A. Huggins, Solid State Ionics, 7, 213-218 (1982).

[21] J. Wang, PhD Thesis, University of Technology, Sydney (1998)..






437.



[26] M. Green, W.C. Smith and J.A. Weener, Thin Solid Films 38 (1976) 89.

130

[27] N.F. Mott, J. Non-Cryst. Solids 1 (1968) 1.

[28] V. Wittwer, O.F. Schirmer and P. Schlotter, Solid State Comm. 25 (1978) 977.

CHAPTER 6

SIMULATION OF ELECTROCHROMIC SWITCHING VOLTAGES

AT ELEVATED TEMPERATURES.

J.P. Matthews, J.M. Bell and I.L. Skryabin

Published: Electrochimica Acta, 46, 1957-1961 (2001).

132




and revised by J.M. Bell before final submission of the manuscript.

133

SIMULATION OF ELECTROCHROMIC SWITCHING VOLTAGES AT

ELEVATED TEMPERATURES.

J.P. Matthews1, J.M. Bell1 and I.L.Skryabin2



2 Sustainable Technologies, Australia,


[email protected]

Abstract

Sol-gel deposited electrochromic WO3/TiO2 films have been reversibly cycled at

temperatures up to 70ºC using a constant-current charge injection technique, in a

stringently dry environment. The resultant switching voltages have been simulated with

a model involving several parameters including the chemical diffusion coefficient (D) of

lithium in the films. The experimental data at various elevated temperatures has been

fitted with the model by varying the diffusion coefficient at each temperature and

holding other parameters constant. The values of D estimated from the simulation of

experimental data are within the range of published values for similar films, however

there are some limitations due to assumptions made in the model.

This paper discusses the application of the model for the prediction of cycling voltages

at high temperatures and the suitability of the model in the estimation of the chemical

diffusion coefficient (D).

Keywords: Electrochromic switching; kinetic behaviour; temperature dependence;

diffusion coefficient; voltage simulation

134

6.1 Introduction

Smart windows based on amorphous thin films of tungsten oxide (WO3) may be used to

reduce the transmittance of both heat and light by changing from a clear state to a deep

blue coloured state. In the coloured state the amount of heat and light transmitted is

significantly reduced because the electrochromic film absorbs a large proportion of the

incoming radiation. This absorption of radiation is accompanied by an increase in the

temperature of the material and it is foreseeable that these devices may be required to

operate at temperatures up to 70ºC [1,2]. The switching characteristics of

electrochromic (EC) devices are dependent on temperature, so it is desirable to predict

this behaviour in order to prevent the application of excessive voltages. The ability to

simulate EC switching voltages over an extended temperature range therefore has

application in the determination of suitable control strategies. An optimal control

strategy would ensure maximum device lifetime and also uniform and consistent

coloration of the devices, regardless of temperature.

The kinetic behaviour of EC materials is affected by many factors including

microstructure [3,4], composition [5], ion mobility [6] (specifically diffusion rates) and

charge transfer resistance [7]. Temperature significantly affects ionic mobility, and the

diffusion rate of the mobile ions within the EC host lattice is observed to increase with

temperature [8]. The increased ease with which ions may move in the EC material at

high temperatures is one reason why the magnitude of the voltages required to colour

and bleach EC materials decreases at high temperatures. An increased diffusion rate

also means that the concentration profile of ions inside an EC film will be flatter, and

135

the surface concentration will then be lower than for the same injected charge density at

a lower temperature. These effects of temperature on ionic diffusion may be described

by a mathematical model, which in turn may be used to simulate the voltages required to

colour and bleach these films.

In this paper the simulation model developed by Wang [9] has been used to model

experimental data between 35ºC and 76ºC and the chemical diffusion coefficient of

lithium in the films has been extracted from the simulation process. The diffusion is

observed to closely follow the expected Arrhenius type activation behaviour, and the

activation energy for diffusion of lithium ions in the films has been calculated.

6.2 Experimental


Mixed tungsten-titanium oxide electrochromic films (mole ratio W:Ti of 4:1), were

deposited onto 10cm x 10cm substrates of LOF TEC8/3 glass. The films were coated

using the sol-gel dip coating method [10], with solutions of tungsten and titanium oxy-

butoxides as the organic precursors. The details of the film preparation have been

described previously [11].


Electrochemical measurements were made using a three electrode cell, the counter and

reference electrodes being a sheet of copper coated platinum (area=85cm2), and a

Ag/AgCl wire respectively. The electrolyte was 1M LiClO4/propylene carbonate and

was dried over molecular sieves prior to use. The complete electrochemical cell was

136

contained inside a dry glovebox, and maintained with a nitrogen atmosphere under a

slight positive pressure. A glass tank was used to hold the electrodes and electrolyte of

the cell, and this was partially submerged in a temperature controlled oil-bath. An

electrical heater/stirrer unit was used to control the temperature of the oil bath, and

hence the electrolyte solution. The electrolyte solution was also stirred during the

experiments to minimise any thermal lag. The atmosphere inside the dry-box was

maintained at less than 1ppm absolute humidity by exposure to phosphorus pentoxide

(P2O5) and recirculation through pre-dried molecular sieves.

The WO3/TiO2 films were cycled using a voltage-limited constant current technique,

described previously [12,13]. The only difference between this method and that

described previously is that the maximum voltage limit for bleaching was not fixed, but

was determined during cycling by measuring the rate of change of voltage. The

constant current bleaching step was terminated when the rate of voltage increase

exceeded 0.5V/s, which ensured that excessive bleaching voltages were not applied at

elevated temperatures.

The results reported are for a film cycled to 15mC/cm2 (Current density = 0.1mA/cm2,

film area = 75cm2) hence the first 150s correspond to the coloration cycle. The film was

bleached after a 2s delay, with a current density of 0.1mA/cm2. The cycles used for the

data simulations were number 42(35.8ºC), 69(46.5ºC), 86(56.2ºC), 100(65.3ºC) and

118(76.4ºC).

6.2.3 Simulation of EC film coloration voltage

The coloration voltage simulations were carried out by fitting experimental data to the

137

equation [9]

−

−++=),0(.1

),0(.ln),0('.)(

tcVtcV

FRTtcVbaRitV

m

mmcca

ν (6.1)

which is a modified form of equations first reported by Nagai et al [14] and Crandall et

al [15]. Rc is the series resistance of coloration, ic is the switching current and a is a

constant related to the free energy of formation of WO3 and free energy changes upon

reduction of W6+ to W5+ and the resultant changes in interactions between the atomic

centres within the film. ν is related to the entropy change associated with the EC

reaction [15]. Vm is the molar volume of the film and

−=

dydEbb' where dE/dy is the

slope of the coulometric titration curve. In practice the parameters a, b' and ν are used

to fit experimental data to equation (6.1). The surface lithium concentration c(0,t) is the

charge density in moles per unit volume at the electrode/electrolyte interface and is

determined from the equations [8,9]

Γ=DnF

jtc 2),0( (6.2(a))

and ππt

Dtlkt

Dtklerfc

Dkl

kk−−−=Γ ∑∑

∞

=

∞

= 1

22

1)][exp(2]2[ (6.2(b))


F is Faradays constant and D is the chemical diffusion coefficient of lithium ions. We

can use Eq. (7.2) to determine the electrode surface lithium concentration during ion

injection for various values of current density, diffusion coefficient and time.

138


6.3.1 Voltage characteristics

Figure 6.1 shows the voltage (relative to a reference electrode) of the mixed oxide EC

film during coloration and bleaching at temperatures from 35.8ºC to 76.4ºC. The

magnitude of the voltages was observed to decrease as temperature increased primarily

due to an increase in the lithium ion mobility and therefore a reduction in the surface

lithium concentration [8].

-1.5

-1

-0.5

0

0.5

1

1.5

0 50 100 150 200 250 300 350 400

35.8oC

46.5oC

56.2oC

65.3oC

76.4oC

Volta

ge (V

)

Time (s)

Figure 6.1 Applied voltage during colouration and bleaching of a sol-gel WO3/TiO2

film to 15mC/cm2/s.

The data for the 35.8ºC cycle was simulated using Eq. (6.1) and (6.2) using a least

squares method to obtain the best fit by adjusting the parameters a, b, ν, Rc and D. The

constants a, b', ν and Rc were assumed to be intrinsic to the electrochromic film being

studied, and hence only the diffusion coefficient varied with temperature. A close fit for

the 35.8ºC data was obtained using the constants a = -1.3V, b' = -0.3V,

139

ν= 4.2, Rc = 40Ω and D = 8.7 x 10-13cm2/s and Figure 6.2(a) shows the experimental and

simulated data at this temperature. The voltage/time data for the coloration at the other

temperatures was then simulated (See Figure 6.2(b)-(e)) using the above values for a, b,

ν and Rc, and only changing the diffusion coefficients.

-1.3

-1.2

-1.1

-1

-0.9

-0.8

-0.7

-0.6

0 40 80 120 160

Exp 35.8 oC

Sim 35.8 oC

Volta

ge (V

)

Time (s)

(a)

-1.2

-1.1

-1

-0.9

-0.8

-0.7

-0.6

-0.5

0 40 80 120 160

Exp 46.5 oC

Sim 46.5 oC

Volta

ge (V

)

Time (s)

(b)

-1.1

-1

-0.9

-0.8

-0.7

-0.6

-0.5

-0.4

0 40 80 120 160

Exp 56.2 oC

Sim 56.2 oC

Volta

ge (V

)

Time (s)

(c)

-1.1

-1

-0.9

-0.8

-0.7

-0.6

-0.5

-0.4

0 40 80 120 160

Exp 65.3 oC

Sim 65.3 oC

Volta

ge (V

)

Time (s)

(d)

-1

-0.9

-0.8

-0.7

-0.6

-0.5

-0.4

-0.3

0 40 80 120 160

Exp 76.4 oC

Sim 76.4 oC

Volta

ge (V

)

Time (s)

(e)


electrochromic film. (a) T = 35.8ºC, D = 8.68x10-13cm2/s, (b) T = 46.5ºC,

D = 1.54x10-12cm2/s, (c) T = 56.2ºC, D = 4.02x10-12cm2/s, (d) T = 65.3ºC,

D = 1.48x10-11cm2/s and (e) T = 76.4ºC, D = 6.00x10-11cm2/s.

140

It is evident from the data simulations shown in Figure 6.2 that the experimental data is

quite well simulated by Eq. (7.1) and (7.2) with a single fitting parameter. The lithium

chemical diffusion coefficients determined from the data simulations were observed to

increase with temperature, as expected from kinetic and thermodynamic considerations

[8]. The diffusion coefficients estimated from simulation of the experimental data (For

values see Figure 6.2) are within the approximate range of values reported in the

literature [3,14,16] for other EC films. The relatively low diffusion coefficients

estimated at the lower temperatures (approximately 1x10-12cm2/s) would indicate a fairly

low ionic mobility, and this conclusion is supported by the fact that relatively large

voltages were required for switching at these temperatures. The difference in the

maximum coloration and bleaching voltage at 35.8ºC was 2.64V, whereas in a previous

experiment [8] this voltage range was only 1.44V at 20.1ºC (D=2.07x10-12cm2/s

extrapolated from data simulation), for the same injected charge density. This behaviour

suggests that the switching voltages are strongly dependent on ionic mobility, with high

mobility allowing EC films to be switched with the application of quite low electrical

potentials.

We would expect the values of the chemical diffusion coefficient (determined from the

simulations) to follow the temperature dependence of

)(

0RTQd

eDD−

= (6.3)

where Qd is the activation energy for diffusion, R is the ideal gas constant, T is absolute

temperature and D0 is a pre-exponential term. An Arrhenius plot of ln D versus 1/T

should then yield a straight line of slope R

Qd− and intercept equal to ln(D0). Figure 6.3

141

shows one such plot for the diffusion coefficients extrapolated from the data simulations

and we can see that equation (6.3) is reasonably well satisfied (R2=0.961) over the

experimental temperature range.

-28

-27

-26

-25

-24

-23

2.85 10-3 2.95 10-3 3.05 10-3 3.15 10-3 3.25 10-3

y = -11497x + 9.052

R2= 0.961

ln D

1/T (K-1)

Figure 6.3 Arrhenius plot showing the variation in estimated diffusion coefficients with

temperature.

The plot begins to deviate from linearity as temperature increases, with the trend that the

diffusion coefficient is becoming very large. This behaviour may be due to a shift from

a diffusion-limited motion to a mechanism limited by charge transfer as temperature

increases, as discussed below.

The activation energy for diffusion, calculated from the slope of the regression line in

Figure 6.3, is 0.99eV, which is approximately twice the value extrapolated from

previous experimental work [8], in which the switching voltages were also found to be

142

significantly smaller. This result is in agreement with the conclusion that the lithium ion

mobility in the films in this experiment was quite low. We have been unable to find

published values of the diffusion activation energy of lithium ions for comparison with

this result (excluding our own previous experimental work). We may however

reasonably expect this value to be of a similar order of magnitude to that for diffusion of

carbon in

α-iron, because this system also involves the movement of relatively small ions within a

lattice of much larger host atoms. The activation energy for the diffusion of carbon

atoms within α-iron is 0.83eV/atom [17], and so compares reasonably well with the

result determined from this experimental work.

The simulation model presented here is based primarily on diffusion limited motion of

ions in the EC film. The diffusion coefficient estimated at 76.4ºC was 6x10-11cm2/s,

which implies that diffusion was very fast at this temperature. At this temperature, the

applied voltages are also quite low which will decrease the rate of the charge transfer

step [18]. It is possible that the EC reaction is limited by diffusion at lower

temperatures (when diffusion is relatively slow and voltages are large) and limited by

charge transfer at the electrode surface at high temperatures (when diffusion is fast and

voltages are low). The absence of a term describing the charge transfer process is

therefore an inherent limitation to the current simulation model presented here, and a

possible explanation for the deviation from linearity of Figure 6.3 towards high

temperatures.

The incorporation of a term describing the rate of the charge transfer step in this

simulation model may help better describe experimental results, and lead to a better

143

understanding of the underlying mechanisms occurring during coloration and bleaching

of EC films and devices.

6.4 Conclusion

The voltage-time characteristics for the coloration of a mixed tungsten/titanium oxide

EC film have been simulated over a wide temperature range, using a model based on the

diffusion limited motion of ions and a series coloration resistance. A good

approximation of the experimental data was made for temperatures between 36º-76ºC by

only changing the diffusion coefficient for each temperature. The relationship between

the extrapolated diffusion coefficients and temperature was close to Arrhenius activation

behaviour, with an activation energy for diffusion of 0.99eV. It has been proposed that

the charge transfer process increasingly becomes a limiting mechanism at high

temperatures and the inclusion of a term describing this in the simulation model will

enable better prediction and understanding of the EC reaction. Lithium ion chemical

diffusion coefficients in the range of 10-11-10-13cm2/s were estimated with this model,

which are in reasonable agreement with values published in the literature.

144

Acknowledgements






145

REFERENCES

[1] C.M. Lampert, A. Agrawal, C. Baertlien and J. Nagai, Solar Energy Materials and

Solar Cells, 56, 449-463 (1999).

[2] M.E. Badding, S.C. Schulz, L.A. Michalski and R. Budziak, Electrochemical

Society Proceedings, 96-24, 369-384 (1996).

[3] J.-G. Zhang, C. Edwin Tracy, D.K. Benson and S.K. Deb, J. Mater. Res., 8, 2649-

2656 (1993).


2667 (1993).



(1980).


[8] J.M. Bell and J.P. Matthews, Temperature Dependence of Kinetic Behaviour of Sol-

Gel Deposited Electrochromics, Solar Energy Materials and Solar Cells (2000),

accepted for publication.







437.

146





1411 (1976).

[16] M. Green, W.C. Smith and J.A. Weener, Thin Solid Films 38 (1976) 89.

[17] W.D. Callister, Materials Science and Engineering: An Introduction, John Wiley

and Sons, Inc., New York (1985).


CHAPTER 7

SELF-BLEACHING, MEMORY EFFECT AND REVERSIBILITY OF

ELECTROCHROMISM AT ELEVATED TEMPERATURES.

J.P. Matthews and J.M. Bell

Status: Submitted to Solar Energy Materials and Solar Cells.

148



the supervision of J.M. Bell. The paper was written by J.P. Matthews and revised by

J.M. Bell before final submission of the manuscript.

149

SELF-BLEACHING, MEMORY EFFECT AND REVERSIBILITY OF

ELECTROCHROMISM AT ELEVATED TEMPERATURES.

J.P. Matthews1, J.M. Bell1



[email protected]

Abstract

Sol-gel deposited WO3/TiO2 electrochromic films have been cycled under various

conditions of moisture and temperature. Films that were coloured and left in a liquid

electrolyte were observed to self-bleach with time, even with the counter electrode

disconnected from the cell. The rate of this self-bleaching was observed to increase

with moisture content and temperature. Charge was able to be injected and extracted

very reversibly under stringently dry conditions, however the time required to bleach the

film under these conditions was unacceptably long.

The results suggest that the water present in the electrolyte and the water present inside

the film play different roles in the self-bleaching mechanism. The addition of water to a

dry electrolyte during film cycling, resulted in significantly lower voltages being

required for bleaching, however the coloration voltage was unaffected. The overall

result is that an electrochromic film has been demonstrated to have an excellent

electrochromic memory and reversibility over a wide temperature range, however this

was accompanied by very long switching times and high voltages required to colour and

bleach.

150

Keywords: Electrochromic memory; reversibility; self-bleaching; water; temperature

dependence

7.1 Introduction

Electrochromic materials undergoes a colour change when ions and electrons are

inserted into them under the influence of an applied electric field or current.

Electrochromic materials have application in Smart Windows, glazings that enable the

amount of heat and light entering a building to be controlled, in order to optimise energy

efficiency [1]. Smart windows will be required to operate at temperatures exceeding

60ºC due to the absorption of solar radiation, inherent to the coloured state [2]. It is then

useful to understand how changes in temperature may affect the electrical and optical

properties of electrochromic materials, especially in terms of reversibility and

electrochromic memory. Several papers have described self-bleaching of

electrochromic films and the way this behaviour is affected by various conditions

including water content [3,4], the type of injected ion [5], film composition [6]

preparation conditions [7] and film roughness [8].

Recent experimental work [9] has shown that the reversibility of the electrochromic

reaction is largely dependent on environmental conditions such as moisture level and

temperature. Irreversible charge injection was observed for temperatures over 30ºC,

whereby a small proportion of the charge injected during colouration was unable to be

extracted during the bleaching process. Chemical analysis revealed that a significant

proportion of the injected ions did indeed remain inside the film after bleaching. Films

that were coloured in a liquid electrolyte in the ambient laboratory environment were

151

also observed to slowly self-bleach with time [9] even when the working electrode was

electrically isolated from the cell. The self-bleaching rate was found to increase

monotonically with temperature, and the authors speculated that the irreversibility

observed was due to the presence of water in the electrolyte. Similar films showed very

reversible behaviour when cycled in a stringently dry environment, at temperatures up to

50ºC, however self-bleaching experiments using a very dry electrolyte have not been

reported.

This paper describes the observation of self-bleaching in electrochromic WO3/TiO2

films cycled under various conditions of moisture and temperature. Films cycled in a

very dry electrolyte were observed to self-bleach, however the rate was much slower

than for a moist electrolyte. A film was fired in a furnace at 250ºC for 8 hours and then

used in self-bleaching experiments in very dry conditions. The film did not self-bleach

significantly over a 30-minute period even at 75ºC, however it was electrochemically

very difficult to bleach the film. The voltage characteristics of the bleaching process

were found to be highly dependent on the level of moisture in the electrolyte.

The electrochromic memory and reversibility of the fired film cycled in a dry electrolyte

were very good, however the slow bleaching response and large voltages required for

switching largely limit the application of such a system.

152

7.2 Experimental


Mixed tungsten-titanium oxide electrochromic films (mole ratio W:Ti of 4:1), were

deposited onto 10cm x 10cm substrates of LOF TEC8/3 glass. The films were coated

using the sol-gel dip coating method [10], with solutions of tungsten and titanium oxy-

butoxides as the organic precursors. The details of the film preparation have been

described previously [11].

7.2.2 Electrochemical Testing

Electrochemical measurements were made using a three electrode cell, the counter and

reference electrodes being a sheet of copper coated platinum (area=85cm2), and a

Ag/AgCl wire respectively. The electrolyte solution was 1M LiClO4/propylene

carbonate, which was dried over molecular sieves prior to use. The complete

electrochemical cell was contained inside a dry glovebox, and maintained with a

nitrogen atmosphere under a slight positive pressure. A glass tank was used to hold the

electrodes and electrolyte of the cell, and this was partially submerged in a temperature

controlled oil-bath. An electrical heater/stirrer unit was used to control the temperature

of the oil bath, and hence the electrolyte solution. The electrolyte solution was also

stirred during the experiments to minimise any thermal lag. The atmosphere inside the

dry-box was maintained at less than 1ppm absolute humidity by exposure to phosphorus

pentoxide (P2O5) and recirculation through pre-dried molecular sieves.

The WO3/TiO2 films were cycled using a voltage-limited constant current technique,

described previously [12,13]. The only difference between this method and that

153

described previously is that the maximum voltage limit for bleaching was not fixed, but

instead was determined in-situ during cycling by measuring the rate of change of

voltage. The constant current bleaching step was terminated when the rate of voltage

increase exceeded 0.5V/s, which ensured that excessive bleaching voltages were not

applied at elevated temperatures.

The self-bleaching experiments involved colouring of films to 15mC/cm2 using a

current density of 0.1mA /cm2 and a film area of 85cm2. The counter electrode was

disconnected from the electrical circuit immediately after completion of the coloration.

Optical and electrical measurements were made at a frequency of 1Hz while the

electrolyte solution was maintained at the appropriate temperature. At the end of the 30

minute self-bleaching period, the counter electrode was reconnected, the film was

bleached and the temperature was changed for the next data set.

The cycling data reported in section 3.3 was recorded in the same fashion as the

experiment described above, however there was no self-bleaching period and the

counter electrode was permanently connected to the circuit.

Optical measurements were made by passing a 1mW, 670nm laser beam through the

working electrode onto a silicon photodiode, and the photocell voltages were used to

calculate the change in optical density (∆OD) of the films during cycling.

154


7.3.1 Observation of self-bleaching at elevated temperatures

Self-bleaching was previously observed when experiments were carried out in the

ambient laboratory environment, and it was proposed that water present in the

electrolyte was reacting with lithium ions at the electrode surface [9]. It was therefore

expected that there would be no self-bleaching when the experiment was repeated in a

very dry electrolyte, but this result was not observed here.

Figure 7.1(a) shows the change in optical density (∆OD) as a function of time during a

self-bleaching experiment carried out under stringently dry conditions. In this

experiment a WO3/TiO2 was coloured by injection of 15mC/cm2 of lithium ions, and the

counter electrode was immediately disconnected from the electrochemical cell. Any

reaction observed is therefore separate to the normal electrochromic process, which

requires an electron current to flow from the back contact of the working electrode to

the counter electrode.

0.45

0.50

0.55

0.60

0.65

0.70

0 5 10 15 20 25 30 35 40

24.3oC

34.8oC

45.2oC

55.1oC

65.3oC

75.3oC

OD

Time (min)

∆

(a)

-0.85

-0.80

-0.75

-0.70

-0.65

-0.60

-0.55

0 5 10 15 20 25 30 35 40

24.3

34.8

45.2

55.1

65.3

75.3

Volta

ge (V

)

Time (min)

(b)


of a WO3/TiO2 film in dry electrolyte.

155

The optical density of the coloured film is seen to decrease with time as the film slowly

bleaches in the liquid electrolyte. The rate of change in ∆OD increases monotonically

with temperature, although the rates are generally slower than for similar experiments

previously carried out in the ambient environment where significant amounts of

moisture were present. Figure 7.1(b) shows the change in film voltage (measured

relative to the reference electrode) with time. The measured voltage is a function of the

surface composition of the film, and this result indicates that the composition of the

electrode surface is still changing even after 30 minutes at 75ºC.

The fact that self-bleaching was observed even in a very dry electrolyte suggested that

there was some moisture present inside the film, even though they had been stored over

P2O5, so a film was fired in a furnace at 250ºC for eight hours to further dry it. This

fired film was then used for a similar self-bleaching experiment, the results of which are

shown in Figure 7.2(a) and (b).

0.35

0.40

0.45

0.50

0.55

0.60

0 5 10 15 20 25 30 35 40

28.6

35.5

43.8

50

58.1

71.7

Time (min)

OD

∆

(a)-1.20

-1.10

-1.00

-0.90

-0.80

-0.70

0 5 10 15 20 25 30 35 40

28.635.5

43.850

58.171.7

Volta

ge (V

)

Time (min)

(b)


of a WO3/TiO2 film, after firing at 250ºC for 8 hours.

156

After firing the film, the rate of self-bleaching decreased dramatically, and very little

change in the optical density is observed with time, even at 58ºC. The large amount of

noise in the optical data at 71.7ºC is a result of maintaining the constant temperature

during this cycle. The heater thermostat switched the heating element on and off

intermittently, resulting in changing refractive indices of the oil bath, and hence

interfering with the optical measurements. The rate of self-bleaching is seen to have

almost no temperature dependence, as the lines of ∆OD as a function of time are not

observed to diverge at long times, as they do in Figure 7.1(a).

Figure 7.2(b) shows the electrical response of the working electrode, which is much

flatter than for the un-fired film, especially at the higher temperatures. The voltage data

for the 28.6ºC cycle appears to significantly deviate from linearity, but it is likely that

insufficient time has passed for diffusion processes to occur. This is supported by the

fact that the slopes of the lines of Figure 7.2(b) become approximately the same for high

temperatures and long times. There is little difference between the slopes of the

voltage-time lines between 35.5 and 71.1ºC also indicating that there is little

temperature dependence of the self-bleaching and that the memory of the fired film is

significantly better than of the un-fired film.

The optical density of the coloured state of the film was not uniform as expected but it is

believed that this is due to incomplete charge extraction under the experimental

conditions used. The optical density of the fired film in the bleached states was

observed to increase with progressive cycling however this was not necessarily due to

charge trapping. There is experimental evidence to suggest that not enough time was

allowed to completely bleach the film even though the bleaching current was very small

157

(~1µA/cm2) when the cycle was terminated. When the same film was repeatedly cycled

at room temperature after the self-bleaching experiment, charge extracted during

successive bleaching cycles was greater than that injected during coloration and the

optical density for the bleached states of the film was observed to decrease again,

approaching its original value. Stated quite simply, the fired film cycled in dry

electrolyte was unable to be bleached effectively within a reasonable time frame. The

difficulty in bleaching the film was exacerbated by leaving the film in the coloured state

for a prolonged period of time. Even though the film was not bleached completely

during cycling, the data trends are the same and the significance of the differences seen

when cycling a fired film is still obvious.

7.3.2 Effects of water on self-bleaching rates at elevated temperatures

The change in optical density is related to the injected charge density by the coloration

efficiency (η) hence the slopes of the ∆OD-time plots are proportional to the rates of

self-bleaching. Linear regression was carried out on the data from Figure 7.1 and Figure

7.2, as well as on previously reported self-bleaching data [9] in order to investigate the

effects of moisture and temperature on self-bleaching rate. The curves are

approximately linear from 5 minutes after the end of coloration, until the initialisation of

bleaching, and so this data was used for the determination of self-bleaching rates.

Figure 7.3 shows the change in self-bleaching rate as a function of temperature under

various conditions, where the rate is equal in magnitude but negative in sign to the slope

of the ∆OD-time plots.

158

0.0000

0.0005

0.0010

0.0015

0.0020

0.0025

0.0030

0.0035

0.0040

20 30 40 50 60 70 80

unfired film/moist electrolyteunfired film/dry electrolytefired film/dry electrolyte

Self-

blea

chin

g ra

te (-

d O

D/d

t)

Temperature (oC)

∆

Figure 7.3 Rate of self-bleaching for WO3/TiO2 films under various conditions.

It is evident from Figure 7.3 that the self-bleaching rate increases with temperature and

also with the amount of moisture present in the system. It also appears that water

present in the electrolyte plays a much larger role in the self-bleaching mechanism, than

the water inside the film which is removed upon firing. The fired film in the dry

electrolyte does appear to undergo some self-bleaching, however the rate is much lower

than for an unfired film in moist or dry electrolytes, and the temperature dependence is

minimal. Although the fired film cycles very reversibly even up to 70ºC and has a good

electrochromic memory, the rate at which the film bleaches is unacceptably slow unless

very high switching voltages are applied.

159

7.3.3 Effects of water on voltage characteristics

Figure 7.4 shows the voltage-time characteristics during room temperature cycling of

WO3/TiO2 films under various experimental conditions.

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

0 2 4 6 8 10 12 14 16


Volta

ge (V

)

Time (min)

Figure 7.4 Voltage characteristics during cycling of WO3/TiO2 films under various

conditions.

The films were cycled using a voltage-limited constant-current injection and extraction

technique described previously [12,13] whereby a constant current was used to colour

and bleach the films, until some voltage limit was reached. The cycling characteristics

of the films shown in Figure 7.4 are all very different, owing primarily to the different

environments used to conduct the experiments. The unfired film cycled in a moist

electrolyte [9] only required relatively small voltages to colour and bleach the film, and

was bleached in approximately 3 minutes. When a similar film was cycled in a very dry

160

electrolyte it took only 30 seconds longer to bleach, however the voltages required for

coloration and bleaching were much higher. The voltage limit during bleaching was

reached when approximately 12mC/cm2 was extracted, and the remainder of the inserted

charge was extracted at this constant voltage. The fired film cycled in the dry

electrolyte required very large voltages to insert and extract charge, and the voltage limit

on bleaching was reached after only 5mC/cm2 of the injected charge was extracted.

This meant that the remaining 10mC/cm2 of lithium ions was extracted under constant

voltage, taking 13 minutes to bleach the film while the current decayed exponentially.

Although these conditions enabled the film to be cycled very reversibly (ie. all of the

injected charge was eventually able to be extracted), the slow response time for

bleaching and the high voltages required for switching make them unpractical for use in

a commercial device.

The lower magnitude of the voltages required to colour and bleach the films when

moisture is present indicates that water plays a key role in the mechanism of charge

injection and extraction. It is interesting to note that the difference in switching

characteristics are more marked for the bleaching process, indicating that water plays a

more significant role in the bleaching mechanism than in the coloration mechanism. It

also seems that the water present in the electrolyte and the water inside the film play

quite separate roles in the electrochromic process.

It is possible that water inside the film assists in the diffusion of ions through the

substrate lattice, by providing a site for charge transfer and by causing swelling and

therefore lattice expansion. Another possibility is that adsorbed water may cation

exchange a proton for a lithium ion, thus forming LiOH and inserting a proton into the

161

lattice. As the protons are much more mobile than the lithium ions, we may reasonably

expect a reduction in the voltage required to maintain a constant current. It is likely that

mixed diffusion of protons and lithium ions is a major cause of the large variation in

switching characteristics between moist and dry conditions.

In order to further investigate the effects of water in the electrolyte on the device

switching characteristics, an unfired film was continuously cycled at room temperature

and water was periodically added to an initially dry electrolyte. Figure 7.5 shows the

way that the cycling characteristics changed when water was added in increments of

approximately 0.01%v/v at 30 minute intervals.

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2.00

0 50 100 150 200 250 300 350

no water

0.01% water

0.02% water

0.03% water

Volta

ge (V

)

Time (s)


concentration.

162

The addition of water to the dry electrolyte had little effect on the coloration voltage, but

considerably changed the bleaching characteristics. This implies that the large

difference between the coloration voltages when cycling unfired and fired films in a dry

electrolyte (Figure 7.4), is due to the presence of water inside the unfired film. We may

reasonably expect water in the electrolyte to change the properties of charge transfer at

the electrode surface. We may also expect water inside the film to affect the diffusion

properties of the lithium ions. As the addition of water to the electrolyte does not affect

the coloration voltage, it is likely that the diffusive motion of ions inside the electrode

limits the electrochromic system discussed here.

The bleaching voltage changes considerably when water is added to the electrolyte,

possibly by reacting with lithium ions at the electrode surface. This hypothesis is

supported by the fact that self-bleaching occurs even when it is not possible for an

electron current to flow, and that some of the injected charge cannot be extracted when

cycling in conditions where moisture is present. It is possible that by the time bleaching

is initiated, some of the charge injected during coloration has already been lost to a

reaction with water in the electrolyte, thereby reducing the optical density of the film

and reducing the quantity of charge remaining to be extracted.

163

7.4 Conclusion

Self-bleaching has been observed in electrochromic films of WO3/TiO2, in a liquid

electrolyte. The rate of self-bleaching was found to increase significantly with

temperature and with increased moisture in the system. Water present in the electrolyte

solution reduced the electrochromic memory of the film, and led to significant

irreversibility of the ion injection process. Water present in the film before cycling was

found to increase the ease with which coloration and bleaching occurred, possibly due to

the mixed diffusion of lithium ions and protons, but also led to some small irreversibility

of the electrochromic process. A fired film was cycled very reversibly in a dry

electrolyte and had an excellent memory, however the voltages required to switch were

very high under these conditions. The time required to bleach the film was also very

excessive, at around 15 minutes.

The results suggest that good electrochromic memory and reversibility are achieved at

the expense of response time, and vice versa. This information may be used in order to

design a device with enough water present to enable reasonably fast switching, while

minimising irreversibility enough to achieve a specific device lifetime.

164

Acknowledgements






165

REFERENCES [1] C. M. Lampert, IEEE Circuits and Devices, 8, 19-26 (1992).


Solar Cells, 56, 449-463 (1999).

[3] J.P. Randin, J. Electron. Mater., 7, 47-63 (1978).

[4] J.A. Duffy, M.D. Ingram and P.M.S. Monk, Solid State Ionics, 58, 109-114(1992).

[5] Q. Zhong, S.A. Wessel, B. Heinrich and K. Colbow, Solar Energy Materials, 20,

289-296 (1990).


[7] M. Burdis and J.R.Siddle, Thin Solid Films, 237, 320-325 (1993).

[8] J.-G. Zhang, D.K. Benson, C.Edwin Tracy, J. Webb and S. Deb, Proceedings of the

SPIE, Vol 2017, 104-112 (1993).

[9] J.P. Matthews, J.M. Bell and I.L. Skryabin, "High Temperature Behaviour of

Electrochromics", Renewables: The Energy for the 21st Century


Sayigh Ed.), 230-235 (2000).






437.


166


CHAPTER 8

GENERAL DISCUSSION

168

GENERAL DISCUSSION

8.1 Introduction and Identification of Knowledge Gaps

This chapter discusses the entire body of results collected in this PhD research as a

single collection, as opposed to as individual publications. The experimental research

was carried out in several stages, and each stage is discussed separately in sections 8.2

to 8.5. These sections are not necessarily presented in chronological order (as they are

in the publications), but are rather presented in the context of a cohesive research project

with specific aims and goals. Several knowledge gaps were identified in the Literature

Review (Chapter 2) and these can be summarised as follows:

1. How does temperature effect

a. Electrical response of EC films

b. Optical response of EC films

2. How is the reversibility of the EC process affected by

a. Temperature?

b. Water?

3. What are the causes of irreversibility and self-bleaching, and how can these

effects be minimised?

4. Can a single model be used to describe EC behaviour over a wide temperature

range (ie. 20ºC < T < 70ºC)?

169

Sections 8.2 to 8.5 therefore serve to discuss the research progress in the context of

addressing and systematically filling these four knowledge gaps in turn. Although this

chapter necessarily repeats discussion and conclusions made in previous chapters, it has

been more succinctly discussed to minimise repetition. The detail given here is then

more brief than for previous chapters, however reference is made to relevant chapters if

more detailed information is sought.

8.2 Initial Characterisation in the Ambient Laboratory Environment

Initial experiments involved cycling of WO3/TiO2 films using a constant current

technique between 20ºC and 50ºC, and Figure 8.1 shows the voltage response during

coloration and bleaching at elevated temperatures. The film was coloured with a

constant current for 150 seconds, and then bleached after a 30 second delay. The

magnitude of the voltages required for coloration and bleaching was reduced by

increasing the temperature, although the same amount of charge was injected each time.

Decreases in switching voltages were greatest for small deviations from room

temperature with the voltage data converging at high temperature. These results

illustrated that small increases from room temperature will result in significant changes

in the kinetic behaviour of the film, and a good switching regime should appropriately

reflect this.

170

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0 50 100 150 200 250 300 350 400

20.6oC

28.4oC

32.3oC

50.0oC

Volta

ge (V

)

Time (s)


bleaching of WO3 thin film electrode and plotted for four temperatures, for an injected


Under these conditions some irreversibility was observed at high temperatures,

identified by the inability to extract all the injected charge. The amount of charge that

could not be extracted increased monotonically with temperature, however the same

optical density was achieved for the bleached states. The small amount of charge

remaining in the films after bleaching at high temperature therefore did not contribute to

coloration. As the reversibility was found to decrease with temperature, a separate

experiment was carried out involving cycling to 5mC/cm2, in an attempt to reduce the

amount of charge trapped at high temperature. The emf for coloured and bleached states

of the film was measured after allowing sufficient time for diffusion processes to occur.

171

The magnitude of the emf was found to decrease with temperature, however there was a

hysteresis effect for results obtained at increasing and decreasing temperatures.

The coloration voltage (Vc), calculated by subtracting the emf from the measured

voltage, did not exhibit this hysteresis, suggesting that the hysteresis effect was

associated purely with the film emf. The maximum coloration voltage (Vc max) was

found to be related to temperature by Arrhenius type behaviour, with log Vc max

approximately proportional to reciprocal temperature as shown in Figure 8.2.

-0.95

-0.90

-0.85

-0.80

-0.75

-0.70

0.0030 0.0031 0.0032 0.0033 0.0034 0.0035

y = -3.0546 + 688.62x

R2= 0.87498

log

|V

|

(V)

1/T (K-1)

c m

ax

Figure 8.2 Log of absolute value of maximum coloration voltage versus reciprocal

temperature for WO3 film, for an injected charge density of 5mC/cm2.

Figure 8.3 shows the change in optical density (∆OD) versus injected charge density for

cycles performed up to 50ºC. The coloration efficiency is observed to decrease slightly

at elevated temperatures and it was hypothesised that effect was related to the issue of

172

irreversible charge injection. As the injected charge density increases during coloration,

the emf decreases. If charge was irreversibly accumulating in the film with progressive

cycling we would expect the magnitude of the emf for a given charge density to slowly

decrease. This was observed in this experimental work, which supports the theory of

irreversible charge injection and helps explain the hysteresis associated with the voltage

data.

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.0 4.0 8.0 12.0 16.0

20.6oC

28.4oC

32.3oC

50.0oC

OD

Charge injected (mC/cm2)

∆


at four temperatures.

This work showed that electrochromic voltage characteristics are heavily dependent on

temperature and also demonstrated that a constant current technique is superior to

constant voltage, for controlling electrochromics over a wide temperature range.

Constant current charge injection allows control of injected charge density and therefore

coloration level, regardless of temperature. Previous reports of electrochromic

173

experiments at high temperature focussed on response times, or only provided limited

data hence this research and paper 1 (Chapter 3) represent a significant contribution to

filing that particular knowledge gap (Knowledge gap 1 in section 8.1).

8.3 Dry-box Experiments

At the time of this initial research the causes of the charge injection irreversibility and

reduction in coloration efficiency were not known although it was suspected that water

may play a key role. All of the work discussed in section 8.2 was carried out in the

ambient laboratory environment. During this time, attempts were made to keep the

electrolyte solution free from water by using molecular sieves and a positive pressure of

dry nitrogen, however the high humidity of the location (Brisbane, QLD Australia)

combined with the highly hygroscopic nature of the electrolyte made this experimentally

difficult. In order to determine the effect of water on reversibility, the experimental

apparatus was moved inside a dry-box.

The dry-box was maintained in a stringently dry condition by exposing the internal

atmosphere to phosphorus pentoxide (P2O5) and recirculating nitrogen through

gas-drying units. Access to the experimental apparatus was made possible via three

glove ports and a load lock chamber. Solenoid valves controlled the flow of gas into

and out of the box in order to maintain a slight positive pressure inside the box at all

times and humidity was kept below 1 ppm absolute.

174

Initial experiments in the dry-box involved cycling of a WO3/TiO2 film to 15mC/cm2

between 20ºC and 50ºC. Films cycled in this fashion exhibited very reversible

behaviour and the coloration efficiency was found to be independent of temperature.

The coloration efficiency was determined to be 38.6cm2/mC, and the plot of ∆OD

(change in optical density) versus injected charge density was very linear with a Pearson

moment correlation coefficient (R2) of 0.99 (See Figure 4.1(b)). The large differences in

reversibility when cycling in ambient and dry conditions are illustrated by Figure 8.4,

which shows the percentage of injected charge unable to be extracted on bleaching. The

amount of irreversible charge clearly increases with temperature under ambient

conditions, and is significant even at 30ºC.

-3

-2

-1

0

1

2

3

4

0 10 20 30 40 50 60

Ambient conditionsDrybox experiment

Cha

rge

trapp

ed p

er c

ycle

(%)

Temperature (oC)


percentage of the injected charge density trapped per cycle.

175

The amount of irreversible charge in dry conditions is very small, and at high

temperatures more charge was removed than inserted for most of the cycles. This was

because the same voltage limits were applied for bleaching at all temperatures, which

has the effect of pushing the bleaching process progressively further as temperature

rises. During pre-experiment training of the WO3/TiO2 film, large amounts of charge

were irreversibly inserted however reversibility quickly improved after a few (~10)

cycles. The fact that the quantity of charge extracted exceeds the injected charge at high

temperatures it likely due to removal of some of the lithium ions inserted during the film

training process.

This body of work showed that the trends in voltage reduction with temperature which

were first observed in the ambient laboratory environment, are the same for very dry

conditions. This work also showed that films may be cycled much more reversibly in

dry conditions, even up to 50ºC, and that the coloration efficiency and ion injection

irreversibility are not dependent on temperature under these conditions. These facts

support the theory that water is responsible for the irreversibility issues observed in

ambient conditions. This work then provides a valuable contribution to filling

knowledge gap 2 (Section 8.1).

176

8.4 Self-bleaching Experiments

The work described in the previous section has shown that irreversibility increases with

water content and temperature. It does not however show why the irreversible charge

did not contribute to coloration and why the coloration efficiency decreased at elevated

temperatures. A set of self-bleaching experiments was devised in an attempt to answer

these questions and fill knowledge gap 3 (Section 8.1). These experiments involved the

coloration of a film in the electrolyte, followed by immediate disconnection of the

counter electrode. Under these conditions the film should not undergo any redox

reaction because there is no path for electron flow from the working electrode. We then

ideally expect the film to maintain a constant level of coloration independent of time, an

effect known as electrochromic memory. Figure 8.5 shows the change in optical density

versus time for a film self-bleaching in electrolyte in the ambient environment.

It is evident from Figure 8.5 that the film was slowly self-bleaching with time, and the

rate increases at higher temperatures. Voltage measurements showed a similar trend

(See Figure 4.3(b)) towards more anodic potentials indicating that the surface

composition of the film was slowly changing with time.

177

0.50

0.55

0.60

0.65

0.70

0.75

0.80

0.85

0 5 10 15 20 25 30 35 40

26.4oC

32.9oC

40.2oC

45.8oC

50.5oCO

D

Time (min)

∆

Figure 8.5 Change in optical density with time for a film coloured to 15mC/cm2 at

various temperatures.

After the 30 minute self-beaching period the counter electrode was connected and the

cell bleached as per usual. The amount of charge that could not be extracted from the

films was proportional to temperature and at 50.5ºC approximately 4.3mC/cm2 (of the

20mC/cm2 injected) could not be extracted from the film. This self-bleaching did not

occur by the normal electrochromic process because the electrode was isolated from the

cell, and so must be attributed to some side reaction. It is likely that this side reaction

involves water and lithium ions and this reaction would be at least partially responsible

for the observed irreversibility.

Chemical analyses were carried out to gain information about the self-bleaching

mechanism and to determine the location of the missing lithium ions. A WO3/TiO2 film

178

that had been used for a self-bleaching experiment was dissolved off the glass/ITO

substrate and the quantity of lithium and tungsten present was determined by inductively

coupled plasma-atomic emission spectroscopy (ICP-AES). The results showed that the

amount of lithium present in the bleached film corresponded to approximately

15mC/cm2. The total accumulation of charge not extracted after many cycles was

almost 130mC/cm2 so only a small portion of the total irreversible charge remained

inside the film. The large remainder of the lithium was then presumably lost to the

electrolyte by reaction of water (in the electrolyte) with lithium at the film surface,

possibly causing some film dissolution. This work showed that both water and

temperature play key roles in the reversibility of the EC process, and that water

promotes self-bleaching and is closely associated with irreversibility.

At this stage, it was desirable to show that self-bleaching did not occur appreciably

under very dry conditions, thereby confirming the connection between water and the

reduction in coloration efficiency and hence electrochromic memory. Self-bleaching

experiments were then carried out in the dry-box with the expectation that no self-

bleaching would be observed, however this result was not seen. Figure 8.6 shows the

change in optical density versus time for a self-bleaching experiment in the dry-box.

179

0.45

0.50

0.55

0.60

0.65

0.70

0 5 10 15 20 25 30 35 40

24.3oC

34.8oC

45.2oC

55.1oC

65.3oC

75.3oCO

D

Time (min)

∆

Figure 8.6 Change in optical density during self-bleaching of a WO3/TiO2 film in dry

electrolyte.

It is obvious that self-bleaching still occurred even in the dry-box and the rate increased

with temperature, however the rates are considerably lower than for in the ambient

environment. The voltage-time data (See Figure 7.1(b)) confirmed that the surface

composition of the film was actually changing with time and the rate was also

proportional to temperature.

The fact that self-bleaching occurred even in a very dry electrolyte suggested that

perhaps some water was still present inside the film, so another film from the same

batch was fired in a furnace at 250ºC for eight hours, and then used in a similar

experiment. The results from self-bleaching of a fired film are shown in Figure 8.7.

After firing the self-bleaching characteristics of the film were very different. The rate of

180

self-bleaching was very small and the optical density remained essentially constant even

at 58ºC.

0.35

0.40

0.45

0.50

0.55

0.60

0 5 10 15 20 25 30 35 40

28.6

35.5

43.8

50

58.1

71.7

Time (min)

OD

∆

(a)-1.20

-1.10

-1.00

-0.90

-0.80

-0.70

0 5 10 15 20 25 30 35 40

28.635.5

43.850

58.171.7

Volta

ge (V

)Time (min)

(b)


of a WO3/TiO2 film, after firing at 250ºC for hours.

The large amount of noise present in the 71.7ºC optical data was due to the heater being

used to maintain the temperature (as described in section7.3.1), and does not accurately

represent the optical properties of the film. The voltage data is free from this noise as

seen in Figure 8.7(b) and the voltages are indeed very flat with time. The slight

curvature of the low temperature data was caused by not allowing sufficient time for

diffusion processes to occur (as discussed in section 7.3.1), a conclusion supported by

the fact that the voltage-time slopes are almost constant for long times. The slopes of

the voltage-time lines for 35.5ºC and 71.1ºC are almost the same, indicating that the rate

of self-bleaching for the fired film is almost independent of temperature. This

conclusion is also supported by the fact that the optical density values do not diverge for

181

long times, as they do for the unfired films (See for comparison Figure 8.5 and Figure

8.6).

Self-bleaching rates were calculated by linear regression of the optical density-time

curves for experiments carried out in the ambient environment, and for fired and unfired

films in the dry-box (See section7.3.2). The dependence of the self-bleaching rates on

water and temperature is illustrated by Figure 8.8.

0.0000

0.0005

0.0010

0.0015

0.0020

0.0025

0.0030

0.0035

0.0040

20 30 40 50 60 70 80


Self-

blea

chin

g ra

te (-

d O

D/d

t)

Temperature (oC)

∆

Figure 8.8 Rate of self-bleaching for WO3/TiO2 films under various conditions and

temperatures.

It is clear from Figure 8.8 that the rate of self-bleaching increases with both water and

temperature. It is also apparent that water present in the electrolyte plays a different role

182

in the self-bleaching mechanism, to water present in the film. Self-bleaching is

detrimental to the normal operation of an EC system, it would seem that the ideal

conditions include very dry electrodes cycled in very dry electrolytes. The fired film

cycled in the dry electrolyte does indeed have very good electrochromic memory and

reversibility even at high temperature, however the response times for bleaching are

excessively long. This effect is demonstrated more clearly by looking at the voltage-

time characteristics for cycles carried out at room temperature, under the various

conditions described above. Figure 8.9 shows the measured voltage during cycling of

three films coloured to 15mC/cm2, under the same set of conditions as for Figure 8.8.

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2.00

2.50

0 2 4 6 8 10 12 14 16


Volta

ge (V

)

Time (min)

Figure 8.9 Voltage characteristics during cycling of WO3/TiO2 films to 15mC/cm2

under various conditions.

The film cycled in the ambient laboratory environment only required very small

183

voltages for coloration and bleaching, and was bleached in approximately three minutes.

Cycling a similar film in a dry electrolyte significantly increased the voltages required

for coloration and bleaching, although the bleaching step only took 30 seconds longer.

Cycling the fired film in dry electrolyte required very high voltages for coloration and

bleaching and the bleaching step took approximately 13 minutes.

The constant current technique used for this cycling involved the application of

0.1mA/cm2 until some voltage limit was reached. In the bleaching step, the voltage

limit was determined in-situ by limiting the rate of change voltage with time, as

described in section 7.2.2. Only 5mC/cm2 of the injected charge was extracted before

the bleaching voltage limit was reached for the fired film in dry electrolyte. This means

that the remaining 10mC/cm2 of charge was extracted under a constant voltage, while

the current decayed exponentially. Cycling in this manner enabled the charge to be

extracted very reversibly (ie. all of the inserted charge could eventually be removed with

time), however the slow response and high switching voltages required, make this

electrochromic system of little practical use.

It is possible that water inside the film assists in diffusion of ions thereby reducing the

voltage required for cycling. It is also possible that water inside the film could cation

exchange a proton, thereby trapping a lithium ion by forming LiOH. The high mobility

of protons would allow much faster switching, thereby requiring smaller voltages to

maintain a given current density. It is then likely that mixed diffusion of protons and

184

lithium ions occurs in electrodes where water is present, contributing to the large

difference in switching voltage characteristics for moist and dry conditions.

Another experiment was carried out in order to determine the effect of electrolyte water

on cycling conditions. An unfired film was cycled continuously to 15mC/cm2 in dry

electrolyte and water was periodically added to the electrolyte in small increments (See

section 7.3.3). Figure 8.10 shows the voltage-time characteristics for the film, for

several additions of water to the electrolyte.

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2.00

0 50 100 150 200 250 300 350

no water

0.01% water

0.02% water

0.03% water

Volta

ge (V

)

Time (s)


concentration.

185

The coloration voltage is essentially unaffected by addition of water to the electrolyte,

however the ease with which bleaching occurs is considerably increased. This implies

that the significant difference in coloration voltages for fired and unfired films in a dry

electrolyte (Figure 8.9) is due to water inside the film, and not caused by electrolyte

water. It is therefore likely that the coloration process for the system investigated is

limited by diffusion, rather than charge transfer at the electrode/electrolyte interface.

It is also probable that the reduction in bleaching voltages observed in Figure 8.10 is due

to reaction of water with lithium ions at the film surface. Self-bleaching occurs even

when it is not possible for an electron current to flow, which further supports this

hypothesis and would also account for the ion injection irreversibility associated with

the self-bleaching process. According to this scheme, at the time when bleaching is

initiated, some of the injected charge has already been lost to reaction with electrolyte

water, thereby reducing the amount of charge available for extraction and reducing the

films optical density.

This work shows a clear correlation between the irreversibility, self-bleaching and the

presence of moisture. Films may be cycled very reversibly, with little self-bleaching

under very dry conditions, however this is achieved at the expense of response time.

This suggests that good electrochromic memory and fast response are mutually

competitive aims, and perhaps some compromise is needed, and this work goes a long

way towards filling knowledge gap 3 (Section 8.1).

186

8.5 Simulation and Estimation of Ionic Mobility

A major objective of this work was to simulate electrochromic voltage response to a

coloration current at elevated temperatures, to enable prediction of EC behaviour and

improve device design and control strategy and hence contribute to filling knowledge

gap 4 (Section 8.1). Simulation of data involves modelling of the ion distribution during

coloration but if the process is not reversible, charge may accumulate and there is no

simple way to tell how much charge is in the film for a given cycle. The results from

initial experiments in the ambient environment were therefore not suitable for

modelling, but dry-box experiments provided good data for this purpose.

The simulation model reported by Wang [1] elaborated on, and combined parts of

previous electrochromic models. Wang used this model to simulate voltage

characteristics for cycling of a film to various charge densities at various rates. The

model simulated the data well at room temperature, but no results were reported for

higher temperatures. The model is discussed in detail in Chapter 5 so this section will

focus more closely on the results of data simulations at high temperature. The

simulation model may be described as a function of the surface lithium concentration

(c(0,t)) by the equation

−

−++=),0(.1

),0(.ln),0('.)(

tcVtcV

FRTtcVbaRitV

m

mmcca

ν (8.1)

which is a modified form of equations first reported by Nagai et al [2] and Crandall et al

[3]. Rc is the series resistance of coloration, ic is the switching current and a, b' and ν

are constants discussed in section 5.2.3. Vm is the molar volume of the film and

187

−=

dydEbb' where dE/dy is the slope of the coulometric titration curve. The surface

lithium concentration c(0,t) is the charge density in moles per unit volume at the

electrode/electrolyte interface and is determined from the equations

Γ=DnF

jtc 2),0( (8.2(a))

and ππt

Dtlkt

Dtklerfc

Dkl

kk

−−−=Γ ∑∑∞

=

∞

= 1

22

1

)][exp(2]2[ (8.2(b))


F is Faradays constant and D is the chemical diffusion coefficient of lithium ions.

All the variables in equation 8.2 were known, except for diffusion coefficient. The

modelling process then involved estimation of diffusion coefficients for lithium at

various temperatures in order to solve equation 8.2(a) and (b) and calculate the surface

lithium concentration for given time (t). The variation in surface lithium concentration

c(0,t) was then used to solve equation (8.1) and estimate the applied voltage as a

function of time.

The experimental data for the modelling component was collected by cycling a

WO3/TiO2 film to 15mC/cm2 at temperatures ranging from 20ºC to 50ºC. A least

squares method was used to determine the best fit to the 20ºC data by adjusting the

constants a, b, ν, Rc and D. The voltage/time data colouration at the other temperatures

was then simulated by holding the values for a, b, ν and Rc constant and only changing

the diffusion coefficient. Figure 8.11 shows the experimental data and simulations for

188

the lowest and highest temperatures studied (Graphs for the other temperatures are

shown in Figure 5.5).

-1

-0.9

-0.8

-0.7

-0.6

-0.5

-0.4

-0.3

0 50 100 150

Exp 20.1oCSim 20.1oC

Volta

ge (V

)

Time (s)

(a)

-0.9

-0.8

-0.7

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

0 50 100 150

Exp 50.0oC

Sim 50.0oC

Volta

ge (V

)

Time (s)

(b)


electrochromic film. (a) Temperature = 20.1ºC, D = 2.07x10-12cm2 and

(b) temperature = 50.0ºC, D = 1.71x10-11cm2.

It is evident from Figure 8.11 that the theoretical voltage characteristics predicted with

the model described above are a very close approximation to the experimental data. The

diffusion coefficients estimated from the simulation process increased with temperature

as we may predict from thermodynamic and kinetic considerations, with an Arrhenius

dependence on temperature. Figure 8.12 shows an Arrhenius plot of lnD versus

reciprocal temperature, which was used to calculate the activation energy for diffusion

of lithium ions of 0.58eV.

189

-27.0

-26.5

-26.0

-25.5

-25.0

-24.5

0.003 0.0031 0.0032 0.0033 0.0034 0.0035

y = -3.585 -6767.8x

R2= 0.94663

ln D

1/T (K-1)


with temperature between 20.1ºC and 50.0ºC.

The plot shown in Figure 8.12 does appear to deviate from linearity at high temperature,

but the cause of this was unclear at the time of this work. The deviation from linearity

may be due to limitations and assumptions inherent to the model as discussed in section

5.5, and it was hoped that data recorded at higher temperature may help understand this

behaviour. This research work represents the first time a relatively simple model was

used to simulate voltage characteristics of an electrochromic system at elevated

temperatures. Other models have failed to describe the large change in switching

voltages when temperature increases, so this work was a significant improvement in

terms of electrochromic modelling. The paper resulting from this research was also the

first to report the estimation of ionic mobility from voltage-time data, which makes this

technique a valuable research tool.

190

Film cycling experiments were repeated between 36ºC and 76ºC in order to test the

model over a more realistic temperature range and also to attempt to resolve the issue of

high temperature deviation from Arrhenius behaviour. Films were again coloured to

15mC/cm2 and voltage characteristics recorded. The results were simulated in a similar

fashion to the experiment described above (See section 6.2.3 for details and plots).

Diffusion coefficients were again within the approximate range expected although the

35.6ºC diffusion coefficient of 2x10-12cm2/s was quite low. This result indicated that the

switching characteristics are very dependent on ionic mobility, with low voltages

required to switch films with high ionic mobility. The estimated diffusion coefficients

again increased with temperature and obeyed an Arrhenius law, as illustrated by

Figure 8.13.

-28.0

-27.0

-26.0

-25.0

-24.0

-23.0

0.0029 0.0029 0.0031 0.0031 0.0032

y = -11497x + 9.052

R2= 0.961

ln D

1/T (K-1)


with temperature between 35.6ºC and 76.4ºC.

191

Figure 8.13 was used to determine the activation energy for diffusion of lithium ions of

0.99eV. This value is significantly higher than that for the previous modelling

experiment, however this result is in accord with the smaller lithium diffusion rates for

the film used this experiment.

The Arrhenius plot shown in Figure 8.13 still deviates from linearity at high

temperature, however this time it is in the other direction. It is likely that the model

used is accurate for lower temperatures because under these conditions diffusion

processes are slower, and then more likely to be the limiting mechanism. The model

used assumes that the coloration process is controlled by the diffusion-limited motion of

ions and a series resistance. As temperature increases, diffusion becomes very fast and

the applied voltages are much lower. Under these conditions, it is likely that charge

transfer is a more significant rate limiting mechanism and this is a possible reason for

deviations from linearity observed in Figure 8.12 and Figure 8.13. It is likely that the

model could be improved by including a term to describe the charge transfer resistance

with temperature and this model would ideally describe the gradual change in rate-

limiting mechanisms as temperature increases.

This research work and resulting paper (Chapter 6) is significant because it is the first to

address the possibility that the rate limiting mechanism of the electrochromic process is

dependent on temperature, and also is an important contribution to knowledge gap 4

(Section 8.1).

192

8.6 Conclusions

The kinetic behaviour of electrochromic tungsten/titanium mixed oxide ion-insertion

electrodes is largely affected by both temperature and water. Temperature affects the

properties in several ways. Firstly it increases ionic diffusion rates, which makes faster

switching possible and requires lower switching potentials for a specific injected charge

density. Increased temperature also increases the ease with which side reactions occur

and therefore may cause detrimental effects such as self-bleaching, ion injection

irreversibility and decreased electrochromic memory, if water is present.

The presence of water promotes ion injection irreversibility and causes self-bleaching to

occur with time, thereby reducing the optical density and progressively lowering the

maximum contrast ratio that may be achieved. The presence of water may also increase

the ease with which bleaching is performed, possibly due to mixed diffusion of protons

as well as the injected lithium ions. Electrochromic films may be cycled very reversibly

at temperatures as high as 70ºC if very dry conditions are maintained, but the bleaching

rate is then significantly reduced.

These results suggest that achieving a good electrochromic memory and high

reversibility may come at the cost of increased response time. Conversely a device with

moisture present may require small voltages for switching and respond very quickly, but

reversibility will be lowered thereby limiting device lifetime. It is possible that a

successful commercial electrochromic device may only be produced when a fine

balance between reversibility and response time is achieved.

193

This work has also shown that a single mathematical model may be used to describe the

kinetics of the coloration process in sol-gel deposited tungsten-titanium oxide films. A

large advantage of this process is that ionic mobility can be estimated in the process.

The ability of the model to accurately describe the electrical response is poorer at

elevated temperatures (>50ºC) possibly due to a change in rate limiting mechanism from

one of diffusion-limited motion of ions, to being charge-transfer limited. Even though

there is some small discrepancy between real and simulated data at these temperatures,

the difference is small (ie. ~10mV). The model is therefore suitable for use in

determination of suitable control strategies for electrochromic devices, over a

temperature range suitable for real practical use.

8.7 Future Research

Several knowledge gaps have been at least partially filled as a result of this doctoral

project, however several more questions have arisen. A large area for future research

relates to specific information regarding the mechanisms of the degradation processes,

in particular irreversible ion insertion and self-bleaching.

The location of ions which may not be extracted is not known, as well as the reason for

these ions not contributing to the overall film coloration. It is possible that the ions

could for example be trapped at grain boundaries, and if this were so it may be possible

to improve the morphology of the film to reduce these detrimental effects. It would be

very useful to know the location of all lithium ions which can not be extracted after film

194

coloration. It is not possible to look for these ions in the electrolyte, because the

concentration of lithium is so high initially (ie. 1MLiClO4 in propylene carbonate). If

indeed the ions are lost to the electrolyte by reaction of water with surface LiWO3 sites,

we would reasonably expect to see some small concentration of tungsten in the

electrolyte solution. This experiment would be relatively simple, yet provide valuable

information regarding the mechanism of the irreversibility.

Perhaps some chemical modification of the film composition may also help stabilise the

film against dissolution, while still allowing reasonably fast ionic insertion and

extraction.

It would also be beneficial to extend the model described here to consider the bleaching

process as well as the charge transfer step at high temperatures. The situation of a

complete electrochromic cell is considerably more complex than for a single film, and a

large amount of work is required to adapt the model for this use.

A future goal would be to use the model to build an electrochromic controller which can

use cycling information and external parameters such as temperature and light levels, to

automatically control EC devices. Such a system could optimally control an EC device

in a manner so that excessive voltages are never applied, and some requirements such as

internal temperature (of a building) or light level are consistently met. It is likely that

EC device will always slowly degrade, but the rate of degradation need only be reduced

sufficiently to achieve a lifetime great enough to for allow successful commercialisation.

195

REFERENCES




1411 (1976).

Jeremy P. Matthews - QUT ePrintseprints.qut.edu.au/15786/1/Jeremy_Matthews_Thesis.pdf · i Abstract...

Documents

Transcript of Jeremy P. Matthews - QUT ePrintseprints.qut.edu.au/15786/1/Jeremy_Matthews_Thesis.pdf · i Abstract...