t
:C . .-

1. 31...“ a a .

.
. ....~,..,.,.2

 

 

 

hunks” . 1?.
#3. am»

 

 

. “In?
.L ._..r.
2.24.. I

 

 

Lu r

.6 «gm.

a; . ,
Y»... sum:
it:
.711; . 3i.
. 15.. 1:1. 3.
5...... .x (5.x .{

. t.
$.41? . .
e. t.
. Jar! . a
.1, 7
I h "0..
.

15%.... .. . .

taut-AIL

. . . . . . , . :2:
, x: I. . .I.
A. 1. ..

I‘ 51 . v-
L: a». in.) 3
3.2 2.112.... .

 

.Ezwvu. .. . .. , Lana... . . . . . . . w .2.......puua.4......

 

‘IrIt‘ ‘1‘...

2:2”)

This is to certify that the
dissertation entitled

INVESTIGATION OF ALUMINUM CORROSION
IN LITHIUM-ION BATTERY ELECTROLYTES
- INFLUENCE OF WATER CONTAMINATION

presented by

Mateusz Lukasz Hupert

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Chemistry

 

 

V yajor Professor’s Signature

12-52 «(3 00;;

 

Date

MSU is an Affirmative Action/Equal Opportunity Institution

‘ v v- V

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN Box to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClRC/DateDuo.p65-p,15

INVESTIGATION OF ALUMINUM CORROSION
IN LITHIUM-ION BATTERY ELECTROLYTES
- INFLUENCE OF WATER CONTAMINATION

By

Mateusz Lukasz Hupert

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Chemistry

2002

ABSTRACT

INVESTIGATION OF ALUMINUM CORROSION IN LITHIUM-ION
BATTERY ELECTROLYTES - INFLUENCE OF WATER
CONTAMINATION

By

Mateusz Lukasz Hupert

The susceptibility of high purity (99.999%) aluminum foils to oxidation (formation
of a passivation layer) and/or corrosion (pitting and metal loss) during room tempera-
ture polarization in LiPF6/ethylene carbonate-dimethyl carbonate (EC-DMC),
LiPF6/propylene carbonate (PC), LiClO4/EC-DMC, and LiClO4/PC was investigated.
The effect of water on the corrosion process was studied through controlled additions
(50-2000 ppm) of the water impurity to the electrolyte/solvent mixture. The testing of
two different metal surfaces was performed, (i) aluminum with the native oxide film
removed by mechanical polishing under an inert, argon atmosphere, and (ii) aluminum
covered with the oxide layer formed by electropolishing in aqueous HBF4 solution. The
susceptibility of aluminum toward corrosion in each of the electrolyte/solvent mixtures
was assessed with electrochemical techniques, including cyclic voltammetry (CV),
chronoamperometry (CA), and open circuit potential measurements (OCP). Microscopic
techniques such as Optical microscopy (OM), atomic force microscopy (AFM), and
scanning electron microscopy (SEM) were used to study morphological changes of the

aluminum surface brought about by the imposed electrochemical conditions. X-ray

photoelectron spectroscopy (XPS), energy dispersive x-ray analysis (EDS), and Raman
spectroscopy were used to probe the chemical composition of the surface before and after
polarization in non-aqueous electrolyte/solvent mixture. Electrochemical quartz crystal
microbalance (EQCM), electrochemical atomic force microscopy (EC-AFM), and
electrochemical optical microscopy CBC-OM) were employed to study
corrosion/passivation processes in situ, during anodic polarization.

Obtained results indicate that none of the investigated electrolyte/solvent systems is
corrosive toward aluminum in the absence of added water (below 50 ppm), at least up to
5 V vs. Li/Li+. A solid passivation layer forms on the surface at such conditions. The
passivation layer is composed primarily of aluminum oxide and either aluminum
oxyfluoride in LiPF6 solutions or aluminum perchlorate in LiClO4 solutions. An oxide
film present on the electropolished aluminum surface improves the corrosion resistance
of the metal.

In the presence of added water, anodic polarization of aluminum in LiClO4/EC-
DMC and LiClO4/PC leads to pitting and progressive corrosion. A thick layer of
corrosion products (mostly Al(ClO4)3 and Al(OH)3) forms on the mechanically polished
aluminum surface. An oxide film, present on the electropolished surfaces, does not
prevent the underlying metal from being corroded in LiClO4 solutions with added water
impurity. At water levels below 500 ppm the corrosion products deposit in and fill the
pits formed under these conditions. At higher water levels (1000 - 2000 ppm) empty pits
form, due to evolution of gaseous oxidation products. Addition of water impurity to
LiPF6/EC-DMC and LiPF6/PC, does not cause pitting and corrosion of aluminum. The
surface remains effectively passivated by an aluminum oxide/oxyfluoride/hydroxy-

fluoride layer.

ACKNOWLEDGMENTS

Last five years were incredible learning experience for me. It would not be possible
without help, encouragement or simply good word from many people I have met during
that time, and whom I would like to acknowledge.

I would like to thank my advisor, Dr. Greg Swain, for the guidance and support
during my studies. Thank you, Sir, for giving me all this freedom and letting me learn on
my own mistakes. This is, possibly, the best knowledge one can get.

I would also like to thank the present members of my committee: Dr. Bruening,
Dr. Ayres, and Dr. Smith. Thank you for all the comments and suggestions concerning
my research. I have special thanks to Dr. Brown, Dr. Bialkowski, Dr. Parker,
and Dr. Dennison who served as my graduate committee during the three years at Utah
State University. Thank you for all the help and scientific discussions, so important at the
beginning of the graduate studies. I would also like to acknowledge Prof. Strojek,
Dr. Zak, and Dr. Koppang for showing me a different way one can look
at electrochemistry.

I would like to thank my wife, Gosik, for her love and support. Thank you for going
with me through all the ups and downs during our first five years, from a teaching
assistant workshop at USU, through winters of Michigan, to our first taste of Louisiana.

Special thanks belong to the former and the present members of the Swain’s research
group: Mike, Chen, Jishou, Jian, Shannon, Prerna, Grace, Gloria, Jason B., Jason S., Jin

Woo, Show, Josef, Zuzka, and Doug. Thanks guys. I really appreciate your friendship

iv

and, of course, a chance to taste good food from all over the world. JB, thanks for
showing me that there is a hope for soccer fans in this country.
I would also like to thank Wiesia and Bogdan J uszczakiewicz for their friendship and
providing this great home-like feeling so many thousands miles away from Poland.
Finally and most importantly, I would like to thank my family. Your love, faith and
never-ending encouragement meant a lot to me during these years. Mom and Dad thank
you for your wisdom. Thank you for sending me to college with a major goal in mind: to

broaden my horizons. I am sure, I did. Thank you.

Mateusz

TABLE OF CONTENTS

 

 

 

page

LIST OF TABLES x

LIST OF FIGURES xii
CHAPTERS

1. Introduction 1

1.1. Lithium-ion battery .................................................................................................. 2

1.1.1. Overview ......................................................................................................... 2

1.1.2. Capacity fade phenomenon ............................................................................. 5

1.1.3. Current collectors ............................................................................................ 7

1.2. Corrosion of aluminum in non-aqueous solvent/electrolyte mixtures commonly

 

employed in lithium-ion batteries — literature review .............................................. 9

1.3. Passivation of metals in solution — review of possible mechanisms ..................... 16
1.4. Outline of the dissertation ...................................................................................... 20

2. Experimental Section 22
2.1. Preparation of aluminum electrodes ...................................................................... 22
2.1.1. Mechanical polishing ..................................................................................... 22
2.1.2. Electropolishing ............................................................................................. 22

2.2. Surface characterization techniques ....................................................................... 23
2.2.1. Atomic force microscopy (AFM) .................................................................. 24
2.2.2. Optical microscopy (OM) .............................................................................. 25

vi

2.2.3. Scanning electron microscopy (SEM) and energy dispersive x-ray micro-

analysis (EDS) ............................................................................................... 28

2.2.4. X-ray photoelectron spectroscopy (XPS) ...................................................... 28
2.2.5. Raman spectroscopy ...................................................................................... 28
2.3. Electrochemical measurements .............................................................................. 29
2.4. Electrochemical quartz crystal microbalance (EQCM) ......................................... 30
2.5. Chemicals ............................................................................................................... 32
2.6. Remaining information .......................................................................................... 32
2.6.1. Data analysis and plotting .............................................................................. 33
2.6.2. Determination of water content inside glove boxes ...................................... 33

3. Electrochemical Investigation of the Corrosion/Passivation of High-Purity

Aluminum in LiClOJEC-DMC, LiClOJPC, LiPFJEC-DMC, and LiPF6/PC .....=34
3. 1. Introduction ............................................................................................................ 34
3.2. Results and discussion ........................................................................................... 36
3.2.1. Pretreatment and characterization of Al electrodes ....................................... 36
3.2.2. Open circuit potential (OCP) measurements ................................................. 39
3.2.3. Cyclic voltammetry in LiPF6/PC and LiPFdEC-DMC ................................. 41
3.2.4. Cyclic voltammetry in LiClO4/PC and LiClO4/EC-DMC ............................ 45
3.2.5. Chronoamperometry ...................................................................................... 48
3.2.6. XPS analysis of aluminum surface ................................................................ 51
3.2.6. Stability of passivation layer ......................................................................... 55

3.2.7. Passivation of aluminum in LiClO4/EC-DMC, LiClO4/PC, LiPFdEC-DMC,

and LiPF6/PC — mechanism ........................................................................... 56

3.3. Conclusions ............................................................................................................ 59

vii

4. Investigations of Aluminum Corrosion in LiPFdEC-DMC, LiPFdPC, LiClO4/

 

EC-DMC and LiClOJPC - Influence of Water Contamination 60
4.1 . Introduction ............................................................................................................ 60
4.2. Results and discussion ........................................................................................... 62

4.2.1. Corrosion of aluminum in LiClO4/EC-DMC/HZO and LiClO4/PC/I-120

mixtures -— mechanically polished surface ..................................................... 62

4.2.2. Corrosion of aluminum in LiClO4/EC-DMC/1120 and LiClO4/PC/HZO

mixtures — electropolished surface ................................................................ 69

4.2.3. Corrosion of aluminum in LiClO4/EC-DMC/I-120 and LiClO4/PC/H20

mixtures — chronoarnperometric studies ........................................................ 76

4.2.4. Chemical composition of the corrosion products in LiClO4/EC-DMC/H20
and LiClO4/PC/HZO mixtures — energy dispersive x-ray microanalysis
(EDS) and Raman spectroscopy studies ........................................................ 81

4.2.5. Mechanism of aluminum corrosion in LiClO4/EC-DMC/H20

and LiClO4/PC/H20 mixtures ........................................................................ 90

4.2.6. Corrosion of aluminum in LiPFg/EC-DMC/HZO and LiPFdPC/I-IZO

mixtures ......................................................................................................... 93

4.3. Conclusions ............................................................................................................ 96

5. Aluminum Corrosion in LiPFdEC-DMC and LiCIOJEC-DMC — Invstigatiom Using

 

Atomic Force Micmsoopy (AFM) 98
5. 1. Introduction ............................................................................................................ 98
5.2. Results and discussion ......................................................................................... 100
5.2.1. Ex-situ atomic force microscopy — dry electrolytes .................................... 100
5.2.2. Ex situ atomic force microscopy - LiPF6/EC-DMC/HZO ........................... 103
5.2.3. Ex situ atomic force microscopy - LiClO4/EC-DMC/I-120 ......................... 106

viii

5.2.4. In situ atomic force microscopy - LiClO4/EC-DMC/HZO .......................... 110

5.3. Conclusions .......................................................................................................... l 17

6. Aluminum Corrosion in IjPFdEC-DMC and LiCIOJEC-DMC - Investigations Using

 

Electrochemical Quartz Crystal Microbalanoe (EQCM) 119
6.1. Introduction .......................................................................................................... 1 19
6.2. Results and discussion ......................................................................................... 124

6.2.1. OCP measurements ...................................................................................... 125

6.2.2. EQCM measurements in LiPF6/EC-DMC and LiClO4/EC-DMC

— mechanically polished surface .................................................................. 126

6.2.3. EQCM measurements in LiPFg/EC—DMC and LiClO4/EC-DMC

— untreated surface ....................................................................................... 132

6.2.4. EQCM measurements in 1M LiClO4/EC-DMC and 1M LiPFg/EC-DMC

 

— influence of water contamination ............................................................. 135

6.3. Conclusions .......................................................................................................... 141

7. Summary 142
REFERENCES 145

 

ix

LIST OF TABLES

Table page

1.1. Characteristics of commonly used rechargeable batteries. a — depends on regular
maintenance; b — at room temperature; c — C-rate is a current that is numerically

equal to the Ah rating of the cell. Data adapted from Buchmann.23 ........................... 4

3.1. Typical results of OCP measurements. ...................................................................... 40

3.2. Summary of the cyclic voltammetric and chronoamperometric polarization
data for both mechanically polished and electropolished aluminum. The cyclic
voltammetric data (a) are for the first scan and the chronoamperometric data

(b) are for a 1h potential step to 5000 mV vs. Li/Li+. ............................................... 50

3.3. Summary of the XPS data for mechanically polished aluminum surfaces before
and after 1 h potentiostatic polarization at 5 V vs. Li/Li+. The following peaks
were used for determination of relative atomic abundance: Al3+ (2p, 75 eV),
A10(2p, 71 eV), Li (ls, 56 eV), C (Is, 285 eV), 0 (Is, 532 eV), C1 (2p, 209 eV),
F (Is, 686 eV), and P (2p, 135 eV). ........................................................................ 52

4.1. Summary of the potentiodynarnic cycling data for mechanically polished
aluminum in LiClO4/EC-DMC and LiClO4/PC with different levels of added
water. Scan rate = 5 mV/s. The potential range scanned was from 2.5 to 5.0 V

vs. Li/Li+. (***) - Pitting was not observed. ............................................................. 65

4.2. Summary of the chronoamperometric data for mechanically polished and
electropolished aluminum. Potential step from 2500 to 5000 mV vs. Li/Li+,
for 1 h. ........................................................................................................................ 81

6.1. Chemical composition and W/z value for possible passivation film species ........... 124

6.2. Summary of OCP-EQCM data for mechanically polished and oxide coated

aluminum electrodes. a) — at l-h mark, b) — during the 1-h period. ........................ 126

xi

LIST OF FIGURES

Figure page
1.1. Schematic diagram of lithium-ion rechargeable battery and illustration of electro-
chemical process occurring during charging. M stands for Ni, Co or Mn. Figure
adapted from Megahed and Scrosati.22 ....................................................................... 3
1.2. Half-cell discharge curves for a lithium-ion battery showing various capacity fade
processes. Adapted from Arora et at?4 ....................................................................... 6
1.3. Schematic diagrams of possible passivation mechanisms for metal (M) in
. 29
a solution. ............................................................................................................... 18
2.1. (A) Experimental setup for the ECAFM measurements including, 1 - glove box;

2.2.

2 — ECAFM microscope head; 3 — Optical microscope; 4 — computer with monitors
5 & 6; 7 — optical microscope monitor. (B) 8 - ECAFM fluid cell; 9 — Li/Li+
reference electrode; 10 — solution inlet with syringe; ll - solution outlet with
built-in Pt-wire auxiliary electrode and collection vessel. (C) Schematic diagram

of how the LilLi+ reference electrode was prepared .................................................. 26

Diagram of the electrochemical cell used for the in-situ optical microscopy.

(A) side view and (B) top view. 1 — microscope objective lens; 2 — metal clamp
with screws; 3 — optical window; 4 — Teflon® body; 5 — lithium reference electrode
port; 6 — solution inlet; 7 — solution outlet; 8 — aluminum disc electrode;

9 — Chemraz® o-ring; 10 - copper current collector; 11 - Teflon® base;

12 — lithium reference electrode; and 13 — platinum wire counter electrode. ........... 27

2.3. The EQCM experimental setup including, 1 — Teflon cell; 2 — spring-loaded

clamp; and 3 - silicone o-ring. ................................................................................... 31

xii

3.1. AFM images of aluminum surfaces (A) as received, (B) after mechanical polishing,

(C) after electropolishing, and (D) after electropolishing with enhanced z-axis as
compared to (C). ........................................................................................................ 37

3.2. XPS depth profiles for an electropolished aluminum surface. ................................... 38

3.3. Cyclic voltammetric i-E curves recorded at room temperature in 1M LiPF6/PC

and 1M LiPF6/EC-DMC for (A) mechanically polished, and (B) electropolished
aluminum electrodes. Scan rate = 5 mV/s. Prior to the CV measurements the

electrodes were kept at OCP for 10 h. ....................................................................... 43

3.4. Cyclic voltammetric i-E curves recorded at room temperature in 1M LiPFg/EC-DMC

for a mechanically polished aluminum electrode. Scan rate = 5 mV/s ...................... 44

3.5. Cyclic voltammetric i-E curves recorded at room temperature in 1M LiClO4/PC

3.6.

and 1M LiClO4/EC-DMC for (A) mechanically polished, and (B) electropolished
aluminum electrodes. Scan rate = 5 mV/s. The electrodes were stored at OCP

for 10 h before the CV measurements. ...................................................................... 46

Chronoamperometn'c i-t curve recorded at room temperature for an electropolished

aluminum electrode in 1M LiPF6/PC. Potential step from 2500 to 5000 mV ................ 49

4.1. Cyclic voltammetric i-E curves recorded for mechanically polished aluminum

4.2.

at room temperature in 1M LiClO4/EC-DMC with 500 ppm of added H20.

Scan rate = 5 mV/s. .................................................................................................... 63

SEM images of the mechanically polished aluminum surface. (A) An untreated

and (B,C, and D) after 3 potentiodynarnic scans from 2000—5000 mV vs. LilLi+

in LiClO4/EC-DMC with 500 ppm of added H20. (C, D) The images are the same
as (B) but the deposited layer has been removed by washing with DMC. ................ 67

4.3. Cyclic voltammetric i-E curves recorded at room temperature for an electropolished

aluminum electrode in 1M LiClO4/PC + 200 ppm of H20 during (A) the first three
scans and (B) scans 5 through 50. Sean rate = 5 mV/s .............................................. 70

xiii

4.4. Photomicrographs of the electropolished aluminum surface after 3 (A) and 50 (B)
potentiodynarnic scans in 1M LiClO4/PC + 200 ppm of H20 (conditions
as in Figure 4.3). ........................................................................................................ 71

4.5. Pit density and average pit diameter as a function of concentration of water impurity
in 1M LiClO4/EC—DMC. Data recorded after 3 potentiodynarnic scans from 2000
to 5000 mV vs. mm“, at 5 mV/s. ............................................................................. 73

4.6. Photomicrographs of the electropolished aluminum surface, recorded in-situ during
the 4th potentiodynarnic scan at 5 mV/s in 1M LiClO4/EC-DMC with 1000 ppm
of added H20. Magnification 2400 x ......................................................................... 75

4.7. Chronoamperometn'c i-t curves recorded at room temperature for an electropolished
aluminum electrode in 1M LiClO4/PC. (A) Curves for a potential step from 2500
to 4500 mV. (B) Curves for a potential step from 2500 to 5000 mV. Solid line -
electrolyte solution with 2000 ppm of water added. Dashed line - electrolyte

with no added water. .................................................................................................. 78

4.8. Photomicrographs of the electropolished aluminum surface after a 1 h potential
step to (A) 4500 and (B) 5000 mV in 1M LiClO4/PC with 2000 ppm of added

water. .......................................................................................................................... 79

4.9. An SEM image and an energy dispersive x-ray analysis spectrum for mechanically

polished aluminum prior to polarization. ................................................................... 83

4.10. An SEM image and an energy dispersive x-ray analysis spectrum for mechanically
polished aluminum after potentiodynarnic polarization (3 scans at 5 mV/s; 2000-

5000 mV vs. Li/Li+) in 1M LiClO4/PC + 500 ppm H20. ......................................... 84

4.11. An SEM image and energy dispersive x-ray analysis spectra for electropolished
aluminum after potentiodynamic polarization (3 scans at 5 mV/s; 2000-5000 mV
vs. Li/Li+) in 1M LiClO4/PC + 200 ppm H20. Spectrum 1 corresponds to a region

inside the pit while spectrum 2 corresponds to a region outside. .............................. 85

xiv

4.12. An SEM image and an energy dispersive x-ray analysis spectrum for electro—
polished aluminum 2 potentiodynarnic scans (5 mV/s; 2000-5000 mV vs. Li/Li+)
in 1M LiClO4/PC + 2000 ppm H20. ......................................................................... 86

4.13. Raman spectrum of the corrosion product in the center of a pit on electropolished
aluminum. The pit and corrosion product were formed after 2 potentiodynarnic scans
from 2500 to 5000 mV vs. Li/Li+ in 1M LiClO4/PC with 500 ppm of added H20.
Scan rate = 5 mV/s. Spectra for the freshly electropolished surface, and solid samples
of LiClO4 and Al(ClO4)3.H20 are shown, for comparison ....................................... 88

4.14. (A) Photomicrograph of a corrosion pit on the electropolished aluminum surface
after 2 cyclic voltammetry scans between 2500 and 5000 mV vs. LilLi+ in 1M
LiClO4jEC-DMC + 500 ppm of water. Scan rate = 5 mV/s. The black line indicates
Raman spectra line profile. (B) A series of Raman spectra obtained along the line

shown in 1 tun increments. Acquisition time = 10 min. ........................................... 89

4.15. Cyclic voltammetry (1St scan) in 1M LiPF6/EC-DMC + 500 ppm H20. (A)

- mechanically polished aluminum; (B) — mechanically polished aluminum after
10 h at OCP in the same electrolyte mixture; (C) electropolished aluminum; (D)

- mechanically polished electrode, no added water. Scan rate = 5 mV/s. ................. 94

5.1. Ex situ AFM images of an electropolished aluminum surface (A) before and after
anodic polarization in (B) 1M LiClOJEC-DMC and (C) 1M LiPFdEC-DMC.

Top images were obtained in the deflection mode, bottom in the height mode. ..... 101

5.2. AFM images of an electropolished aluminum surface after 3 potentiodynarnic scans
in 1M LiPFdEC-DMC with 500 ppm of added H20 (5 mV/s; 2000—5000 mV
vs. Li/Li+). (A) Deflection mode image. (B) Height mode image .......................... 105

5.3. AFM images of a mound formed on the electropolished aluminum surface after
2 potentiodynarnic scans (5 mV/s; 2000 — 5000 mV vs. Li/Li+) in 1M LiClO4/
EC-DMC with 500 ppm of added H20. (A) Deflection and (B) height mode

images of the same mound. ...................................................................................... 108

XV

5.4. AFM images of the pit formed on the surface of electropolished aluminum after
2 potentiodynarnic scans (5 mV/s; 2000 -— 5000 mV vs. Li/Li+) in 1M LiClO4/
EC-DMC with 1000 ppm of added H20. (A) Deflection and (B) height mode

images of the same pit .............................................................................................. 109

5.5. Cyclic voltammetric i-E curves for electropolished aluminum recorded at 3 mV/s
in 1M LiClO4/EC-DMC with 1000 ppm of added H20. The measurements were
made in the AFM fluid cell. ..................................................................................... 111

5.6. In situ ECAFM measurements of an electropolished aluminum surface in 1M
LiClO4/EC-DMC with 1000 ppm of added H20. (A) Deflection mode images
of a mound and (B) height mode profiles for a cross section through the middle
of the corrosion mound, as shown on the top image. The electrode potentials are
indicated on the AFM images and the cross sectional profiles, and they correspond

to the i-E curve presented in Figure 5.5 (3rd scan). ................................................. 113

5.7. In situ ECAFM measurements in 1M LiClO4/EC-DMC with 1000 ppm of added
H20. (A) Deflection mode images of the mound and (B) cross sectional profiles
through the middle of the mound. The electrode potentials are indicated on the
AFM images and the cross sectional profiles, and they correspond to the i-E

curve presented in Figure 5.5 (4th scan). ................................................................. 114
5.8. Horizontal and vertical development of the corrosion product deposit. .................. 115

6.1. Time dependent open circuit potential and mass change measurements for

mechanically polished aluminum during exposure to 1M LiPF6/EC-DMC ............ 125

6.2. (A) Cyclic voltammogram, and (B) mass change (Am) and normalized charge
change (AQn with W/z = 11) versus potential profiles for a mechanically polished
aluminum in 1M LiPF6/EC-DMC during the first two potentiodynarnic cycles.

Scan rate = 5mV/s. ................................................................................................... 128

6.3. (A) Cyclic voltammogram, (B) mass change (Am) and normalized charge change
(AQn with W/z = 60) diagram, and (C) W/z change with potential diagram for

xvi

mechanically polished aluminum in 1M LiClO4/EC-DMC during the first two

potentiodynarnic cycles. Scan rate = 5mV/s. ........................................................... 131

6.4. (A) Cyclic voltammogram, and (B) mass change (Am) and normalized charge
change (AQn with W/z = 11) profiles for untreated aluminum in 1M LiPF6/
EC-DMC during the first two potentiodynamic cycles. Scan rate = 5mV/s ............ 133

6.5. (A) Cyclic voltammogram, and (B) mass change (Am) and normalized charge
change (AQn with W/z = 22) profiles for untreated aluminum in 1M LiClO4/

EC-DMC during the first two potentiodynamic cycles. Scan rate = 5mV/s. ........... 134

6.6. (A) Cyclic voltammogram, and (B) mass change (Am) and normalized charge
change (AQn with W/z = 15) profiles for untreated aluminum in 1M LiPF6/
EC-DMC with 500 ppm of added water during the first two potentiodynamic

cycles. Scan rate = 5mV/s. ....................................................................................... 138

6.7. (A) Cyclic voltammogram, and (B) mass change (Am) profile for untreated
aluminum in 1M LiClO4/EC-DMC with 500 ppm of added water during the

first two potentiodynarnic cycles. Scan rate = 5mV/s .............................................. 139

6.8. (A), (B) Photomicrographs of the circular pits formed on the untreated aluminum
surface after 2 potentiodynarnic scans in 1M LiClO4/EC-DMC with 500 ppm of
added water. Scan rate = 5mV/s. (C) AFM deflection mode image of one of the
pits. (D) AFM height mode image of the edge of the pit. Height of the edge is
equal to 200 nm. ....................................................................................................... 140

xvii

CHAPTER 1

1. Introduction

Corrosion/passivation of aluminum and its alloys in aqueous solutions has been one
of the most widely studied problems in electrochemistry for several decades. This is not
surprising considering the degree to which aqueous environments surround us and the
importance of aluminum as a construction material in practically all areas of industry,
transportation and every day life.1 Unlike in aqueous environments, corrosion/passivation
of aluminum in non-aqueous media has received much less attention in the literature.
This situation began to change in recent years and can be credited, in part, to the use of
aluminum as a current collector material for modern lithium-ion rechargeable batteries}-
13 Other applications of aluminum and its alloys in non-aqueous media include aluminum
batteries”.16 high performance electrolytic capacitors,l7-19 and tanks for alcohol-rich
fuelszo’21

Since this work is focused on understanding processes that govern corrosion and
passivation of aluminum in the electrolytes commonly employed in lithium-ion batteries,
the aim of the following paragraph is to familiarize the reader with the concept of the
lithium-ion battery and its importance in a modern, portability-oriented market. This is

followed by discussion of the function of current collectors in a battery system, along

with a description of the desired properties of current collector materials, in general, and

in lithium-ion batteries, in particular. In the next two paragraphs, a review of the
literature regarding corrosion/passivation of aluminum in lithium-ion battery electrolytes
and review of general mechanisms by which the metal can passivate in solution, are

given. Finally, the outline of this dissertation is presented.

1.1. Lithium-ion battery

1.1.1. Overview

In the last few years, lithium rechargeable batteries have become a very popular and
convenient energy source for portable electronics. Their success and ubiquitous nature is
due to their low weight, small size, and high energy storage capacity. The properties of
the Li-ion cells have improved over time, largely due to the introduction of materials that
can serve as lithium intercalation hosts: carbon-type (amorphous coke or crystalline
graphite) anodes, and lithiated metal oxide (LiNiOz, LiCoOz, LiMn204) cathodes.
During operation, lithium ions migrate from the cathode to the anode. Such design
eliminates the presence of metallic lithium and enhances cell stability, safety of
operation, and the number of charge-discharge cycles that can be achieved. A schematic
diagram of a lithium-ion battery during charging and the overall charge-discharge

reaction are presented in Figure 1.1.

 

 

 

 

'- I.
1°3 5
"E a
t _ E
5 a

1 Li+ conducting

, electrolyte .

LI‘C“ and separator LI (1") M30”
Charge
C6 + LilVIyOzy LixCG + LiHMyOzy
Discharge

Figure 1.1. Schematic diagram of a lithium-ion rechargeable battery and illustration of
the electrochemical processes occurring during charging. M stands for Ni, Co or Mn.

Figure adapted from Megahed and Scrosati.22

Typical characteristics of the lithium-ion battery and other rechargeable batteries are
presented in Table 1.1. It is clear that lithium-ion batteries provide some excellent
features in comparison to other rechargeable battery systems. For example, the
gravimetric energy density obtained from the lithium-ion cell is approximately twice that
of the standard NiCd and ~ 1.5 times that of Ni-MH batteries. This means that products
powered by lithium-ion batteries can be made lighter without sacrificing run time. The

lithium-ion cell also provides very high cell voltage (3.6 V), which simplifies the

manufacturing process, since one cell, instead of a stack of cells, can be used to achieve
desired voltage. The lithium-ion battery requires low maintenance, an advantage that
most other rechargeable battery systems cannot claim. There is no memory effect, and no
scheduled cycling is required to prolong the battery’s life. In addition, the self-discharge

capacity loss is less than half that of Ni-Cd and Ni-MH batteries.

 

 

 

 

 

 

 

 

 

 

Ni-Cd Ni-MH Lead Acid Li-ion
G‘aVimemc Energy 45-80 60-120 30-50 110-160
Density, [Wh/kg]
vommemc Energy 110 180 180 300 80 110 220 360
Density, [Wh/dm3] - -
Cell Voltage, [V] 1.2 1.2 2.0 3.6
Cycle Life soc-1500a soc-1000a zoo-300a 500-1000
Self Discharge/Monthb 15-20% 20-30% <5% 54%
Overcharge Tolerance moderate low high very low
Peak Load CurrentC 20 C 5 C 5 C 2 C
Fast Charge Time, [h] 0.25-1 2-4 8-16 2-4

 

 

 

 

 

 

Table 1.1. Characteristics of commonly used rechargeable batteries. a — depends on
regular maintenance; b — at room temperature; c — C-rate is a current that is numerically

equal to the Ah rating of the cell. Data adapted from Buchmann.23

All these highly advantageous features have made lithium-ion batteries the fastest
growing sector of the battery industry and led to their dominance in the portable
electronics market (e.g., cell phones, notebooks, and camcorders). The use of Li-ion

technology in other areas, such as power tools, electric vehicles, and medical and space

applications, has been limited, so far. This is due to the fact that despite its overall
advantages, Li-ion technology has its drawbacks. First of all, Li-ion cells cannot be
charged and discharged as quickly as Ni-Cd or Ni-MH batteries, which precludes their
use in high-power applications, such as power tools and electric vehicles. Also, aging is a
serious concern with most Li-ion batteries. Some capacity deterioration is commonly
observed after one year, whether the battery is in use or not. After two to three years, the
battery frequently fails. This considerably limits its use in medical and space
applications.4’23

Therefore improvements are needed if lithium-ion batteries are to be successfully
used in these advanced applications. A good understanding of the processes occurring in
the battery, as well as the factors that influence its operation, is necessary if improve-
ments are to be made. One of the most widely studied problems in the lithium-ion battery

- 24 d f th '
field over the last decade has been the capacrty fade phenomenon. an re erences em"

1.1.2. Capacity fade phenomenon

Capacity fade is a general term used to describe all side reactions and processes that
occur in a battery during operation and lead to a loss of capacity or premature failure of
the battery system. These processes can occur during normal operation or result from
abuse of the battery system by subjecting it to overcharge or overdischarge, a condition
that is especially common in stacks of cells, where potentials of the individual cells
cannot be controlled. Undesirable reactions that occur can involve both active materials
(anode, cathode, and electrolyte) and cell hardware (current collectors, separators, and

cell housing).

It is not surprising that there has been a significant interest in understanding capacity
fade mechanisms. An extensive review of this subject was presented recently by Arora,
White, and Doyle.24 The authors described several undesirable processes that can lead to
capacity fade. These include lithium deposition, electrolyte and solvent decomposition,
phase changes in the insertion electrode materials, active material dissolution, current
collector corrosion, and passive film formation over the electrode and current collector
surfaces. Over time, these processes commonly lead to the loss of active material and an
increase in the internal impedance of the cell. Figure 1.2 summarizes the various capacity
fade processes. For clearer illustration of the conditions at which specific process usually

occur, half-cell discharge curves are also presented in this figure.

    
 
    
  

 

 

> oxygen evolution, solvent/electrolyte
\o > . . . .
—-: N, ox1dation, anode dissolution,
Posrtive current collector corrOSion
__, Electrode
Discharge
g ”S cathode phase changes
an Li
W .
> 'é’ current collector dissolution
Discharge Negative
__’ Electrode
B, > <— . . .
Q q Charge Li(s) deposuion,
'1‘ O solvent/electrolyte reduction,
Capacity [Ah/kg] solid electrolyte interface formation

Figure 1.2. Half—cell discharge curves for a lithium-ion battery showing various capacity

fade processes. Adapted from Arora et (11.24

There is already much understanding of the capacity fade phenomena involving the
cell active materials.24 and references were“! This is not surprising since these are the
components that determine the overall cell characteristics such as storage capacity,
maximum voltage, and charge and discharge rates. However, much less is known about
the capacity fade phenomena involving cell hardware, in general, and current collectors,
in particular. More detailed understanding of these processes is needed. This is especially
true for long-term applications of the lithium-ion batteries (space, medical, and military)
as well as for introducing new, active materials that can operate at higher voltages. This
dissertation is exclusively focused on the mechanisms that lead to oxidation (formation of

a passivation layer) and/or corrosion (pitting and metal loss) of aluminum current

collectors in non-aqueous solvents/electrolytes.

1.1.3. Current collectors

Current collectors are important hardware components, which serve two functions
(see Figure 1.1). They act as a substrate for the active material and transfer charge to and
from the material. Thus, current collectors must provide good adhesion for active
material and offer low electrical resistance for charge transfer. Also, from the
electrochemical point of view, the current collector must withstand the same potential
extremes the active material does during charging and discharging, without any
deterioration or corrosion over the cell life. Lastly, since the mass and volume of current
collectors are included in the specific energy density, it is important that they are made in
a thin-film form and have sufficient strength to withstand stresses generated during the

9

cell fabrication.2

Selecting the right current collector material is a very important part of the battery
design process. Failure to do so properly may lead to one of several undesirable reactions.
These include electrolyte decomposition, passive film formation, and localized or general
corrosion. In long-term applications, these unfavorable processes can affect the cell
characteristics through an increase in the internal impedance during cycling, loss of
adhesion between the active material and the current collector, changes in electrolyte
composition and/or formation of corrosion products that can attack or passivate the active
electrode materials. These phenomena are very harmful to the performance of the battery
and often lead to premature failure of the whole battery system.4’24’25

Fulfilling the requirements for the current collector presents a special challenge in
the case of lithium-ion batteries. This stems from two major facts: (i) the extreme
reactivity of active materials used in lithium-ion batteries (e.g., lithium salts, lithiated
metal oxides, lithium intercalated carbon) and (ii) extremes of the electrochemical
potentials applied: 0 to 4.2 V vs. Li/Li+, or - 3 to 1.2 V vs. NHE.

The negative electrode current collector must resist lithium insertion to form alloys,
because alloy formation would deplete lithium from the electrode and also degrade the
mechanical properties of the current collector. Copper is commonly used since it offers
good resistance toward alloy formation with lithium at negative potentials. It is also
relatively cheap, less harmful than other collector materials like nickel, highly
conductive, and easily rolled into thin foils.25’26

On the other hand, the positive electrode current collector must be resistant to

oxidation in the presence of the electrolyte anion. This is a difficult condition to satisfy,

since the potential of the positive electrode is usually above the dissolution potential of

all the metals. Also, since electrolyte anions are designed for high solubility and
dissociation of their lithium salts, they also form highly soluble salts with other metals
(e.g., aluminum). The same is true for the solvent - a high donor number stabilizes not
only lithium complexes, but metal salt complexes in general. Therefore, a passivating
film is essential for metallic positive electrode current collectors. Aluminum is the most
common choice due to its low cost, low density, ductility, and good resistance to

. . . . 25
ox1dation due to the natural, protective ox1de layer.

1.2. Corrosion of aluminum in non-aqueous solvent/electrolyte mixtures

commonly employed in lithium-ion batteries - literature review

Several investigations of the corrosion/oxidation of aluminum in Li-ion battery
electrolytes have been performed in recent years. Both liquids and solid polymers have
been tested as the supporting electrolyte. A considerable amount of research has been
focused on understanding corrosion/passivation of aluminum in new lithium salts that
could overcome the hydrolytic and thermal instability of LiPF6, currently the most
commonly used electrolytes-8 These include lithium trifluoromethanesulfonate
(LiCF3SO3), lithium tris(trifluoromethylsulfonyl)methide (LiC(CF3SOg)3), and three
members of the lithium imide family: lithium bis(trifluoromethylsulfonyl)imide
(LiN(CF3802)2), lithium (trifluoromethylsulfonylnonafluorobutylsulfonyl)imide
(LiN(CF3S02)(C4F9802)), and lithium bis(pentafluoroethylsulfonyl)imide
(LiN(C2F5SOZ)2). Some researchers have investigated methods to suppress the

corrosion/oxidation of aluminum in these and other electrolytes. For example, various

aluminum alloys, fluorocarbon-based polymeric coatings and electrolyte additives (e.g.,
HF) have been tested for this purpose. The following literature review highlights the
crucial findings in the area of aluminum current collectors corrosion.

Kanamura et al. investigated the electrochemical reactions taking place on aluminum
electrodes during anodic polarization in propylene carbonate (PC) containing 1M
LiClO4, LiBF4 or LiPF6.2 Small oxidation currents (< 0.1 mA/cmz) were observed
during the potential sweep (50 mV/s) up to 5.5 V vs. Li/Li+ in all three electrolytes. At
higher potentials, the anodic current increased significantly in LiClO4 (up to 1 mA/cm2
at 6.5 V), whereas in LiBF4 and LiPF6, the currents remained below 0.2 mA/cm2 and 0.1
mA/cmz, respectively, up to 7 V vs. Li/Li+. The results indicated that aluminum is most
stable in LiPF6. XPS analysis of the aluminum surface after potentiostatic polarization at
5.5 V in LiPF6/PC revealed the presence of an AlF3/Ale3 layer. XPS of the aluminum
surface polarized in LiClO4/PC indicated the presence of an A1203 layer rich in C104-
and C1-. The authors also investigated the resistance of PC toward oxidation using in situ
FI‘IR spectroscopy. Spectra of the electrolyte solution, taken in the vicinity of the
electrode, revealed the onset potentials for PC oxidation to be 5.2, 4.6, and 4.4 V for
LiPF6, LiBF4, and LiClO4, respectively. The oxidation was not significant, therefore the
authors concluded that the anodic charge originated primarily from the oxidation of
electrolyte and the aluminum surface. This was different from what they observed for a
nickel electrode, at which the oxidation current corresponded well with extent of PC
decomposition.27

Long-term studies of Al current collector corrosion were performed by Braithwaite

and coworkers.4 Aluminum current collectors were aged in LiPF6/propylene carbonate-

10

diethyl carboate (PC-DEC) and LiPF6/propylene carbonate-dimethyl carbonate (PC-
DMC) by continuously applying a simulated electrical cycle analogous to that employed
in low-earth-orbit (LEO) aerospace applications. Using impedance spectroscopy, the
authors discovered that aluminum became more passive during cycling. XPS and Auger
electron spectroscopy measurements indicated that Li was the predominant surface
species. Also, twice as much surface F was observed than would be associated with the
deposition of LiPF6, suggestive of electrolyte decomposition. SEM studies of the surface
structure revealed significant localized pitting. However, the pits were filled with
corrosion products forming three-dimensional structures (nodules and mounds). The
authors attributed the pitting to possible contamination of electrolyte! solvent mixture
with chloride, which is well known to induce pitting in aqueous environments. ’

Krause et al. investigated the stability of aluminum in newly synthesized
LiN(C2F5SOZ)2 and LiN(CF3S02)(C4F9802) electrolytes, and compared the behavior to
that in LiPF6, LiCF3SO3, and LiN(CF3SOz)2.5 Potentiostatic polarization of aluminum
at 4.2 V vs. LilLi+ led to development of significant oxidation currents, in excess of
10 mA/cmz, in LiCF3SO3/PC and LiN(CF3SOz)2/PC, whereas in other three electrolytes
(PC solvent), small, less than 10 riA/cmz, oxidation currents were observed.
Potentiodynamic cycling of aluminum electrodes showed evidence for pitting corrosion
with protection or repassivation potentials increasing in the order of LiCF3SO3 (2.78 V)
< LiN(CF3S02)2 (3.55 V) < LiN(C2F5802)2 (4.50 V) < LiN(CF3SOg)(C4F9SOz)
(4.62 V). The authors related the increased stability of aluminum in heavier analogues of

LiN(CF3SOz)2 to the larger molecular size of the anion.

ll

Improved passivation of aluminum in LiN(CF3SOz)(C4F9802)/PC, as compared to
LiCF3SO3/PC and LiN(CF3SOz)2/PC, was also observed by Kanamura et al.6 The
results of potentiodynamic cycling, XPS analysis, and in situ FTIR showed that
aluminum corrodes significantly at potentials more positive than 3.8 V vs. Li/Li+ in
either LiCF3SO3/PC and LiN(CF3SOz)2/PC. Both solvent and electrolyte decomposition
were also observed at such conditions. In LiN(CF3SOz)(C4F9802)/PC, aluminum was
stable up to 4.6 V. The authors attributed the improved passivation properties of the
LiN(CF3SOz)(C4F9802) salt to better distribution of negative charge on the anion, which
can stabilize the aluminum surface more efficiently.

Yang et al. used the electrochemical quartz crystal microbalance (EQCM) to study
aluminum corrosion in electrolytes encountered in present and possible future lithium-ion
batteries, including LiPF6, LiBF4, LiClO4, LiCF3SO3, LiC(CF3SOz)3, and
LiN(CF3SOp_)2.7 They found that the aluminum, with its protective oxide film intact, is
relatively corrosion-resistant in these electrolytes. Mechanical damage of the oxide film
leads to significant corrosion in PC containing LiCF3SO3, LiC(CF3SOz)3, and
LiN(CF3SOz)2. The authors observed a decrease of the electrode mass during anodic
polarization up to 5 V vs. LilLi+ in each of these electrolytes. More detailed EQCM
studies of aluminum corrosion in LiN(CF3SOz)2/PC revealed that the corrosion process
involves the formation of an adsorbed aluminum compound or film that subsequently
desorbs. The authors also found that potentiodynarnic polarization of aluminum in
LiPF6/PC and LiBF4/PC leads to formation of a protective film after the first
potentiodynarnic scan. An increase in the electrode mass was observed in such cases. In

LiClO4/PC, formation of the oxidation product layer was also observed during the first

12

potentiodynamic scan. This layer did not, however, prevent surface oxidation during
subsequent scan.

Walker, Cox and Salomon studied the stability of LiC(CF3SOz)3/EC-DMC-MF (EC-
DMC-methyl forrnate) on aluminum, platinum and glassy carbon electrodes.8 Results of
potentiodynarnic polarization up to 4.3 V vs. Li/Li+ indicated that the electrolyte is most
stable at aluminum as the lowest oxidation currents were observed. The authors attributed
this stability to the formation of a passivation layer on the aluminum surface.

Behl and Plichta used controlled potential coulometry to investigate aluminum
degradation in electrolytes composed of an EC/PC/DMC (20:20:60 vol.%) mixture and
LiN(CF3SOz)2 and LiC(CF3802)3.9 They observed that aluminum foils were quite
stable in the LiC(CF3SOz)3 solution at potentials up to 4.5 V vs. Li/Li+, but underwent
severe corrosion in LiN(CF3SOz)2 below 3.5 V vs. Li/Li+.

An interesting approach toward understanding corrosion/passivation of aluminum in
the (CF3SOZ)2N' (rm') and (C2F5302)2N' (Ben) was taken by Goldman and
McEwen.10 The authors compared the electrochemical properties of these two anions in
solvent-free ionic liquids created with the 1—ethyl-3-methyl imidazolium (EMI) cation:
EMIIm and EMIBeti. They found that aluminum passivates easily in both neat ionic
liquids. Similar oxidation currents were recorded in both EMIIm and EMIBeti. The
addition of 10% of propylene carbonate (PC) to either of the ionic liquids had a dramatic
effect on aluminum sability. A significant increase in the oxidation current was observed,
with the EMIIm being more corrosive than EMIBeti. These results implicate PC as active
participant in the corrosion process, at least in these electrolytes. The authors could not,

however, determine whether or not the increased corrosion observed in the presence of

13

PC is due to the solubility of the passivating layer in this solvent, or if the PC is directly

involved in the corrosion process.

Corrosion of aluminum current collectors in polymer electrolytes was also
studied.1 H3 Munshi and coworkers reported serious corrosion of aluminum current
collectors in Li/(PEO)3—LiCF3803N6013) during battery overcharging.ll The results of
Choe et al. showed that aluminum substrates are stable up to 4.2 V vs. LilLi+ in
polyacrylonitrile-based (PAN) electrolytes containing LiAsF6 and LiPF6, and only up to
3.7 and 3.5 v with LiN(CF3SOz)2 and LiCF3SO3, respectively.” Chen et al.
investigated the corrosion susceptibility of aluminum current collectors in Li/(PEO)-
LiN(CF3SOz)2N2013 or Tisz cells, during anodic polarization.13 The authors observed
that aluminum is resistant to uniform corrosion. It, however, undergoes localized pitting
during overcharge conditions (4 V vs. Li/Li+), as was shown through SEM studies. The
authors also found that aluminum evaluated in a standard electrochemical cell is less
prone to the corrosion than when used in a real battery, polarized at the same conditions.
They have attributed this to inhomogeneous current flow through the current collector in
the battery (e.g., due to the non-uniform electrical resistance of the cathode material),
which leads to local breakdown of the passive film on aluminum at sites of locally high
current density. This was a very important finding, indicating that if susceptibility of
aluminum toward corrosion is evaluated in an electrochemical cell, it should be
investigated at potentials higher than overcharge conditions in order for the results to be

meaningful.

l4

Some researchers also investigated various methods to suppress the corrosion of
aluminum. One of the proposed approaches involves the use of multi-component
electrolytes. For example, Behl and Plichta reported that the addition of LiBF4 to IM
LiN(CF3SOz)2/EC-PC-DMC significantly reduces the corrosion current observed during
potentiostatic polarization of aluminum.9 They found that complete passivation of
aluminum occurs when the concentration of LiBF4 is higher than 0.5M. A steady state
oxidation current of less than 10 |.iA/cm2 was observed in 1M LiN(CF3SOz)2/EC-PC-
DMC with 0.5M LiBF4, at 4.5 V vs. Li/Li+. This current was two orders of magnitude
lower than the one measured in solution without LiBF4.

A similar beneficial effect of added LiBF4 and LiPF6 on the stability of aluminum in
LiN(CF3SOz)2/PC was observed by Yang et al.7 The authors learned that aluminum
passivates in the 0.5M LiBF4 (or 0.5M LiPF6) + 0.5M LiN(CF3SOz)2 in PC, and that the
passivation layer is formed from LiBF4 (or LiPF6) as the W/z (molecular weight to
charge ratio) obtained from the EQCM measurements were the same as for pure
1M LiBF4/PC or 1M LiPF6/PC.

Other authors investigated the role of additives and anti-corrosion coatings to
improve the stability of aluminum. Kanamura et al. studied the effect of low levels of HF
(0.01 mol/dm-B) on the behavior of aluminum in LiCF3SOg/PC and LiN(CF3SOz)2/PC.6
The authors observed improved passivation in the presence of HF in both electrolytes.
XPS data showed the presence of a fluorine-rich passivation layer on the surface,
suggestive of the conversion of native A1203 to AlF3 and AlOF. The composition of this
layer was independent of the electrolyte type, which implies that it was formed solely

from the HF additive. In another study, Braithwaite et al. investigated the effect of low

15

levels of water impurity (20 ppm) on the long-term stability of aluminum in 1M
LiPthlPC-DEC.4 Based on impedance data, the authors observed improved passivation
in the presence of water. In their studies, they also evaluated the influence of two types of
carbon-fluorocarbon-based conductive coatings on the corrosion resistance of aluminum
in 1M LiPFg/PC-DEC. They found improved corrosion resistance, which they attributed
to sealing of the aluminum surface by coating.

Another approach to reduce the corrosion rate of aluminum was investigated by
Chen and coworkers.13 These researchers proposed the use of a tungsten-aluminum alloy
as the current collector material. They found that aluminum surfaces impregnated with
~20 atom % tungsten, by ion implantation, exhibited enhanced resistance to pitting

corrosion in poly(ethylene oxide)/LiN(CF3SOZ)2.

1.3. Passivation of metals in solution — review of possible mechanisms

The mechanisms by which a metal can passivate in solution were reviewed by Kelly
and Moran, and can be classified into four types: (1) air formed film, (2) salt film, (3)
chemisorption of the solvent, and (4) oxide/oxyhydroxide formation.29 For a metal to
passivate by mechanism 1, the air-formed film must be stable in the solution under the
experimental conditions, and thereby form a barrier to corrosive attack of the metal. The
type 2 passivation mechanism requires active dissolution of a metal surface at a high
enough rate that supersaturation in the solution boundary layer near the metal surface
occurs, with respect to the salt composed of the metal cations and electrolyte anions. This

salt then precipitates onto the surface to form an adherent, protective layer. The

16

protective ability of the salt film depends strongly upon its solubility and rate of
dissolution. Lower solubility and dissolution rates require lower metal oxidation charge
to flow to replace the metal lost by dissolution, thus providing better passivation. When a
metal passivates by mechanism 3, solvent molecules must attach to the metal surface
irreversibly by either chemical or electrochemical adsorption. Presumably, this adsorbed
film protects the metal by raising the activation energy for metal dissolution. The
chemical adsorption process typically results in only monolayer coverage and is limited
to potentials below the solvent oxidation. Finally, for a metal to passivate by mechanism
4, a source of oxygen atoms must be present in solution, either in the form of atoms,
molecules, or some anions containing oxygen. Either through direct chemical reaction, or
via an electrochemical reaction, oxygen or oxygen containing anions bind to metal
cations at the surface, forming an adherent, three-dimensional oxide/oxyhydroxide film.
This last mechanism is the one by which most metals are thought to passivate in aqueous
solutions, with water providing the oxygen atoms. However, in anhydrous organic
solutions this mechanism is unlikely due to the fact that even if oxygen is present as a
structural component of organic solvent, it is strongly bound to the carbon atom(s) and
therefore cannot readily participate in the formation of an oxide film. All four

mechanisms are presented schematically in the Figure 1.3.

17

(1) air-formed film

 

air-formed metal oxide layer

///, metal

(2) salt film formation

M ——> M’1+ + ne‘
mMn+ + nAm' —> Mm(A)n~L

salt film
‘ metal

 

(3) solvent adsorption

(solv) ——> (solv)adS
(solv) —> (solv)adS + e“

: adsorbed solvent layer
' metal

 

 

 

(4) oxide/oxyhydroxide formation

M —) M211+ + 2ne'
M2n+ + nHZO ——> Monl + 2nH+
M2n+ + 2nH20 —> M(0H),nl + 2nH+

 

- — metal oxide/oxyhydroxide layer
/% metal

Figure 1.3. Schematic diagrams of possible passivation mechanisms for metal (M) in a

. 29
solution.

18

Work by Scanlon, Kruger and Moran on the passivation of iron in 0.5M
LiAsF6/dimethoxyethane, indicated a fifth possible passivation mechanism -
electropolymerization of the solvent.30 The authors showed that significant suppression
of corrosion can occur if a layer of polymerized solvent forms on the metal surface.
However, as pointed by the authors, this mechanism is limited to a select few
solvent/electrolyte/metal systems and, thus, should not be treated as general passivation
mechanism in non—aqueous environments.

A very good example of how different mechanisms may provide passivity to the
metal surface is the work by Kelly et al. on the passive behavior of iron in 0.5M
LiClO4/PC.31 The authors showed that under different circumstances, competition
between different mechanisms of passivation or a domination of one of the mechanisms,
may occur. They observed that in anhydrous (< 1 ppm) PC solutions, below the oxidation
potential of PC, bare Fe passivated via oxidative chemisorption of solvent molecules.
However, at potentials above 3.6 V vs. Li/Li+, PC was oxidized to C02, and therefore
adsorbed solvent film did not provide protection to the metal surface anymore. At this
point bare Fe remained passivated through the formation of salt film composed of
Fe(ClO4)2. The presence of this film was confirmed by both x-ray microanalysis and
XPS. They also observed that the air-formed film on the metal surface provided a
protection over the wide range of potentials. High stability of the air formed film was
attributed to the lack of aggressive species in anhydrous solution. The authors postulated
that adsorbed solvent film formed on the top of the air-formed film during anodic

polarization, although they could not directly prove their postulate.

l9

1.4. Outline of the dissertation

This dissertation is devoted to understanding the mechanisms that lead to oxidation
(formation of a passivation layer) and/or corrosion (pitting and metal loss) of aluminum
in non-aqueous electrolytes commonly used in lithium-ion batteries. Important
parameters studied include the role of the electrolyte type (e. g., LiClO4 vs. LiPF6), the
organic solvent type (e. g., propylene carbonate (PC) vs. a mixture of dimethyl carbonate
and ethylene carbonate (DMC-EC», the effect of water contamination (SO-2000 ppm),
and the physicochemical properties of the aluminum surface (e. g., bare vs. oxide coated).
An overview of each chapter is given below along with a synopsis of the findings.

A description of the experimental setup and protocols used during the measurements
is presented in Chapter 2. Chapter 3 describes potentiodynarnic and potentiostatic
examination of the susceptibility of aluminum to corrosion/passivation in M (<50 ppm
H20) LiClO4/EC-DMC, LiClO4/PC, LiPF6/EC-DMC, and LiPF6/PC. It was determined
that aluminum passivates well in all four electrolyte/solvent mixtures. XPS analysis of
the surfaces revealed that passivation occurs through different mechanisms depending on
the electrolyte. The results presented in that chapter serve as a comparison for the data
presented in Chapter 4, which reflect a study of the effect of water contamination of the
solvent on the corrosion/passivation process.

In Chapter 4, the effect of water contamination on the corrosion/passivation of the
aluminum in LiPFdEC-DMC, LiPF6/PC, LiClO4/EC-DMC, and LiClO4/PC is presented.
Controlled water additions of 50 — 2000 ppm were investigated. It was determined that

water does not have a significant influence on corrosion in LiPF6, as the aluminum

20

surface remains passivated, even in the presence of relatively high (500 ppm) water
levels. On the other hand, the presence of water is detrimental to the passivation of
aluminum in LiClO4. Electrochemical and microscopic (optical microscopy and scanning
electron microscopy) measurements showed evidence of significant surface pitting.
Raman spectroscopy and energy dispersive x-ray analysis (EDS) indicated that
Al(ClO4)3 is the major corrosion product.

Chapter 5 presents results of a detailed examination of the passivation and/or
corrosion of aluminum using in situ and ex situ atomic force microscopy (AFM). The
formation of a solid passivation layer was observed in all dry solvent/electrolyte
mixtures. Roughening of the electrode surface was observed in LiPF6, when water was
present. It was also found that pitting in LiClO4 solutions containing water leads to
formation of either mounds (pits filled with corrosion products) or empty pits, depending
on the amount of water present. Mounds were observed when the water level was below
500 ppm, whereas empty pits were found at higher water levels.

Chapter 6 examines the corrosion of aluminum in LiPF6/EC-DMC and LiClO4/EC-
DMC with in situ electrochemical quartz crystal gravimetry (EQCM). Correlations
between electrochemical charge and corresponding electrode mass change were used to
determine the products of the corrosion/passivation processes. Finally, a summary of all

the data is given in the Chapter 7.

21

CHAPTER 2

2. Experimental Section

2.1. Preparation of aluminum electrodes

Aluminum discs (9.5 or 14 mm in diameter) were cut from high purity foil (1 mm
thick, 99.999%, Goodfellow Cambridge Ltd.) with a bench press and ferrule. The discs
were cleaned by rinsing with hexane, acetone, methanol, and deionized water. One side
of each disc was then either mechanically polished in hexane, or mechanically polished

and electropolished in order to obtain a low-oxide or oxide-coated surface, respectively.

2.1.1. Mechanical polishing

The discs were mechanically polished in hexane using successively finer silicon
carbide (SiC) polishing paper (320, 600, 1200, 2000 grit). Each polishing step was
followed by a 10 min sonication in hexane. The discs were then transported to an Ar-
purged glove box for a final polish with a 2000 grit SiC paper. After the final polish, the
discs were soaked in hexane inside a sealed container, taken out of the glove box for a 15

min sonication, and placed back inside a glove box for electrochemical evaluation.

2.1.2. Electropolishing
Prior to electropolishing, the discs were mechanically polished, as described above,

to remove the native oxide film and decrease the surface roughness. Electropolishing was

22

performed in a 3% aqueous solution of HBF4, at a constant current - a procedure adapted
from an Alzak commercial electropolishing process.32 No solution stirring was involved.
The disc was mounted in a home-built Teflon® cell by means of a Viton® o-ring and
spring loaded clamp. The disc served as the anode and was placed in parallel to a cathode
made from high purity aluminum foil. The distance between the electrodes was 25 mm.
Constant current polarization was conducted in two steps, first at 36 mA/cm2 for 30 min
and then at 9 mA/cm2 for 10 min. The constant current was applied using a HP 6209B
DC power supply (Hewlett-Packard, Rockville, MD), and the potential difference
between the electrodes was monitored with a Servogor 124 strip chart recorder (NGI
Norma Goerz Instruments GmbH, Austria). After electropolishing, the discs were rinsed
with deionized water, soaked in KOH (7.5% w/w; 20 s) and H2804 (30% w/w; 30 s),
thoroughly rinsed with deionized water and dry ethanol, soaked in dry ethanol for 30 min,

rinsed with hexane, and stored in hexane until use.

2.2. Surface characterization techniques

Changes in the aluminum surface morphology brought about by the imposed
electrochemical conditions were studied ex-situ with optical microscopy (OM), atomic
force microscopy (AFM), and scanning electron microscopy (SEM). Changes in the
morphology were also studied in situ with electrochemical optical microscopy (ECOM)
and electrochemical atomic force microscopy (ECAFM). The chemical composition of
the surface, before and after polarization, was investigated with energy dispersive x-ray

microanalysis (EDS), x-ray photoelectron spectroscopy (XPS) and Raman spectroscopy.

23

2.2.1. Atomic force microscopy (AFM)

The atomic force microscopy (AFM) was performed with a Nanoscope IIIa
instrument (Digital Instruments, Veeco Metrology Group; Santa Barbara, CA) in the
contact mode. Si3N4 tips mounted on gold cantilevers (200 um legs, 0.12 N/m spring
constant) were used to acquire topographical images in air. Two types of data were
collected for each image: height and deflection. Height data are recorded at constant
cantilever deflection and correspond to the change in piezo height. They were used for
determination of height of the structures present on aluminum surface. Deflection data,
on the other hand, are collected at constant piezo position and come from the differential
signal of the top and bottom photodiode segments. They provide a sensitive edge-
detection technique.

ECAFM was performed with a Nanoscope II microscope (Digital Instruments,
Veeco Metrology Group; Santa Barbara, CA). Images were collected in the contact mode
with Si3N4 tips mounted on gold cantilevers (100 um legs, 0.58 N/m spring constant). A
commercial electrochemical fluid cell (Digital Instruments) was employed, which
consisted of a glass body, a silicone o-ring, an aluminum disc working electrode, a Li/Li+
reference electrode, and platinum wire counter electrode. The geometric area of the
exposed disc was 0.4 cm2. The microscope head was placed inside a glove box specially
designed for the ECAFM studies, which was constantly purged with N2 allowing
removal of water down to the 500 ppm level (procedure for determination of water
content inside a glove box is described in paragraph 2.6.2.). A LilLi+ reference electrode,
protected from direct contact with N2, was constructed. It consisted of the polyethylene

body, a glass wool plug, a piece of lithium metal, a nickel wire as a contact to lithium,

24

and a “hot glue” seal. The electrode was assembled inside an Ar-filled glove box and
stored in vial with the specific electrolyte solution. The typical equilibrium potential of
this reference electrode was within 5 mV of freshly scratched lithium rod in the same
electrolyte solution.

Figure 2.1 shows pictures of (A) the entire ECAFM experimental setup, (B) liquid

cell, and (C) a schematic diagram of how the LilLi+ reference electrode was prepared.

2.2.2. Optical microscopy (OM)

Optical micrographs of the aluminum discs were obtained using an Olympus BH-2
microscope supplied with a DPll digital camera (1712 x 1368 pixels). 10x and 80x
objectives were used giving total magnifications of 300x and 2400x, respectively. Images
of at least three different, most representative spots on aluminum surface were collected.

For the ECOM work, the microscope was placed inside an Ar-filled glove box (Coy,
Grass Lake, MI; < 100 ppm H20). The measurements were performed in a specially
designed, home-built cell, which consisted of a Teflon® base and body, a glass optical
window and a 4-screw metal clamp (Figure 2.2). The aluminum disc was mounted on the
bottom of the cell using a Chemraz® o-ring. The geometric area of the exposed surface
was ~ 0.4 cm2, and the cell volume was approximately 0.1 cm3. A platinum wire served
as the counter electrode and freshly scratched lithium rod was used as the reference

electrode.

25

 

“hot-glue” plug

 

nickel wire

micropipctic tip

 

lithium metal

glass wool fritt wettcd
. with electrolyte solution

 

Figure 2.1. (A) Experimental setup for the ECAFM measurements including, 1 — glove
box; 2 — ECAFM microscope head; 3 - optical microscope; 4 - computer with monitors 5
& 6; 7 — optical microscope monitor. (B) 8 - ECAFM fluid cell; 9 — Li/Li+ reference
electrode; 10 — solution inlet with syringe; ll — solution outlet with built-in Pt-wire
auxiliary electrode and collection vessel. (C) Schematic diagram of how the Li/Li+

reference electrode was prepared.

26

 

 

Figure 2.2. Diagram of the electrochemical cell used for the in-situ optical microscopy.
(A) side view and (B) top view. 1 - microscope objective lens; 2 - metal clamp with
screws; 3 — optical window; 4 — Teflon® body; 5 — lithium reference electrode port;
6 — solution inlet; 7 — solution outlet; 8 — aluminum disc electrode; 9 — Chemraz® o-ririg;
10 - copper current collector; 11 - Teflon® base; 12 — lithium reference electrode; and
13 - platinum wire counter electrode.

27

2.2.3. Scanning electron microscopy (SEM) and energy dispersive x-ray micro-
analysis (EDS)

SEM and EDS were performed with JEOL-6400V electron microscope. SEM images
were collected at a 2048 x 1536 pixel resolution with an ADDA H digital image
acquisition system and AnalySIS Pro 3.2 image acquisition software (both Soft Imaging
System GmbH, Munster, Germany). EDS data were collected with a Vantage DI
microanalysis system equipped with a light element detector (ThermoNORAN,
Middletown, W1). A 30° take off angle and 20 kV accelerating voltage were used. Data

analysis was performed with Vantage 1.5.1 analysis software (ThermoNORAN).

2.2.4. X-ray photoelectron spectroscopy (XPS)
XPS analysis was performed with a PHI 5400 ESCA (Physical Electronics, Eden
Prairie, MN). Data were collected and analyzed with PHI MultiPak V 6.0A software.

The area analyzed on each sample was approximately 250 umz.

2.2.5. Raman spectroscopy

Raman spectra were obtained in a back-scattered collection mode with a Raman
2000 spectrometer equipped with a 2080 Raman Microprobe attachment and a 2090
Direct Video Module (all from Chromex, Inc., Albuquerque, NM). The intermediate
resolution 1200 grove/mm grating was used. The excitation light was from a frequency
doubled Nd-YAG laser (30 mW, 532 nm), focused to a spot size of approximately 5 pm.
This yielded an estimated power density of 150 kW/cmz. Spatial distribution of the

Raman features was accomplished using an in-line imaging mode. In this mode, the laser

28

light was focused into a line ~60 pm in length and ~2 pm in width. The estimated power
density in line-imaging mode was 25 kW/cmz. Spectra were collected every lttm along

the line.

2.3. Electrochemical measurements

The cyclic voltammetric (potentiodynarnic) and potential step (potentiostatic)
measurements were performed using a CS-2000 computerized potentiostat (Cypress
Systems, Inc., Lawrence, KS). Open circuit potentials were recorded with either the
CS-2000 potentiostat or with a Servogor 124 strip chart recorder (NGI Norma Goerz
Instruments GmbH, Austria) that was fitted with an operational amplifier (CA3140A,
Intersil) in a voltage follower configuration, in order to increase the input resistance to
1.5 TO.

All electrochemical measurements were made in a single compartment, glass cell.
Freshly scratched lithium rods (98+%, Aldrich) were pressed against nickel wires and
served as the auxiliary and reference electrodes. The aluminum disc working electrode
was mated to the bottom of the cell with a Chemraz® 514 o-ring (Ace Glass, Inc.) and a
metal clamp. The geometric area of the exposed electrode was 0.2 cm2, and all currents
reported are normalized to this area. The cell was assembled and operated in an Ar-

purged glove box (Coy, Grass Lake, MI) with a moisture level below 100 ppm.

29

2.4. Electrochemical quartz crystal microbalance (EQCM)

Electrochemical quartz crystal microbalance (EQCM) studies were performed with
EQCN-701 quartz crystal microbalance and a PS-205B potentiostat (both Elchema,
Potsdam, NY). The system was interfaced to a PC using an AT-MIO-16X, 16-bit data
acquisition card (National Instruments, Austin, TX) and Voltscan software (Elchema).
All measurements were conducted in a single compartment, home built, Teflon® cell. 10
MHz, commercially-available quartz crystals, with aluminum evaporated on the surface
(Elchema), were used as the working electrode. A theoretical mass sensitivity of 4.4
ng/Hzcm2 was possible with these crystals. The quartz crystal resonator was mated to the
bottom of the cell with two silicone o-rings (Digital Instruments, Santa Barbara, CA) and
a spring-loaded clamp. The piezoelectrically-active area of the crystal was determined by
size of the metal electrodes deposited on the surface and was equal to ~ 0.2 cm2. The
electrochemically-active area was determined by the silicone o-ring and matched
piezoelectrically-active area. All currents reported were normalized to the 0.2 cm2 area.
Freshly scratched lithium rods were pressed against nickel wires and served as the
auxiliary and reference electrodes. The cell was assembled and operated in an Ar-purged
glove box (Coy, MI) with a moisture level below 100 ppm. A scheme of the experimental

setup is presented in Figure 2.3.

30

Li Counterige’ctrode

Li/l:i+ Reference El. Potentiostat

Working E1.
'|

  
   
   

Frequency Counter

 

 

 

 

 

 

Oscillator

 

 

 

 

 

.
.lil'Il',

 

 

 

 

 

Computer 1

 

Electrochemical
Cell

Figure 2.3. The EQCM experimental setup including, 1 - Teflon cell; 2 — spring-loaded
clamp; and 3 — silicone o-ring.

The instrument was validated using the reversible electrodeposition of Cu on Au.
This was done by potentiodynarnic cycling of the gold QCM electrode (Elchema) at 50
mV/s in 5 mM Cu(NO3)2 + 0.1 M HNO3 from 500 mV to —200 mV vs. Ag/AgCl. The
charge recorded during the deposition was equal to the stripping charge with a value of
650 11C. Using a value of n = 2, the deposition charge corresponds to deposition of 214
ng of Cu. This value is in good agreement with the 218 ng value calculated from the

measured frequency change, using the Sauerbrey equation (see Chapter 6, Eq. 6.1).

31

2.5. Chemicals

LiClO4 (Aldrich) was dried for at least 12 h at 120°C under a flow of high purity Ar.
LiPF6 (99.9%; Aldrich) was used as received. Ethylene carbonate (EC) (Aldrich) was
vacuum distilled and mixed with freshly distilled dimethyl carbonate (DMC) (Aldrich) in
a 1:1 volumetric ratio. Propylene carbonate (PC) (Aldrich) was used without any
additional purification. Both the ECzDMC mixture and PC were stored over freshly
activated (500°C, 5 h, Ar atmosphere) 4 A molecular sieves for at least 48 h prior to use.
The water content in the solvents was measured by GC-MS (Hewlett—Packard G18OOB,
with El detector). Standard additions of water to the solvent were made and the intensity
of m/z = 18 signal was monitored. The water content was estimated to be < 50 ppm in
both solvents.

l M solutions of electrolytes in organic solvents were prepared in a Nz-purged glove
box (Coy, Grass Lake, MI). The glove box atmosphere was constantly circulated through
a drying column filled with Drierite (W. A. Hammond Drierite Company, Xenia, OH).
The humidity inside the glove box was maintained below 500 ppm. After preparation, the
solutions were purged with Ar for 30 min and tightly sealed in containers. The containers
were then transferred to the Ar-filled glove box and placed inside a dessicator until use in

the electrochemical measurements.

32

2.6. Remaining information

2.6.1. Data analysis and plotting
Mathematical operations on numerical data and plotting of numerical data were
performed using PSI-Plot 4.57 data analysis software (Poly Software International, Salt

Lake City, UT).

2.6.2. Determination of water content inside glove boxes
The relative humidity (RH; [%]) inside the glove boxes was measured using
a CMM80 digital therrnohygrometer (Mannix, Lynbrook, NY). Assuming an atmospheric
pressure of 1 atm, the estimated water content (Cw; [ppm]) was calculated according to
the following equation:
4
Cw = RH-pw. 10

where pW is a water vapor pressure (0.023 atm, at 20°C).33

33

CHAPTER 3

3. Electrochemical Investigation of the Corrosion/Passivation
of High-Purity Aluminum in LiClOJEC-DMC, LiClO4IPC,
LiPFdEC-DMC, and LiPF6/PC

3.1. Introduction

In the last few years, lithium rechargeable batteries have become a popular and
convenient energy source for portable electronic devices. Their success and ubiquitous
nature is due to their low weight, small size, and high-energy storage capacity. Due to
these highly attractive attributes, there is considerable interest in using the lithium-ion
technology for new applications such as electric vehicles, medical devices, and space
technology. These new applications require batteries that are reliable over prolonged
periods of time. Therefore, the long-term stability of all battery components becomes an
issue. Corrosion of current collectors is of particular importance, considering their role in
the rechargeable battery systems. The current collector serves two major functions. It acts
as a substrate for the active electrode material, and transfers charge to and from the
material. The current collector must offer low resistance to charge transfer, provide good
adhesion for active material, and be sufficiently durable to withstand the stresses

produced during the battery fabrication. It also, from an electrochemical point of view,

34

must withstand the same potential extremes the active material does during charging and
discharging, without any deterioration or corrosion over the cell life.

Aluminum and copper are the most commonly used current collector materials for
the positive and negative electrodes, respectively, in lithium-ion batteries. Aluminum is
used mainly because of its low cost, low density, ductility, and good resistance to
oxidation due to its natural, protective oxide layer. Copper, on the other hand, besides
being inexpensive and ductile, offers good resistance toward the formation of alloys with
lithium at negative potentials. Unfortunately, both copper and aluminum are prone to
corrosion and/or oxidation in the electrolyte/solvent mixtures typically used in lithium-
ion batteries.24’25 and references therein

Several investigations of the corrosion/oxidation of aluminum in non-aqueous
electrolytes have been performed. Both liquidsz'10 and solid polymersll-l3 have been
tested as the supporting electrolyte. Much attention has also been paid to several new
electrolyte salts that might overcome the hydrolytic and thermal instability of LiPF6,
which is currently the most commonly used electrolyte. These new salts include

5-7,9,10 . .
In general, consrdering the state

LiCF3SO3, LiN(CF3SOz)2 and LiC(CF3SOz)3.
of the aluminum electrode surface after anodic polarization, results from these studies can

be classified in two ways: (i) pitting corrosion or (ii) surface passivation.

This chapter examines the corrosion/oxidation of high purity aluminum (99.999%) in
LiPF6/EC-DMC, LiPF6/PC, LiClO4/EC-DMC, and LiClO4/PC. Testing was performed
using two different metal surfaces, (i) aluminum with the native oxide film removed by

mechanical polishing under an inert, argon atmosphere, and (ii) aluminum, first

35

mechanically polished, and then electropolished in an aqueous solution of HBF4. The
second surface preparation was chosen to mimic the oxide film that likely forms during
fabrication of commercial batteries. Aluminum current collectors in the batteries are
commonly pretreated by acid-base etching to improve the adhesion of the active cathode
material with the surface.24 Electropolishing was also chosen because it produces a very
smooth and reflective surface, which was necessary for the AFM measurements
discussed in Chapter 5. Electrochemical techniques, including cyclic voltammetry (CV),
chronoamperometry (CA), and open circuit potential measurements (OCP) were used to
asses the susceptibility of aluminum toward corrosion/passivation in each of the
electrolyte/solvent mixtures. X-ray photoelectron spectroscopy (XPS) was used to study
the elemental composition of any passivating film formed during electrochemical

polarization.

3.2. Results and discussion

3.2.1. Pretreatment and characterization of Al electrodes

The Al disc electrodes were either mechanically polished in hexane under a blanket
of Ar or electropolished prior to evaluation, as described in Chapter 2. Figure 3.1 A-D
presents contact mode AFM images of the (A) as received, (B) mechanically polished,
and (C, D) electropolished aluminum surfaces. It is evident that both polishing methods
smooth the as received surface. For example, the surface roughness decreased from

greater than 150 run over a 100 um2 area for the as received aluminum, to less than 40

36

nm for the mechanically polished surface, and to less than 3 nm for the electropolished

surface.

1000 nm
1000 nm

   

1000 nm

   

Figure 3.]. AFM images of aluminum surfaces (A) as received, (B) after mechanical
polishing, (C) after electropolishing, and (D) after electropolishing with enhanced z-axis

as compared to (C).

XPS analysis of the mechanically polished surface revealed the presence of Al (2p,
74.5 eV) and O (ls, 532 eV) peaks with an atomic ratio of CIA] = 1.6, consistent with

A1203 (O/Al = 1.5). A lower binding energy Al (2p, 71 eV) peak, consistent with Al

37

metal, was also observed. Thus, the A1203 film thickness must be less than 10 nm, as this
is a typical escape depth for the photoelectrons in the instrument used. XPS analysis of
the electropolished surface also revealed the presence of an A1203 layer (O/Al = 1.6).
The film thickness was estimated to be 5-10 nm, based on a depth profiling results
presented in Figure 3.2. Traces of B (Is, 188 eV) and F (Is, 686 eV) were also detected
on the surface of the oxide film, but not in the film itself, indicating that the
electropolishing did not incorporate the tetrafluoroborate anion. Carbon impurities were

also detected on both polished surfaces. This is most likely a result of the exposure to

 

 

 

hexane.
100 _

E 80 1

8 : A1213 Surface atomic %
g L 018 41

g 60 , C15 28

g -. ~. A12s 26

8 40 ‘i‘ 015 Fls 4

.2 I ‘\ BIS 1

s

2 20 ‘ CIS ‘~-~.,_.__~

" Fls Bl; ______________
07“ a. , .2 A - .A -1... . Aw..-
0 100 200 300 400 500 600
Sputter depth [A]

Figure 3.2. XPS depth profiles for an electropolished aluminum surface.

38

3.2.2. Open circuit potential (OCP) measurements

OCP measurements were conducted to assess the susceptibility of aluminum toward
corrosion and/or oxidation. It is generally accepted that OCP measurements for active
metals covered with a passivating oxide layer can be quite variable from sample-to-
sample, even if all experimental parameters are kept constant.34 Thus, it was not our
objective to evaluate the absolute magnitude of the OCP, but rather to study the general
trends of magnitude and dynamics associated with reaching a pseudo steady-state value.
Atanasoski reported OCP values for oxide-covered aluminum in the KAlCl4/PC in the
range of 2.4-2.8 V vs. K/K+ (2.5-2.9 V vs. Li/Li+).35 When the underlying aluminum
surface was exposed, in situ, by rotation against a stainless steel blade, the OCP
decreased to 1.6-1.7 V vs. K/K+ (1.7-1.8 V vs. Li/Li+). After removal of the scrapping
blade, the OCP slowly increased, approaching 2.3-2.4 v vs. K/K+ (2.4-2.5 v vs. Li/Li+)
during a 10 h solution exposure. Clearly, the surface passivated over time and the author
concluded that a passivation reaction occurs between bare aluminum and PC because the
OCP values were independent of the electrolyte composition and concentration.

Typical results in LiPF6/EC-DMC, LiPF6/PC, LiClO4/EC-DMC, and LiClO4/PC,
for mechanically polished and electropolished aluminum, are presented in Table 3.1
Values in parenthesis represent the rate of OCP change recorded over a l h period at the
beginning and end of the 10 h measurement. OCP values for the mechanically polished
electrodes are initially more active (negative) than they are for the electropolished
surfaces, in both solvent systems, particularly for LiClO4. The OCP also increases
rapidly at the beginning of the exposure, as the rate of change is the highest. The more

active potential and the rapid positive increase are consistent with a low oxide,

39

incompletely passivated surface. Over time, the OCP shifts to more positive (noble)
values before approaching a pseudo steady-state value. This trend is consistent with the
formation of a passivating layer on the surface. These trends are similar to those reported
by Atanasoski, suggesting that the mechanical polishing procedure produces a low oxide

. . . . 35
(re, defective), active aluminum surface.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Mechanically Polished Electropolished
ocp [mV vs. Li/Li+] OCP [mV vs. Li/Li+]

(rate of change [mV/h]) (rate of change [mV/h])

Electrolyte Initial 10 hours Initial 10 hours
EC-DMC
LiClO4 1800 (150) 2120 (7) 2250 (20) 2380 (1)
LiPF6 2050 (130) 2380 (11) 2230 (20) 2340 (7)
PC

LiClO4 1850 (180) 2220 (8) 2450 (4) 2510(3)
LiPF6 2170 (100) 2400 (10) 2380 (5) 2400 (1)

 

 

 

 

 

 

 

Table 3.1. Typical results of OCP measurements.

Further analysis of the data presented in Table 3.1. indicates that the OCP values in
LiPF6 are ~200 mV higher than in LiClO4 for the mechanically polished aluminum in
both solvents. Also, the OCPs in LiPF6 are practically independent of the solvent,
whereas the values in LiClO4 solutions show some solvent dependence. For example, the
values in PC are 100 - 150 mV more positive than in EC-DMC. These observations
probably reflect a slightly different passivation mechanism in each electrolyte. Another

observation is that the presence of the oxide layer on the electropolished surface causes

40

the initial OCP values to be 200400 mV higher. This is likely because the passivating
layer is less defective. In summary, the mechanically polished aluminum surface shows a
greater tendency toward corrosion/oxidation than does the electropolished surface. Also,

both surfaces show a greater tendency toward corrosion/oxidation in LiClO4 than in

LiPF6.

3.2.3. Cyclic voltammetry in LiPF6/PC and LiPFdEC-DMC

Cyclic voltammetric i-E curves for mechanically polished aluminum (5 mV/s) in 1M
LiPF6/PC and 1M LiPF6/EC-DMC, after a 10 h soak at OCP conditions, are presented in
Figure 3.3A. A similar response is seen for both solvents. The oxidation current begins to
increase at ~ 2800 mV. During the first positive scan, there is significant anodic current
between 3000 and 5000 mV, with three oxidation peaks around 3300, 3850, and 4300
mV vs. Li/Li+. The anodic charge passed during the first scan was 21 mC/cm2 and 20
mC/cmz, respectively, in EC-DMC and PC. There is very little charge passed during
subsequent scans indicating that anodic reactions occurring during the first scan are
electrochemically irreversible, and the reaction products passivate the surface. No
significant current flows after the first scan, in either solvent, at potentials up to 5000 mV
vs. Li/Li+.

Figure 3.3B presents cyclic voltammetric i-E curves for electropolished aluminum
(5 mV/s) in 1M LiPF6/PC and 1M LiPF6/EC-DMC. A response, similar in shape to that
for the mechanically polished surface is observed. The oxidation current begins
to increase near 3000 mV in both solvents. However, the oxidation charge during the first

scan, between 3000 and 5000 mV, is much lower (about 5 times) than the charge

41

observed during the first scan for the mechanically polished surface. The anodic charge is
3.4 mC/cm2 and 3.9 mC/cmz, respectively, for EC-DMC and PC. This is due to the fact
that the electropolished surface is covered by a less defective oxide film, which provides
protection to the surface. Interestingly, only the first anodic peak at 3300 mV vs. Li/Li+ is
present. This suggests that the two other peaks observed for the mechanically polished
surface are associated with oxidation of the electrolyte/solvent species at the unprotected
metal sites.

Figure 3.4 presents cyclic voltammetric i-E curves for mechanically polished
aluminum (two different electrodes) in 1M LiPF6/EC-DMC immediately after being
placed in the cell, and after 10 h at OCP. A significant positive shift in the onset potential
for the oxidation current is evident for the conditioned electrode, being about 600 mV
more positive than for the unconditioned one. The oxidation charge for the conditioned
electrode is 22 mC/cm2 and for the unconditioned electrode is 26 mC/cmz; ~ 20% more.
This observation is consistent with the positive shift of the OCP during the extended
soaking, and indicates that passivation progressively occurs. The aluminum surface is
effectively passivated to potentials as positive as 5000 mV due to reactions with the
electrolyte/solvent system that occur during open circuit exposure, and during anodic

polarization between 3000 and 5000 mV vs. Li/Li+.

42

7o_
60; —LiPF6/EC-DMC
so;- --- LlPFé/PC
40

3o_
20_
10:

 

  
  

 

V ' 1 V

 

Current density [uA/cmz]

    

 

   

I

3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

_10 a l
2000 2500

 
 
  

 

 

 

 

 

14
$12". —LiPF6/EC-DMC
E. 10'. ---LiPF6/PC
2 l
3 8r
3,. .
”a 6_
5
'0 4L
0 0
2 .

 

 

 

2000 2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

Figure 3.3. Cyclic voltammetric i-E curves recorded at room temperature in 1M

LiPF6/PC and 1M LiPF6/EC-DMC for (A) mechanically polished, and (B)

electropolished aluminum electrodes. Scan rate = 5 mV/s. Prior to the CV measurements

the electrodes were kept at OCP for 10 h. Numbers correspond to scan number.

43

70

 

 

 

 

 

 

,— E —afier10hatOCP
NE 60 F
E 50 - - - after 1 min at OCP _________ -..----
3. 40 E” o ’1’"
E? I 1""
re 30 _' "f
'8 20 E ,I’
5 ‘0?
L) 0 h ".3"
-10 m

2000 2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

Figure 3.4. Cyclic voltammetric i-E curves recorded at room temperature in 1M
LiPF6/EC-DMC for two different, mechanically polished aluminum electrode. Scan rate
= 5 mV/s.

These results are in agreement with those of other researchers who have studied
aluminum polarization in LiPF6.2’4’5’7 Although, it is difficult to directly compare results
from laboratory-to-laboratory due to differences in experimental conditions, all of these
studies have shown that aluminum passivates during the initial stages of anodic
polarization. Probably, the most comparable data with our work can be found in the
studies of Kanamura et al.2 and Yang et al.7. Kanamura et al. investigated the
electrochemical reactions taking place on aluminum electrodes during anodic polarization
in LiPF6/PC.2 Small oxidation currents (< 0.1 mA/cmz) were observed during the
potential sweep at 50 mV/s up to 7 V vs. Li/Li+. XPS studies of aluminum after anodic

polarization revealed the presence of a passivation layer composed of AlF3 and A1203.

Yang and coworkers used cyclic voltammetry and electrochemical quartz crystal

44

gravimetry (EQCM) to study the passivation of aluminum in I_.iPF6/PC.7 On a sputtered
aluminum surface, that was mechanically polished prior to the measurement, they
observed current densities in the range of 60 p.A/cm2 during the first potentiodynarnic
scan at 50 mV/s. Very little charge was observed during the second scan indicating that
passivation of the surface occured. EQCM data showed an increase of the electrode mass,
consistent with the buildup of a passivation layer. The authors were unable, however, to
assign the apparent mass-to—charge ratio (W/z) of 26, obtained from EQCM
measurement, to any single oxidation product (for example, the W/z for A1203 is 8, for

AlF3 is 19, and for Al(PF4)3 is 145).

3.2.4. Cyclic voltammetry in LiClO4/PC and LiClO4/EC-DMC

Cyclic voltammetric i-E curves for mechanically polished aluminum (5 mV/s) in 1M
LiClO4/PC and 1M LiClO4/EC-DMC, after 10 h at OCP, are presented in Figure 3.5A.
As was the case for LiPF6, similarly shaped polarization curves are observed in both
solvents. An anodic charge of 13 mC/cm2 and 11 mC/cm2 is observed during the first
scan between ~ 2500 and 5000 mV vs. Li/Li+ in EC-DMC and PC, respectively. There is
little charge passed during subsequent scans, indicating the anodic reactions are
electrochemically irreversible and the reaction products passivate the surface. In contrast
to the LiPF6-based electrolyte/solvent mixtures, the current begins to flow at a more
negative potential of 2500 — 2700 mV and rises slowly until reaching a maximum

at ~ 4500 mV vs. Li/Li+.

45

 

    

 

   

(T. 40} —LiClO4/EC-DMC
§ --- LiClO4/PC

i 30_

é‘ 20’

CI

0)

“O

a 10

g ,

U 0 h ,,,,,,

 

 

2000 2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

 

   

 

 

 

 

B
15
5; —LiClO4/EC-DMC
g - - - LiClO4/PC
a. 10 "
.Q
m
8
“c 5 -
a
05) t
O 0
2000 2500 3000 3500 4000 4500 5000

Potential [mV vs. Li/Li+]

Figure 3.5. Cyclic voltammetric i-E curves recorded at room temperature in 1M
LiClO4/PC and 1M LiClO4/EC-DMC for (A) mechanically polished, and (B)

electropolished aluminum electrodes. Scan rate = 5 mV/s. The electrodes were stored at
OCP for 10 h before the CV measurements.

46

Figure 3.53 presents the cyclic voltammetric i-E curves for electropolished
aluminum (5 mV/s) in 1M LiClO4/PC and 1M LiClO4/EC-DMC. Again, similar
behavior is observed in both solvents. The onset potential for the oxidation current is
shifted positive, as compared to the mechanically polished surface, to 3000 mV. The
electrode is also passivated by the oxidation reaction products after the first scan. The
oxidation charge for the first scan (3.8 mC/cm2 in both solvents) is much lower (about 4
times) than the charge observed for the mechanically polished electrode. This is due to
protective properties of a thicker and less defective surface oxide layer after
electropolishing.

Mechanically polished aluminum, stored at OCP in either LiClO4/PC or LiClO4/EC-
DMC for 10 h, showed only a slightly smaller (~ 5 %) oxidation charge during the first
scan when compared to electrodes that were tested right after placement in the
electrochemical cell. These results agree well with the OCP measurements, which
indicated much weaker passivation in LiClO4-based electrolyte/solvent mixtures (~200

mV lower OCP) compared to the LiPF6 solutions.

Other researches have studied the polarization of aluminum in LiClO4/PC. Our
results are comparable to those of Kanamura et al. who showed the oxidation charge is
associated primarily with the formation of a mixed aluminum oxide layer on the
surface?"3 The authors observed oxidation currents < 100 ltA/cm2 during the potential
sweep, at 50 mV/s up to 5.5 V vs. Li/Li+ in 1M LiClO4/PC. Our results, however, are in
contrast with those of Yang et al. who showed significant oxidation currents, reaching 0.8
mA/cm2 at 5 V vs. Li/Li+ during the first potentiodynamic cycle, at 50 mV/s, in 1M

LiClO4/PC.7 The EQCM data for this scan showed a large mass increase of the electrode,

47

which indicated the formation of a thick layer of oxidation products. This layer was not
well passivating as an oxidation current of 0.5 mA/cm2 was recorded during a subsequent
scan, along with a further increase of the mass of the electrode. It is interesting, however,
that the shapes of the i-E curves recorded by Yang et al. show characteristics of pitting
corrosion, and are very similar to our results obtained in LiClO4-based mixtures
containing added water impurity. This is discussed further in Chapter 4 of this

dissertation.

3.2.5. Chronoamperometry

The passivation of aluminum in LiClO4/PC, LiClO4/EC-DMC, LIPFfi/PC, and
LiPF6/EC-DMC was also studied with chronoamperometric measurements. Figure 3.6
presents chronoamperometric i-t curve for a mechanically polished electrode in 1M
LiPF6/PC, during an anodic step from 2500 to 5000 mV. The current rises quickly and
then decreases as the double layer charging followed by the oxidation process are
completed. The current decays to a low and steady value of ca. 0.8 itA/cmz. The charge
passed was 7.3 mC/cmz. Very low steady state current and charge indicate quick
passivation of the surface.

Similarly shaped chronoamperometric i-t curves were observed in all electro-
lyte/solvent systems. A summary of the chronoamperometric data for both mechanically
polished and electropolished electrodes is presented in Table 3.2. The oxidation charge
and the onset potential for the oxidation current, recorded during the first
potentiodynamic scan, are also summarized in the table. These results show that the
oxidation charge is independent of the type of solvent for both mechanically polished and

electropolished surfaces. A larger oxidation charge was observed in LiPF6 than in

48

LiClO4 solutions for the mechanically polished electrodes. This indicates that the
passivation process occurs through different mechanism in either electrolyte. On the other
hand, the oxidation charge for the electropolished surface was much smaller and
practically the same in all four electrolytes/solvent systems. This is due to protective
properties of less defective surface oxide layer produced by electropolishing.

The mass of aluminum oxidized during potentiostatic polarization is also reported in
Table 3.2. The m0x values were calculated using Faraday law with assumption that the
whole anodic charge corresponded to the formation of Al3+. It is clear that the loss of
aluminum is negligible even in the case of mechanically polished electrodes for which
m0x < 2.2 rig/cm2 was observed. Such weight corresponds to uniform loss of ~ 10 nm of

aluminum.

30.0

 

1M LiPF6/EC-DMC

MN
pot
00

15.0_

10.01

Current density [uA/cmz]

.U‘
o

 

 

 

 

I 1 1 1 I 1 1 1 I 1 1 L

30 40 50 60
Time [min]

.0
o
r

 

O
O
N
0

Figure 3.6. Chronoarnperometric i-t curve recorded at room temperature for an
electropolished aluminum electrode in 1M LiPF6/PC. Potential step from 2500 to 5000 mV.

49

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Eon...” Q... (CVf‘) 0.. (CA?) I... (CAP) mox (CA)
[mV vs. Li/Li+] [mC/cmz] [mC/cmz] [llA/cmz] [rig/cmz]
mechanically polished
EC-DMC
LiClO4 2750 13 18 2.5 1.7
LiPF6 3050 21 24 4.0 2.2
PC
LiClO4 2800 1 1 20 3.0 1.9
LiPF6 3030 20 23 4.0 2.1
electropolished

EC-DMC
LiClO4 3010 3.8 7.8 0.80 0.73
LiPF6 3100 3.4 6.5 0.65 0.60
PC
LiClO4 2970 3.8 7.0 0.75 0.65
LiPF6 3180 3.9 7.3 0.75 0.68

 

Table 3.2. Summary of the cyclic voltammetric and chronoamperometric polarization

data for both mechanically polished and electropolished aluminum. The cyclic

voltammetric data (a) are for the first scan and the chronoamperometric data (b) are for a
1h potential step to 5000 mV vs. Li/Li+.

It is important to indicate at this point that mechanical polishing or electropolishing

did not generate fully reproducible surfaces. The variations in the currents for

experiments performed at virtually identical conditions were in the order of 10-30%.

However, the shapes of potentiodynarnic and potentiostatic curves were reproducible and

indicative of the passivation of the surface.

50

 

3.2.6. XPS analysis of aluminum surface

XPS was used to probe the chemical nature of the passivating layer formed on
aluminum in LiClO4 and LiPF6. Table 3.3 presents results for mechanically polished
aluminum (i) prior to any electrochemical evaluation, (ii) after a 10 h soak at OCP in 1M
LiPF6/EC-DMC or IM LiClO4/EC-DMC, and (iii) after 1 h at 5 V vs. Li/Li+ in these two
electrolytes. The treated electrodes were washed with DMC and dried in a flow of Ar
before the XPS measurements. All samples were stored in hexane during transport to the
XPS chamber.

Two peaks for Al (2p, 71 and 75 eV) were observed for all mechanically polished
samples. The major peak at 75 eV was assigned to A1203 and/or AlF3, depending on the
electrolyte, and the minor peak at 71 eV was assigned to Al metal. The intensity ratio of
these two peaks is presented in parentheses. The fact that an Al metal peak (71 eV) is
observed suggests that the passivating film is thinner than 10 nm. The carbon impurity
detected (ls, 285 eV) is consistent with CC or C-H bonding. It probably resulted from
storing electrodes in hexane during transport and/or contamination of the XPS chamber
(pump oil).

As expected, only Al and 0 were detected on mechanically polished aluminum. The
relative atomic abundance (18 and 29 atomic %) of these two elements indicates that the
surface was covered by a thin-film of A1203. Similar results were obtained for
mechanically polished aluminum soaked in LiClO4/EC-DMC indicating that the surface
does not react with the electrolyte at open circuit. 0 and Al were also the major elements
detected on the surface after anodic polarization in LiClO4/EC-DMC. However, the
atomic ratio at which these two elements are present (O/Al = 2.9) does not correspond

directly to the formation of A1203 (O/Al = 1.5). A possible explanation for the increased

51

O is the presence of Cl (2p, 209 eV), consistent with a C104 surface moiety. This
suggests that Al(ClO4)3 is a component in the passivating layer. Since Al/O/Cl atomic
ratio in Al(ClO4)3 is equal 1/12/3, the relative atomic abundance of Cl (2p, 209 eV) equal
to 4 % indicates that 16 % atomic O and 1.3 % atomic Al detected on the surface is
bound as Al(ClO4)3. Thus, 30 % atomic O and ~ 15 % atomic Al is present in some other
form. Atomic ratio of CIA] in the remaining species is equal to 2 and is still higher than
the value expected for pure A1203. This may indicate that the mixture A1203/Al(OH)3 is
present on the surface rather than pure A1203. The C1 (2p, 209 eV) signal was also
accompanied by much weaker (~15 % of the peak at 209 eV) peak for a Cl (2p, 200 eV)

that is characteristic of Cl-. Presence of Cl- indicates partial electrooxidation of C104:

 

 

 

 

 

 

 

 

 

Relative atomic abundance [%]
Ala) ,
Treatment (Alp/A13.) Li C O C] F p
mechanically 18
polished (0. 10) “d 53 29 nd nd nd
soaked in 22
LiClO4/EC-DMC (0.10) "d 40 33 nd nd nd
1 hr at 5 V in 16
LiClO4/EC-DMC (0.01) “d 34 46 4 nd nd
soaked in 13
LiPFdEC-DMC (0.17) 15 3° 15 nd 26 1
1 hr at 5 V in 14
LiPFdEC-DMC (0.06) 7 27 14 nd 36 2

 

 

 

 

 

 

 

Table 3.3. Summary of the XPS data for mechanically polished aluminum surfaces
before and after 1 h potentiostatic polarization at 5 V vs. Li/Li+. The following peaks
were used for determination of relative atomic abundance: A13+ (2p, 75 eV), AlO (2p, 71
eV), Li (ls, 56 eV), C (Is, 285 eV), 0 (ls, 532 eV), Cl (2p, 209 eV), F (Is, 686 eV),
and P (2p, 135 eV). a) - total aluminum detected.

52

 

Previously, Eggert and Heitbaum investigated the electrooxidation of LiClO4 in PC
at Pt electrodes by differential electrochemical mass spectrometry (DEMS).36 Later,
using the same technique, Cattaneo and Ruch investigated the oxidation of LiClO4 at
Mn02 electrodes.37 Both groups found that the oxidation of perchlorate starts at ~ 4.5 V
vs. Li/Li+, and leads to formation of C102 and HCl. The following mechanism was

assumed for this process:

Clo4' —-> 'ClO4 + c' —> C102 + 2 cad + e' [3.1]
C102 + H+ + e' —> HCl + 02(g) [3.2]

2 0ad —> O2(g) l3-3]

02(g) and Oad are formed as reaction products and both can react with aluminum to form
A1203. The formation of A1203 significantly improves the passivation of Al in LiClO4
electrolyte/solvent systems.

Previously, Kanamura et al. investigated the oxidation of aluminum in
LiClO4/PC.2’3 Based on the XPS analysis of aluminum polarized at 5.5 V vs. LilLi+ for
l b, they concluded that the surface is covered with highly porous A1203. They also
detected the presence of C104- and C1. in the surface oxide film, which they assigned to

electrochemical oxidation of C104:

Significant amounts of Li and F, in addition to Al and 0, were detected on the
surface after both soaking at the OCP, and polarizing the aluminum in LiPFg/EC-DMC.

A very small amount of P was also detected indicating that the majority of F did not

53

come from the electrolyte anion (PF6') but rather from its decomposition products. It is

well known that LiPF6 undergoes slow hydrolysis in the presence of trace water,

. . . 24 and references therein
according to the followrng reactions:

LiPF6 Z PF5 +LiFl [3.4]

PF5 + H20 —-> 2HF + POF3 [3.5]

Hydrolysis products may then react with the aluminum oxide on the surface to form a
mixed aluminum fluoride and oxyfluoride passivation layer. Fluoride may also
precipitate as LiF or migrate into the oxide film and react electrochemically with the

aluminum metal surface. Possible reactions are as follows:

A1203 + 6HF —> 2A1F3l + 3H20 [3-6]

A1203 + 2111: —> 2A10Fl + H20 [3-71

Li+ + F' —+ LiFl [3.8]

A1 + 3F' —> A1F3l + 3c' [39]
If one assumes that, on the anodically polarized surface, all the P was present as PF6-
and all the Li as LiF, then the Al/O/F atomic ratio is equal 1/ 1/1.2. This indicates
formation of the mixed aluminum oxide/oxyfluoride/fluoride layer on the surface. On the

other hand, significant amounts of Li that were detected on the surface of mechanically

polished aluminum after soaking in the LiPF6/EC-DMC, and the relative abundances of

54

other elements indicate that fluorine was predominantly present as LiF and the proportion
of F in oxyfluoride layer was much lower than on the surface that was anodically
polarized.

These observations seem to resolve a discrepancy between results reported by
Krause et al.5 and Kanamura et al.2 Kanamura and coworkers, using XPS, identified
A1203 and A1F3 as the main components of a passivation layer formed after polarization
at 5.5 V vs. Li/Li+ in LiPF6/PC. Contrastingly, Krause et al. found primarily A1203 and
LiF on a surface polarized at 4.2 V vs. Li/Li+ in the same solution. A significant
difference in the polarization potential used by both groups may explain the difference in
the chemical composition of the passivation layer. At lower potentials, precipitation of
LiF dominates, whereas at higher potentials electrochemical oxidation of aluminum and
formation of A1F3 occurs.

The formation of a surface layer in both LiPF6/PC and LiClO4/PC has been also
confirmed by electrochemical quartz crystal microbalance (EQCM).7 It was observed
that an increase in the mass of the aluminum electrode occured during potentiodynarnic
cycling in both electrolytes. The authors were unable, however, to assign the values of
apparent mass to charge ratio (26 for LiPF6 and 220 for LiClO4), to any single product

formed on the surface.

3.2.6. Stability of passivation layer
The stability of the passivation layer formed in LiClO4/EC-DMC and LiPF5/EC-
DMC was investigated. Three potentiodynarnic scans were performed for mechanically

polished aluminum in both electrolytes. After the third scan, the electrodes remained at

55

open circuit for 1 h, after which time a fourth potentiodynarnic scan was recorded. The
oxidation charge during the fourth scan in LiClO4/EC-DMC increased approximately 2
times over that for the third scan, from 2.9 to 5.5 mC/cmz. On the other hand, the fourth
scan recorded in the LiPF6/EC-DMC was not significantly different from the third scan.
A similar observation was made when the passivated electrodes were left at OCP for 5
min during which time the solution was agitated by the flow of Ar. These results indicate
a greater stability for the passivation layer formed in LiPF5/EC-DMC than for the one
formed in LiClO4/EC-DMC. These results also support the passivation mechanism
proposed above. The Al(ClO4)3 formed, is soluble in EC-DMC, and is stable only when
supersaturation conditions are met. Both mixing and open circuit soaking cause some
dissolution of the layer, leading to the increased anodic charge during the subsequent
polarization scan. On the other hand, the fluoride/oxyfluoride film formed during
polarization in LiPF6/EC-DMC is insoluble in the organic solvent. It is thus stable toward
both the hydrodynamic conditions of mixing and slow dissolution at open circuit. This is
important information indicating that, although aluminum is stable in LiClO4 electrolytes
at highly anodic potentials, multiple charge-discharge cycles may lead to corrosion of the

surface.

3.2.7. Passivation of aluminum in LiClOJEC-DMC, LiClO4/PC, LiPFg/EC-DMC,
and LiPF6/PC - mechanism

In the present research, OCP measurements, as well as voltammetric and
chronoamperometric data, show that aluminum passivates in all four electrolyte/solvent

systems. However, the mechanisms by which passivation occurs seem to be different in

56

either electrolyte. First of all, OCPs measured for mechanically polished (low oxide)
aluminum surface in LiPF6, are significantly more noble (~ 200 mV) than these observed
in LiClO4. The OCPs are also solvent independent in LiPF6 solutions whereas some
solvent dependence is observed for LiClO4 with PC passivating better than EC-DMC
mixture. Also, XPS data indicate a significant amount of fluorine and lithium present on
the surface of aluminum after 10 h soaking in LiPF6/EC-DMC, whereas no lithium or
chlorine were detected on the aluminum soaked in LiClO4/EC-DMC. These results imply
that at OCP conditions, aluminum passivates by solvent adsorption in LiClO4 (see
mechanism (3) p. 18), and by salt film formation (see mechanism (2) p. 18) in LiPF6
solutions. Of course, it is also possible that both mechanisms (3) and (2) are operative in
LiPF6.

Passivation of aluminum in LiClO4/EC-DMC and LiClO4/PC by the solvent
adsorption mechanism is also supported by research of other authors. Previously,
Atanasoski showed that the OCP values of aluminum in PC were independent of the type
of electrolyte (KCl vs. KAlC14) and electrolyte concentration.35 Thus, the author
concluded that a passivation reaction occurs between the aluminum surface and PC
molecules. Passivation by solvent adsorption of iron and stainless steel in LiClO4/PC was
also proposed.3l’38’39 It was shown that such a mechanism is operative for potentials up
to the oxidation potential of PC and suggested that an adsorbed solvent layer raises the
activation energy for metal dissolution, thus providing protection to metal surface.

A different passivation mechanism seems to be operating in LiPF6 solution at OCP
conditions. XPS data indicate the formation of a salt film as the passivation layer. XPS

data suggest, however, that the salt film is not a simple precipitate of A1(PF6)3. The

57

amount of phosphorus relatively to amount of fluorine detected on the surface, is too
small to support this mechanism. The salt film is rather a result of the reaction of
aluminum and/or aluminum oxide with the products of electrolyte hydrolysis, as
described earlier in this chapter.

Results of potentiodynamic cycling and potentiostatic polarization in dry
electrolyte/solvent mixtures indicate that the aluminum surface remains passivated in all
examined solutions, at least for the potentials up to 5 V vs. Li/Li+. Passivation by solvent
adsorption at such highly anodic potentials is not very likely due to possible solvent
oxidation. Although it has been shown by Kanamura et al.2’3 that bulk oxidation of PC
does not occur on aluminum at potentials up to 5.0 V vs. Li/Li+ in LiClO4 and 5.4 V vs.
Li/Li+ in LiPF6 due to presence of the oxide layer, significant oxidation of PC on
platinum was shown to take place at potentials as low as 3.6 V vs. Li/Li+.4O It is thus not
likely for the solvent molecules adsorbed on the surface to protect metal at highly anodic
potentials. Kelly et al. showed previously that at potentials higher than the oxidation
potential of PC, iron passivates in LiClO4/PC through formation of salt film
(Fe(ClO4)2).31 A similar mechanism in which Al(ClO4)3 is formed, can be envisioned for
passivation of aluminum in LiClO4 solutions. This mechanism is supported by the XPS
results. XPS data show the presence of aluminum oxide and perchlorate on the surface of
aluminum after anodic polarization. This mechanism is also supported by AFM and
EQCM data presented in the Chapters 5 and 6, respectively.

Anodic polarization in LiPF6/PC and LiPF6/EC-DMC also leads to passivation of
the surface through formation of the salt film. XPS data indicate formation of mixed

aluminum fluoride/oxyfluoride layers rather than Al(PF6)3. The very low solubility of

58

aluminum fluoride/oxyfluoride deposits provides highly stable corrosion protection for
aluminum in PC and EC-DMC. These results are in agreement with the results of other

9

authors.

3.3. Conclusions

The stability of aluminum during anodic polarization in LiClO4/PC, LiClO4/EC-
DMC, LiPF6/PC, and LiPF6/EC-DMC was studied. It was found that aluminum
passivates in all four electrolyte/solvent mixtures. The susceptibility to
corrosion/oxidation was greater for the mechanically polished than for the electropolished
surface due to thinner and more defective A1203 film on the former. Different
mechanisms of passivation are found in the two electrolytes. At OCP conditions,
aluminum is passivated by both an A1203 layer and adsorbed solvent molecules in
LiClO4 solutions. In LiPF6, the passivation layer consists of A1203 plus some salt
precipitation (e.g., LiF). After anodic polarization, XPS measurements indicate the
formation of highly stable, fluoride rich passivation layer in LiPFG, composed of a
mixture of A1203, AlOF, A1F3, and LiF. On the other hand, the passivation layer formed
in LiClO4 is composed mostly of A1203 and Al(ClO4)3. This layer is the less stable of
the two, seemingly due to slow dissolution of A1(ClO4)3. The potentiostatic and
potentiodynarnic polarization do not indicate any significant effect of the solvent

composition (PC vs. EC-DMC) on the extent of oxidation processes.

59

CHAPTER 4

4. Investigations of Aluminum Corrosion in LiPFdEC-DMC,
LiPF6/PC, LiClOJEC-DMC and LiClO4/PC - Influence

of Water Contamination

4.1. Introduction

The results presented so far indicate that aluminum is effectively passivated in
solution of either LiClO4 or LiPF6 in EC-DMC or PC. Any localized corrosion that
occurs during the first few polarization scans to 5000 mV vs. Li/Li+ is effectively
inhibited by deposition of the oxidation products. However, an interesting question to
address is, what is the eflect of water contamination on the stability of the aluminum?

Interestingly, there has not been much attention paid to the role of water impurity on
the corrosion susceptibility of aluminum in Li-ion battery electrolytes. It is especially
surprising, considering the fact that water is the most common contaminant in the non-
aqueous electrolyte/solvent systems.29 This is because it is extremely difficult to remove
water from a polar solvent down to trace levels. It is also important to realize that water
may form during battery operation as a product of solvent decomposition, or may enter

. . . . . . 24
the battery interior, in the form of atmospheric morsture, due to degraded sealing. Thus,

60

understanding the effect trace amounts of water have on the corrosion/oxidation of
aluminum in non-aqueous electrolytes seems to be an important issue.

The influence of water on the passivation of metals in organic solvents was reviewed
by Kelly and Moran.29 The authors reported that most metals, in neutral anhydrous
solvent/electrolyte mixtures, passivate by either solvent adsorption or salt film formation.
The presence of low levels of water (100 — 1000 ppm) interferes with both of these
processes and leads to increased surface reactivity. Water may affect the adsorbed solvent
by adsorbing competitively on the metal surface with its smaller size relative to organic
solvent being advantageous. Once adsorbed, it may participate directly in the oxidation of
the metal or increase corrosion/oxidation rate by effectively solvating newly created
metal cations. Water also interferes with formation of the salt film by increasing its
solubility. Only at high water concentrations (70—90 mole %) is passivation restored due
to formation of a stable, oxide/hydroxide layer.

Corrosion/passivation of aluminum in lithium-ion battery electrolytes in the presence
of water contamination has received limited attention in the literature. Braithwaite et al.
investigated the effect of low levels of water contamination (20 ppm) on the long-term
stability of aluminum in 1M LiPF6/PC-DEC.4 Based on impedance data, the authors
observed improved passivation in the presence of water. They did not, however,
investigate the reason for the improvement. Kanamura et al. studied the effect of low
levels of HF (0.01 mol/dma) on the stability of aluminum in LiCF3SO3/PC and
Li(CF3SOz)2N/PC.6 Improved passivation was observed. Since HF was added as an

aqueous solution, the effect of water could not be separated in their research.

61

Presently, the effect of water contamination on the corrosion/oxidation of high purity
aluminum (99.999%) in LiPFdEC-DMC, LiPF6/PC, LiClO4/EC—DMC, and LiClO4/PC
was investigated. 50 — 2000 ppm levels of water were studied through controlled
additions of water to the solvent/electrolyte mixture. Testing was performed using two
different metal surfaces, (1) aluminum with the native oxide film removed by mechanical
polishing in hexane under an inert, argon atmosphere, and (ii) aluminum, first
mechanically polished, and then electropolished in an aqueous solution of the HBF4.
Cyclic voltammetry (CV), chronoamperometry (CA), and open circuit potential
measurements (OCP), were used, in addition to optical microscopy (OM), atomic force
microscopy (AFM), and scanning electron microscopy (SEM), to study morphological
changes brought about by the imposed electrochemical conditions. Energy dispersive x-
ray analysis (EDS), and Raman spectroscopy were used to probe the chemical

composition and structure of the surface before and after polarization.

4.2. Results and discussion

4.2.1. Corrosion of aluminum in LiClOdEC-DMC/Hzo and LiClOJPC/HzO
mixtures - mechanically polished surface

Figure 4.1 shows cyclic voltammetric i-E curves (5mV/s) for mechanically polished
aluminum in 1 M LiClO4/EC-DMC with 500 ppm of added water. The voltammetric
response is much different in the presence of water impurity than in the dry electrolyte
(Chapter 3). The first positive scan resembles, to some extent, the one observed for the

dry electrolyte (see Figure 3.3A). A gradual increase in the anodic current starts at 2700

62

mV and continues until ca. 4600 mV at which point the current increases significantly.
There is current crossover at the positive limit of 5000 mV such that the current during
the reverse scan is larger than during the forward scan. This is characteristic of a
corrosion process in which breakdown of a passivation layer occurs on the forward scan.
In case of the anodic polarization of aluminum, this behavior usually indicates pitting
corrosion.28 At the potential Epit, or ca. 4600 mV, the passive layer breaks down and
current flows due to the oxidation of the exposed aluminum. During the reverse scan, a

large anodic current flows for the same range of potentials until repassivation occurs at

Epass, at ca. 3600 mV.

350

 

300
250.
200

150-
100*
50’

Current density [llA/cmz]

 

 

 

 

O n l 1 l I 1
2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

 

Figure 4.1. Cyclic voltammetric i-E curves recorded for mechanically polished
aluminum at room temperature in 1M LiClO4/EC-DMC with 500 ppm of added H20.

Scan rate = 5 mV/s.

63

The second and third potentiodynarnic scans also show characteristics of pitting
corrosion with current crossovers. The current decreases with each scan, however,
indicating a reduction in the active area. Also, Epit and Epass change. The onset potential,
or Epit, for the anodic current is ca. 4000 mV. There is a peak current, at 4200 mV and
4300 for 2nd and 3rd scans, respectively, during the forward sweep. The current continues
the decrease after reversing the scan at 5000 mV with a small current crossover near 4000
mV. The current and charge associated with the crossover decrease with each scan.
Repassivation, or Epass, occurs near 3500 mV for all the scans. This behavior suggests
that water contamination causes localized pitting of the aluminum, but the oxidation
products that form passivate the surface to some extent. However, even after the 3rd
cycle, there is still a significantly greater anodic charge passed during the forward scan
than is present in the dry electrolyte. Therefore, this level of water impurity appears to
cause extensive corrosion and oxidation of the aluminum surface. Similar results were
obtained when PC was used as a solvent.

Table 4.1 summarizes the potentiodynarnic polarization results for 1M LiClO4/EC-
DMC and 1M LiClO4/PC with different levels of added water. OCP for dry and for 2000
ppm solutions are also presented. Each measurement was performed on a freshly,
mechanically polished surface (hexane and inert Ar atmosphere). The OCP shifts more
negative (active) with added water in both electrolyte/solvent systems, with the larger
shift seen in LiClO4/PC. Also, and most importantly, the anodic charge passed during the
lSt scan increases pr0portionally with the added water. This indicates that water is an
active participant in the corrosion process. Interestingly, at 50 ppm of added water, the
current crossover in the i-E curves was not observed during the first two scans. This
implies that water is essential for activation of the surface and only after the activation
occurs, does the surface undergo corrosion. At lower water levels activation time is

longer, and thus pitting is not observed during the initial scans.

.cotomno Ho: 83 magi - fit Mgr— .m> > o6 8 Wm 8on me? @0538 owqfi

3:5qu 05. 838 m u 28 98m .683 page we m~o>2 82:56 5:5 “EEO—US pee UEQOEVQUE
E BEES? 3523 3335588 How 83 mam—93 oggcoueoeom 05 mo .Cafifism 4.9 039—.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3 N N: r... r... r... r... t... t. 88 o
N S 8 88 r... r... 8:. e... r... - 8

8 8 R 88 88 88 88 88 88 - o8

o: o: a: 8m... 88 88 88 88 83. 8:: 88

9806:

o.N 3. 2 e... r... r... r... t... r... 88 o
m N: 8 8:. r... s... 8:. r... t... - 8

8 mm 8 88 88 8mm 8:. 88 8:. - o8

8. 8: ON 88 88 88 88 88 8:. 88 88

02908.66: 0N:
88o
m N H N N 2 m N 2 F53 9 >3 mo 88
.5956 zoom .325: Sam 398:: 58. 3 cs 8H

 

 

made. so

 

Cd: .9 25 use

 

Cd: as >5. see

 

 

 

 

65

Similarly to what was observed in dry electrolytes, mechanical polishing did not
generate fully reproducible surface as the variations in the currents for experiments
performed at virtually identical conditions, were in the order of 20-40%. However, the

shapes of voltammograms and pitting and passivation potentials were reproducible.

The formation of a partial passive layer after the 3rd scan was observed by SEM.
Figure 4.2 presents representative SEM images of a mechanically polished aluminum
surface prior to (Figure 4.2A) and after 3 potentiodynarnic scans from 2000—5000 mV
vs. LilLi+ in 1M LiClO4/EC-DMC with 500 ppm of added H20 (Figure 4.2B, C, and D).
A relatively smooth surface is observed for aluminum before polarization. Multiple
grooves, which originated from mechanical polishing, are clearly visible. On the other
hand, a thick, highly porous layer of the corrosion product covers the surface after
polarization. The deposit is not strongly attached to the aluminum surface as it can be
easily removed by washing with DMC. Figure 4.2C shows the structure of the aluminum
surface after the surface layer was removed. A new, highly damaged surface is observed.
Multiple cracks and pores indicate significant corrosion and dissolution of aluminum.
Images of the cracks obtained at higher magnification show clearly that damage reaches
deep into aluminum electrode. This is well visible on the image of one of the cracks
presented in Figure 4.2D. A similar, weakly passivating layer and highly damaged
aluminum surface were also observed after polarization in the LiClO4/PC containing 500
ppm of added H20.

SEM images confirm what the voltammetric data indicated, that corrosion of the
aluminum surface occurs when water is present in the LiClO4/EC-DMC or LiClO4/PC.

The resulting corrosion/oxidation product deposits do not completely passivate the

66

surface due to a porous and defective morphology.

 

Figure 4.2. SEM images of the mechanically polished aluminum surface. (A) An
untreated and (B,C, and D) after 3 potentiodynarnic scans from 2000—5000 mV vs. Li/Li+
in LiClO4/EC-DMC with 500 ppm of added H20. (C, D) The images are the same as (B)
but the deposited layer has been removed by washing with DMC.

The stability of the passivating layer was also investigated in the following
experiments. First the mechanically polished electrode was exposed to five
potentiodynarnic cycles in 1M LiClO4/EC-DMC + 500 ppm H20. The oxidation charge
recorded during the fifth cycle was equal to 35 mC/cmz, and Epit and Epass were 4000
and 3550 mV vs. Li/Li+, respectively. The electrolyte was then turbulently mixed by

argon bubbling, and three consecutive potentiodynarnic scans were recorded between

67

2500-5000 mV. The oxidation charge increased to 72 mC/cm2 for the sixth (total) scan
and then again decreased for the consecutive scans. Epit initially decreased to 3800 mV
for the sixth scan and then increased to 3950 mV with further scans. ElmS was practically
the same for all scans. The increased oxidation charge and decreased pitting potential
after disruption of the passivating layer, indicates that indeed, the passivation layer
formed on the aluminum surface in the presence of water does not provide a good
protection to the surface, and only diminishes corrosion rates through blockage of some
of the active area. After disruption of this layer, the corrosion proceeds at increased rates
until a build-up of a new corrosion product layer occurs.

Second, 50 potentiodynamic scans at 5 mV/s, in 1M LiClO4/EC-DMC with 500 ppm
of added H20, were performed on the mechanically polished electrode to achieve a stable
oxidation charge of 16 mC/cmz, Bpit = 3850 V, and Epass = 3550 mV. The electrode was
then soaked in the solution for 1h at OCP. This was followed by a single potentiodynarnic
scan. The oxidation charge for this scan increased to 29 mC/cmz. Interestingly, Bpit also
increased by 200 mV, whereas EpaSS remained the same. The increased Epit indicates
some passivation of the electrode occurs at OCP. This may be explained by reaction of
aluminum metal with water. It is well known that bare aluminum reacts spontaneously
with water to form an aluminum oxide/hydroxide layer, which gives aluminum its
passivity in aqueous solutions.1 Such a reaction, although hindered, can also take place in
non-aqueous solutions with added water. It leads to coverage of the active aluminum sites
by an oxide/hydroxide layer. At OCP, water diffuses to the metal surface, reacts with
metal, and forms a stable film that increases the initial corrosion resistance of metal.
However, during anodic polarization, aluminum dissolves rapidly. Due to the limited

amount of water present, a well passivating oxide/hydroxide layer cannot form.

68

4.2.2. Corrosion of aluminum in LiClOdEC-DMC/HzO and LiClOJPC/Hzo
mixtures — electropolished surface

It is well known that a layer of aluminum oxide passivates the surface and increases
the corrosion resistance of the underlying metal. To determine if the A1203 layer can
prevent corrosion in LiClO4/EC-DMC/I-120 and LiClO4/PC/HZO mixtures, anodic
polarization of electropolished aluminum was performed. Figure 4.3 shows
potentiodynarnic scans recorded at 5 mV/s, in 1 M LiClO4/PC + 200 ppm H20. The
aluminum oxide layer seems to passivate the surface during the first three scans, as the
current density is similar to that observed in the electrolyte mixture without added water.
However, closer analysis of the results of the third scan reveals characteristic crossover of
the i-E curve, suggestive of pitting corrosion. Further cycling clearly reveals the
incidence of corrosion (Figure 4.3B). In addition to current crossover, the anodic charge
continuously increases during cycling. This is an indication of an increasing
electrochemically active area due to corrosion.

The surface of the electropolished aluminum electrode after anodic polarization was
investigated by optical microscopy. Figure 4.4 presents representative photomicrographs
of the surface after the 3rd and 50th (conditions the same as in the Figures 4.3 A and B)
potentiodynarnic scans in 1 M LiClO4/PC + 200 ppm of H20. During the initial stages of
corrosion, multiple circular pits (l - 6 um diameter) are randomly formed on the surface.
The pits grow in diameter and new pits develop over time, with neighboring pits
coalescing as the corrosion proceeds. Eventually, the oxide film is completely destroyed,
with only isolated regions of the film (smooth areas) intact after the 50th scan. This

observation explains the progressively increasing current density observed during the

69

early stages of the potentiodynarnic scans, as presented in Figure 4.33. Similar results

were also obtained when EC-DMC was used as the solvent.

A

N
O

 

H
KI!
I

 

 

 

Current density [ ILA/01112]
ES

 

 

 

OPT: ' l 4 l r l r l r i .
2000 2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

 

oo
o
w

 

50‘“

\I
O
l

 

Ln 0
O O
I'V‘IUI I

N OJ
0 O
UTI'V'IUVI

Current density [liA/cmz]
'5‘ 8

 

 

 

 

0 C
_10 ' . i m 1 1
200 2500 3000 3500 4000 4500 5000

Potential [mV vs. Li/Li+]

 

l n l n l n

 

Figure 4.3. Cyclic voltammetric i-E curves recorded at room temperature for an
electropolished aluminum electrode in 1M LiClO4/PC + 200 ppm of H20 during (A) the
first three scans and (B) scans 5 through 50. Sean rate = 5 mV/s

70

 

Figure 4.4. Photomicrographs of the electropolished aluminum surface after 3 (A) and 50
(B) potentiodynarnic scans in 1M LiClO4/PC + 200 ppm of H20 (conditions as in Figure
4.3).

71

Pit size and density were also evaluated as a function of the amount of water
impurity present in the electrolyte/solvent mixture. For this purpose, 3 potentiodynarnic
scans in electrolyte solutions with different amounts of added water were performed on
electropolished aluminum electrodes. Optical micrographs of three different areas on
each electrode were recorded after polarization. The number of pits and their sizes were
then estimated from the micrographs. Figure 4.5 presents results of such measurements in
1M LiClO4/EC-DMC with 200, 500 and 1000 ppm of added water. The total anodic
charge recorded after the 3rd potentiodynarnic scan increased in the order of increasing
amount of water (21, 59, and 123 mC/cnn2 for 200, 500, and 1000 ppm of added water,
respectively), indicating more extensive corrosion at higher water levels. The flow of
anodic charge was accompanied by a significant pitting of the aluminum surface.
Interestingly, average pit diameter increased with increasing water concentration,
whereas at the same time, pit density decreased. Average pit diameters were 3.0 :1: 1.3
(43 % RSD), 5.3 :l: 2.1 (39 % RSD), and 9.2 i 3.1 (34 % RSD) run, and pit densities were
8500 :l: 800, 4400 :1: 600, and 2600 :l: 500 pits/mm2 at 200, 500, and 1000 ppm of water,
respectively. These results indicate that at higher water concentrations, corrosion
proceeds primarily through the development of pits formed during the first scan, whereas
at lower water levels, development of already existing pits is accompanied by formation
of the new ones. It is possible that at lower water levels pits partially repassivate through

deposition of corrosion products and thus corrosion through pit development is inhibited.

72

 

p—i
O
O
O
O
tit—1
h

 

 

 

 

—I:I— Pit density l 12
RE 8000 _ .........0 Pit diameter . '3
(D
g l. .1 10050;
=3 6000 - ‘ ‘”
3‘ ‘ 8 E‘
.5 . a:
§ 4000 - j6 g
~53 - 4 ft
2000. . £1
_ 1:
.2 a.

O r a r l . . . 1 r 1 n l . 1 1 1 I O

0 200 400 600 800 1000 120
Water concentration [ppm]

C

Figure 4.5. Pit density and average pit diameter as a function of concentration of water
impurity in 1M LiClO4/EC-DMC. Data recorded after 3 potentiodynarnic scans from
2000 to 5000 mV vs. Li/Li+, at 5 mV/s.

Higher magnification (x 2400) microscopic evaluation of the pits formed on the
aluminum surface after potentiodynarnic cycling in LiClO4 in either PC/HZO or EC-
DMC/HZO showed another interesting trend. In general, pits formed at lower water levels
(< 500 ppm), were filled with corrosion products, indicating the deposition of the
products inside the pit. On the other hand, some of the pits formed at higher water levels
(1000ppm) seemed to be free of any corrosion products. Deposition of corrosion products
at lower water levels was also observed by the EDS-SEM and Raman spectroscopy data,
which are presented in the following section of this chapter, as well as by AFM imaging,

which is discussed in Chapter 5.

73

Development of the corrosion pits and their coalescence during the polarization
scans was also observed by in-situ optical microscopy. A Teflon cell with an optical
window was constructed for this purpose, as described in Chapter 2. Measurements were
performed in such a way that the aluminum electrode was first subjected to three
potentiodynarnic cycles from 2500 to 5000 mV vs. Li/Li+, at 5 mV/s, so the formation of
the pits could be easily observed on the surface with a lower magnification lens (x 300).
A higher magnification lens (x 2400) was then used to image the pitted area during a
subsequent potentiodynarnic scan. The potentiodynamic i-E curves showed clear
evidence for corrosion with Epit = 4050 and Epass = 3450 mV, and with shapes similar to
those presented in Figure 4.3B. A sequence of images, acquired at different potentials
during the 4th potentiodynarnic scan, is presented in Figure 4.6. The first picture (top,
left) presents an image recorded at the beginning of the 4th potentiodynarnic scan, at
potential of 2500 mV vs. Li/Li+. Three, similarly shaped circular pits (~ 8 pm in
diameter) are clearly visible on the surface. As the potential sweep proceeds, grth of
the pits is observed. The last picture (bottom, right) presents an image recorded at the end
of the potentiodynarnic scan, at 2500 mV. It is evident that pit diameter increased nearly
two times, to ~ 15 um, during the span of potential cycle. A coalescing of the
neighboring pits is also visible. The lighter areas around the pits are due to lifting of the
oxide layer by the growing corrosion products. This is further discussed in Chapter 5.

It is clear that the evolution of pits occurs only in the range of potentials where the
oxidation current flows. For example, the size of the pits remained constant in the
potential range 2500 — 4000 mV during the forward scan, and 3300 — 2500 mV during

the reverse scan. The pits grew, however, above Epit, where significant oxidation current

74

flowed and did not cease until Epass. This provides direct evidence that aluminum

corrosion and dissolution occur only at the highly oxidizing potentials.

 

20 um

Figure 4.6. Photomicrographs of the electropolished aluminum surface, recorded in-situ
during the 4th potentiodynarnic scan at 5 mV/s in 1M LiClO4/EC-DMC with 1000 ppm
of added H20. Magnification 2400 x.

75

4.2.3. Corrosion of aluminum in LiClO4/EC-DMC/HzO and LiClO4/PC/HzO
mixtures — chronoamperometric studies

Breakdown of the passivating A1203 film on electropolished surfaces was also
observed in chronoamperometric measurements. Figure 4.7 presents chronoamperometric
i-t curves in 1M LiClO4/PC with 2000 ppm of added water, for the anodic steps from
2500 to 4500 and 5000 mV. For comparison, the i-t curves obtained in dry electrolyte, at
the same potentials, are shown as dashed lines. In the dry electrolytes, the current rises
quickly and then decreases as the electrode double layer charging is completed. The
current decays to a low and steady value of ca. 0.6 and 0.8 liA/cm2 for potential steps to
4500 and 5000 mV, respectively. The charge passed is in the range of 5 - 7 mC/cmz. The
low currents result from the passivation of the surface by the oxide layer. Completely
different i-t behaviors are observed in the electrolyte with added water. Breakdown of the
oxide film is clearly visible during the potential step to 4500 mV (Figure 4.7A). The
current quickly rises and decays as the double layer charging is completed. However, at
the 2 min mark, the current begins to increase before reaching a maximum at the 30 min
mark. Thereafter, the current remains fairly constant. The current at the end of the step is
55 uA/cm2 and is much larger than the 0.6 uA/cm2 seen for the dry solvent. The anodic
charge passed is 170 mC/cmz. The i-t curves for the 5000 mV potential step have similar
shapes (Figure 4.7B). The inset shows the i-t curves for the first 3 min. Once again, a
second current increase is observed. At the higher applied potential, the induction period
is greatly reduced, from 2 min to 0.2 min, and the maximum current is reached after
about 0.5 min. The current decreases at much higher rate after the second current

maximum, than was seen for the 4500 mV step. The current at the end of the step is ca.

76

50 uA/cmz, about the same as for the 4500 mV step. The total charge is, however, much
larger, 360 mC/cmz. Clearly, the extent of the oxidation is greater at the higher potential,
but the rate of oxidation at the 60 min mark is the same for both potentials. The curves
indicate that exposure to potentials of 4500 mV or greater, causes localized breakdown of
the passivating oxide film. Pit growth occurs at a diffusion-limited rate, as evidenced by

the near steady state or decreasing current density after the second current maximum.

Photomicrographs of the electropolished aluminum surface after 1 h potential steps
to 4500 and 5000 mV in 1M LiClO4/PC with 2000 ppm of added water are presented in
Figure 4.8 A and B. Complete destruction of the smooth oxide film is observed after
exposure to 5000 mV, whereas some areas of intact oxide layer are visible on the surface
after exposure to 4500 mV. Interestingly, compared to the micrograph of the surface after
50 potentiodynarnic scans (Figure 4.4), a much higher pit density is observed, with an
overall lower pit diameter. This indicates that during potentiodynarnic cycling, corrosion
proceeds primarily through development of already existing pits. There is enough time
for the electrolyte/solvent mixture to diffuse to the pitted area when electrode potential is
below Epit- Since the corroded area is more prone to further corrosion than a neighboring
oxide-coated surface, corrosion proceeds by further development of the existing pits.
Under potentiostatic conditions, the greater charge causes the oxide film to break down at

multiple locations. A higher pit density and smaller pit diameter are thus observed.

77

 

 

 

 

 

 

 

 

 

 

 

 

 

 

150
”a
0
E100 -
.E‘
U)
c:
Q)
.‘i. 50-
C
Q)
t:
:3
U
0 M"
0 10 20 3O 40 50 60
Time [min]
B
._. 800 800.
5 70° e600.
2 600 i) 400:
3 500 5"
g3 200
g 400;
<12 ‘ o ‘ .
“U 300,L 0.0 0.5 1.0 1.5] 2.0 2.5 3.0
E : t[min
g 200:
:3 i
U 100t
0}‘_.__l____.h__-_1___._l____1__—‘
O 10 20 3O 4O 50 60
Time [min]

Figure 4.7. Chronoarnperometric i-t curves recorded at room temperature for an
electropolished aluminum electrode in 1M LiClO4/PC. (A) Curves for a potential step
from 2500 to 4500 mV. (B) Curves for a potential step from 2500 to 5000 mV. Solid line
- electrolyte solution with 2000 ppm of water added. Dashed line — electrolyte with no

added water.

78

 

Figure 4.8. Photomicrographs of the electropolished aluminum surface after a l h
potential step to (A) 4500 and (B) 5000 mV in 1M LiClO4/PC with 2000 ppm of added

water.

79

Mechanically polished aluminum electrodes were also evaluated by
chronoamperometry. Table 4.2 summarizes the chronoamperometric results in
1M LiClO4/EC-DMC and 1M LiClO4/PC with different levels of added water. Each
measurement was performed on a freshly, mechanically polished surface (hexane and
inert Ar atmosphere). The potential was stepped from 2500 to 5000 mV vs. Li/Li+ for 1
h. Clearly, both the anodic charge and the current at the end of the step in the two
solvents, increase with increasing amount of added water. This increase is not directly
proportional, however. There is a much larger increase when the amount of water is
quadrupled from 500 to 2000 ppm, than when it is increased ten times, from 50 to 500
ppm. This may indicate that, at higher water content, some other process takes place on
the surface. A possible process is increased solubility of the Al(ClO4)3 layer formed on
the surface, as it was proposed by Kelly for iron in LiClO4/PC. This result is in
agreement with previous voltammetric results, which indicated increased corrosion at
water levels higher than 500 ppm. Also, both the oxidation current and charge at the 1 h
mark are higher in EC-DMC than in PC. This may be explained by higher viscosity of PC
and thus lower mobility of water in this solvent.

The chronoamperometric results for electropolished aluminum in 1M LiClO4/EC-
DMC and 1M LiClO4/PC with 2000 ppm of added water are also presented in Table 4.2,
for comparison. Although only the 2000 ppm water level was investigated on
electropolished surfaces, it is clear that both the oxidation current and charge at the l h
mark are much lower on the electropolished surface than on the mechanically polished
one, at the same level of added water impurity. This is in agreement with our
voltammetric results, which indicated some protective properties of the aluminum oxide

layer.

80

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

LiClO4/EC-DMC LiClO4/PC
ppm of 2 2 2 2
added H20 Qox [mC/cm ] on [uAlcm ] Q0x [mC/cm ] 10X [p.A/cm ]
mechanically polished

2000 750 105 600 80

500 265 40 200 30

50 245 35 180 25

O 13 2-5 20 3.0

electropolished
2000 420 55 360 50

 

 

 

 

 

 

 

Table 4.2 Summary of the chronoamperometric data for mechanically polished and
electropolished aluminum. Potential step from 2500 to 5000 mV vs. Li/Li+, for l h.

Oxidation charge and current data were collected at 1 h mark.

4.2.4. Chemical composition of the corrosion products in LiClOJEC-DMC/HZO and
LiClO4/PC/H20 mixtures — energy dispersive x-ray microanalysis (EDS) and
Raman spectroscopy studies

The elemental composition of the precipitated corrosion products was investigated
with energy dispersive x-ray microanalysis (EDS). Figures 4.9 and 4.10 present
representative x-ray analysis data for the mechanically polished aluminum surface before
and after polarization (3 potentiodynarnic scans at 5 mV/s; 2000-5000 mV vs. Li/Li+) in
1M LiClO4/PC with 500 ppm of added H20. Black crosses on the SEM images indicate
the sampling points. Before polarization, Al, O and traces of C were the only elements

detected on the surface. A highly porous layer of corrosion products forms on the surface

81

during polarization, as evidenced in the SEM image in Figure 4.10. A large signal for C1
is present and the 0 signal increases over that observed for the surface prior to
polarization. These results suggest that the perchlorate anion plays an active role in the
corrosion of aluminum.

Somewhat similar results were also obtained for the electropolished surface. Figure
4.11 (top) presents a SEM image of the surface after 3 potentiodynarnic scans (5 mV/s;
2000-5000 mV vs. Li/Li+) in 1M LiClO4/PC with 200 ppm of added H20. Corrosion pits
(~ 4 pm in diameter) are clearly observed in SEM image. Energy dispersive x-ray
analysis spectra acquired inside and outside of the pit are also present in Figure 4.11.
Clearly, the pits present are filled with corrosion products. The elemental composition of
the corrosion products within the pits is similar to the composition of the corrosion
products on the mechanically polished surface, based on the peak intensity ratios in the
spectra. On the other hand, neither Cl nor increased amounts of 0 were detected on the
undamaged A1203 regions (spectrum 2). This indicates that the pitting corrosion process
is very localized. However, for higher water concentrations (2000 ppm), clean pits were
often observed. EDS indicated much lower amounts of C1 or O for such a case (Figure
4.12). This observation suggests that at higher water concentrations, the corrosion
products are highly soluble and afford little protection to the corroding metal. Analogous

results were also obtained for LiClO4/EC-DMC/I-120 mixtures.

82

 

 

2000 1

Counts

 

 

 

0.000 keV 10.240

 

 

Figure 4.9. An SEM image and an energy dispersive x-ray analysis spectrum for

mechanically polished aluminum prior to polarization.

83

 

 

2000J

Cl

Counts

 

C CI
0.000 keV 10.240

 

 

 

 

 

Figure 4.10. An SEM image and an energy dispersive x-ray analysis spectrum for
mechanically polished aluminum after potentiodynarnic polarization (3 scans at 5 mV/s;
2000-5000 mV vs. Li/Li+) in 1M LiClO4/PC + 500 ppm H20.

84

 

 

 

 

 

 

 

 

 

 

100d M 1
c:
a o
E
U
C]
o C
0.000 k_eV 10.240
40d A] 2
:9
E
U
C A]
A "1'?th 'v‘:“‘ r; 45.1... -
" 0.000 keV 10.240

 

 

 

Figure 4.11. An SEM image and energy dispersive x-ray analysis spectra for
electropolished aluminum after potentiodynarnic polarization (3 scans at 5 mV/s; 2000-
5000 mV vs. Li/Li+) in 1M LiClO4/PC + 200 ppm H20. Spectrum 1 corresponds to a

region inside the pit while spectrum 2 corresponds to a region outside.

85

 

 

—

400

Counts

Cl

‘Y‘ wmv' "WW TAN-‘IL' Y1.A_'-.'._ ... .4 A‘
""V - Vr ‘F

keV

 

 

 

 

 

0 0. 000

 

Figure 4.12. An SEM image and an energy dispersive x-ray analysis spectrum for
electropolished aluminum after 2 potentiodynarnic scans (5 mV/s; 2000-5000 mV vs.

Li/Li+) in 1M LiClO4/PC + 2000 ppm H20.

86

The deposited corrosion products were also investigated by Raman spectroscopy.
Figure 4.13 shows a Raman spectrum for corrosion product in a pit formed on an
electropolished aluminum surface during anodic polarization in 1M LiClO4/PC with 500
ppm of added H20. The spectrum was collected in air, after the electrode was removed
from the electrochemical cell, washed with DMC, and dried in flow of Ar. Three sharp
peaks are present in the spectrum. They were identified as modes characteristic of C104:
The peaks at 457 cm.1 and 630 cm'1 are associated with the v2(e) and V4(f) vibrations,
respectively, and the peak at 933 cm-1 is due to a symmetric stretch v1(a1) of the
perchlorate anion.41 Two small peaks at 715 and 845 cm.1 as well as a group of three
peaks centered at 2940 cm'1 were assigned to PC residue.41 The peaks at 715 and 845
cm.1 , are associated with out-of-plane C=O ring bending and ring breathing, respectively.
Three peaks around 2940 cm.1 are associated with C-H stretches. The broad peak
observed at 3500 cm-1 is most probably due to the presence of mixed aluminum
hydroxides and is associated with (OH)-stretching.42 For comparison, Raman spectra of
the unpolarized electropolished surface, bulk LiClO4 and Al(ClO4)3-H20 crystals are
presented in Figure 4.13. A red shift of the symmetric v1(a1) stretch (933 vs. 955 cm-1)
for the corrosion product is seen, and this indicates free C104: A similar shift is also
observed for hydrated perchlorate salts.43 Analogous results were also obtained for
LiClO4/EC-DMC/I-120 mixtures.

The presence of surface C104- was also observed for the mechanically polished
electrodes after polarization. This indicates that the mechanism of the corrosion process
is the same for both electropolished and mechanically polished electrodes. The difference

is the extent of the corrosion and the amount of corrosion product on the surface.

87

 

23.1(c104)3

T I I

2000 L LiClO4

l' T

j

1000 ~— Pit area

: “Newly polished

0 . 1 . n . .
O 500 1000 1500 2000 2500 3000 3500 4000

Raman shift [cm'l]

 

 

 

Figure 4.13. Raman spectrum of the corrosion product in the center of a pit on
electropolished aluminum. The pit and corrosion product were formed after 2
potentiodynarnic scans from 2500 to 5000 mV vs. Li/Li+ in 1M LiClO4/PC with 500
ppm of added H20. Scan rate = 5 mV/s. Spectra for the freshly electropolished surface,
and solid samples of LiClO4 and Al(ClO4)3-H20 are shown, for comparison.

The spatial distribution of Raman features across a pitted area is shown in the line-
imaging profile in Figure 4.14. A photomicrograph of the pit formed and the line profile
are shown in Figure 4.14A. Figure 4.14B presents the set of 14 Raman spectra collected
along the profile. Clearly, C104- is present only inside the pit. The presence of C104- in
the corrosion product is probably caused by ion migration into corroding area and

formation of solid Al(ClO4)3.

88

1000

S

\:j

l

\ U!
8
Kirsuoun untrue};

 

 

Figure 4.14. (A) Photomicrograph of a corrosion pit on the electropolished aluminum
surface after 2 cyclic voltammetry scans between 2500 and 5000 mV vs. Li/Li+ in 1M
LiClO4/EC-DMC + 500 ppm of water. Scan rate = 5 mV/s. The black line indicates
Raman spectra line profile. (B) A series of Raman spectra obtained along the line shown

in 1 um increments. Acquisition time = 10 min.

89

4.2.5. Mechanism of aluminum corrosion in LiClOdEC-DMCIHZO
and LiClO4/PCIHZO mixtures

The results presented in Chapter 3 show that aluminum passivates in dry LiClO4/EC-
DMC and LiClO4/PC solutions. Two mechanisms are operational. At OCP conditions,
aluminum passivates by solvent adsorption, and during anodic polarization by formation
of a salt film, composed primarily of A1(ClO4)3. Also, the oxide film, formed on the
surface during electropolishing, is stable during anodic polarization and additionally
passivates the surface. On the other hand, the results of electrochemical, microscopic, and
spectroscopic investigations presented in this chapter indicate that water interferes
strongly with the passivation of aluminum in LiClO4/EC-DMC and LiClO4/PC. Severe
corrosion of either mechanically polished or electropolished surfaces is observed when

the solvent is contaminated with water, particularly at levels 2 500 ppm.

Previously, Kelly et al. studied the influence of water on the passivity of iron in
LiClO4/PC.31’44 They observed that water interferes with all three passivation
mechanisms (solvent adsorption, air-formed oxide film, and salt film formation)
commonly observed in non-aqueous electrolyte/solvent mixtures. The authors showed
that water levels as low as 100 ppm are detrimental to the stability of iron in LiClO4/PC
solutions. On bare iron surfaces, water molecules compete with the solvent molecules for
adsorption sites on the surface. The presence of surface water lowers the activation
energy for metal dissolution, destroying the passivation provided by a chemisorbed

solvent layer. The presence of an air-formed oxide film on the surface did not prevent

iron from corroding in electrolyte solutions with added water. This was attributed to the

90

oxidation of water and acidification of electrolyte solution. Under such conditions,
localized breakdown of the air-formed film occurs. Finally, the presence of water led to
increased solubility of the Fe(ClO4)2 layer formed on the surface during polarization,
thus destroying the passivity provided by salt film.

A similar mechanism in which water interferes with the passivation of aluminum in
LiClO4/EC-DMC and LiClO4/PC is proposed and is consistent with the experimental
observations. First of all, water can interfere with the adsorption of the solvent molecules
at OCP conditions. This seems to be true as the OCP values were lower (i.e., more active)
in both LiClO4/PC and LiClO4/EC-DMC in the presence of water. The effect was
especially pronounced in PC, in which the addition of 2000 ppm of water lowered the
OCP by ~400 mV. This can be explained by strong competition between water and
organic solvent molecules for the adsorption sites on the metal/metal oxide surface.
Zelenay and coworkers observed previously that water displaces PC molecules and
adsorbs on the platinum electrode surface.45 The maximum measured surface
concentration of water reported by the authors was ~ 1.4 x 1014 molecules/cm2 (in 0.1M
LiClO4/PC with 1500 ppm of added water).

During anodic polarization, water may react with aluminum to form a mixed

aluminum oxide and hydroxide layer according to the following reactions:

2A1 + 31120 -) A1203 + 6 H+ + 6 e- [4.1]
A1 + 31120 —> Al(OH)3 + 3 Hr + 3 e’ [4.2]
A1 + 21120 —> AlOOH + 4 H+ + 4 e' [4.3]

91

Such a bilayer greatly improves passivity of aluminum in aqueous solutions.1
However, since only limited amounts of water are present in the non-aqueous
electrolyte/solvent mixture, the oxide/hydroxide layer is not continuous, which
significantly limits its protective capabilities. In fact, Kelly et al. showed that corrosion of
iron occurs in LiClO4/PC when water at the levels of 5 mM (75 ppm) to 17 M (26 %) are
present in the electrolyte/solvent rr1ixture.31’44 Only at water levels exceeding 17 M limit,
is stable passivation restored due to the formation of a continuous oxide/hydroxide layer.

A non-uniform oxide film was observed in present studies. SEM imaging shows a
highly porous passivation layer on the surface of mechanically polished aluminum after
anodic polarization in LiClO4/EC-DMC/HzO and LiClO4/PC0120.

Reactions 4.1-4.3 also indicate a possible explanation for the low resistance of the
oxide film present on the electropolished surface toward pitting corrosion in LiClO4/EC-
DMC/1120 and LiClO4/PC0120. These reactions lead to localized acidification of the
electrolyte mixture. It is well known that, in aqueous solutions, pitting of aluminum is not
likely in neutral solutions. It occurs, however, in acidic environments.l

Another reaction is also likely to induce pitting of the electropolished surface. XPS
results for the aluminum surface after anodic polarization in LiClO4/EC-DMC, presented
in Chapter 3, indicate that at high anodic potentials, oxidative decomposition of C104-
occurs and Cl- forms. The corrosive properties of C1. are well established in aqueous
solutions. Cl- as a small ion with high diffusivity, tends to penetrate the oxide layer at
weak points (flaws) enhancing dissolution of the base metal, thus leading to pitting.l’28 A
similar effect of the Cl- can be expected in PC/HzO and EC-DMC/HZO solutions.

Interestingly, pitting of the aluminum surface is not observed in dry electrolytes which

92

may be explained by the low activity of C1- caused by strong ion pairing observed in
organic carbonates.46 The synergism between water and Cl- was also observed before, by
Kelly, for corrosion of iron in LiClO4/PC.44

Finally, water can increase dissolution of the salt film formed during anodic
polarization. As XPS, EDS, and Raman spectroscopy data show, the Al(ClO4)3 is a
product of aluminum dissolution during anodic polarization in either dry or water
contaminated solvents. The formation of Al(ClO4)3 in dry solvents effectively passivates
the surface due to supersaturation and precipitation of the salt deposit at the active sites
on the metal surface. The presence of water increases the solubility of Al(ClO4)3, thus,
diminishing the passivation. In the present studies, this effect was especially visible at

higher water concentrations (> 500 ppm) as empty pits were observed on the surface.

4.2.6. Corrosion of aluminum in LiPFdEC-DMC/HZO and LiPFdPC/HZO mixtum
The addition of water to UPFdEC-DMC or LiPF6/PC did not have as much of an
effect on the susceptibility of aluminum to corrosion, as it did in the LiClO4-based
electrolyte/solvent systems. Passivation of the aluminum was observed to occur after the
first potentiodynamic scan, similar to what was seen for the water free electrolytes.
Figure 4.15 presents potentiodynarnic i-E curves (5 mV/s) for the first cycle in 1M
LiPF6/EC-DMC with 500 ppm of added H20. Curve (A) is the response for the
mechanically polished electrode, (B) the response for the mechanically polished electrode
after 10 h at OCP, and (C) the response for the electropolished electrode. For comparison,
the response for the mechanically polished electrode during polarization in solution

without intentionally added water is showed in (D). Since consecutive cycles did not

93

show any signs of further oxidation or corrosion of the aluminum surface, they are not

presented here for clarity.

 

 

 

 

 

60,
“E50:
0 P
2 4o
3:, :
330i
U)

S .
.g 20_
E

<1)

1: 10
:3 .
U 0

-10+ - e

 

 

2000 2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

Figure 4.15. Cyclic voltammetry (1St scan) in 1M LiPF6/EC-DMC + 500 ppm H20. (A)
— mechanically polished aluminum; (B) — mechanically polished aluminum after 10 h at
OCP in the same electrolyte mixture; (C) electropolished aluminum; (D) — mechanically

polished electrode, no added water. Scan rate = 5 mV/s.

As can be seen, oxidation current and charge recorded during the first cycle in
electrolyte solution with added water are smaller than in dry electrolyte. This indicates
that the LiPF6/EC-DMC/HzO solution better passivates the surface. There is also
significant difference in the peak height observed for the oxidation peak at 4300 mV vs.
Li/Li+ for the electrode cycled right after placement in the cell and for one that was kept

at OCP for 10 h before cycling. This may indicate that the peak at 4300 is due to

94

oxidation of water or oxidation of aluminum in the presence of water. It is also possible
that this peak arises due to interaction of aluminum with products of the hydrolysis of
LiPF6. It is well known that LiPF6 is not stable in the presence of water and is

hydrolyzed quickly according to reactions:24

LiPF6 Z PF5 + LiFl [4.4]
PF5 + H20 —> 2HF + POF3 [4.5]

Fast hydrolysis of LiPF6 was also observed during our studies. For example,
preparation of the solution with higher than 500 ppm water content caused gas evolution,
etching of the glass cell and fast corrosion of lithium electrodes.

XPS analysis of the surface revealed the atomic presence of the following elements:
Al (2p, 75 eV) — 17 atomic %, Li (Is, 56 eV) - 10 atomic %, C (ls, 285 eV) - 19 atomic
%, F (Is, 686 eV) — 31 atomic %, O (Is, 532 eV) — 22 atomic %, and P (2p, 135 eV)
— 1 atomic %. These results are similar to those obtained in electrolytes with no added
water indicating the mechanism of passivation is similar, and is based on the improved
passivity of the aluminum oxyfluoride film over the aluminum oxide film in the non-
aqueous electrolytes.

The important information gained is that aluminum passivates in LiPF6/EC-DMC,
even in the presence of water. This also means that the presence of HF, a hydrolysis by-
product, does not induce corrosion. Passivation of aluminum in the presence of water and
HF was previously observed by Kanamura et al.6 The authors successfully used small
additions of aqueous hydrofluoric acid in order to suppress corrosion in otherwise

strongly corrosive LiCF3SOg/PC and LiN(CF3SOz)/PC mixtures. Results of their XPS

95

studies indicated that the passive layer was a mixture of AlF3 and AlOF. The chemical
composition of the passivation layer was independent on the electrolyte, indicating that
its passivating properties resulted directly from the presence of fluoride in the
electrolyte/solvent mixture.

The surface of the electropolished aluminum, after polarization in LiPF6/EC-DMC +
500 ppm H20, was visually different from the surface before polan'zation. The normally
very reflective surface appeared “milky” at the area that was exposed to the electrolyte
solution. OM and SEM, however, did not reveal any noticeable changes in the surface
structure. AFM microscopy, on the other hand, showed a uniform, surface roughening

(nm range). AFM results are described in detail in the next chapter.

4.3. Conclusions

The presence of added water inhibits the passivation of aluminum in LiClO4/EC-
DMC and LiClO4/PC at highly oxidizing potentials (5 V vs. Li/Li+). Pitting and
progressive corrosion is accompanied by the formation of a highly porous, non-protective
layer of the corrosion products on the surface. The native oxide film does not prevent the
underlying aluminum metal from being corroded in LiClO4-based solvent/electrolyte
mixtures when water impurity is present. Significant pitting of the surface occurs under
such conditions. Due to the limited solubility of the corrosion products (mostly
Al(ClO4)3), the pits fill with the corrosion products, forming mounds on the surface.

On the other hand, water does not significantly affect the passivity of aluminum in

LiPF6/PC and LiPF6/EC-DMC. This is most probably due to formation of the

96

oxyfluoride layer that is very effective at protecting aluminum surfaces during anodic

polarization.

97

CHAPTER 5

5. Aluminum Corrosion in LiPFdEC-DMC and LiClOJEC-DMC
— Investigations Using Atomic Force Microscopy (AFM)

5.1. Introduction

Monitoring changes in the surface structure of an electrode is an essential component
of research concerned with understanding electrochemical processes, such as corrosion,
passivation, and deposition. Over the years, various microscopic techniques have been
used for this purpose. Historically, optical microscopy (OM), scanning electron
microscopy (SEM), and transmission electron microscopy (TEM) have been the most
often used techniques. The introduction of scanning tunneling microscopy (STM) in
1982,47 and development, in the subsequent years, of atomic force microscopy (AFM)
provided researchers with new, powerful tools for the analysis of surface structure.48

AFM is one of the most important and widely used forms of scanning probe
microscopy. It provides. very high lateral resolution, superior to OM and SEM, and in
some cases, can achieve atomic resolution. AFM can also produce topographical images
of electrically insulating surfaces, materials that cannot be probed with STM. Another
powerful feature of AFM is its ability to image surface structure while immersed in a
liquid.49 This feature has provided a revolutionary approach for the in situ

characterization of electrode surfaces and electrochemical processes. The combination

98

of AFM with various galvanostatic and potentiostatic electrochemical techniques,
called electrochemical atomic force microscopy (ECAFM), makes possible the study
of electrode surface morphology while electrochemical reactions are underway.
The technique has been used for studying such phenomena as underpotential deposition

’

(upd),50’51 potential-driven adsorption and desorption,S and formation and self-
organization of monolayers.54
AFM has been also shown to be very effective for studying electrode processes at
much larger than atomic or molecular dimensions. Three-dimensional topographical
imaging of a surface, at the nanometer to micrometer scale, is particularly useful for in
situ monitoring of processes such as electrochemical corrosion (general and localized),55-
7 passivation,58.60 and metal plating and strippingm'63 In many cases, ECAFM has
enabled the verification of reaction mechanisms that otherwise could only be based on
electrochemical measurements and ex situ microscopy. For example, ECAFM studies of
pitting corrosion of metals in aqueous environments have confirmed the importance of
the surface inhomogeneities (grain boundaries, inclusions, and second-phase precipitates)
in assisting passivation layer breakdown and initiation of corrosion.5 ’
Metal corrosion processes in non-aqueous media have not been extensively studied
with AFM or ECAFM. There are only a few examples of such studies in the literature.
Bellucci et al. investigated the passivation of Ni in 0.1M HZSO4/DMF and 0.1M
H2804/acetonitrile solutions.64 The electrochemical measurements indicated the
formation of a thin, poorly passivating layer and the possibility of salt precipitation onto
the metallic surface. AFM images showed that in both solvents, the sample surface was

very inhomogeneous with flakes and fractures, which confirmed the results of

electrochemical measurements. ECAFM has also been extensively used to investigate the

99

corrosion/passivation of lithium in non-aqueous electrolytes. This is mostly due to
developments in lithium and lithium-ion batteries.65

To the best of my knowledge, there are no reports on the use of AFM to study
aluminum corrosion in non-aqueous media. The use of AFM and ECAFM to monitor
topographical changes in the aluminum surface during anodic polarization in 1M
LiClO4/EC-DMC and 1M LiPFdEC-DMC, is described herein. The imaging modes were
also used to study the corrosion mounds and pits formed during anodic polarization in
LiClO4/EC-DMC/HZO mixtures. All measurements were made using electropolished
aluminum electrodes. Electropolishing produces a very smooth surface, which proved to
be essential for the observation of small deposits formed on the surface. It also produces a
very reflective surface, which was very useful during the studies of corrosion pit
formation, as the pits could easily be located with an optical microscope to allow for

proper positioning of the AFM tip.

5.2. Results and discussion

5.2.1. Ex-situ atomic force microscopy — dry electrolytes

Results presented so far indicate that aluminum passivates in dry electrolyte/solvent
mixtures. Both electrochemical measurements and XPS data indicate that a passivation
layer forms on the surface, protecting the metal from further corrosion. The formation of
the passivation layer could not be confirmed by either OM or SEM. Thus, a goal was to
see if it could be observed with AFM, which provides superior resolution, both vertically

and horizontally, as compared with OM and SEM

100

Figure 5.1 presents representative contact mode ex situ images of (A) an untreated
electropolished surface, and electropolished surfaces after anodic polarization in (B)
LiClO4/EC-DMC and (C) LiPFdEC-DMC. The images were acquired after the
electrodes were cycled in the corresponding electrolyte/solvent mixture (3 scans; 5 mV/s;
2500 — 5000 mV vs. Li/Li+), rinsed with copious amounts of DMC, and dried in a flow of
argon inside the dry box. The formation of the 50—200 nm deposits, evenly distributed
over the aluminum surface, is observed for both electrolytes. Their average height is only
5 — 10 nm, which explains why they were difficult to observe by OM or SEM.
Interestingly, the deposit morphology and surface coverage are similar in both
electrolytes indicating that their formation is directly influenced by the structure of the
oxide film. The formation of localized deposits implies that the initial oxidation processes
take place at pores and defects in the oxide film, as is the case in aqueous
environments.28 The corrosion products are largely insoluble in organic solvent and they
deposit in the localized regions where they are formed. The deposition of the corrosion
products effectively passivates the surface. It was also observed that these deposits
strongly adhere to the surface as they could not be removed by ultrasonication in DMC or
by moderate scratching with the AFM tip. This also proves that these deposits were not
some kind of precipitate from the electrolyte solution, but rather were products of the
reaction between electrolyte and the aluminum surface.

Interestingly, similar surface deposits were also observed on electrodes simply
soaked in LiPF6/EC-DMC for a period of 10 h, whereas no deposits were observed on
electrodes soaked in LiClO4/EC-DMC. These observations are consistent with the
electrochemical and XPS data discussed in the Chapter 3. In other words, the aluminum

surface reacts with the LiPF6 hydrolysis products (e. g., HF), even at OCP.

101

.088 Emma: 08 8 8928 8825 can 288 couoouou 08 8 385% 203 woman: non. .OEQOm—EPEE
S: ADV Ea 029-0%055 S: cm: 8 Pg:— .m> >8 08w 1 8mm 658 m 3.58 8 8535—09
0605 not“ 23 A3 283 3883 8:885? team—$05020 5 mo 838m 2”? 8...“. Rm .m.m 0.53...—

Se
66%.

 

102

5.2.2. Ex situ atomic force microscopy - LiPFdEC-DMC/Hzo

Data presented in Chapter 4 indicate that the passivation of aluminum in 1M
LiPFdEC-DMC occurs even in the presence of added water. Previously, Braithwaite et
al. observed, with electrochemical impedance spectroscopy, that aluminum passivation in
1M LiPF5/PC-DEC improves when the water is present as a contaminant.4 The improved
passivation was attributed to stabilization of the passivation layer by water molecules. On
the other hand, Kanamura et al. observed improved passivation in non-aqueous
electrolytes, in the presence of HF.6 They attributed the improved passivation to reaction

of HF with aluminum oxide and the formation of A1F3 and AlOF according to reactions:

A1203 + 6HF —9 2A1F3l + 3H20 [5.1]

A1203 + 2HF —> 2AlOF~L + H20 [52]

They confirmed the existence of such a layer with XPS, and concluded that AlOF and
AlF3 have better passivating properties in non-aqueous electrolytes than does A1203.

HF was not purposely added to the electrolyte/solvent mixture in present research. Its
presence is, however, expected considering the fact that HF is a product of LiPF6
hydrolysis (Chapter 3, reactions [4] and [5]). Taking this into account, a passivation layer
of different structure than the one observed in dry electrolytes is expected. Again, as in
the case for the dry electrolyte, the formation of the passivation layer could not be
confirmed by OM or SEM. It was observed, however, that the appearance of the naturally
shiny, and very reflective electropolished surface, changed. The surface became “milky”

in the area exposed to the electrolyte solution during polarization. If this change in

103

appearance was accompanied by different surface structure, it should be possible to
observe such changes with AFM.

Figure 5.2 presents the AFM images of the electropolished aluminum surface after 3
cyclic voltammetric scans in 1M LiPF5/EC-DMC with 500 ppm of added H20. The
potential was scanned between 2000 — 5000 mV vs. LilLi+ at 5 mV/s. After the last scan
was completed, the electrode was washed with DMC and dried in the flow of argon.
The 10 x 10 um2 deflection mode image (Figure 5.2 A) resembles that for the electrode
cycled in LiPF6/EC-DMC without added water (Figure 5.1 C top). Multiple, 50 to 200
nm deposits are randomly formed on the surface, indicating highly localized dissolution
of aluminum and passivation of the surface. However, a higher magnification, height
mode image (Figure 5.2 B) indicates a significant increase in the surface roughness. Such
uniform surface roughening can be explained by reaction of HF, formed from the
hydrolysis of the LiPF6, with the aluminum oxide film on the surface, and formation of a
new, oxyfluoride layer. A similar reaction has been proposed previously by Kanamura et
al.6

Roughening of the aluminum surface in the presence of fluoride is also commonly
observed in aqueous solutions. For example, Hurlen and coworkers observed that the
presence of F- stimulates the anodic dissolution (hence, roughening) of passivated
aluminum, without completely destroying the passivation.68 The authors concluded that
this process occurs mostly by a transport-limited reaction of fluoride with aluminum ions

dissolving at the metal surface. A similar conclusion was reached by Richardson et al.,

who investigated the passivation of aluminum in aqueous fluoride solutions with

104

, 69 . . . .
impedance measurements. Therr results were consrstent wrth rapid removal of the

surface oxide film and its replacement by a complex oxyfluoride film.

A

20nm

 

Figure 5.2. AFM images of an electropolished aluminum surface after 3 potentiodynarnic
scans in 1M LiPF6/EC-DMC with 500 ppm of added H20 (5 mV/s; 2000-5000 mV vs.
Li/Li+). (A) Deflection mode image. (B) Height mode image.

105

5.2.3. Ex situ atomic force microscopy - LiClO4/EC-DMC/HzO

The electrochemical and microscopic observations of aluminum corrosion in
LiClO4/EC-DMC and LiClO4/PC, contaminated with water (Chapter 4), showed that the
impurity is detrimental to the passivation of the metal. Significant pitting of the
aluminum surface was observed. SEM, EDS, and Raman spectroscopy results have
shown that oxidation products deposit in the corrosion pits when the water level is kept
below 500 ppm. Figure 5.3 presents AFM images of a pit formed on the aluminum
surface after 2 potentiodynamic scans (5 mV/s; 2000 — 5000 mV vs. Li/Li+) in 1M
LiClO4/EC-DMC with 500 ppm H20. Clearly, the formations on the surface are mounds
(pits filled with corrosion products). Figure 5.3A presents a deflection mode image of the
mound. The square-shaped, 1 pm high mound is easily visible. Figure 5.3B presents a
height mode image of the same mound. Lateral progress of the corrosion can be clearly
observed as two distinct rectangular shapes, one inside the other. It seems that the
LiClO4/EC-DMC/H20 mixture is not corrosive toward the entire oxide film, as intact
regions are present at the edges of the mound. This observation indicates that the
corrosion process starts locally, probably at the pores or defects in the oxide film. It then
proceeds at the interface between aluminum and aluminum oxide layer. As aluminum is
being dissolved, formation of Al(ClO4)3 probably occurs, as indicated by EDS and
Raman data (Chapter 4). Since the Al(ClO4)3 is soluble in both PC and EC-DMC, its
precipitation indicates that local supersaturation conditions are met. Breakage of the
oxide film occurs, as the solid corrosion products deposit and lift the oxide film up from

the surface.

106

Such a mechanism would require relatively fast migration of the C104' through the
oxide film. Since corrosion does not occur in dry LiClO4 solutions, as the surface
passivates during the first scan, then one of the roles of water in the corrosion process
might be increasing the C104' partitioning rate through the oxide film. This may occur
due to the changes in the solvation sphere of the C104’ in the presence of water. In
anhydrous solvent, C104' is solvated by ethylene carbonate and dimethyl carbonate (or
propylene carbonate) molecules. ClO4' - EC/DMC (PC) complexes have a structure in
which non-polar groups are sticking out, thus partitioning of ClO4' into a highly
hydrophilic aluminum oxide film is hindered. It is supposed that the addition of water to
the solvent makes the solvation sphere more polar, thus allowing it to penetrate oxide

film more easily.

EDS and Raman measurements indicated that in solutions of higher water
concentration (> 500 ppm), C104" was not present, or present at much lower levels, inside
the pits, than in solutions with lower water content. This might indicate formation of the
pits without corrosion products deposited inside. AFM images presented in Figure 5.4 are
the examples of a pit formed on the electropolished surface after 2 potentiodynarnic scans
(5 mV/s; 2000 — 5000 mV vs. Li/Li+) in 1M LiClO4/EC-DMC with 1000 ppm of added
H20. A large pit, roughly 2.5 mm deep and 5 pm across, is clearly visible. This pit is not
filled with the corrosion products, as was observed for electrolyte with lower water
content. There is also a significant amount of the loose solid material present on the
surface around the corrosion pit. The “empty” pit may indicate that at the higher water
levels, solubility of A1(ClO4)3 increases, and thus, it does not accumulate inside the pit.

To investigate this possibility further, in-situ ECAFM measurements were performed.

107

1600 nm

 

Figure 5.3. AFM images of a mound formed on the electropolished aluminum surface
after 2 potentiodynarnic scans (5 mV/s; 2000 — 5000 mV vs. Li/Li+) in 1M LiClO4/EC-
DMC with 500 ppm of added H20. (A) Deflection and (B) height mode images of the

same mound.

108

4000 nm

 

\Q 939

   

Figure 5.4. AFM images of a pit formed on the surface of electropolished aluminum after
2 potentiodynarnic scans (5 mV/s; 2000 — 5000 mV vs. Li/Li+) in 1M LiClO4/EC-DMC
with 1000 ppm of added H20. (A) Deflection and (B) height mode images of the same

pit.

109

5.2.4. In situ atomic force microscopy - LiClO4/EC-DMC/H20

ECAFM is known to be a very useful technique for studying corrosion of metals in-
situ.55’56 It allows for direct correlation between the charge passed and the changes in
electrode surface structure brought about by the electrochemical reactions. The objective
of these measurements was to learn more about the pit formation and development in the
LiClO4/EC-DMC/HZO.

Previously presented data showed that corrosion of the electropolished aluminum is
highly localized and that pits form randomly over the surface. Since it is not possible to
predict where the pits will form, a priori, and due to the fact that images are restricted to
15 x 15 um (AFM scanner limitations) or less, the following procedure was used. First,
the electrode was subjected to 2 potentiodynarnic cycles from 3300 to 4500 mV vs.
Li/Li+ at 3 mV/s in 1M LiClO4/EC-DMC with 1000 ppm of added H20, so that pit
formation could be observed with an optical microscope. Then, the AFM tip was
positioned manually over the pitted area and an image was acquired to confirm presence
of a pit/mound. The electrode was then subjected to additional potentiodynarnic scanning
(3rd and 4th scan) and the AFM images were recorded in situ, during the polarization.

Figures 5.5 through 5.8 present the data for 1M LiClO4/EC-DMC with 1000 ppm of
added H20. Figure 5.5 presents the cyclic voltammetric i-E curves recorded at 3 mV/s for
the two scans during which AFM images were collected. A response indicative of pitting
corrosion is clearly visible for both scans with Epit equal to 4100 and 4200 mV vs. Li/Li+
for the 3rd and 4th scan, respectively, and Epass = 3450 mV vs. LilLi+ for both scans. The
current at 4500 mV for both scans was 155 and 175 11A, respectively. Both, the current at

this potential and the anodic charge increase with cycle number, consistent with an

increasing electrode area involved in the corrosion process.

110

 

 

 

 

 

 

 

31rd scan
‘5.
O
/ :9:
4th scan
<
/ ..
52
3300 3500 3700 3900 4100 4300 4500

Potential [mV vs. Li/Li+]

Figure 5.5. Cyclic voltammetric i-E curves for electropolished aluminum recorded
at 3 mV/s in 1M LiClO4/EC-DMC with 1000 ppm of added H20. The measurements
were made in the AFM fluid cell.

Figure 5.6A presents a sequence of the 9 x 9 pm2 AFM images (deflection mode) of
the corrosion mound recorded at the beginning (top), the mid point (middle), and the end
(bottom) of the 3rd potentiodynarnic scan. A circular formation, ~ 100 nm high
and ~ 3.5 um in diameter is visible on the surface at the beginning. There are also two
smaller circular pits present to the upper left of the circular mound (indicated by an
arrow). This kind of small pit is often observed on the electropolished surface, and
originates from the electropolishing process itself.32 Since, there was no evidence that
these pits take an active part in the corrosion in 1M LiClO4/EC-DMC/HZO, they

provided a convenient reference point on the aluminum surface.

111

Figure 5.6B presents a sequence of the cross sectional profiles obtained from the
height mode images of the developing mound recorded at different potentials. Each
profile was taken through the middle of the circular mound (see Figure 5.6A top).

Analysis of the profiles shows that the growth of the corrosion pit and redeposited
products are directly related to the applied potential, and thus the anodic charge passed.
There is little change in the height or diameter of the deposits between 3300 and 4200
mV during the positive scan. At 4300 mV, the pit width and both the horizontal and
vertical size of the deposit begin to increase until the repassivation potential (3450 mV) is
reached during the reverse scan. There also appears to be a temporary collapse of the
growing mound at 4200 mV during the reverse scan, after which the mound continues
growing until repassivation potential is reached.

Figure 5.7A presents a sequence of the 11 x 11 um2 deflection mode ECAFM
images of the same region of the surface during the 4th potentiodynarnic scan. Figure
5.7B presents a sequence of image cross sections taken at different applied potentials
through the middle of the corroding area. Interestingly, the corrosion mound developed
during the 3rd potentiodynarnic scan, fully collapses at the beginning of the 4th scan. It
remains in the collapsed state until the Epit (4500 mV) is reached, at which potential it
starts growing again. The mound then continues growing until a potential of 4200 mV is
reached during the reversed scan. At this potential, the mound suddenly breaks open and
detaches from the surface, leaving behind a ~1 um deep pit and large number of loosely-
bound, solid corrosion products, that give rise to the streaks in the AFM image (last

image, Figure 5 .7A).

112

 

1:
g 3300 V
ALJ AM}
4U/ 399.1"):
~1L,J 4000 mV
4200 mV

/
L]! w

    

 

 

 

 

 

'J
/ 4200 mV

,_v
1/ NW

[#f/

g/om \s... .v
l 33 mV

/
9 um 1 9pm 6 2 4 '6 1';

pm

 

Figure 5.6. In situ ECAFM measurements of an electropolished aluminum surface in 1M
LiClO4/EC—DMC with 1000 ppm of added H20. (A) Deflection mode images of a
mound and (B) height mode profiles for a cross section through the middle of the
corrosion mound, as shown on the top image. The electrode potentials are indicated on
the AFM images and the cross sectional profiles, and they correspond to the i-E curve

presented in Figure 5.5 (3rd scan).

113

 

LIMV

WV
WV

/\ 4300 mV
my

45993!

4200 mV

A
'—£oo mV"

(1 2.5 s 715 111
um

 

 

 

 

 

 

Figure 5.7. In situ ECAFM measurements in 1M LiClO4/EC-DMC with 1000 ppm of
added H20. (A) Deflection mode images of the mound and (B) cross sectional profiles
through the middle of the mound. The electrode potentials are indicated on the AFM
images and the cross sectional profiles, and they correspond to the i-E curve presented in

Figure 5.5 (4th scan).

The potential dependent horizontal and vertical growth of the corrosion pit and
deposited products, during both potentiodynamic scans, is depicted in the Figure 5.8. The
data for the 3rd scan were obtained from the cross section profiles presented in Figure
5.6B. The data for the 4th scan were obtained from the profiles presented in 5.7B and two
profiles, not shown in Figure 5.7B, recorded at 3800 and 3300 mV during the reverse
potential sweep of the 4th scan. It appears that after reaching Epita the deposited products
grow vertically and this growth is later followed by horizontal expansion. This result can
be explained by the following mechanism in which, at Epit, corrosion starts at an exposed
aluminum surface, as this surface is more active than an oxide-coated one. Corrosion
undercuts the oxide film, and the growth of the solid corrosion product deposit lifts the

oxide film. This process is observed as a horizontal growth of the deposit.

 

 

     

 

 

 

 

 

 

 

 

 

 

 

 

 

240 d 6.0
L r 4
220. 3 scan .-"\ .55
H ’ A’M-—’,\. In .\ ‘ .
E 200’ 5.0 g
.. 180- / a.-
8 i _________,_/ 4.5 =r
g 160 E:
140; “V i4°°~—a
120g ----height —width :35
100-l--*1+4A'-'~3.0
3200 3400 3600 3800 4000 4200 4400 4600
400, m ‘12
200;4...§f§: 111
0: T“ """"" ~10
.__. -400~ ""height ‘8 g
in '600? — width 37 g
5:9 -800: . .1
-1000- “.6
-1200j ‘5
_14OOA.-.-....-..-4
3200 3400 3600 3800 4000 4200 4400 4600

Potential [mV vs. Li/Li”]

Figure 5.8. Horizontal and vertical development of the corrosion product deposit.

115

The sudden collapse of the deposit and its detachment from the surface, leaving an
exposed pit, cannot be explained by increased solubility of the corrosion products in the
presence of water. If deposit solubility was the cause, it should be relatively constant
during the course of the entire potentiodynamic cycle, not a sudden transition as was
observed. A reasonable possibility is the evolution of gaseous products, primarly 0;,

formed from the oxidative decomposition of water according to reaction:

2H20 —> 02 + 4 H+ + 4 e” (4.27 v vs. Li/Li+) [1]

The evolution of gaseous oxidation products can also explain another observation
made from the analysis of Figure 5.8. It seems that the horizontal grth is much faster
after the sudden collapse of the mound. This can be explained by the fact that once gas
escapes from the corroding area it forms a vacuum that is quickly filled with the fresh
electrolyte solution. As a result of an increase in the electrochemically active area, the

corrosion proceeds at a higher rate.

Considering all the information, the following mechanism can be envisioned for the
corrosion process in LiClO4/EC-DMC with added water impurty. In the first step, very
localized corrosion starts at defects in the oxide film. At high anodic potentials (>4000
mV), decomposition of water, as well as dissolution of aluminum occurs. The corrosion
products deposit in the newly formed pit, forming a mound on the surface. At low water
levels (< 500 ppm), the evolution of oxygen is not significant and the corrosion site

remains filled with and partially passivated by the corrosion products. At high water

116

impurity levels, evolution of oxygen occurs to a greater extent, and this leads to the
formation of the gas bubbles. These bubbles coalesce into a larger bubble that eventually
releases from the surface, raising the product layer. This process may lead to formation of
empty pits on the surface. This process destroys the passivation layer provided by the
deposited corrosion products leading to a larger area of aluminum exposed to the solution
and an increased rate of corrosion. Kelly et al. observed that passivity of iron in
LiClO4/PC mixtures depends strongly on the amount of water present in the electrolyte/
solvent mixture.31 The authors reported that at water levels less than 500 ppm,
passivation through formation of salt film consisting of Fe(ClO4)2, occurs. At higher
water levels, the iron surface lacks passivity and significant corrosion occurs at potentials
greater than 4.1 V vs. Li/Li+. They attributed the lack of passivation to increased
solubility of Fe(ClO4)2 and local acidification of the electrolyte/solvent mixture
as a result of water oxidation. Data presented in this work show that a similar mechanism
of corrosion may be occuring on aluminum surfaces. ECAFM results show, however, that
higher water levels may also lead to increased corrosion through mechanical disruption of

the deposited salt film (Al(ClO4)3) by evolution of gaseous oxidation products.

5.3. Conclusions

AFM imaging shows clearly that passivation of aluminum in LiPF6/EC-DMC
and LiClO4/EC-DMC is accompanied by formation of a layer of solid deposits layer. In
LiPF6, this layer forms both during anodic polarization and at open circuit conditions.

Addition of water to LiPF6/EC-DMC leads to surface roughening due to attack of the

117

fluoride on the oxide film. This process does not destroy passivity due to formation of an
oxyfluoride layer. The addition of water to LiClO4/EC-DMC leads to significant
localized corrosion of aluminum surface. Pits, filled with solid corrosion products
(mounds), form in the electrolyte at added water levels below 500 ppm. At higher water
levels, “empty” pits are observed. Their formation may be assigned to disruption of the

deposits by gaseous products evolving during oxidation.

118

CHAPTER 6

6. Aluminum Corrosion in LiPFJEC-DMC and LiClOJEC-DMC
— Investigations Using the Electrochemical Quartz Crystal
Microbalance (EQCM)

6.1. Introduction

Electrochemical quartz crystal microgravimetry (EQCM) is an analytical technique
that combines electrocherrristry with an ultra-sensitive mass measurement provided by a
quartz crystal microbalance (QCM). The technique allows a direct correlation to be made
between the charge passed through an electrode and the mass change induced on the
surface as a result of the charge passed. It allows for the investigation of the mechanisms
of electrochemical reactions that induce a mass change on the electrode. The EQCM has
provided important information about electrochemical phenomena such as underpotential
deposition, electropolymerization, double layer stucture, electrolyte adsorption,
electrochemically-driven self-assembly, and corrosion-’0.72 and references therein

The EQCM measurement involves two simultaneously measured variables, mass and
charge. Measurement of charge can be accomplished in straightforward way by

employing a potentiostat or galvanostat, whereas the mass changes are determined

indirectly from the measurement of the oscillation frequency change of a quartz crystal

119

resonator (QCR). The QCR consists of a piezoelectric quartz crystal, usually in the form
of disc that is sandwiched between two metal electrodes. During the operation, a high
frequency AC electric field is applied between the electrodes, and across the thickness of
the crystal. This causes the OCR to oscillate in the mechanically resonant shear mode.
The frequency of the oscillation is very sensitive to interfacial mass changes at the
electrode surface. The relationship between the change in QCR mass (Am) and the
change in oscillation frequency (Af) was first recognized and described by Sauerbrey,73

and is contained in the following equation:

_ 2
znf" Am=—KAm [6.1]

A =___
f AJI‘QPQ

where f0 is the fundamental frequency of the quartz resonator [Hz], #Q is the shear
modulus of AT-cut crystal (2.947-1011dyn/cm2), pQ is the density of quartz (2.648
g/cm3), A is the piezoelectrically active area [cm2] and n is the overtone number; K - is
the sensitivity factor that includes all the values that are constant for a given experimental
setup.

The negative sign in the Sauerbrey equation indicates that a mass decrease is
accompanied by an oscillation frequency increase, whereas a mass increase leads to a
decrease in the oscillation frequency. This equation also shows the origin of the
extraordinary sensitivity of mass measurements performed with QCM. For example,
operating at a resonant frequency of 10 MHz provides a theoretical sensitivity of 0.22
Hz-cmzongJ. Since the frequency can be measured very accurately to :l: 1 Hz, a 10 MHz
EQCM can detect a mass change of 4.4 ng-cm-z. Such sensitivity corresponds to

detecting less than 2% of a monolayer of Pb atoms deposited on gold.74

120

Further interpretation of EQCM data is accomplished in following manner. Since the
electrochemical charge represents the total number of electrons transferred in a given
electrochemical process (e.g., deposition), it corresponds to a mass change on the
electrode surface. This mass change can be easily calculated from the Faraday Law

according to the equation:

F

Z

where: W is the apparent molecular weight [g/mol]; z is the number of electrons; Q is the
charge passed [C]; F is the Faraday constant (96486 C/mole).
By substituting equation [6.2] into equation [6.1] and rearranging, the following

equation is obtained:
— = —— . — [6.3]

It is evident that all the parameters on the right side are either a constant (F and K) or
are determined during the experiment (4f and AQ). Thus, this relationship allows for the
determination of the molecular weight per electron (W/z), which is characteristic for the
electrochemical process that leads to the observed mass change.

Several limitations of the Sauerbrey equation have to be recognized, however, for
data analysis to be meaningful. A major assumption is that the mass added or lost at the
oscillator surface does not experience any shear deformations during oscillation. This is a
good approximation for thin, rigid films. It may however lead to many errors when
investigating formation of the thick, soft materials (e.g., polymers).7o’71 It is also
important to recognize the fact that, as implied by Sauerbrey equation, the QCR is

sensitive to mass changes per unit area of oscillation. Thus the QCR is most useful when

121

probing processes that occur uniformly across its surface — processes that can be
represented as a change in mass per unit area. Recognizing this limitation is very
important, as many electrochemical corrosion processes lead to very localized dissolution
of the surface, thus limiting evaluation of the results to qualitative, rather than
quantitative conclusions. Other factors that may influence the operation of the QCR are
the electrode surface roughness and the total mass (frequency) change experienced by the
QCR. Rough surfaces may lead to entrapment of solvent molecules, which are than
recognized by the QCM as a mass increase. This is not usually a serious problem as long
as the polished crystals are used and significant surface roughening does not occur during
measurement.71 Also, if the total frequency change does not exceed a few % of the
fundamental frequency of the crystal, the dependence between interfacial mass change

. . . . 71
and oscrllation frequency change remains linear.

The corrosion/passivation of aluminum in lithium-ion battery electrolytes was
recently studied with the EQCM by Yang et (11.7 The authors investigated various
presently used or possible lithium-ion battery electrolytes, including LiPF6, LiBF4,
LiClO4, LiCF3SO3, LiC(CF3SOp_)3, and LiN(CF3SOg)2. They found that aluminum
corrodes significantly in PC containing LiCF3SO3, LiC(CF3SOg)3, and LiN(CF3802)2.
The authors observed mass loss from the electrode after anodic polarization in these three
electrolytes. More detailed EQCM studies of aluminum corrosion in LiN(CF3SOz)2/PC
revealed that the corrosion process entails formation of an adsorbed aluminum compound
or non-adherent film that subsequently desorbs. It was also found that potentiodynarnic
polarization of aluminum in LiPF6/PC and LiBF4/PC leads to formation of a protective

film after the first potentiodynarnic scan. An increase in the electrode mass was observed

122

in such cases. In LiClO4/PC formation of the oxidation product layer was also observed
during the first potentiodynamic scan. This layer, however, did not prevent surface

oxidation during subsequent scans.

In this chapter, the corrosion/oxidation of aluminum in LiPF6/EC-DMC and
LiClO4/EC-DMC was investigated with the EQCM. The experimental setup is described
in Chapter 2. The measurements were conducted using commercial, 10 MHz, polished
QCRs, with aluminum electrodes evaporated on the surface. Theoretical mass sensitivity
of the crystals was 4.4 ng/Hzcm2 and linearity of the Sauerbrey equation was assumed.
The aluminum electrode thickness was measured with SEM to be ~ 200 nm. Although
the exact procedure by which aluminum electrodes were evaporated on the crystal surface
is not known, it is anticipated that an oxide layer formed on the surface with a thickness
of l — 4 nm, which is typical for air-formed aluminum oxide films.32 Prior to the
measurements, the aluminum QCRs were rinsed with DMC and dried in a flow of Ar.
These electrodes are referred to as untreated. For some measurements, the native
aluminum oxide was damaged by uniform mechanical polishing with 0.1 pm size
diamond powder, in hexane, under an inert Ar atmosphere, rinsed with hexane, and dried
in a flow of Ar. Complete removal of the native oxide layer could not be ensured as the
polishing was limited, due to thinness of the aluminum electrodes. These electrodes are

referred to as mechanically polished.

123

6.2. Results and discussion

Results presented in the previous chapters indicate that aluminum passivates in dry
LiClO4/EC-DMC and LiPF6/EC-DMC. AFM images indicated the formation of a solid
passivation layer on the surface. XPS data revealed that in LiPF6/EC-DMC the
passivation layer is rich in fluoride, and in LiClO4/EC-DMC rich in perchlorate. The
formation of a solid passivation layer should be evidenced by increased electrode mass
after polarization. Also, by comparison of the charge passed during polarization with the
change in mass, the apparent mass per electron (W/z) characteristic of the oxidation
process can be determined (see Equation 6.3). Table 6.1 summarizes the various possible

oxidation products along with the W/z values characteristic for their formation.

 

 

 

 

 

 

 

 

 

 

 

Chemical composition W/z (g/equiv)
A1203 8
AlOF 1 1.7
A120F4 15.3
Al(OH)3 l7
Al(OH)2F 17.7
Al(OH)F2 18.3
A1F3 19
Al(ClO4)3 99.5
Al(PF6)3 145

 

 

 

Table 6.1. Chemical compositions and W/z values for possible passivation film species.

124

6.2.1. OCP measurements

Figure 6.1 presents results of open circuit potential (OCP) measurements in 1M
LiPF6/EC-DMC for a mechanically polished electrode. The mass change of the electrode
was also monitored. Both the potential and mass of the electrode increase during the 1-h
period consistent with increased passivation of the surface by reaction deposits (e.g.,
AlF3). The rate of increase is the highest at the beginning and then leads toward a more

constant value.

 

 

 

 

_ 2500 _, ,700
+ i -‘_,.-o-"'-- :
E 2450." 1600 Z
'5 2400- 3 8
r " ‘ 500
US ,a” . O
> 2350 ' ." 4 :r
> . I; 1400 g
E 2300 * I 00
__. - / 1300 a»
E 2250' I, 1 g
8 2200- .o’ ----Am 120008
E 2150-" —E @1003»
2100 r ' ‘ A ‘ 0
O 500 1000 1500 2000 2500 3000 3500 4000

Time [s]
Figure 6.1. Time dependent open circuit potential and mass change measurements for

mechanically polished aluminum during exposure to IM LiPF6/EC-DMC.

Potential and mass change curves similarly shaped to these were also observed in 1M
LiClO4. Table 6.2 summarizes the results of OCP measurements in 1M LiPF6/EC-DMC
and 1M LiClO4/EC-DMC for mechanically polished and untreated aluminum electrodes.
It is evident that the passivation layer forms in either electrolyte and on both types of
surfaces. Interestingly, the OCP values observed on the untreated aluminum surface were

more positive (noble) than the ones reported in Chapter 3 for the electropolished surface.

125

This is probably due to a different, less defective aluminum oxide film present on

evaporated aluminum.

 

 

 

 

 

 

 

 

 

OCP [mV vs. Luff) Am [ng/cm2]b)
Electrolyte
Mechanically Polished
LiClO4 2200 i 80 400 :1: 100
LiPF6 2350 :l: 50 500 i 100
Untreated
LiClO4 2600 i 50 400 i 150
LiPF6 2800 i- 100 700 i 150

 

 

 

 

 

Table 6.2. Summary of OCP-EQCM data for mechanically polished and oxide coated
aluminum electrodes. a) — at 1-h mark, b) — during the 1-h period.

6.2.2. EQCM measurements in LiPFdEC-DMC and LiClO4/EC-DMC
- mechanically polished surface

Formation of a passivation layer on aluminum electrodes was also investigated with
cyclic voltammetry combined with QCM. The voltammograrn and associated change in
mass (Am) for a mechanically polished aluminum electrode in 1M LiPF6/EC-DMC, after
a l-h soak at OCP, are presented in Figure 6.2. During the first scan, there is an anodic
charge of 3 mC/cm2 passed between 2700 and 5000 mV. Little charge is passed during
the subsequent scan indicating that passivation of the aluminum surface occurs. The
anodic charge recorded during potentiodynarnic cycling is accompanied by increase in
the electrode mass. The mass gain was equal approximately 320 and 60 ng/cm2 for the

lst and 2nd potentiodynarnic cycles, respectively. These results indicate the formation of

126

a solid passivation layer on the surface. The mass gain is best matched to that caused by a
charge transfer with a mass increase per electron (W/z) of 11. The normalized charge
change (AQn, which is WAQ/(zF)) with W/z = 11 is presented in Figure 6.2B. The result
suggests the formation of AlOF (W/z = 11.7). It is also possible that mixed oxidation
products with an average mass-to-charge ratio of 11 are formed during anodic
polarization (Table 6.1). This result is in agreement with results of the XPS analysis
presented in Chapter 3. The XPS data indicated formation of a fluoride-rich passivation
layer (e.g., A1F3, AlOF).

Recently, Yang et al. performed an EQCM investigation of aluminum passivation in
1M LiPF6/PC.7 The authors observed a similar oxidation charge and passivation of the
electrode during the first potentiodynarnic scan. They also observed an increase in the
electrode mass, which indicated the formation of a solid passivation layer on the surface.
The apparent mass-to-charge ratio reported in their studies was equal 26. This number
would appear to indicate that a product with a higher W/z was formed. The authors did
not assign any specific product to the observed W/z value. The analysis of their results,
however, suggests a possible explanation for the observed difference in W/z. AQn values
presented by the authors did not track Am well, especially for the second
potentiodynarnic cycle. There was significant, constant mass increase observed in their
studies even at the potentials where no current was observed in voltammetric data. This
indicates that the electrochemical oxidation observed in their studies was accompanied by
another process that, independently from polarization, leads to a mass increase. The
authors did not comment on this issue at all. Rough correction of their data to eliminate

this constant mass increase yielded a W/z value of approximately 18 g/equiv. This result

127

is within the range of formation of mixed aluminum fluoride/oxyfluoride layers (see

Table 6.1), as proposed in this and other studies.2’4

 

 

 

 

 

2500 A 3000 3500 4000 I 4500 A 5000
Potential [mV vs. Li/Li+]
B

_ _‘-- -‘
—_

_ _ - -
4.- [1 Ill- __ ‘ --- -‘~

Q [ng/cm2
N U)
U: C
O O _
\\‘

A
p—n
\ll
0
er fiffi
\

O
I
'
'
C
'
I
'c
o“‘
C
o-"
-
--‘-

 

 

2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

Figure 6.2. (A) Cyclic voltammogram, and (B) mass change (Am) and normalized charge

change (AQn with W/z = 11) versus potential profiles for mechanically polished

aluminum in 1M LiPF6/EC-DMC during the first two potentiodynamic cycles. Scan rate
= 5mV/s.

128

Cyclic voltammetric i-E curves and corresponding mass change profiles for
mechanically polished aluminum in 1M LiClO4/EC-DMC, after 1h at OCP, are presented
in Figure 6.3. Similarly to what was observed in LiPF6/EC-DMC, passivation occurs
during the first potentiodynamic scan, as little anodic charge is passed during the
subsequent cycle. The oxidation charge is also accompanied by a mass increase on the
surface. The mass gain is best matched to that caused by a charge transfer with W/z of
~60. The normalized charge change (AQn) with W/z = 60 is presented in Figure 6.2B.
Interestingly, calculated AQn does not track the measured Am over the whole potential
range, in contrast to what was observed for LiPF6/EC-DMC. At the beginning of the
initial scan, the mass increase is lower than the calculated AQn. This indicates that two or
more different processes take place at different potentials. To investigate this more, the
apparent mass per electron (W/z = F-Am lAQ) was plotted against potential and is
presented in Figure 6.3C. At the beginning of the first positive sweep, W/z stays below
17 up to 4000 mV, indicating the formation of mixed aluminum oxide/hydroxide product.
Such a product may form from the reaction between aluminum and trace water present in
the electrolyte/solvent mixture. W/z gradually increases up to 24 by 4800 mV. This
indicates an increasing proportion of some heavier oxidation product in the overall
reaction. At potentials close to the positive limit (5000 mV), W/z rises quickly and
reaches values over 60, indicating the formation of an even heavier product.

XPS data reported in Chapter 3 indicated Al(ClO4)3 as a possible oxidation product.
The formation of pure Al(ClO4)3 on the surface would be evidenced by a W/z of 99.5.
The observed W/z was much lower, however. There are two possible explanations for

this observation. First, a mixture of aluminum oxide/hydroxide and aluminum perchlorate

129

forms on the surface, thus the observed W/z is an average value for a mixed oxidation
product. On the other hand, it is possible that the Al(ClO4)3 that forms is not strongly
attached to the surface, and thus, it does not give rise to a signal characteristic of its
formation. Both possibilities seem feasible. The first one is supported by formation of a
lighter (A1203/Al(OH)3) oxidation product that is visible at the beginning of the first
positive scan. The second one is evidenced by the slight mass decrease observed at
positive limits during the second scan. Mass loss indicates that at highly anodic potentials
partial detachment of the passivating layer occurs. It is possible that due to an increased
oxidation rate at positive potentials, dissolution of aluminum and formation of A1(ClO4)3
interferes with the already formed passivating layer and leads to observed mass loss.
Yang et al. briefly investigated corrosion/passivation of mechanically polished
aluminum in 1M LiClO4/PC.7 They found that potentiodynarnic polarization leads to the
deposition of a heavy product, which is characterized by a W/z of 220. This deposit,
however, was not effective at passivating the surface as substantial currents were
observed during the second potentiodynarnic scan. For example, the oxidation charge
observed during the second cycle was only ~ 25% lower than the charge recorded for the
first cycle. Interesting, however, is the fact that the i-E curves recorded by the authors
showed characteristics of pitting corrosion. The authors did not address this observation
and they also did not assign the W/z value to any oxidation product, or mixture of

products.

130

 

8.0 i
6.0 _
4.0 _

i [uA/cmz]

 

2500

1400
1200
1000
800
600
400
200

Am or AQn [ng/cmz]

E
2.0 P
0.04

 

M

 

 

L

3000 .3500 4000 14500 i 5000
Potential [mV vs. L1/L1+]

B

AQ. /

3000 .3500 4000 _ 4500 5000
Potentral [mV vs. L1/L1+]
C

 

W/z

 

0 _
2500

 

 

 

 

3000 i 3500 A 4000 A 4500 A 5000
Potential [mV vs. Li/Li“]

Figure 6.3. (A) Cyclic voltammogram, (B) mass change (Am) and normalized charge

change (AQn with W/z = 60) diagram, and (C) W/z change with potential diagram for
mechanically polished aluminum in 1M LiClO4/EC-DMC during the first two

potentiodynarnic cycles. Scan rate = 5mV/s.

131

6.2.3. EQCM measurements in LiPFdEC-DMC and LiClO4/EC-DMC - untreated
surface

Passivation of untreated aluminum surfaces in LiClO4/EC-DMC and LiPF6/EC-
DMC was also investigated. Figures 6.4 and 6.5 present EQCM data recorded during
potentiodynarnic cycling at 5 mV/s in LiPF6/EC-DMC and LiClO4/EC-DMC,
respectively. These figures present also the calculated normalized charge change (AQn)
with corresponding W/z ratios. Clearly, passivation of the surface occurs in both
electrolytes. It is important to point out that, compared to mechanically polished surfaces,
the onset potential for the oxidation current is shifted positively (to ~ 3800 mV) in both
electrolytes. This observation agrees with the much higher OCP values recorded for
untreated surfaces (Table 6.2).

Figure 6.4B presents a normalized charge change (AQn) calculated with W/z = 11 for
untreated surface in LiPF6/EC-DMC. This value is the same, as that observed for
mechanically polished surfaces, indicating that the oxidation processes are of the same
nature and involve formation of aluminum oxyfluoride layer.

The normalized charge change for LiClO4/EC-DMC is presented in Figure 6.5B.
The best fit of electrochemical charge data to measured electrode mass gain was obtained
for W/z = 22. This value is much lower than the average value obtained on mechanically
polished surface and indicates a much lower proportion of Al(ClO4)3 in the passivation
layer. The passivation layer is mostly composed of mixed aluminum oxide and aluminum

hydroxide.

132

 

121)
10.0;

[uA/cmz]
9°
C r 1 .

 

Current density

Am or AQn [ng/cmz]

F’ :8 9‘
c: c: c:

53
c:

2500

U:
O

0"
2500

 

     

     

   

 

 

3000 3500 4000

Potential [mV vs. Li/Li“]

4500

5000

B

 

1 M ..............................

I—

‘4

b
P"::l‘. I \ -~ . A — » A - --

} W/z= 11

, -—--Am
T "—110“

 

I
I ‘4‘.

3000 "33000 4000
Potential [mV vs. Li/Li+]

 

I T
.\

4500

 

5000

Figure 6.4. (A) Cyclic voltammogram, and (B) mass change (Am) and normalized charge
change (AQn with W/z = 11) profiles for untreated aluminum in 1M LiPF6/EC-DMC

during the first two potentiodynarnic cycles. Scan rate = 5mV/s.

133

 

99‘.“
c> c> c>

:8
<3

 

Current density [uA/cmz]

_1 ’ 7 4 1 . 1 1 1 1 1
2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

P

 

9’ r: F’ 9’
c> c> c> c> c:

 

 

400 I;

‘------~-
-
.-

 

b)
O
O

 

 

N
O
0

Am or AQn [ng/cmz]
S
O

 

 

 

0 1 L 1 l r L 1 L L
2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

 

 

 

Figure 6.5. (A) Cyclic voltammogram, and (B) mass change (Am) and normalized charge
change (AQn with W/z = 22) profiles for untreated aluminum in 1M LiClO4/EC-DMC

during the first two potentiodynarnic cycles. Scan rate = 5mV/s.

134

6.2.4. EQCM measurements in 1M LiClOdEC-DMC and 1M LiPFdEC-DMC
- influence of water contamination

The potentiodynarnic i-E curves and associated mass change (Am) profiles for
untreated aluminum in 1M LiPF6/EC-DMC with 500 ppm of added water are presented
in Figure 6.6. A large oxidation peak current of 18 uA/cm2 is observed at ~ 4200 mV.
The oxidation charge during the first cycle was similar to one observed in dry electrolyte
(~2 mC/cmz). The surface becomes passivated during the first potentiodynamic cycle as
only a very small current was flowing during the second cycle. Passivation was
accompanied by deposition of solid oxidation product on the surface as indicated by the
mass increase (Figure 6.6B). Clearly, the addition of water does not cause significant
corrosion or mass loss of aluminum in LiPFGIEC-DMC. The normalized charge change
(AQn) is best fitted to the Am with a W/z = 15. This indicates the formation of slightly
heavier oxidation product compared to dry electrolyte (W/z = l 1). It may be explained by
formation of a passivation layer with a higher content of fluoride (e.g., A120F4) due to
the oxidation in the presence of increased amounts of F- formed from the hydrolysis of
LiPF6, as was proposed in Chapter 4. Formation of A1(OH)3 (W/z =17), according to

reaction [4.2], may also explain higher W/z values observed in the presence of water.

Completely different behavior was observed in LiClO4/EC-DMC with added water.
Figure 6.7 presents voltammetric i-E curves and associated mass change profiles for
untreated aluminum in 1M LiClO4/EC-DMC with 500 ppm of added water. It is evident
that the surface passivation does not occur during the first cycle. The onset potential for

the oxidation current is ca. 3200 mV. The mass change also begins at this potential.

135

The oxidation charge recorded during the second cycle (8.1 mC/cmz) is more than twice
that for the first cycle (3.5 mC/cmz). There is also clear evidence for pitting corrosion as
the i-E curve forms characteristic crossover, with the current during the reverse scan
being higher than during the forward scan. Epit and Epass are equal, respectively, 3200
and 3150 mV for the first cycle and 3150 and 3100 mV for the second cycle. The values
for both Epit and EpaSS are much lower than the values reported previously in Chapter 4.
This may be due to the different forms of aluminum used in both studies. Crystalline
aluminum metal is apparently much more corrosion resistant than a film of small
aluminum aggregates evaporated on a quartz surface.

The QCM data show that during the initial stage of potentiodynarnic polarization, up
to ~4500 mV, an increase in the electrode mass occurs. This increase is characterized by
W/z = 17, consistent with the formation of Al(OH)3. This is in agreement with the EDS
and Raman spectroscopic results reported in Chapter 4. However, at potentials more
positive than 4200 mV 3 mass loss is observed. The mass decreases until the
repassivation potential of ~ 3200 mV is reached during reverse sweep. Similar behavior
is seen during the second scan with a much larger mass loss. If one assumes that the
oxidation charge is associated exclusively with dissolution of aluminum, the mass loss
corresponding to the 11.6 mC/cm2 charge would be 1080 ng/cm2 (W/z = -9). This value
is roughly comparable to the total mass loss of ~800 ng/cm2 observed after the second
scan.

Dissolution and loss of aluminum is in contradiction to the results presented in
Chapters 4 and 5, which revealed the deposition of an Al(ClO4)3-rich layer during anodic

polarization with 500 ppm of added water. Also, the current and charge observed in the

136

i-E potentiodynarnic curves continuously increases with scan number at potentials more
positive than Epit- The i-E curves reported in Chapter 4 showed a clear current maximum
and diffusion limited current decrease. These observations indicate different corrosion
mechanisms and may result from the type of aluminum used — bulk aluminum vs. a thin-
film evaporated on the quartz. Optical microscopy and AFM were used to image the
surface after potentiodynarnic polarization to further investigate the nature of aluminum

corrosion.

Figure 6.8 presents photomicrographs (top) and AFM images (bottom) of the
untreated aluminum surface after 2 potentiodynarnic cycles in 1M LiClO4/EC-DMC with
500 ppm of added water. Round pits, from 20 to 150 um in diameter, are visible on the
electrode surface. The deflection mode AFM image (C) shows that the pit is essentially
empty of corrosion products with only small deposit visible in the middle of the pit. The
height mode AFM image taken at the pit edge shows a height of ~200 nm, which is the
thickness of the aluminum film on the quartz crystal. This indicates complete dissolution
of the aluminum inside a pit. This observation agrees with the EQCM data showing
aluminum dissolution. It also indicates a reason for the corrosion current increase at all
potentials positive of Epit- As corrosion proceeds only laterally in the thin film, the
corrosion products do not block the access to the corroding surface, thus no current peak
and diffusion-limited current decrease are observed. It is important to recognize that the
QCM data can be treated only qualitatively and not quantitatively, because linearity of
Sauerbrey equation cannot be assumed at highly localized corrosion where dissolution of

. 70
alurrunum occurs.

137

20X)

 

—n

S"

O
l

 

 

 

Current density [uA/cmz]
S
O
\

 

 

2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

 

 

 

 

 

 

\NVz=15

r—tN

UIO

CO
.,.,.

100' —-——Am
_ —AQ,,

O LE=_--=’—r ‘ - 1 wrw ‘ 1 - 1 .
2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

(J1
O

 

 

Am or AQn [ng/cmz]

 

 

Figure 6.6. (A) Cyclic voltammogram, and (B) mass change (Am) and normalized charge
change (AQn with W/z = 15) profiles for untreated aluminum in 1M LiPF6/EC-DMC

with 500 ppm of added water during the first two potentiodynarnic cycles. Scan rate =
5mV/s.

138

20.0

 

— 1”t scan ‘1
- - - ' 2"d scan .

fl
5"
O
t
1
1
l
I
I‘.
‘~

 

Current density [uA/cmz]
o. S
O O

 

 

 

0.0 - 1 ‘ . . . - . _
2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

 

400

 

200 1

 

 

O

 

[qg/cmzl
N
O

. ‘r"

— 18t scan ,.

Am
.L
O
O

- - - 2“" scan

3»

o

o

\

\
\
‘\

 

 

0 — . 1 __..---“' 1 . 1 . 1 .
2500 3000 3500 4000 4500 5000
Potential [mV vs. Li/Li+]

 

Figure 6.7. (A) Cyclic voltammogram, and (B) mass change (Am) profile for untreated
aluminum in 1M LiClO4/EC-DMC with 500 ppm of added water during the first two

potentiodynarnic cycles. Scan rate = 5mV/s.

139

 

 

 

Figure 6.8. (A), (B) Photomicrographs of the circular pits formed on the untreated
aluminum surface after 2 potentiodynarnic scans in 1M LiClO4/EC-DMC with 500 ppm
of added water. Scan rate = 5mV/s. (C) AFM deflection mode image of one of the pits.
(D) AFM height mode image of the edge of the pit. The height of the edge is equal to 200

nm.

140

6.3. Conclusions

Corrosion/passivation of aluminum in LiClO4/EC-DMC and LiPF6/EC-DMC was
studied with EQCM. The state of the aluminum surface (untreated vs. mechanically
polished) and the influence of water contamination were investigated. During anodic
polarization in electrolyte/solvent mixtures with no added water, aluminum passivates in
both electrolytes. Surface passivation is accompanied by deposition of the oxidation
products. The passivation layer formed in LiPF6 is characterized by an average apparent
molecular weight per electron (W/z) of 11 equiv/g, consistent with the formation of
aluminum oxyfluoride (AlOF). The passivation layer formed in LiClO4/EC-DMC is
probably a mixture of A1203, Al(OH)3, and Al(ClO4)3. The proportion of Al(ClO4)3 in
the layer seems to be higher on the mechanically polished (W/z = 60) than on the
untreated (W/z = 22) surface.

Water contamination leads to passivation of aluminum in LiPF6/EC-DMC. An
average W/z of 15, consistent with a fluoride and/or hydroxide rich passivation layer is
observed. In LiClO4/EC-DMC, the presence of water leads to catastrophic pitting

corrosion and loss of aluminum from the thin film.

141

CHAPTER 7

7. Summary

Corrosion susceptibility of aluminum in LiPF6/EC-DMC, LiPF6/PC, LiClO4/EC-
DMC, and LiClO4/PC in the absence and presence of added water impurity was
investigated in the present studies. The electrochemical measurements (CV, CA, and
OCP) were combined with microscopic imaging (OM, SEM, and AFM) and surface
analysis techniques (XPS, EDS and EQCM) to understand processes that govern the

stability of aluminum in selected non-aqueous electrolytes. It was found that:

1. None of the four investigated electrolyte/solvent systems is corrosive toward
aluminum in the absence of added water (below 50 ppm), at least up to 5 V vs.
Li/Li+. A solid passivation layer forms on the aluminum surface at such conditions.
A1F3, AlOF, A1203, and LiF were found as the major components of a passivation
layer formed in LiPF6-based solutions, whereas the layer formed in LiClO4

consisted mostly of Al(ClO4)3, A1203, and A1(OH)3.

2. Anodic polarization in LiClO4/EC-DMC and LiClO4/PC, in the presence of added
water, lads to pitting and progressive corrosion. Passivity of aluminum is disturbed at

added water levels as low as 50 ppm. Water levels higher than 500 ppm significantly

142

increase the rate of corrosion. This is most probably due to increased dissolution of
Al(ClO4)3 and the evolution of gaseous oxidation products, which disturb the layer
of solid corrosion products and enhance transport of solvent and electrolyte toward

the metal surface.

Anodic polarization in LiPF6/EC-DMC and LiPF6/PC in the presence of added water
does not cause pitting and corrosion. The surface remains effectively passivated by

the highly insoluble aluminum oxyfluoride/hydroxyfluoride layer.

An oxide layer, present on the electropolished aluminum, offers additional corrosion
protection in the electrolyte/solvent mixtures without added water impurity. It does
not, however, protect aluminum from pitting that occurs in LiClO4/EC-DMC and
LiClO4/PC in the presence of added water impurity. At water levels lower than 500
ppm, pits that form on the surface are filled with the deposited corrosion products
(mostly Al(ClO4)3 and Al(OH)3). Higher water levels lead to increased corrosion
rates and formation of empty pits. This is due to evolution of gaseous oxidation

products from the pit interior.

Considering the mechanism of aluminum passivation it was found that at open
circuit conditions aluminum passivates through salt film formation in LiPF6 and
solvent adsorption in LiClO4. During anodic polarization, aluminum passivates by
salt film formation in both electrolytes. In LiClO4, it is a precipitate of Al(ClO4)3,

whereas in LiPF6, a salt film forms from the reaction between aluminum and F-

143

formed from the hydrolysis of PF6-. The passivation layer formed in LiClO4 is less
stable than in LiPF6. This is due to slow dissolution of the Al(ClO4)3 layer in
organic solvent. Water interferes with both solvent adsorption and salt film

formation.

Finally, the electrolyte type (LiClO4 vs. LiPF6) not the solvent (PC vs. EC-DMC)

has the most significant effect on the corrosion susceptibility of aluminum.

144

10.

11.

12.

13.

REFERENCES

R. W. Revie, Uhlig's Corrosion Handbook, Wiley-Interscience, New York,
(2000).

K. Kanamura, T. Okagawa, and Z.-I. Takehara, J. Power Sources, 57, 119 (1995).

K. Kanamura, S. Toriyama, S. Shiraishi, and Z.-I. Takehara, J. Electrochem. Soc.,
142, 1383 (1995).

J. W. Braithwaite, A. Gonzales, G. Nagasubrarnanian, S. J. Lucero, D. E. Peebles,
J. A. Ohlhausen, and W. R. Cieslak, J. Electrochem. Soc., 146, 448 (1999).

L. J. Krause, W. Larnanna, J. Summerfield, M. Engle, G. Korba, R. Loch, and R.
Atanasoski, J. Power Sources, 68, 320 (1997).

K. Kanamura, T. Umegaki, S. Shiraishi, M. Ohashi, and Z.-I. Takehara, J.
Electrochem. Soc., 149, A185 (2002).

H. Yang, K. Kwon, T. M. Devine, and J. W. Evans, J. Electrochem. Soc., 147,
4399 (2000).

C. W. Walker, Jr., J. D. Cox, and M. Salomon, J. Electrochem. Soc., 143, L80
(1996).

W. K. Behl and E. J. Plichta, J. Power Sources, 72, 132 (1998).
J. L. Goldman and A. B. McEwen, Electrochem. Solid-State Lett., 2, 501 (1999).

M. Z. A. Munshi, R. Gopaliengar, and B. B. Owens, Solid State Ionics, 27, 259
(1988).

H. S. Choe, B. G. Carroll, D. M. Pasquariello, and K. M. Abraham, Chem. Mater.,
9, 369 (1997).

Y. Chen, T. M. Devine, J. W. Evans, O. R. Monteiro, and I. G. Brown, J.
Electrochem. Soc., 146, 1310 (1999).

145

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

Y. Matsuda, Y. Ouchi, and H. Tamura, J. Appl. Electrochem, 4, 53 (1974).

L. Bai and B. E. Conway, J. Electrochem. Soc., 137, 3737 (1990).

S. Licht, G. Levitin, C. Yarnitzky, and R. Tel-Vered, Electrochem. Solid-State
Lett, 2, 262 (1999).

M. Ue, F. Mizutani, S. Takeuchi, and N. Sato, J. Electrochem. Soc., 144, 3743
(1997).

M. Ue, K. Shima, and S. Mori, Electrochim. Acta, 39, 2751 (1994).

M. Morita and Y. Matsuda, J. Power Sources, 60, 179 (1996).

L. A. Pawlick and R. G. Kelly, J. Corros. Sci. Eng. [Electronic Publication:
http://www. cp. umist. ac. uk/J C SE/Vol 1/PAPER4/Paper4. htm I , l, ( 1995).

R. Umebayashi, N. Akao, N. Hara, and K. Sugimoto, J. Electrochem. Soc., 149,
B75 (2002).

S. Megahed and B. Scrosati, Interface, 4, 34 (1995).

I. Buchmann, Batteries in a Portable World, Cadex Electronics Inc., Richmond,
Canada, (2001).

P. Arora, R. E. White, and M. Doyle, J. Electrochem. Soc., 145, 3647 (1998).

J. R. Owen, Chem. Soc. Rev., 26, 259 (1997).

M. Zhao, S. Kariuki, H. D. Dewald, F. R. Lemke, R. J. Staniewicz, E. J. Plichta,
and R. A. Marsh, J. Electrochem. Soc., 147, 2874 (2000).

K. Kanamura, S. Toriyama, S. Shiraishi, and Z.-I. Takehara, J. Electrochem. Soc.,
143, 2548 (1996).

G. S. Frankel, J. Electrochem. Soc., 145, 2186 (1998).

R. G. Kelly and P. J. Moran, Corros. Sci, 30, 495 (1990).

146

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

45.

J. F. Scanlon, J. Kruger, and P. J. Moran, J. Electrochem. Soc., 140, 1268 (1993).

R. G. Kelly, P. J. Moran, J. Kruger, C. Zollman, and E. Gileadi, J. Electrochem.
Soc., 136, 3262 (1989).

S. Wernick, The Surface Treatment and Finishing of Aluminium and Its Alloys,
Finishing Publications Ltd., Middlesex, England, (1990).

R. C. Weast, ed., CRC Handbook of Chemistry and Physics, CRC Press, Inc.,
Boca Raton, FL, (1980).

P. A. Malachesky, in Encyclopedia of Electrochemistry of the Elements, Vol. VI
(A. J. Bard, ed.), Marcel Dekker, Inc., New York, 1976.

R. T. Atanasoski, J. Serb. Chem. Soc., 57, 935 (1992).

G. Eggert and J. Heitbaum, Electrochim. Acta, 31, 1443 (1986).

E. Cattaneo and J. Ruch, J. Power Sources, 44, 341 (1993).

D. A. Shifler, P. J. Moran, and J. Kruger, Corros. Sci, 32, 475 (1991).

D. A. Shifler, P. J. Moran, and J. Kruger, J. Electrochem. Soc., 139, 54 (1992).

P. Novak, P. A. Christensen, T. Iwasita, and W. Vielstich, J. Electroanal. Chem.
Interfacial Electrochem., 263, 37 (1989).

D. Battisti, G. A. Nazri, B. Klassen, and R. Aroca, J. Phys. Chem, 97, 5826
(1993).

K. A. Rodgers, Clay Miner., 28, 85 (1993).

W. H. Leong and D. W. James, Aust. J. Chem, 22, 499 (1969).

R. G. Kelly, P. J. Moran, E. Gileadi, and J. Kruger, Electrochim. Acta, 34, 823
(1989).

P. Zelenay, M. Winnicka-Maurin, and J. Sobkowski, J. Electroanal. Chem.
Interfacial Electrochem., 278, 361 (1990).

147

y

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

J. N. Butler, D. R. Cogley, and E. Grunwald, J. Phys. Chem, 75, 1477 (1971).

G. Binning, H. Rohrer, C. Gerber, and E. Weibel, Phys. Rev. Lett., 49, 57 (1982).

G. Binnig, C. F. Quate, and C. Gerber, Phys. Rev. Lett., 56, 930 (1986).

B. Drake, C. B. Prater, A. L. Weisenhom, S. A. Gould, T. R. Albrecht, C. F.
Quate, D. S. Cannell, H. G. Hansma, and P. K. Hansma, Science, 243, 1586
(1989).

S. Manne, P. K. Hansma, J. Massie, V. B. Elings, and A. A. Gewirth, Science,
251, 133 (1991).

N. Ikemiya, S. Miyaoka, and S. Hara, Surf. Sci, 327, 261 (1995).

B. J. Cruickshank, D. D. Sneddon, and A. A. Gewirth, Surf. Sci, 281, L308
(1993).

P. Mrozek, Y.-E. Sung, M. Han, M. Gamboa-Aldeco, A. Wieckowski, C.-H.
Chen, and A. A. Gewirth, Electrochim. Acta, 40, 17 (1995).

C. A. Alves and M. D. Porter, Langmuir, 9, 3507 (1993).

K. Kowal, J. DeLuccia, J. Y. Josefowicz, C. Laird, and G. C. Farrington, Mater.
Res. Soc. Symp. Proc., 409, 201 (1996).

R. M. Rynders, C. H. Paik, R. K6, and R. C. Alkire, J. Electrochem. Soc., 141,
1439 (1994).

G. Bertrand, E. Rocca, C. Savall, C. Rapin, J. C. Labrune, and P. Steinmetz, J.
Electroanal. Chem, 489, 38 (2000).

U. Kamachi Mudali and Y. Katada, Electrochim. Acta, 46, 3735 (2001).

J. Li and D. J. Meier, J. Electroanal. Chem, 454, 53 (1998).

N. Ikemiya, T. Kubo, and S. Hara, Surf. Sci, 323, 81 (1995).

J. Zak and M. Kolodziej-Sadlok, Electrochim. Acta, 45, 2803 (2000).

148

62.

63.

65.

66.

67.

68.

69.

70.

71.

72.

73.

74.

K. Kowal, L. Xie, R. Huq, and G. C. Farrington, J. Electrochem. Soc., 141, 116
(1994).

J. Y. Josefowicz, L. Xie, and G. C. Farrington, J. Phys. Chem, 97, 11995 (1993).

F. Bellucci, G. Capobianco, A. Deganello, A. Glisenti, T. Monetta, and G.
Moretti, Mat. Sci. Forum, 289-292, 1311 (1998).

Y. S. Cohen, Y. Cohen, and D. Aurbach, J. Phys. Chem. B, 104, 12282 (2000).

D. Aurbach, B. Markovsky, M. D. Levi, E. Levi, A. Schechter, M. Moshkovich,
and Y. Cohen, J. Power Sources, 81-82, 95 (1999).

D. Aurbach, A. Zaban, Y. Ein-Eli, I. Weissman, O. Chusid, B. Markovsky, M.
Levi, E. Levi, A. Schechter, and E. Granot, J. Power Sources, 68, 91 (1997).

T. Hurlen and K. H. Johansen, Acta Chem. Scand., Ser. A, A39, 545 (1985).

J. A. Richardson and G. C. Wood, J. Electrochem. Soc., 120, 193 (1973).

M. R. Deakin and D. A. Buttry, Anal. Chem, 61, 1147A (1989).

M. D. Ward, in Physical Electrochemistry (I. Rubinstein, ed.), p. 293, Marcel
Dekker, New York, 1995.

M. D. Ward and D. A. Buttry, Science, 249, 1000 (1990).

G. Sauerbrey, Z. Physik, 155, 206 (1959).

M. R. Deakin and O. Melroy, J. Electroanal. Chem. Interfacial Electrochem.,
239, 321 (1988).

149