development of wettable cathode for aluminum …€¦ · developing new materials, as alternative...

HAMED HEIDARI

DEVELOPMENT OF WETTABLE CATHODE FOR

ALUMINUM SMELTING

Thèse présentée

à la Faculté des études supérieures et postdoctorales de l’Université Laval

dans le cadre du programme de doctorat en génie des matériaux et de la métallurgie

pour l’obtention du grade de Philosophiae doctor (Ph.D.)

DÉPARTEMENT DE GÉNIE DES MINES, DE LA MÉTALLURGIE ET DES MATÉRIAUX

FACULTÉ DES SCIENCES ET DE GÉNIE UNIVERSITÉ LAVAL

QUÉBEC

2012

© Hamed Heidari, 2012

Résumé

Le procédé de l’électrolyse Hall-Héroult demeure la méthode principale pour la production

mondiale de l’aluminium primaire depuis son invention en 1887. L'utilisation de cathodes

mouillables au lieu de cathodes de carbone usuelles a été proposée afin de réduire de plus

de 10% de la consommation d'énergie électrique du procédé, ce qui constitue plus de 35%

de coûts de la production de l'aluminium. Cependant, à cause des conditions sévères qui

prévalent dans le bain d'électrolyse, la fabrication d'une bonne cathode mouillable a été un

défi au cours des 60 dernières années et aucune cathode mouillable commerciale n’est

encore disponible sur le marché mondial. Dans ce projet, une nouvelle céramique poreuse a

été développée par frittage sans pression de TiB2 avec additif de Ti-Fe pré-allié. Ce

matériau possède les propriétés requises pour servir de cathode mouillable.

Dans cette étude doctorale, le frittage en phase liquide sans pression a été choisi comme

méthode de consolidation permettant la fabrication de grandes pièces à un coût relativement

bas. Des essais ont été réalisés afin de comprendre l'effet de différentes conditions de

traitement y compris la composition d'additif, la température de frittage, le temps de

broyage, et la pré-alliage des additifs sur les propriétés physiques, mécaniques et

métallurgiques ainsi que le comportement de mouillage et la stabilité des spécimens dans

l'aluminium liquide.

Après l'ajustement des paramètres de procédé, le matériau sélectionné a été fabriqué par le

mélange de TiB2 en poudre avec 10% en poids d'additif 7Ti-3Fe pré-allié dans le broyeur à

billes à haute énergie pendant 30 min, suivi par un pressage à 150 MPa et un frittage sous

atmosphère de Ar/H2 à 1650°C pendant 1 h.

Une microstructure sans fissures avec une distribution uniforme de pores, une densité

maximale relative de 91%, une résistance à la flexion de 300 MPa et une résistivité

électrique de 54 μΩ.cm ont donc été obtenues. Une goutte d’aluminium a très bien mouillé

la surface de l'échantillon et une solidification isotherme s'est produite lors de sa

pénétration due à l'interaction avec les additifs métalliques et la formation des phases TiAl3

et Fe4Al13.

ii

Malgré la dissolution des additifs métalliques, le matériau développé a montré une

excellente stabilité après exposition dans l'aluminium fondu à 960°C pour une durée

maximale de 5 jours tout en maintenant sa forme, et aucun signe d’expansion ou de

gonflement n’ont été observés. Les analyses microstructurales ont révélé la formation de

ponts de TiB2 entre les particules, en présence de phase liquide de Ti-Fe au cours du

frittage, et donc la formation d’un squelette TiB2 qui est la cause de la stabilité du matériau

développé dans l'aluminium liquide. Par conséquent, ce matériau est proposé en tant qu’un

candidat fiable pour l'application en tant que cathodes mouillables dans la production

d'aluminium.

iii

Abstract

Hall-Héroult electrolysis process has been the major method for world production of

primary aluminum since its invention in 1887. The use of wettable cathodes instead of

usual carbon cathodes has been proposed to reduce more than 10% of the electrical energy

consumption of the process which constitutes more than 35% of the aluminum production

costs. However, due to the severe conditions of the electrolysis bath, the fabrication of a

proper wettable cathode has been a challenge during the last 60 years and no commercial

wettable cathode is available in the world market yet. In this project, a novel porous

ceramic by pressureless sintering of TiB2 with pre-alloyed Ti-Fe additives was developed.

This material showed to meet the required properties to be used as wettable cathode. In this

doctoral study, the pressureless sintering in the presence of liquid phase was selected as the

consolidation method allowing the fabrication of large parts at relatively lower

temperatures and costs. Experimental efforts were made in order to understand the effect of

different processing conditions including additive composition, sintering temperature,

milling time and pre-alloying of additives on the physical, mechanical and metallurgical

properties as well as wetting behavior and stability in liquid aluminum of specimens. Based

on the results of the adjustment of processing parameters, the selected material was

fabricated by mixing of TiB2 powder and 10 wt% pre-alloyed 7Ti-3Fe additive in high

energy ball mill for 30 min, compacting under the pressure of 150 MPa to prepare the green

parts, and sintering under Ar/H2 atmosphere at 1650C for 1 h. Uniform crack-free

microstructure with even distribution of pores as well as maximum relative density of 91%,

bending strength of 300 MPa and electrical resistivity of 54 µΩ.cm were accordingly

obtained. Aluminum drop wetted the surface of the specimen very well and isothermal

solidification occurred during its penetration due to the interaction with the metallic

additives and the formation of TiAl3 and Fe4Al13 phases. Despite of the dissolution of

metallic additives, this material showed excellent stability after being exposed to molten

aluminum at 960C for up to 5 days by maintaining its shape and no sign of expansion or

swelling was observed. Microstructural investigation revealed the precipitation of inter-

iv

particle bridges of TiB2 nature in the presence of Ti-Fe liquid phase during sintering

forming a TiB2 skeleton, which is the cause of the stability of the developed material in

liquid aluminum. This material is proposed as a reliable candidate for application as

wettable cathodes in aluminum smelting.

Preface

This doctoral thesis is presented to the department of mining, materials and metallurgical

engineering of Laval university and reports the works carried out as part of the project:

Development of wettable cathode for aluminum smelting. The project was supported by a

Collaborative Research and Development (CRD) grant from The Natural Science and

Engineering Research Council of Canada (NSERC) and Hydro Québec as a part of “Energy

Efficient Programs”.

Hall-Héroult electrolysis process has been the major method for producing primary

aluminum in the world since its invention in 1887. A lot of improvements and

modifications have been made to this process in order to reduce the energy consumption

from more than 50 kWh/kg Al in year 1900 to less than 15 kWh/kg Al in 2010. Despite the

considerable increase of the energy efficiency of the process, it is still the major consumer

of electrical energy in Canada as the third largest producer of primary aluminum in the

world. Developing new materials, as alternative for common carbon cathodes, which are

wettable by liquid aluminum has been proposed as a solution for further increase of the

energy efficiency of this process.

TiB2 is one of the few materials, with good electrical conductivity, which could withstand

the severe conditions of electrolysis cell for long durations. However, the fabrication of

large TiB2 cathode parts is encountered by several scientific and technical issues, which

made the development of wettable TiB2-based cathodes a 60-years challenge for aluminum

industry.

This thesis reports firstly the review of the previous works reported on this subject,

followed by materials selection and proposed methodology. Efforts made during this

doctoral study include the design, development and evaluation of a new candidate material

for wettable cathode. Numerous preliminary experiments were performed at the beginning

of the study to choose the best combination of materials and process parameters for this

challenging application.

vi

In this context, we propose, for the first time, the use of a pre-alloyed Ti-Fe additive for

pressureless sintering of TiB2-based ceramics at the presence of liquid phase. Fabrication of

ceramic materials which meet the property requirements of wettable cathode with the use of

commercial grade TiB2 powder, short milling time of 30 min, compacting pressure of 150

MPa and pressureless sintering at temperature of 1650°C make the developed method a

potential choice for fabrication of large cathode parts at relatively low price in industrial

scales.

This doctoral project was carried out under the direction of Professor Houshang Alamdari

and co-direction of Professor Dominique Dubé from Laval University and collaboration of

Dr. Robert Schulz from Hydro Québec research institute (IREQ). This thesis has been

prepared as an article insertion thesis and includes four peer-reviewed articles (submitted

and/or published), which report the results obtained from collaborative works of the authors

in Laval University and Hydro-Québec during the different stages of the project.

My contribution to these articles was the identification of the objective of each article,

preparation of the plan of experiments, design and assembly of the experimental set-ups

and performing the experiments including: consolidation of specimens, metallographic

preparations, microstructural studies, mechanical and electrical evaluations, and chemical

stability investigations. I performed the density and electrical resistivity measurements as

well as wettability evaluations with the help of Dr. Robert Schulz and Mr. Sylvio Savoie at

IREQ. I subsequently prepared the first draft of the articles, which were revised by the co-

authors before submission.

The first article titled: Pressureless sintering of TiB2–based composites using Ti and Fe

additives for development of wettable cathodes, co-authored by Prof. Houshang Alamdari,

Prof. Dominique Dubé and Dr. Robert Schulz, was presented at TMS 2011 international

conference and was published in the journal: Light Metals 2011, P. 1111-1116 1. In this

article the effect of composition and processing parameters on the microstructure and

physical properties of consolidated TiB2 parts was studied.

vii

The second article titled: Investigating the potential of TiB2-based composites with Ti and

Fe additives as wettable cathode, co-authored by Prof. Houshang Alamdari, Prof.

Dominique Dubé and Dr. Robert Schulz, was presented in Thermec 2011 international

conference and was published in both the journal: Materials Science Forum, 2012, Vol.

706-709, P. 655-660 2; and the journal: Advanced Materials Research, 2012, Vol. 409, P.

195-200 3. In this paper the microstructure, physical properties and wettability of TiB2-

based specimens with different milling time were compared. The interaction of specimens

with liquid Al when exposed to molten Al up to 24 h was investigated as well.

The third article titled: Interaction of molten aluminum with porous TiB2-based ceramics

containing Ti-Fe additives, co-authored by Prof. Houshang Alamdari, Prof. Dominique

Dubé and Dr. Robert Schulz, was published in: Journal of the European Ceramic Society,

2012, Vol. 32, Issue 4, P. 937-945 4. In this paper the interaction of TiB2 ceramics

consolidated using Ti-Fe additive with aluminum drop was investigated. First the

interaction of surface oxide layer with liquid Al was studied. Then the mechanism of

penetration of Al inside the specimen and its interaction with the metallic phases in the

inter-particle spaces were discussed.

The fourth article titled: Pressureless sintering of TiB2-based ceramics with Ti-Fe additive:

sintering mechanism and stability in liquid aluminum, co-authored by Prof. Houshang

Alamdari, Prof. Dominique Dubé and Dr. Robert Schulz, was published in: Advanced

Engineering Materials journal, 2012, Vol. 14, P. 802–880 5. In this paper the effect of

different sintering additives including Fe, Ti and Ti-Fe on the stability of consolidated parts

in molten aluminum was investigated. A complete discussion was presented to explain how

Ti-Fe additive could promote the formation of TiB2 phase as the inter-particle bridges

during the sintering process which is responsible for the stability of this material in molten

aluminum at prolonged exposure time.

viii

Acknowledgments

I would like to express my appreciation to my thesis director Prof. Houshang Alamdari, and

my co-director, Prof. Dominique Dubé, for having confidence in me to conduct this project,

for their great availability for meetings and discussions and for their valuable comments

and suggestions. Their encouragement, guidance, knowledge and advices were very helpful

and appreciated all through my studies. I am also thankful to my co-director in Hydro

Québec Research Institute, Dr. Robert Schulz, for his insight, support, and his valuable

comments and discussions throughout this project. Moreover, I grateful to him for

providing me with the experimental setups, materials and equipment in IREQ laboratories,

which greatly helped me evaluate the developed materials and progress my project. This

dissertation would not have happened without you all.

Special thanks goes to Sylvio Savoie for his proficiency and expertise in the fabrication of

the experimental set-ups I used for my experiments and for his technical assistance to help

me realize the analyses.

I would like to acknowledge Prof. Carl Blais and his team: Philippe Lapointe, Nicolas

Giguère, Bernard Tougas for their kind help and for generously allowing me to use their

laboratory and equipment.

Thanks to all of the staff of Mining, Metallurgical and Materials Engineering Department

of Laval University for their help and support. I am grateful to Maude Larouche, Marc

Choquette André Fernand and Jean Frenette for their help with microstructural analyses,

Daniel Marcotte and Marie-Josée Bouchard for their availability and technical assistance.

Thanks also to my friends and colleagues Mohammad Ghasdi, Kamran Azari, Francois

Chevarin, Milad Mardan, and all my friends at the department for making a lively and

joyful environment and for the greats moments we had together.

ix

I would like to thank my parents for their support, patience, and sacrifice in all and every

stage of my life, and my brother Majid who has made me feel having my family beside me

here in Québec.

Last, but not least, I am very deeply grateful to my lovely wife, Maryam, for her love,

dedication, understanding, patience, support and encouragement. My special thanks for her

guidance, technical discussions and for kindly revising my manuscripts during my doctoral

study.

To my loved ones

Table of contents

Résumé .......................................................................................................................................... i

Abstract ....................................................................................................................................... iii

Preface.......................................................................................................................................... v

Acknowledgments .................................................................................................................... viii

Table of contents ........................................................................................................................ xi

List of tables ............................................................................................................................ xvii

List of figures ......................................................................................................................... xviii

Chapter 1: Introduction ............................................................................................................... 1

1.1 General context .............................................................................................................. 2

1.2 Project motivations ........................................................................................................ 5

1.3 Structure of thesis .......................................................................................................... 6

Chapter 2: Literature review ....................................................................................................... 9

2.1 Aluminum..................................................................................................................... 10

2.1.1 Aluminum production industry ........................................................................... 10

2.1.2 Hall-Héroult process ............................................................................................ 10

2.2 Issues with carbon cathode in Hall-Héroult cell ........................................................ 18

2.2.1 Non-wettable for liquid aluminum ...................................................................... 18

2.2.2 Penetration of electrolyte and liquid aluminum ................................................. 18

2.3 Wettable drained cathode ............................................................................................ 20

xii

2.3.1 General property requirements for an ideal cathode material ........................... 20

2.3.2 Choice of material ................................................................................................ 21

2.4 Titanium diboride ........................................................................................................ 24

2.4.1 Phase diagram, crystal structure and properties of TiB2 .................................... 24

2.4.2 Synthesis of TiB2 .................................................................................................. 26

2.5 TiB2-based cathode technologies ................................................................................ 27

2.6 TiB2-based cathodes .................................................................................................... 30

2.6.1 Use of metallic sinter additives ........................................................................... 31

2.6.2 Use of ceramic sinter additives............................................................................ 34

Chapter 3: Thesis outline .......................................................................................................... 37

3.1 Objectives ..................................................................................................................... 38

3.2 Choices of sintering additives ..................................................................................... 39

3.3 Originality of the project ............................................................................................. 41

Chapter 4: Materials and methods............................................................................................ 43

4.1 Introduction .................................................................................................................. 44

4.2 Experimental procedures ............................................................................................. 44

4.2.1 Starting powders ................................................................................................... 44

4.2.2 Powder processing................................................................................................ 45

4.2.3 Forming method ................................................................................................... 45

xiii

4.2.4 Consolidation and sintering ................................................................................. 45

4.3 Characterisation methods ............................................................................................ 47

4.3.1 Bulk density and apparent porosity measurements ............................................ 47

4.3.2 Mechanical properties evaluation ........................................................................ 48

4.3.3 Measurement of electrical resistivity .................................................................. 49

4.3.4 Wettability in liquid aluminum ........................................................................... 50

4.3.5 Chemical stability and durability to liquid aluminum ....................................... 50

4.3.6 Scratch test ............................................................................................................ 51

4.3.7 Microstructural characterisation .......................................................................... 51

4.3.8 Compositional and phase analysis....................................................................... 52

Chapter 5: Pressureless sintering of TiB2–based composites using Ti and Fe additives for

development of wettable cathodes ........................................................................................... 53

5.1 Résumé ......................................................................................................................... 54

5.2 Abstract ........................................................................................................................ 54

5.3 Introduction .................................................................................................................. 56

5.4 Materials and methods................................................................................................. 58

5.5 Results and discussion ................................................................................................. 60

5.5.1 Effect of additive composition ............................................................................ 60

5.5.2 Effect of sintering temperature ............................................................................ 61

xiv

5.5.3 Effect of pre-alloying additives ........................................................................... 62

5.5.4 Effect of milling time ........................................................................................... 62

5.5.5 Wettability and stability in liquid aluminum ...................................................... 70

5.6 Conclusions .................................................................................................................. 71

5.7 Acknowledgement ....................................................................................................... 71

Chapter 6: Investigating the potential of TiB2–based composites with Ti and Fe additives as

wettable cathode ........................................................................................................................ 72

6.1 Résumé ......................................................................................................................... 73

6.2 Abstract ........................................................................................................................ 73

6.3 Introduction .................................................................................................................. 74



6.6 Conclusions .................................................................................................................. 83

Chapter 7: Interaction of molten aluminum with porous TiB2–based ceramics containing

Ti-Fe additives ........................................................................................................................... 84

7.1 Résumé ......................................................................................................................... 85

7.2 Abstract ........................................................................................................................ 85

7.3 Introduction .................................................................................................................. 86



xv

7.5.1 Sessile drop tests .................................................................................................. 88

7.5.2 Early Stage of Interaction .................................................................................... 90

7.5.3 Later Stage of Interaction..................................................................................... 93

7.5.4 Reaction Mechanism ............................................................................................ 99

7.6 Conclusion.................................................................................................................. 102

7.7 Acknowledgements ................................................................................................... 103

Chapter 8: Pressureless sintering of TiB2-based ceramics with Ti-Fe additives: sintering

mechanism and stability in liquid aluminum......................................................................... 104

8.1 Résumé ....................................................................................................................... 105

8.2 Abstract ...................................................................................................................... 105

8.3 Introduction ................................................................................................................ 106

8.4 Materials and methods............................................................................................... 107

8.5 Results and discussion ............................................................................................... 110

8.5.1 Influence of additives on physical and metallurgical properties ..................... 110

8.5.2 Interaction with a liquid aluminum drop .......................................................... 113

8.5.3 Stability in liquid aluminum .............................................................................. 118

8.5.4 Scratch test .......................................................................................................... 119

8.5.5 TEM analysis ...................................................................................................... 120

8.6 Conclusions ................................................................................................................ 122

xvi

8.7 Acknowledgements ................................................................................................... 123

Chapter 9: General discussion and conclusions .................................................................... 124

Chapter 10: Perspectives of the project ................................................................................. 130

References: .............................................................................................................................. 133

Appendix 1: Ceramic fabrication ........................................................................................... 140

List of tables

Table 2–1- Electrical resistivity of some ceramics (in Ω cm) 27

............................................ 22

Table 2–2- Physical properties of some RHMs 22

................................................................... 23

Table 4–1- Starting powders specification .............................................................................. 44

Table 5–1- Experimental conditions used for consolidation of specimens ........................... 59

Table 5–2- Relative density of specimens with separate (T7F3M10) and pre-alloyed

(T7F3PM10) additives...................................................................................................... 62

Table 8–1- The starting composition of sintered specimens, their relative density and 3-

point bending strength. ................................................................................................... 110

List of figures

Figure 2–1- Schematic of a typical Hall-Héroult cell for electrolytic smelting of aluminum

13 ......................................................................................................................................... 11

Figure 2–2- Voltage distribution in a Hall- Héroult cell 10

..................................................... 16

Figure 2–3- Binary Ti-B phase diagram 32

.............................................................................. 25

Figure 2–4- The structure of TiB2 in a projection along the hexagonal axis (right) and a

perspective view (left) 34

.................................................................................................. 25

Figure 2–5- Transmission electron micrograph of the specimen with 0.5 wt% Cr and 0.5

wt% Fe and energy dispersive spectra of X-ray microanalysis at a triple junction 51

. . 32

Figure 3–1- Binary phase diagram of Fe-Ti 64

. ....................................................................... 40

Figure 4–1- Diagram of sintering cycle ................................................................................... 46

Figure 4–2- Schematic of three-point loading for measuring the flexural strength of

ceramics 68

. ........................................................................................................................ 48

Figure 4–3- Schematic of setup used for sessile drop test and wettability investigation ..... 50

Figure 5–1- Comparison of the relative density as a function of sintering temperature and

composition of sintering additives (T8F2M10: TiB2+8%Ti+2%Fe; T7F3M10:

TiB2+7%Ti+3%Fe). .......................................................................................................... 60

Figure 5–2- Backscattered SEM micrograph of T8F2M10 (TiB2+8%Ti+2%Fe) specimen,

sintered at 1400°C for 1 h (The arrows show segregated phases containing the

additives). .......................................................................................................................... 61

Figure 5–3- Influence of milling time on relative density of green and sintered specimens

(TiB2+7%Ti+3%Fe) using pre-alloyed additive and sintered at 1650°C for 1h........... 63

xix

Figure 5–4- Influence of milling time on bending strength of specimens

(TiB2+7%Ti+3%Fe) prepared using pre-alloyed additive and sintered at 1650°C for

1h. ...................................................................................................................................... 64

Figure 5–5- Effect of milling time on the particle size distribution for powder mixtures

containing the 70%Ti and 30%Fe pre-alloyed additive. ................................................ 65

Figure 5–6- SEM micrograph of a large agglomerate formed after 240 min milling in

T7F3PM240 powder (TiB2+7%Ti+3%Fe). .................................................................... 66

Figure 5–7- XRD analysis of powders containing pre-alloyed additives after 10 and 240

min milling. (Cu Kα). ....................................................................................................... 67

Figure 5–8- Backscattered SEM micrograph of T7F3PM10 specimen (TiB2+7%Ti+3%Fe)

milled for 10 min and sintered 1 h at 1650°C. ................................................................ 68

Figure 5–9- Backscattered SEM micrograph of T7F3PM30 specimen (TiB2+7%Ti+3%Fe)

milled for 30 min and sintered 1h at 1650°C. ................................................................. 69

Figure 5–10- Backscattered SEM micrograph of T7F3PM120 specimen

(TiB2+7%Ti+3%Fe) milled for 120 min and sintered 1 h at 1650°C............................ 69

Figure 5–11- Behavior of liquid Al drop over T7F3PM30 specimen (TiB2+7%Ti+3%Fe)

during the wettability test at different time. (The time from beginning of test are

reported in minutes) .......................................................................................................... 70

Figure 6–1- Cumulative particle size distribution diagram of pure TiB2, pre-alloyed 7Ti3Fe,

and mixtures after different milling times (M10 e.g. means mixed powder milled for

10 min)............................................................................................................................... 77

Figure 6–2- Micrograph of specimens milled for a) 10 min, b) 30 min, c) 240 min and

sintered at 1650°C for 1h. (Phases in the microstructure are A: TiB2 ; B: TiFe ; C: α-Ti

; D: TiFe2) .......................................................................................................................... 78

xx

Figure 6–3- Photo taken from liquid Al drop on the surface of M30 specimen. .................. 79

Figure 6–4- Wetting angle of liquid Al on M30 and M240 specimens as a function of

elapsed time ....................................................................................................................... 80

Figure 6–5- Micrograph of M30 specimens exposed to liquid Al at 960°C: for 1h: a) BSE

micrograph of contact area, b) Mapping of Al element, c) Mapping of Fe element; and

for 24h: d) BSE micrograph of contact area, e) Mapping of Al element, f) Mapping of

Fe element ......................................................................................................................... 81

Figure 6–6- Micrograph of M240 specimens exposed to liquid Al at 960°C: for 1h: a) BSE

micrograph of Al penetration zone, b) Mapping of Al element, c) Mapping of Fe

element; and for 24h: d) BSE micrograph of Al penetration zone, e) Mapping of Al

element, f) Mapping of Fe element.................................................................................. 82

Figure 7–1- Photos from the contact between a liquid Al drop and a polished specimen’s

surface during a sessile drop test ..................................................................................... 89

Figure 7–2- Average contact angle versus time during sessile drop tests for the as sintered

and polished specimens. ................................................................................................... 90

Figure 7–3- BSE micrograph from the cross section of a specimen with partial penetration

of Al (S3-PP) ..................................................................................................................... 91

Figure 7–4- Partially penetrated test (S3-PP) a, b) BSE micrograph and EDX analysis of

TiAl3 particles formed inside Al drop, c,d) BSE micrograph and EPMA maps of Fe

element within Al drop ..................................................................................................... 92

Figure 7–5- Elemental distribution of aluminum and oxygen at the drop-specimen interface

after the partially penetrated test (S3-PP)........................................................................ 93

Figure 7–6- SEM micrograph of the cross section of a S1-P specimen after the sessile drop

test ...................................................................................................................................... 94

xxi

Figure 7–7- Elemental line scans carried out through the thickness of the specimen after a

sessile drop test at 960°C. (Depth=0 is corresponds to the aluminum-specimen

interface at the beginning of the test) .............................................................................. 95

Figure 7–8- Mapping of aluminum and titanium showing the transition between zone 1 and

zone 2 ................................................................................................................................. 96

Figure 7–9- Mapping of iron and titanium revealing the transition between zone 2 and zone

3.......................................................................................................................................... 96

Figure 7–10- Mapping of aluminum and titanium between zone 3 and zone 4 .................... 97

Figure 7–11- BSE micrographs of zone 2 and zone 3. Arrows show the presence of TiAl3

(P1) and Fe-Al compound (P2) ........................................................................................ 98

Figure 7–12- TiAl3 precipitate in zone 2, a) Transmission electron micrograph, b) SAED

pattern from the [010] zone axis of TiAl3 ....................................................................... 98

Figure 7–13- Fe4Al13 phase precipitated in zone 3, a) Transmission electron micrograph, b)

SAED pattern from the [010] zone axis of Fe4Al13 ........................................................ 99

Figure 7–14- Isothermal section of the Al-Fe-Ti phase diagram at 1000°C 88

................... 101

Figure 7–15- SEM micrograph of cross section of specimen subjected to two subsequent

sessile drop tests .............................................................................................................. 102

Figure 8–1- BSE micrographs from the microstructure of as-sintered specimens: a) 10T, b)

7T3F, c) 10F. Arrows in (a) indicate the presence of metallic titanium between TiB2

particles............................................................................................................................ 112

Figure 8–2- BSE micrograph of the cross section of specimens after reaction with

aluminum drop: a) 10T, b) 7T3F, c) 10F. Numbers on images refer to the different

zones formed as a result of aluminum infiltration. ....................................................... 114

xxii

Figure 8–3- BSE micrograph and EDX spectrum of phase A formed in the Al infiltrated

area (zone 2 in Figure 8–2-a) of 10Ti specimen, A: TiAl3 phase, B: Ti rich phase. .. 115

Figure 8–4- a) BSE micrograph from the expanded portion of 10F specimen (zone 1 in

Figure 8–2-c) after reaction with the liquid aluminum drop (A: TiB2, B: Al, C: Fe4Al13

phase, the black areas correspond to mounting resin), b) EDX spectrum of phase C.

.......................................................................................................................................... 117

Figure 8–5- SEM micrograph of the 7T3F specimen after immersion into molten Al for 5

days at 960C revealing the solid TiB2 skeleton. The metallic phases were dissolved in

a NaOH solution. ............................................................................................................ 119

Figure 8–6- Comparing the scratch resistance (Fx) between specimens (AS: as sintered, 5

days: 7T3F after 5 days of Al exposure, Ceradyne: Hot pressed TiB2 provided from

Ceradyne Inc.; The reported Fx is associated with 7T3F specimen after 5 days of Al

exposure.) ........................................................................................................................ 120

Figure 8–7- TEM micrographs from boundary between two TiB2 grains forming a TiB2

bridge: a) selected area for FIB, b-c) Interface of G1 and G2 grains, d) SAED of G1, e)

SAED of G2, f) Inter-particle bridge ............................................................................. 121

Chapter 1:

Introduction

2

1.1 General context

The primary aluminum production in Canada, has been increased from 1.6 Mt at 1990 to

3.0 Mt in 2009 6 which makes it the third largest aluminum producer in the world

7. Almost

90% of primary aluminum (about 2.6 Mt) of Canada is produced in the province of Québec

8. The electrolysis process used to produce aluminum requires large quantities of electric

power and consequently, the aluminum industry is the largest industrial consumer of energy

in this province. Even though Quebec has no bauxite, this province is one of the main

aluminum producers due to its competitive cost of electric power and its proximity to the

United States. Over 95% of the electrical energy used for electrolysis of aluminum in

Quebec is clean and renewable hydroelectric energy 8.

In 2009, Canada’s aluminum industry consumed about 164,700 terajoules (TJ) of electrical

energy. With an electricity consumption of less than 15 kWh/kg Al, the Canadian

production is ranked as one of the most efficient in the world 6. Although the quantity of

energy required to produce primary aluminum has been reduced by 15% in Canada from

1990 to 2009, still more than 35% of its production cost is devoted to electric consumption.

Aluminum industry takes great care in reducing energy consumption, which is so essential

to its survival. Several technological and scientific research and development programs

have been defined to improve the energy efficiency of electrolytic potlines like: wettable

cathode and drained cell, inert anode, cell and anode insulation modification, etc. It has

been proposed that using a wettable cathode instead of the common carbon cathodes could

reduce the electrical energy consumption by more than 10% 9.

The aluminum reduction process consists of electrolytic reduction of alumina dissolved in

molten cryolite salt (Na3AlF6) at 960C. The theoretical minimum electrical energy

requirement for conventional aluminum electrolysis is approximately 6 kWh/kg Al 10

.

Compared to the theoretical value, modern plants are operating at roughly 40% energy

efficiency. The energy consumed in an electrolysis cell is a function of the current

efficiency and operating voltage. The modern Hall-Héroult cells operate at 95% current

3

efficiency. The applied voltage to the electrolysis cell is consumed by different voltage

components of cell circuit. The theoretical minimum voltage required for the reduction

reaction is the cell reaction voltage which is a function of temperature and at 960C is 1.2

V DC 10

.The overall cell voltage is however about 4.6 V DC for new cells which is the sum

of required potential for cell reaction, overvoltage, bath, cathode, anode, connectors and the

voltage drops due to different resistances such as polarization. A considerable amount of

this voltage drop is due to the bath resistance, which accounts for more than 38% of the

total voltage drop and is directly related to the anode-cathode-distance, so called ACD.

The conventional electrolysis cells use cathodes made from carbonaceous materials which

are commercially available in three main types: amorphous based, semi-graphitic and

graphitized. None of these types of cathodes is wetted by liquid aluminum resulting in a

bad electrical contact and inhomogeneous current distribution at carbon cathode/aluminum

interface 10

. In addition, the non-wetted cathode surface increases the risk of cryolite

penetration at the interface which might cause cathode degradation and extra voltage drop

10. The aluminum pad on the top of carbon cathode should therefore be thick enough to

make a better contact and to avoid the risk of cryolite penetration within the interface. The

thickness of the metal pad is typically kept between 20 and 25 cm.

The liquid aluminum pad is a cathodically charged surface for the reduction reaction and

acts as real cathode in the process. Since there is a high current flow through the cell,

typically around 0.9 A/cm2, the resulting electromagnetic fields induce Lorentz forces

causing turbulent motion of the liquid pad. This turbulence deforms the aluminum-cryolite

interface. The joint gap between carbon cathode blocks can produce additional turbulences

in the moving pad. The combinations of these movements form waves in the metal pad

surface, which can approach the anode and result in electrical short-circuit. These short-

circuits are the cause of major loss of power and productivity and affect the stability of the

cell 10

. The turbulences also produce erosion of the carbon lining and decrease the cell life.

To avoid contact between the liquid aluminum pad and the anode, the ACD must be kept

4

large enough which in turn increases the power consumption of the cell. The typical ACD

in conventional cells is about 4.5 cm 11

.

The concept of using “wettable drained cathode cell” is an approach to overcome the

difficulties associated with thick metal pad. In this approach, the bulk of the liquid metal is

drained continuously from the cell while the cathode surface remains wet by a thin metal

layer resulting in a good electrical contact and acts as a penetration barrier between cathode

and cryolite. By using this concept, it would be possible to operate the cell with a thinner

metal pad, which will considerably reduce the risks associated with short-circuits or

inhomogeneous current distributions. Consequently, the reduction of ACD leads to a

significant improvement of energy efficiency and huge energy gain in this industry.

The major requirement for drained cathode cell is to replace conventional carbon cathodes

with new type of materials, which could be properly wetted by liquid aluminum. It should

be noted that beside the wettability for molten aluminum, the candidate materials should

also possess some other properties close or superior to those of usual carbon cathodes.

These properties include: high electrical conductivity, satisfactory mechanical properties,

low solubility and reaction with molten aluminum, good thermal shock resistance,

capability of being fabricated economically into desired shapes, and acceptable resistance

to penetration and corrosion by molten cryolite and specially sodium ions which will be

discussed in details in chapter 2.

Since almost all refractory metals are attacked by molten aluminum or molten cryolite at

the cell operating temperature, the candidate material should be selected among ceramics.

Most of ceramics, however, are either dielectric or have low electrical conductivity.

Borides, carbides and nitrides of the transition metals in the fourth to sixth groups of the

periodic table have interesting properties including good electrical conductivity, which

make them potential candidate materials for replacement of carbon cathodes. These

materials are known as “refractory hard metals (RHM)”.

5

The material that has attracted most attention in terms of chemical inertness with respect to

molten aluminum and of electrical properties is titanium diboride. Several research works

have been performed on the production and characterization of TiB2-based cathodes.

However, the manufacturing process of TiB2-based cathodes is still facing challenges such

as: high fabrication cost related to high processing temperatures, size limitations related to

the special manufacturing processes e.g. hot isostatic pressing (HIP), and low chemical

stability due to the reaction of grain boundaries with molten aluminum.

Many attempts have been made and fully dense TiB2 parts are successfully obtained at

relatively low processing temperatures using metallic sintering aids. Although the

consolidated parts are suitable for some applications such as cutting tools, they are not

proper choices for wettable cathodes as the grain boundaries are attacked by liquid

aluminum resulting in the disintegration of TiB2 grains and degradation in electrolysis cell.

As a result, none of these works led to the commercial implementation of these ceramics as

wettable cathode in aluminum electrolysis 12

.

1.2 Project motivations

The development of a suitable and low cost wettable cathode for aluminum industry, which

could fulfill all the requirements mentioned previously, is therefore a real challenge and

seems insurmountable. However, the potential gain in energy savings of at least 35 TWh, in

Canada alone, is worth the effort and justifies any additional investment.

With the aim of developing wettable cathodes and taking advantage of huge potential of

energy savings in this field, the objectives of this project were defined. The general

objective of the project was to develop TiB2-based ceramic materials, which meet the

property requirements of wettable cathode. The specific objectives were to consolidate the

TiB2-ceramic parts using new sintering aids offering the possibility to consolidate them at

relatively low temperatures while being able to withstand severe smelting environments. To

meet these goals, a series of practical sintering aids were proposed and the effect of their

composition and the processing parameters on the characteristics of consolidated

6

specimens, including physical, mechanical and metallurgical properties as well as their

behaviour and stability in liquid aluminum were investigated.

The technical objective of the project was, on the other hand, to consider the technological

and economical aspects in order to provide applicable, robust and optimized fabricating

process, which has the ability to produce large cathode block parts in reasonable production

costs. The guideline to achieve these goals was to consolidate the specimens via pressure-

less sintering while keeping the sintering temperature as low as possible.

1.3 Structure of thesis

This thesis is presented in eight chapters. The first chapter is the general introduction of the

thesis and presents the idea and problem identification and objectives of the project. The

structure of the thesis is presented at the end of this chapter as well.

The literature review on the basic knowledge and previous efforts on this subject is

presented in chapter 2. This chapter contains a brief review about the primary aluminum

production and Hall-Héroult process for better understanding the issues with common

carbon cathodes, the idea of wettable drained cathode as a solution for decreasing energy

consumption of primary aluminum production. A detailed review on the properties,

material selection and fabrication process of wettable cathodes then follows.

Chapter 3 emphasises the thesis outlines and the originality of the project.

Chapter 4 is assigned to the general materials and methods used in the experimental parts of

the project describing starting materials, processing and consolidation process and materials

characterization and evaluation methods.

This thesis has been prepared as an article insertion thesis. The results obtained during the

doctoral study are published in the form of four scientific conference and journal articles

with review committee. These articles show the evolution of the project toward the

objectives to develop a reliable material as a wettable cathode. These articles are presented

7

in chapters 5-8, which report the experimental approach as well as results of this thesis. In

all articles I have acted as the principle researcher and the first author.

The first article is reported in chapter 5. In this article the effect of composition and

processing parameters such as preparation of sintering additives, milling conditions and

sintering temperature on the final microstructure and physical properties of TiB2-based

parts consolidated pressureless sintering method was studied. It was shown that pre-

alloying of additives and milling time had significant effect on the final properties of

specimens.

In the second article, chapter 6, the effect of milling time on the particle size distribution of

the starting powder mixture as well as the properties of sintered specimens and especially

their behaviour in molten aluminum was investigated. In this work, the microstructure,

physical properties and wettability of specimens with different milling time were

compared. The interaction with liquid Al when exposed in molten Al up to 24 h was also

studied. It was concluded that the specimen milled for 30 min had the most promising

properties for further investigations.

Chapter 7 is dedicated to the third article. The main purpose of this article was to study the

interaction of TiB2-based specimen with molten aluminum drop. In this paper, first the

interaction of the surface oxide layer with liquid aluminum was studied. Then the

mechanism of the penetration of aluminum drop inside the specimen and its interaction

with the metallic phases in the inter-particle spaces was discussed. It was found that despite

the penetration of aluminum, there were no signs of swelling and expansion of the

specimen.

The fourth article is presented in chapter 8 of this thesis. In this paper, the effect of different

sintering additives including Fe, Ti and Ti-Fe on the stability of consolidated parts in

molten aluminum was investigated. It was shown that only the specimen with Ti-Fe

additive had good stability in liquid aluminum after prolonged immersion time. TEM

analysis of the exposed specimen revealed that the formation of TiB2 phase in the inter-

8

particle bridges is responsible for the stability of specimens in molten aluminum. A

complete discussion was made to explain the way Ti-Fe additive promotes the formation of

TiB2 phase as inter-particle bridges during the sintering process. The results reported in this

article confirm that the developed material could be a reliable candidate, considering the

scientific, technical and economic aspects, to be used as wettable cathode.

In chapter 9, concluding remarks of the project are presented. Some perspectives for further

progress of this work are suggested in chapter 10.

Chapter 2:

Literature review

10

2.1 Aluminum

Aluminum is an essential material for modern manufacturing. It is a lightweight, structural,

and excellent atmospheric corrosion resistant metal with high electrical and thermal

conductivities, and easy to recycle 10

.

2.1.1 Aluminum production industry

Aluminum is the most abundant metal in the Earth's crust, and the third most abundant

element therein, behind oxygen and silicon. However, it does not exist in nature in metallic

form due to its high reactivity. It is typically found as one of the several aluminum oxides

or silicates mixed with other minerals 10

. The most common aluminum ore from which

aluminum is produced is bauxite. It consists largely of the minerals such as gibbsite

Al(OH)3, boehmite γ-AlO(OH) and diaspore α-AlO(OH) 13

. Aluminum production is the

largest consumer of energy on a per-weight basis and is the largest electric energy

consumer of all industries 10

. Most new (primary) aluminum is produced by a sequence of

two processes: a) Bayer process, which extracts aluminum oxide (Al2O3) powder by

refining bauxite and b) Hall-Héroult process during which the aluminum oxide is reduced

by electrolysis 13

.

2.1.2 Hall-Héroult process

The Hall-Héroult process (developed in 1886) is an electrolytic process, which has

undergone modifications and improvements over a century. A modern Hall-Héroult

reduction cell (pot) (Figure 2–1) is a rectangular steel shell of 9 to 12 m long, 3 to 4 m wide

and 1 to 1.5 m deep 10

. It has an inner lining of carbon surrounded by refractory thermal

insulation, which keeps it isolated thermally, and electrically from the steel shell. The

capacity of the commercial cells ranges from 450 to 4,000 kg of aluminum per day

requiring electrical current of 60,000 A to more than 500,000 A, respectively 10

.

11

Figure 2–1- Schematic of a typical Hall-Héroult cell for electrolytic smelting of

aluminum 13

A typical electrolytic cell consists of prebaked anodes (positive electrode) and a cathode

(negative electrode) both made of carbon, and an electrolyte. Cryolite (Na3AlF6), which has

limited solubility of aluminum oxide, was found to be the best molten salt as an electrolyte.

Usually, the cell operates with 1–6.0% aluminum oxide in the electrolyte 14

. Additions of

fluorides of Ca, Al, Li, and Mg reduce the melting temperature of pure cryolite from

1012C to 950–980C and increase the current efficiency 13, 14

. Once the refined alumina is

dissolved in the electrolyte, its ions are free to move around. The reduction reaction is

continuous and alumina must be supplied to the bath at a controlled rate to maintain

constant conditions. This is accomplished with automatic feeders that break the frozen

surface crust and deposit alumina into the molten bath where it is dissolved and distributed

by convection currents. Alumina is also used to cover the carbon anodes and the frozen

bath surface and it serves as thermal insulator and a protective cover to reduce air burning

of the anode 10

.

Direct current enters the cells through the anodes, passes through the electrolyte carried

primarily by sodium ions, passes through the molten aluminum and exits the cell through

the cathode and steel current collector bars. The positively charged aluminum ions migrate

12

to the electrically negative cathode and pick up electrons to yield aluminum metal. The

reaction at the cathode is:

Cathodic Reaction: Al3+

(in electrolyte) + 3 e− → Al (l) ( 2-1)

Molten aluminum layer with 3-25 cm thick is accumulated at the bottom of the cell and is

siphoned at intervals into an external crucible. The purity of produced aluminum is about

99.5–99.8% and the major impurities are silicon and iron. It is then transferred to other

units of plant for further alloying preparations prior to casting into ingots or forming into

other shapes 13, 14

.

The oxygen ions migrate to the anodes, react with carbon and evacuated from the system in

the form of carbon dioxide.

Anodic reaction: O2−

(in electrolyte) + C (s) → CO (g) + 2 e− ( 2-2)

The gas molecules accumulate into large bubbles that are collected and move across the

anode surface to escape around the edge of anode. The buoyancy of gas creates movement

which contributes to the motion of the bath and a molten metal layer 10

.

Overall reaction: Al2O3 (in electrolyte) + 3C (s) → 2Al (l) + 3CO (g) ( 2-3)

Approximately 0.45 kg of carbon anode is consumed for each kilogram of aluminum

produced. The carbon anodes provide a large part of the energy required for aluminum

reduction (about 45% of energy requirement of reaction) 10

. Two types of anodes are in use:

pre-baked carbon blocks and self-baked anodes (known as Soderberg type) 13

. The

consumable carbon anodes are periodically replaced about every four weeks in modern

plants 10

. The pot cover, which is part of the gas collection system, is removed, the used

anode is pulled out from the frozen surface crust, and the new anode is inserted. This has to

be accomplished without significant pot crust breakage or alumina falling into the bath.

Anode changing results in thermal, current, and magnetic disturbance in cell operation 10

.

13

Since the carbon anodes are consumed in the reduction cell and must therefore be replaced

regularly they are mounted on the busbars so that the distance between the anode and the

metal pad can be adjusted and kept constant during the whole process.

The carbonaceous cell linings are composed of the cathode at the bottom and the sidewalls.

The carbon cathode carries the current from molten aluminum layer to the steel collector

bars 13

. The major constituents of carbon cathode blocks are Anthracite and to some extent

graphite, and metallurgical coke 14

. Some cathode blocks contain a higher content of

graphitic carbon in attempt to reduce the resistance. However, less resistive graphite is also

less wear resistant and this compromises the life of the cathode 10

. The commonly used

types of cathodes are: amorphous carbon, semi-graphite and fully graphitized 13

. The

cathode blocks are bonded in the cell by a pitch-carbon paste rammed between the joint 13

.

Since cathode replacement requires the complete dismantling of a cell, the cathode life

generally determines the cell life which maybe last for up to 10 years 14

.

Molten cryolite has low viscosity and interfacial tension that allows it to easily penetrate

any porosity in the carbon lining 14

. The carbon lining could absorb fused electrolyte up to

its own weight 14

. To protect the lining, the thermal insulation is adjusted to provide

sufficient heat loss to freeze a protective coating of the electrolyte, known as “ledge,” on

the inner walls. The cell is never tapped completely dry of molten aluminum in order to

prevent the direct contact between electrolyte and cathode. The carbon cathode must

remain bare for good electrical contact with the aluminum pad and for this reason it is

essential that no alumina or frozen ledge form under the metal pad 10

. Unlike the anodes,

the cathodes are not oxidized because they are protected by the liquid aluminum inside the

cells. Nevertheless, cathodes do corrode and erode, mainly due to electrochemical

processes and metal movement 15

. One of the main causes suggested for the cathode failure

is the accumulated expansion and damage caused by sodium penetration. The sodium

attack causes the swelling and cracking of carbon lining 15

. Cell failure occurs either when

the bath penetrates through the cathode material or the sidewall carbon. In either case, the

cell can no longer contain the electrolyte or the molten metal and must be shut down and

14

rebuilt. Normally the electrical resistance of the cathode increases with time and the lining

is rebuilt after 1200 to 3000 days operation 16

.

There are three main factors that can compromise cell life 12

. The first factor is starting up

and shutting down the cell multiple times. When starting up a cell, significant stresses

occur in the cathode lining because of the significant changes in temperature, from room

temperature to more than 400°C. These stresses can cause cracking that lead to cathode

failure. Cells are rarely completely shut down, in part because of this danger.

The second compromising factor is sludge accumulation under the metal pad. Sludge is a

mixture of undissolved bath and alumina that sinks to the bottom of the cell. Sludge forms

when the bath temperature is not sufficiently above the liquidus temperature or when the

alumina is fed too rapidly. The sludge promotes the erosion of the cathode, which can lead

to cathode failure. Thus, it is important to feed the cell at an appropriate rate and to

maintain the cell within an acceptable temperature range.

The third factor that can shorten the cell life is the melting of the sidewall ledge, exposing

the sidewall carbon to the corrosive molten electrolyte. If this protective layer melts, the

electrolyte will begin to dissolve the sidewall carbon, leading to sidewall failure. Thus, it is

imperative to operate the cell within the proper temperature range to maintain the sidewall

freeze.

Several plant designs have been proposed over the years to minimize electrical energy loss.

Modern plants convert alternating current with a silicon rectifier into 600–900 V DC 13

.

Each electrolytic cell operates at 4.6 V DC, so roughly 150 to 180 cells are connected

electrically in series of long rows called “potlines.” They are placed as close as possible to

each other while maintaining sufficient room for anode changing, alumina feeding, and

reasonably low electromagnetic interference. Current density is calculated by dividing the

amperage supplied to an anode by its geometric face area. It is generally expressed in

amperes per square centimetre (A/cm2) and is considered as an indicator for the

productivity of a cell. Most potlines operate in the range of 0.8 to 1.0 A/cm2 10

. Depending

15

on the length of the potline, the total current flow can be between 50 and 360 kA 13

. Thus,

the electrodes must be good electrical conductors capable of carrying high currents. The

quantity of aluminum produced per cell increases with increasing current density. On the

other hand when the current density increases, its efficiency decreases, resulting in higher

energy consumption per unit of metal produced. Lower current densities are more energy

efficient, but increase capital and labour costs per unit of output 10, 13

.

The energy consumed in an electrolytic reaction is a function of the voltage used and the

current efficiency of the operating cell. The approximate voltage components of a

conventional cell are shown in Figure 2–2. The electric current flows through the cell and

the cell voltage components can be described as a set of resistors in series. The total

resistance of a series circuit is equal to the sum of the resistances of the individual

components in the circuit. The current is the same everywhere in the series circuit.

According to the Ohm's law, potential difference (V) is equal to current (A) × resistance

(Ω). So the sum of the potential drops equals to the sum of the voltage on each component.

E total = E cell reaction + E overvoltage + E bath + E cathode + E anode + E connectors ( 2-4)

The cell reaction voltage is a function of temperature, which is 1.2 V DC at 960oC.

16

Figure 2–2- Voltage distribution in a Hall- Héroult cell 10

The electrical energy consumed in a primary aluminum cell (excluding energy generation

and transmission losses) varies from less than 13 kWh/kg Al, for the modern plants, up to

more than 20 kWh/kg (for older Söderberg facilities) 10

. The theoretical minimum energy

requirement for carbon anode aluminum electrolysis is approximately 6 kWh/kg Al 10

. Heat

dissipated during aluminum smelting is about 8 kWh/kg of Al. The electrodes are

obviously key components in aluminum smelting and play a major role in efficiency as

well as pollutant emissions. Since 10% of the total cell voltage drop is in the carbon

cathode blocks, it is important that they have good electrical conductivity 10

.

The anode-cathode distance (ACD) is the distance between top surface of the aluminum

pad and the lower face of the anode. The space between anode and metal pad is occupied

by electrolyte bath. Decreasing the ACD reduces the cell voltage and the energy

consumption of the bath. The operating ACD should kept short to lowering the bath

17

resistance while at the same time large enough to let the rich alumina electrolyte to reach

the charged surfaces, let the reactant gas bubbles to escape and prohibit the contact of

disturbed liquid metal pad with anode surface 16

.

A part of heat that is required to keep the cell at operating temperature is supplied by the

electrical resistivity of bath as current passes through it. The amount of heat developed

depends on ACD and changing the ACD is one method of controlling the operation

temperature. The ACD is typically in the range of 4 to 5 cm 10

.

The liquid aluminum pad that forms at the bottom of the cell is the cathodically-charged

surface for the reduction reaction. When the large amperage passes through the cell, it

creates a huge electromagnetic field. The electromagnetic forces cause local liquid

aluminum stirring. Cells are designed to minimize this melt stirring. Nowadays melt

velocities of about 5 cm/s are observed, that is three times less than what was common

about fifty years ago 10

. Movement of the aluminum pad is also caused in smaller extent by

the interfacial drag of the bath fluid. Joint discontinuity of cathode carbon blocks also

creates additional disturbance flow in metal pad. The combination of all these forces

produces waves at the surface of the pad. These waves can reach the anode surface and

result in electrical short circuit at the cell. The current that flows during this shorting

produces no aluminum and results in major loss of energy and productivity. The motion of

the aluminum pad also makes erosion on carbon lining and reduces the cell life. The ACD

is constantly changing as a result of waving metal pad. Designing systems to minimize

metal pad movement is a key factor to reduce the ACD and accordingly increasing in cell

efficiency 10

. A better concept is to drain the bulk of the liquid metal to a sump and the

cathode is left wetted only by a thin layer of metal. Essentially the “drained cathode cell” is

a concept to approach to get rid of the difficulties associated with keeping the metal pad

stable.

18

2.2 Issues with carbon cathode in Hall-Héroult cell

2.2.1 Non-wettable for liquid aluminum

The carbon cathodes are not wettable by molten aluminum pad. Therefore, there is always a

gap between the metal pad and the carbon lining 10

. This gap creates an electrical junction

causes a small voltage drop and if the gap becomes thinner, the junction voltage decreases.

Therefore, the higher thickness of metal pad keeps at the bottom of the cell which applies a

greater weight of liquid metal on the cathode surface and lowers the junction voltage drop

10. As a result the metal pad must have a minimum thickness in order for the cell to operate

smoothly 12

. In the other hand, the electromagnetic forces create movement and standing

waves in the aluminum pad and to prevent the shorting between anode and the molten

metal, the ACD must be kept at a safe range of 4 to 5 cm. For traversing the ACD by the

current, the voltage drop in the range of 1.3 to 2.0 V is occurring, compared to the

theoretical voltage of 1.2 V required for the electrochemical reduction of alumina (using

carbon anodes) 10

.

Replacing the carbon cathode with new materials to be wetted by liquid aluminum is

proposed as a solution to reduce the energy loss in the cell. It decreases the junction voltage

drop and would allow decreasing the thickness of metal pad. A thinner pad would be more

hydrodynamically stable and would not be affected by electromagnetic forces. Therefore,

there will be no waves, swirls and electrical shorting. The ACD could considerably be

reduced resulting lower cell voltage. This, combined with a lower cathode voltage drop will

keeping down the energy consumption and will improve the power efficiency 10, 13

.

2.2.2 Penetration of electrolyte and liquid aluminum

The NaF/AlF3 ratio, which is called cryolite ratio (CR), is 3 in pure cryolite. Aluminum

fluoride is usually added to the electrolyte in excess of the Na3AlF6 composition. Although

lowering the cryolite ratio can increase the current efficiency and metal production, it

increases the volatility of the electrolyte 14

. It also adversely affects the electrical

19

conductivity of the bath and the solubility of alumina. The characteristic CR value used in

the aluminum industry lies between 2 and 3 14

.

During cell start-up and operation, electrolytically produced sodium enters the carbon

matrix porosity. Sodium is the by-product of the electrolysis process from the primary

deposition of aluminum melts. It is absorbed by the carbon cathode according to the

following reaction 17

.

Al (l) + 3NaF (in electrolyte) = 3Na (in C) + AlF3 (in electrolyte) ( 2-5)

Since metallic sodium is present in the carbon cathode, the reaction proceeds to the right as

long as the carbon matrix is not saturated with sodium and the sodium activity rises to unity

17. After penetration of sodium, the lining wetting conditions by the electrolyte changes,

thereby allowing the electrolyte to penetrate the carbon cathode. The penetration of both

sodium and electrolyte into the carbon materials causes various cathode defects, e.g.

swelling and crack propagation, cathode heaving and disruption 18

.

Another complication of carbon cathode in cell is the reaction between aluminum and

carbon and formation of aluminum carbide at the surface of carbon cathode:

4 Al (1) + 3 C (s) = A14C3 (s) ( 2-6)

According to Worrell 19

, the Gibbs energy for the reaction is -147 kJ/mole at 970°C, so that

this reaction is thermodynamically favoured at the temperatures of electrolytic aluminum

production. The reaction proceeds and a solid layer of Al4C3 is formed, whereupon the

diffusion-controlled reaction virtually stops. According to solubility data of A14C3 in the

electrolyte and liquid aluminum, solid Al4C3 might be dissolved and be reoxidized by the

anode gas:

Al4C3 (s) + 9CO2 (g) = 2Al2O3 (S) + 12CO (g) ( 2-7)

This reaction leads to under-saturation and dissolution of carbide films formed. Hence, a

steady consumption of carbon cathode may result 20

. In case of an uncontrollable

20

consumption of carbon cathode, the molten aluminum may reach the steel bars resulting in

excessive iron content in the product. In more serious cases, it may tap-out the cell.

It is because of the abovementioned problems related to the carbon cathodes that the

aluminum industry is seeking alternative materials for cathode. A cathode that is wetted by

aluminum would allow a dramatic reduction of ACD, which would decrease losses from

electrical resistance. A consequent increase in electrical efficiency would be a major

breakthrough.

2.3 Wettable drained cathode

Researches in the field of wettable drained cathode cells started in the 1950’s under the

leadership of Charles E. Ransley 21

who obtained patents with the British Aluminum Co. At

that time, he started experimental works on various carbides and borides of transition

metals and cell designs. Since then, lots of research efforts have been undertaken having

been focused on two main areas: a) the development of practical wettable cathode material

b) the design of a reduction cell to utilise such a material. Extensive review about inert

cathodes for aluminum electrolysis was published by Billehaug and Oye in 1980 22

. After

that, only the scarce information on trials on an industrial scale was available in the open

literature and most of the information was taken from the patent literatures 12

. Zhang et al.

15 have reviewed the development of wettable cathodes up to the beginning of 1990s. For

replacing common carbon cathodes, the new material must meet some properties to be

selected as a proper wettable cathode.

2.3.1 General property requirements for an ideal cathode material

It was thought that the non-carbon cathode is capable of providing a longer service life and

a significant power saving by narrowing the electrode gap and decreasing the cathode

voltage 23

. Based on the working conditions of the cathode in Hall-Héroult cell and the

existing problems with carbon materials, following properties are desirable for an ideal

cathode material 12, 15, 22, 24

:

21

1. High electrical conductivity (electrical resistivity < 500 μΩcm)

2. Low solubility and reaction with molten aluminum at 960°C

3. Wettability to molten aluminum

4. Good resistance to penetration and to corrosion by sodium and molten electrolyte

5. Adequate mechanical strength and resistance to cracking due to thermal or chemical

forces

6. Capability of being fabricated economically into desired shapes

7. Reliable electrical contact with current collectors

8. Resistance to abrasion/erosion from the metal and electrolyte

9. Good resistance to oxidation and to corrosion by any reactive gases to which it may

be exposed, particularly at elevated temperature at the exterior of the cell.

It is clear that no conventional carbon and carbon composite can comply with all these

requirements to be an ideal cathode. It is also obvious that few materials, if any, could meet

these requirements. The selected material should have these property requirements equal or

better than carbon cathodes.

2.3.2 Choice of material

High electrical conductivity and low solubility and reaction with molten aluminum at

operating conditions are two essential properties required for a material to be selected as a

wettable cathode. Metals generally have a high electrical conductivity because of their

metallic bond and the “sea of electrons” and some of them have high melting point to be

operated at cell temperature (about 960°C) but none of them could be a candidate to use as

wettable cathode due to the lack of durability in molten aluminum. Most of metallic

elements readily dissolve in liquid aluminum at cell temperatures 25

.

Ceramics are a unique class of materials that are distinguished from common metals and

plastics by their a) high hardness, stiffness and good wear properties; b) ability to withstand

high temperatures; c) chemical durability; and d) electrical properties 26

. They could be

electrical insulators, semiconductors, or ionic conductors. Most of ceramics are

22

predominantly electrical insulators. Some ceramics, like SiC, B4C and TiN, are semi-

conductor and can be used as heating elements or electrodes. Only a few groups of them

have good electrical conductivity in room temperature. The electrical resistivity values of

several ceramics are listed in Table 2–1 27

.

Table 2–1- Electrical resistivity of some ceramics (in Ω cm) 27

Material Electrical resistivity

(Ωcm-1

) at 25oC

Material Electrical resistivity

(Ωcm-1

) at 25oC

TiB2 10-5

MoSi2 2×10-5

ZrB2 10-5

Graphite 10-3

TiC 7×10-5

Si3N4 107-10

12

WC 2×10-5

ZrO2 109-10

11

B4C 0.1-100 Al2TiO5 >1011

SiC 0.1-10 Diamond 1012

TiN 0.5 Al2O3 1014

AlN 1013

-1015

BeO 1014

BN 1011

-1013

MgO 1014

According to the values in Table 2–1, the ceramics with high electrical conductivity are:

TiC, WC, TiB2, ZrB2 and MoSi2. They all belong to a group of ceramics so called

Refractory Hard Metals (RHM).

The RHM are in general defined as borides, carbides, silicides and nitrides of the transition

metals in the fourth to sixth group in the periodic system 28

. Reviews of research works on

wettable cathodes 12, 15, 22

show that most of efforts have been also focused on the use of

RHM, either as a single compound or a mixture of compounds. Some physical properties of

selected RHMs are listed in Table 2–2.

23

Table 2–2- Physical properties of some RHMs 22

Compound* Melting

temperature oC

Density

g/cm3

Electrical

resistivity

25°C

μΩcm

Electrical

resistivity

1000°C

μΩcm

Thermal

conductivity

Wm-1K-1

Thermal

expansion

Coeff.

K-1 108

Elastic

modulus

25°C

GPa

TiB2 2850 – 2980 4.52 9 – 15 60 24 – 59 4.6 253 – 550

ZrB2 3000 – 3040 6.09 – 6.17 7 – 16.6 74 24 5.9 343 – 491

MoB2 2100 – 2140 7.8 – 8.1 20 – 40

B4C 2450 2.52 0.2× 104–7× 10

4 High 29 4.5 448

TiC 3067 – 3250 4.92 – 4.95 51 – 250 119 17 – 21 5.5 – 7. 74 269 - 462

WC 2600 – 2870 15.7 – 15.8 17 – 22 106 – 118 29 – 121 4.5 669 – 710

Si3N4 1870 – 1885 2.37 – 3.19 1019

High 2.5

AlN 2400 decomp. 3.25 High High 30 5.6 345

BN 3000 decomp. 2.25 – 2.27 1.7 × 1018

High 15 7.5

TiN 2950 5.39 – 5.44 21.7 – 53.9 130 17 9.35

Graphite** 3500 subl. 570 – 1170 7.8 6.4 – 13.7 * Commercial polycrystalline grade

** Not RHM, but included for comparison

Titanium diboride (TiB2) has been always the most interesting RHM for the efforts on

development of wettable cathodes fot both economic and material property reasons 12, 15, 22

.

It is well established that TiB2 and ZrB2 have excellent wettability with molten aluminum

29. Ta and Nb borides also have interesting properties but compared to TiB2 they are too

expensive to be use for this application at commercial scale 22

. ZrB2 appeared to be at least

the technical equivalent of TiB2 but was rejected by industry because it is more expensive

30. Other RHMs, such as nitrides (AIN, BN and Si3N4), carbides (TiC, ZrC) are either poor

electrical conductors or have greater solubility in liquid aluminum. The presence of weak

reactions of TiC, ZrC and especially TiC 31

with molten aluminum has been detected 29

.

Many carbides and specially TiC have high melting points and are cheaper to produce but

they are very sensitive to structural defects such as vacancies both on the metal and non-

metal sites 15

.

The oxidation rates of the transition metal borides are low at lower temperature where a

protective layer of B2O3 glass appears on the surface. At temperatures above 1100°C or in

presence of water vapour the oxidation rate increases rapidly 22

.

24

2.4 Titanium diboride

2.4.1 Phase diagram, crystal structure and properties of TiB2

The Ti–B binary equilibrium phase diagram is presented in Figure 2–3 32

. As can be

observed, three intermetallic phases including orthorhombic TiB, orthorhombic Ti3B4 and

hexagonal TiB2 has been confirmed to exist. While Ti3B4 has been shown to be in a line

compound, both TiB and TiB2 exhibit a small composition variation. TiB and Ti3B4

decompose peritectically at 2180°C and 2200°C, respectively, TiB2 melts congruently at

3225°C (Figure 2–3). TiB2 exists over a stoichiometry range of 65.5 – 67.0 at.% B at

ambient temperature 32-34

.

TiB2 crystallizes with a hexagonal close packed structure (HCP) with P6/mmm space group

(a = b = 0.3028 nm, c = 0.3228 nm; α = β = 90, γ = 120) 28

. These dimensional parameters

lead to a density of 4.52 g/cm3, which is identical to the pycnometrically determined values

28. The hexagonal unit cell of TiB2 single crystal is shown in Figure 2–4

34. It is a simple

hexagonal lattice in which HCP Ti layers alternate with graphite-like B layers. By choosing

appropriate primitive lattice vectors, the atoms are positioned at Ti(0,0,0), B(1/3,2/3,1/2)

and B(2/3,1/3,1/2) in the unit cell 35

. Each boron atom has three boron neighbours in a

trigonal planar arrangement, forming a strong covalently bonded hexagonal network

structure. The high hardness of TiB2 and its chemical resistance are attributed to its inherent

crystal structure and atomic bonding 32-34, 36

.

However the relatively strong covalent bonding of the constituents results in low self-

diffusion rates, it gives also a high melting point and stable chemical composition 33

. A

higher mass fraction of TiB2 in composites yields a higher value of elastic modulus E.

When the mass fraction of TiB2 in composites exceeds 90 %, the value of the elastic

modulus appears to converge to 565 GPa at 23oC as the density increases towards 4.5 g/cm

3

33.

25

Figure 2–3- Binary Ti-B phase diagram 32

Figure 2–4- The structure of TiB2 in a projection along the hexagonal axis (right) and a

perspective view (left) 34

26

2.4.2 Synthesis of TiB2

Borides are generally produced by reaction of metals with boron or suitable boron

compounds 37

. The following processes are of practical interest 32, 37

:

1. Synthesis by fusion of metal (or metal hydrides) and boron

2. Synthesis by reactive sintering of metal (or metal hydrides) and boron at

temperatures below the melting point

3. Reducing the mixtures of metal oxide with B2O3, e.g., by aluminum, magnesium,

silicon, or carbon

4. Reaction of metal, metal oxide, or metal hydrides with boron carbide, with or

without the addition of B2O3

5. Electrolysis of fused-salt baths containing metal oxide and boron oxide

6. Deposition of boride layers from vapour phase, e.g. hydrogen reduction of boride

halides in the presence of metal or its halides

Among various synthesis methods, electrochemical synthesis and solid-state reactions have

been developed to produce finer TiB2 powder in large quantity. Borothermic reaction is an

example of solid-state reaction, which can be illustrated by the following reaction:

2TiO2 + B4C + 3C –> 2TiB2 + 4CO ( 2-8)

Typical oxygen and carbon content in as-synthesized TiB2 is less than 0.5 to 0.6 wt%

respectively. Synthesised TiB2 powder is observed to have finer sizes with particle size

distribution with a D50 value around 1.1 mm 32

. It is worthwhile mentioning here that it is

now possible to produce large quantity (kilogram scale) phase pure TiB2 powders on both

laboratory and commercial scale using borothermic reduction process.

A potential way to produce submicron-sized TiB2 powder is mechanical alloying of a

mixture of elemental Ti and B powders. The elemental powders were found to react to form

stable TiB2. It was noted that the size of the transition metal and the heat of formation of

borides greatly affected the mechanical alloying time, while producing finer size TiB2 32

.

27

Self-Propagating High-Temperature Synthesis (SHS) is an advanced technology for the

synthesis of a wide variety of inorganic compounds, either as powders or near net-shape

products. In this method, synthesis progresses by exploiting the heat-energy released by the

exothermic reaction of raw materials via a self-sustaining combustion wave, which

propagates, from one end of the specimen to the other. Ultrafine (nanometric) TiB2 powder

could also be produced through a SHS process involving addition of varying amounts of

NaCl. As the amount of NaCl (diluents) increases, the particle size of TiB2 was found to

decrease, reaching 26 nm in case of 20 wt% NaCl addition. The ignition temperature for the

stoichiometric mixture of TiO2, H3BO3 and Mg was found to be as low as 685°C 32

.

2.5 TiB2-based cathode technologies

As mentioned, titanium diboride and composites containing a major fraction of TiB2 have

appeared to be the best candidate for wettable cathodes. The use of pure TiB2 as a wettable

cathode has suffered from several problems 12, 15, 22, 30

:

Low sinterability because of its low diffusion coefficient

Expensive fabrication costs

Brittle nature and poor thermal shock resistance

Being subjected to inter-granular corrosion by molten aluminum

Difficulty of retrofit into the existing cells

Due to the production difficulties and high fabrication costs the amount of RHM used has

considered to be kept at minimum in some efforts. Some researchers tried to cover the

surface of carbonaceous cathodes with coatings, tiles or pieces sticking out from the surface

12. Acceptable cost, ease of application and strong and stable adhesion to the substrate are

required to make these cathodes interesting to the industry.

Raynolds Company tested hot-pressed and sintered TiB2 bars in a pilot scale cell on 1962

30. They claimed to have produced flawless TiB2 parts with good quality and high density.

However, even these parts had the major problem of crack propagation, associated with

28

inter-granular corrosion. Impurities in TiB2 powder such as carbon and oxygen precipitate

at grain boundaries resulting in subsequent inter-granular penetration of sodium and

aluminum which initiated slow-developing cracks. According to McMinn 30

, the task to

keep the tiles or plates in place on the substrate proved to be really difficult and efforts to

solve this problem were never entirely successful and breakage of TiB2 parts frequently

occurred during the operating in the electrolysis cell.

Raynolds also developed TiB2-AlN-Al cermets, as well as graded cermets with the metal

phase enriched away from the pure TiB2 surface 30

. Although these composites improved

the abovementioned problem caused by impurities in TiB2, they increased the electrical

resistivity of the components and did not completely eliminate the slow crack and crack

propagation problem.

In 1980-1990s, more emphasis was put upon technical solutions that apply TiB2-carbon

coatings or fabricated TiB2-graphite (TiB2-G) composites 12

. Provided that a high enough

TiB2 content is present in the mixture, it was expected that such materials have acceptable

electrical conductivity, partly inert towards electrolyte and molten aluminum and proper

wetting by the aluminum. However, these materials have problems majorly due to the

formation of aluminum carbide in service, leads to degradation and failure of the cathode

parts 12

.

TiB2-carbon coatings with high porosity and permeability have a problem following metal

and salt penetration, as aluminum carbide build-up at the coating-carbon interface causing

the degradation of electrical contact 12

.

In 1984, Martin Marietta Aluminum has reported 38

the application of a 1 cm layer of TiB2-

carbon paste on carbon cathode substrates. It was followed by curing and carbonization to

temperatures of 600-1000°C. The author reported that throughout the test period muck did

not adhere to the coated cathode surface and was more easily dispersed. During normal

operation of the test cells the current efficiency increased approximately 2%. The

29

technology then was acquired by Comalco and the development work of this coating has

been continued in Australia 30

.

TiB2-G composites developed by Great Lakes and Reynolds have been tested in pilot cells

since 1985 12

. They selected these composite materials over hot-pressed TiB2 parts to

reduce materials and processing costs and to increase thermal shock resistance. Exposure of

TiB2-G components in liquid aluminum indicated a gradual dissolution/erosion caused by

penetration of molten aluminum into the bulk porosity of the fabricated shapes but was

limited to the thin outer skin. Aluminum penetration only progressed deeper into the body

when loose material was removed from the outer edges 12

.

Pilot reduction cell tests were also conducted with the TiB2-G composite (90 wt% TiB2-10

wt% graphite) fabricated into mushroom-shaped cathode elements 30

. The elements were

stood out, wetted and drained above the metal pool. However, they had the same problem

as for the TiB2-G composite parts.

It is possible to apply a relatively thin layer of TiB2 coating with the thickness of about 1

mm on the carbon cathode substrate by various techniques including electro-deposition 39

,

plasma spraying 40

, glazing with a powder layer, or chemical vapour deposition. Most

coating techniques, however, are generally not suitable for utilization on an industrial scale.

In addition, it is essential that the coating prevent sodium and carbon interaction and

sodium related buckling damages, while a thin coating is susceptible to crack and

delaminates from the carbon during cell operation 41

.

Some efforts have been focused on the development of relatively inexpensive wettable

cathode coatings which contain large amount (20 to 50 wt%) of a carbonaceous pitch or

resin binder 42-44

. However such coatings also deform and swell because of the absorption

of sodium 41

.

The degradation of TiB2 materials with the TiC secondary phase is caused by the

penetration of Al and its reactions with TiC. The reaction forms Al4C3 phase with increased

molar volume, which builds up internal stresses and possibly induces crack formation and

30

by the way increases the rate of degradation. Such effect has been reported for the reaction

4Al + 3TiC = Al4C3 + 3Ti 45

.

In an effort to avoid the presence of carbon phase in the composition, which results in

formation of aluminum carbide, a TiB2-based non-carbonaceous coating for cathode blocks

has been developed and tested on number of cells. This coating contains a small amount

(less than 10 wt%) of a non-carbonaceous inorganic binder which is mostly colloidal

alumina 41

. The coating can be applied by painting one to several layers on the cathode

surface of the lined cell. An aluminum sheet is placed on the top of the coating before

preheating and allowing it to melt, prior to adding the cryolite and starting the cell

operations. Although the wettability and beneficial physical properties of the coating have

been claimed, there is still the sodium-trapping and the formation of sodium aluminate

during the initial stages of cell operation. Sodium penetrates the pores and partially reacts

with the alumina contained in the coating. However, the absorption is considerably less

than what the pitch in the carbon cathode can absorb 41

.

Having investigated several components and methods, which have been proposed as

wettable cathodes, it was concluded that the proposed material should mainly consist of

TiB2 (TiB2-based). In addition, the components used as sintering aids should promote the

formation of electrically conductive phases with low reactivity and solubility in molten

aluminum at grain boundaries.

2.6 TiB2-based cathodes

Sintering of TiB2 as a polycrystalline bulk material for industrial applications is a challenge

due to its high melting point (~3000°C) and low diffusion coefficient. For this reason, very

high temperatures are required for the activation of the transport mechanisms in TiB2-based

materials. Along with the lower porosity obtained from increasing the sintering

temperatures, an exaggerated grain growth also takes place, degrading the mechanical

properties. The presence of titanium oxides, on the surface of particles enhances the

material transfer at temperatures above 1700°C resulting in a dramatic grain growth.

31

Controlling grain growth is important since microcracking is known to occur above a

critical grain size 45

. Therefore, the consolidation of TiB2-based materials using commercial

powders is limited to temperatures below 1700°C 46

. A very fine powder, sintering aids or

hot pressing are necessary in order to obtain high density and at the same time control grain

growth during sintering.

Different techniques have been used for sintering pure titanium diboride such as hot-

pressing 47-49

, hot isostatic pressing 50

, pressureless sintering 47, 51, 52

, reactive electric

discharge sintering 53

and spark plasma sintering 54

.

Most studies used hot pressing to achieve a significant improvement in densification 47, 55,

56. Some others used pressureless sintering, but the sintering temperature was higher than

1800°C 47, 51, 52

. Hot pressing is an effective densification process and has been a major

fabrication route for dense TiB2 bodies. However, it is relatively expensive and not proper

for the production of large and complex shapes 47

. The cost of fabrication, size limitations,

machining difficulties and some mechanical problems have made hot-pressed TiB2 solution

less attractive in recent years 12

.

Pressureless sintering is a less expensive route for fabricating of near net-shaped large

parts, but high-temperature sintering of 2000°C causes exaggerated grain growth that

results in a decrease of the mechanical properties and thermal shock resistance. Therefore

the use of low-melting-point sintering additives was proposed which allows the

pressureless sintering of TiB2-based materials at relatively lower temperatures through

liquid-phase sintering.

2.6.1 Use of metallic sinter additives

The rather low self-diffusion coefficient of TiB2 caused it’s low sinterability 52

. Therefore,

various sintering additives have been used to facilitate its sintering and increase the

densification. Metals such as Fe, Ni, Co, Cr, Mn have appropriate melting points, above

operating temperature of cell and well below the melting point of TiB2, and possess very

good wetting properties on TiB2 to promote liquid phase sintering 52

. They enhance the

32

densification of TiB2 liquid phase sintering by the presence of eutectic reactions at

relatively low temperature 57, 58

. It has been suggested that this eutectic reactions are based

on the reaction of metallic additive (Ni, Co, Cr, etc.) with TiB2 forming various metal

borides with low melting temperature e.g. Fe3B 58

. These borides also exhibit good wetting

behaviour. Sintering experiments using metallic additives have demonstrated that the very

dense TiB2 parts with very high hardness can be achieved by liquid phase sintering 32

.

Kang and Kim 51

investigated the effect of Cr and Fe on densification and mechanical

properties of TiB2 ceramics fabricated by pressureless sintering. They used 0.5 wt% Cr and

0.5 wt% Fe and sintered the specimens at 1800 and 1900°C for 2 h and the densities of

97.6% and 98.8% of theoretical density were achieved respectively. Figure 2–5 shows the

triple junction of grains and its related energy dispersive spectra (EDS). The authors

suggests liquid phase sintering mechanism. They were suggested that Ti dissolves into the

Cr–Fe solution at the sintering temperature and a Ti-rich liquid phase was formed which

may have a good wettability with TiB2. They believed that this Ti-rich liquid phase

enhanced mass transfer and this accelerated densification. The microstructure of the

specimen sintered at lower temperature showed equiaxed grains with uniform size

distribution but sintering at 1900°C led to an excessive grain growth.

Figure 2–5- Transmission electron micrograph of the specimen with 0.5 wt% Cr and 0.5

wt% Fe and energy dispersive spectra of X-ray microanalysis at a triple junction 51

.

33

However, the fracture toughness of TiB2 containing metallic additives and sintered at

1800°C is as low as that of monolithic TiB2, since the binder phases obtained after sintering

are mainly composed of brittle phases 57

.

Liquid phase sintering with metallic additives enhances mass transport but simultaneously

exaggerated grain growth of TiB2 crystals at much lower processing temperatures than that

of binder-less TiB2 due to dissolution precipitation phenomena 59

. This exaggerated grain

growth results in spontaneous microcracking of TiB2 during cooling, since high residual

stresses are developed among larger TiB2 grains due to their highly anisotropic thermal

properties 32, 57

.

Addition of grain growth inhibitor such as metal borides (TaB2 and W5B2) and carbides

(WC or TiC) has shown to avoid exaggerated grain growth. However, these additives do

not significantly improve mechanical properties due to the dominant effect of the binder

phase. Einarsrud et al 52

studied the effect of adding relatively small amounts (1-5%) of Ni,

NiB and Fe as sintering aids to promote liquid sintering of TiB2. They also added carbon to

some samples to reduce the amount of oxygen impurities from the starting powder. The

samples were sintered both under vacuum and argon atmosphere at the sintering

temperature varied between 1300 and 1700°C. High densities (>94% theoretical density)

were obtained for samples sintered at temperatures higher than 1500°C. Sintering under

vacuum resulted in the higher weight loss and lower density of specimens. A significant

grain growth was observed in the specimens containing Ni, NiB and Fe during sintering at

1700°C. The exaggerated grain growth 52

was observed to be related to the oxygen content

of the samples and to temperature. They proposed a mechanism that is dependent on

surface diffusion through an oxide layer and they consider that the surface diffusion is

enhanced by a titanium oxide rich surface layer 52

. The addition of carbon strongly reduced

the oxygen content and thereby, inhibited grain growth, but it increased the porosity as well

47.

34

2.6.2 Use of ceramic sinter additives

Apart from to the metallic additives, various non-metallic additives were also used to

enhance the sinterability and mechanical properties of TiB2 parts. Although the addition of

metallic additives improves the sinterability of TiB2, they generally degrade some

properties of the ceramics e.g. the fracture toughness. Therefore, non-metallic additives are

used for reasons such as to improve sinterability without grain growth or to retain good

oxidation resistance at high-temperatures 32, 48, 55, 60

.

It has been reported that some non-metallic additives could also promote the liquid phase

sintering of TiB2 and significantly enhance the density and reduce the grain size by the

secondary phase formation. It would be considered that the quantity of additives has

important effect on final density and grain size which must be investigated in each

particular case 32, 55, 60

.

Non-metallic additives are typically added in higher amount to densify TiB2, while a much

smaller amount of metallic additives is used to obtain dense TiB2 32

. A combination of high

Vickers hardness (20–27 GPa) and moderate indentation toughness (4–7 MPa.m1/2

) is

obtainable with the use of a large variety of non-metallic sinter additive (added in an

appropriate amount). A modest flexural strength of 500 MPa or higher is also measured,

depending on the density of sintered TiB2 32, 48, 55, 59, 60

.

Recently, Murthy et al. 55

reported that MoSi2 enhances the densification of TiB2 via liquid

phase sintering (LPS) and the formation of TiSi2 when hot-pressed at 1700°C for 1 h in

vacuum. The addition of MoSi2 does not degrade the mechanical properties and also exhibit

better wear resistance against bearing steel when compared with monolithic TiB2. They

detected the presence of a SiO2-rich layer on the surface. In their next work 56

, the addition

of TiSi2 as a sintering aid was explored. The hot-pressing experiments were conducted at

temperatures between 1400 and 1650°C on the samples containing 0–10 wt% TiSi2 for 1 h.

The sample with 5 wt% TiSi2 and hot-pressed at 1650°C showed the optimal results of high

hardness of 25 GPa, an elastic modulus of 518 GPa, an indentation toughness of ~ 6

35

MPa.m1/2

, a four-point flexural strength of more than 400 MPa, and an electrical resistivity

of 10 μΩ.cm.

Similarly, Torizuka et al. 60

observed the formation of grain boundary liquid phase

(amorphous SiO2), when SiC was used as an additive. They achieved densities greater than

95% of theoretical density by adding 2.5-5.0 wt% SiC and sintering temperature of 1700°C.

They also reached 99% of theoretical density by hot isostatic pressing (HIP) of TiB2- 5

wt% SiC at 1700°C. Optimising the amount of binder and sintering temperature is critical

for obtaining higher densification and improved mechanical properties through finer grain

size. They suggested that the improved sinterability of TiB2 resulted from the SiO2 liquid

phase that was formed during sintering when the raw TiB2 powder had 1.5 wt% of oxygen.

The effect of SiC and ZrO2 on sinterability and mechanical properties of titanium diboride

was investigated by Torizuka et al. 61

. The combined addition of ZrO2 and SiC were found

to be effective in improving the sinterability and mechanical properties of TiB2. The

density of TiB2 and TiB2–20ZrO2 (wt%) after sintering at 1700°C was 70% of theoretical

density. The addition of ZrO2 alone had therefore, little effect in improving the sinterability

of TiB2. However, the addition of SiC was found to be effective in improving the sintered

density. For example, the density of TiB2–19.5ZrO2–2.5SiC (wt%) was 97% of theoretical

density. It was reported that TiO2, existing on the surface of TiB2 powder, reacted with SiC

and formed TiC and SiO2.

Park et al. 48

investigated the effect of hot pressing temperature (1500–1800°C) on the

densification behaviour of TiB2–2.5Si3N4 (wt%). A considerable increase in density at

1500–1600°C is attributed to the formation of silica (SiO2) during hot pressing. Unlike the

density, the average grain size increased steadily as the sintering temperature is increased.

This result is in contrast to the case of transition metal additions, where extensive grain

growth occurs during densification. In the TiB2–Si3N4 composite system, the presence of

reaction products such as TiN and BN has been observed. The fracture toughness decreased

steadily as the Si3N4 addition was increased 48

.

36

When a small amount of AlN (5 wt%) was added to TiB2, the rutile phase (TiO2), present

on the TiB2 powder surface was eliminated by a reaction with AlN to form TiN and Al2O3.

The elimination of TiO2 markedly improved the sinterability and consequently the

mechanical properties of TiB2. It should be pointed out that large AlN addition (>10 wt%)

decreased the sinterability and mechanical properties, apparently owing to the

remaining/unreacted AlN 62, 63

.

As it was stated in this part, there are some ceramic materials, which have the potential to

be used as proper additives for sintering of TiB2. The use of ceramic sinter additives could

be considered as an alternative of the metallic additives for certain application of TiB2-

based materials.

Chapter 3:

Thesis outline

38

3.1 Objectives

The objectives of this project were defined with the aim of developing wettable cathodes

with regard to the huge potential of energy savings in this field. A review of the basic

knowledge of primary aluminum production and earlier efforts on the development and

application of suitable wettable cathodes for aluminum electrolysis was presented in the

previous chapter. After reviewing the conditions of electrolysis bath and the issues with

common carbon cathodes, particularly the energy loss caused by non-wettability for liquid

aluminum, the importance of the development of wettable cathodes as a solution becomes

more clear. TiB2 has been always the most trusted materials in the researches to develop

wettable cathodes because it fulfills the property requirements of this application. However,

fabrication of suitable TiB2-based cathode parts encounters some difficulties and despite a

lot of efforts on the development of TiB2-based wettable cathode since 1950’s, no proper

commercial product is available yet. Reviewing the previous works provides a lot of useful

ideas and insights for further investigation on this subject.

The objective of this project was to develop TiB2-based ceramic materials meeting the

property requirements of a suitable wettable cathode. In other words, the developed

materials must have the physical, chemical, mechanical and metallurgical properties equal

to or superior to the current carbon cathodes. They should also possess good wettability for

molten aluminum (<90) and the electrical resistivity lower than 500 cm and most

importantly, the grain boundaries of such materials must have physical and chemical

stability in molten aluminum at 960C to keep the integrity of the part in electrolysis cell

during the cathode life cycle.

In addition to the material characteristics, the process proposed for the fabrication of the

developed wettable cathodes is of great importance. Although methods such as HP, HIP

and SPS are already used for the production of TiB2-based materials, they are not suitable,

either technically or economically, for the fabrication of large cathode parts in industrial

scale. In this project, pressureless sintering was selected as it has the capacity to produce

large-scale near net shape parts with lower investment and operating costs. The important

39

concern with sintering of TiB2-based parts is that sintering at temperatures higher than

1700°C results in exaggerated grain growth and hence low mechanical properties and

thermal shock resistance due to the hexagonal crystalline structure of TiB2. Therefore, the

addition of metallic sintering aids was suggested in order to provide liquid phase and to

promote sintering at lower temperatures. In addition, metallic additives do not have

negative effect on the electrical conductivity. These additives should promote the formation

of phases in grain boundaries with low reactivity and solubility in molten aluminum

without deteriorating the wetting properties of TiB2. The composition of sintering additives

as well as processing parameters has a great influence on the final properties of the sintered

parts. Therefore, the investigation of the effect of additives composition and processing

parameters on the characteristics of consolidated specimens as well as their behaviour and

stability in liquid aluminum was also part of the project objectives.

3.2 Choices of sintering additives

As mentioned in previous sections, TiB2-based ceramic is the most trusted material for

application as wettable cathode in aluminum electrolysis. However, monolithic TiB2

ceramic has low sinterability. It has also poor resistance to thermal shock due to its thermal

expansion anisotropy. Addition of sintering aids could enhance the sinterability and the

thermal shock resistance of TiB2 32

. Metallic additives such as Fe, Co or Ni improve the

densification of TiB2 via liquid phase sintering by the presence of eutectic reactions at

relatively low temperatures 51, 52

. However, the residual binder phases after sintering are

mainly comprised of phases which could be attacked by molten aluminum and cryolite and

consequently decrease the resistance and stability of the specimens in the bath conditions.

Studies have shown that the grain boundary chemistry is more important than the density

and the strength of TiB2 material used as cathode during Al electrolysis 45

.

In the preliminary experiments, we have introduced boron filament in molten aluminum

and we found that above the certain Ti concentration, the filament will have chemical

stability and will not dissolve in molten aluminum. TEM investigation revealed the

formation of TiB2 layer on the surface of boron filament, which protects the filament from

40

dissolution in molten Al. In fact, when the concentration of Ti in liquid aluminum is more

than 200 ppm of Ti, segregation of TiB2 on the surface of boron is energetically favourable.

Based on this result, we proposed that the controlled addition of Ti to the powder mixtures

as sintering aid might promote the segregation of TiB2 phase at the grain boundaries, which

is durable in liquid aluminum. In addition, the binary phase diagram of Fe-Ti, Figure 3–1,

shows that the Fe-Ti system forms a liquid phase over a wide range of compositions at

temperatures higher than 1450°C. They specially have a eutectic point at around 30 wt% Fe

with a melting point of 1085°C. Fe has been suggested as a proper sintering additive for

liquid phase sintering of TiB2 51

. As conclusion, the mixture of Ti and Fe was selected as

sintering additive to enhance the sinterability by forming the liquid phase during

pressureless sintering of TiB2-based wettable cathode parts.

Figure 3–1- Binary phase diagram of Fe-Ti 64

.

41

3.3 Originality of the project

Several works have been previously reported on development of TiB2-based ceramics as

wettable cathode for aluminum electrolysis industry 12, 15, 22

. However, none of these

materials has been widely used to replace common carbon cathodes in industrial scale. The

majority of works on the fabrication of dense TiB2 cathodes have focused on costly and

complicated fabrication methods. The main disadvantage of the proposed fabrication

methods is their inability to form the designed final shape of cathode lining due to their

technical limitations. Another limitation of the previous researches is the addition of

components to TiB2 which react with the bath compounds mainly molten aluminum and

cryolite 30, 45

. The products of such reactions can destroy the TiB2 grain junctions by

penetration and swelling which lead to separation of TiB2 particles and thereby cathode

abrasion 41

. Some efforts have been focused on using carbonic additives to fabricate TiB2-

based composites or coatings, but the problem of sodium penetration and swelling still

exists 41

. Use of colloidal alumina bonding for coating of TiB2 was also investigated.

Although the wettability and beneficial physical properties of the coating have been

claimed, there is still the sodium-trapping and the formation of sodium aluminate during

the initial stages of cell operation 41

.

In this project, for fabrication of TiB2-based cathode, inexpensive pressureless sintering

method has been used which permits the production of large-scale cathode parts with the

designed final shape at relatively lower costs. The use of Ti as a major metallic additive has

been also investigated. The effect of the excess amount of titanium on sintering of TiB2-

based cathodes had not been previously studied. The processing parameters including

starting powder particle size, sintering additives composition and content, milling

conditions, compacting conditions, sintering cycle, temperature and atmosphere were

investigated and modified to achieve dense parts with required properties. Besides

performing general physical, mechanical and metallurgical characterizations of the

developed materials, other properties for wettable cathode application such as interactions

and wettability by liquid aluminum, stability in molten aluminum and the nature of grain

42

boundaries and inter-particle bridges were studied as well. The majority of literatures found

on the fabrication of consolidated TiB2-based materials do not contain most of these

characterizations, which are essential for cathode application.

Another important achievement of this project was the obtaining of uniformly distributed

pores within the microstructure of the developed TiB2-based material, which could

potentially help prevent crack propagation and failure of the parts during service due to the

probable thermal shock and molar volume changes, caused by the interaction of

components with liquid aluminum. This aspect is proposed to considered in the target

microstructure of TiB2 ceramics for wettable cathode application in future investigations

and in further modifications of the fabrication process parameters.

Chapter 4:

Materials and methods

44

4.1 Introduction

In this project, starting powders were prepared by mixing of TiB2 powder with sintering

additives using high-energy ball mill. Compaction performed in uniaxial die to form the

green parts. They were then sintered under the flow Ar/5%H2 reducing atmosphere for 1 h

at temperatures between 1400 and 1650C. The physical and mechanical properties of

specimens were measured and the microstructures were studied. Their wettability for

molten aluminum was evaluated and their stability in molten aluminum at 960°C was

investigated. The general information about the experimental procedure as well as

characterization methods is explained in this chapter. Further details are provided at the

materials and methods section of each article.

4.2 Experimental procedures

4.2.1 Starting powders

The control of the powder quality has a great influence on the production of green body and

the final sintered microstructure. The powder characteristics of greatest interest are the size,

size distribution, morphology, degree of agglomeration, chemical composition, and purity

65. The specifications of starting TiB2 and additive powders used during the project are

shown in Table 4–1. These materials were purchased from Atlantic Equipment Engineers

Inc. in powder form.

Table 4–1- Starting powders specification

Substances Particle size Purity

TiB2 2-10 micron > 99.7%

Ti < 20 micron > 99.7%

Fe 1-9 micron > 99.9%

45

4.2.2 Powder processing

From the results of preliminary experiments the total amount of 10 wt% of metallic

additives was selected to be added to TiB2 powder for preparing the powder mixture. The

additives were constituted of Ti and Fe powders, either separately or in pre-alloyed form.

Different mixtures of additives were investigated to determine the proper composition for

desired application.

To achieve homogeneous compositional distribution, the powder blends were mixed in

high-energy ball mill (SPEX 8000, Spex Industries, Inc.) using hardened steel vial and balls

with a ball-to-powder weight ratio of 4:1. The particle size distribution was determined

using Laser Scattering Particle Size Distribution Analyser (LA-900, HORIBA).

Pre-alloying of Ti and Fe powders was performed, first by mixing titanium and iron in a

70:30 Ti:Fe weight ratio and then by pressing the powder mixture at 400 MPa using an

uniaxial steel die. The compacted specimens were then sintered at 1150°C for 1 hour. The

resulting pellets were subsequently crushed and milled using high-energy ball milling for 1

h to obtain the pre-alloyed additive in powder form.

4.2.3 Forming method

Uniaxial dies were used to compact the powder mixtures into green parts. The compressing

pressure of 150 MPa was selected from the results obtained during preliminary

experiments. Three cylindrical dies with 13 mm, 16 mm and 25 mm diameters and one

rectangular die with the dimensions of 35 mm × 15 mm were used.

The RHM powders do not exhibit plastic flow even at high pressure; therefore, special care

should be taken during compaction process in order to achieve relatively uniform compact

density in the specimen and to prevent residual internal stresses.

4.2.4 Consolidation and sintering

Sintering of specimens was performed in the temperatures ranging between 1400 and

1650C. Green parts were placed on a high alumina plate (Al2O3 > 99.5%, Anderman

46

Ceramics Co.) and were inserted in a high temperature tube furnace (INP15-20, Norax

Canada inc.). The flow Ar-5%H2 reducing atmosphere was used during sintering process.

The heating rate was about 6C /min and the specimens were maintained at sintering

temperature for 1 h. The specimens were then furnace cooled. Figure 4–1 shows an

example of sintering profile used in the experiments.

Figure 4–1- Diagram of sintering cycle

47

4.3 Characterisation methods

The physical and mechanical properties of specimens including as density, electrical

conductivity and flexural strength were determined in order to study the effect of

fabrication parameters on their properties with regard to their application as wettable

cathode for aluminum electrolysis. The wettability behaviour of specimens for liquid

aluminum was also investigated. Chemical stability of selected specimens in molten

aluminum at 960°C was then studied. Morphology and microstructure of specimens were

characterized using optical microscope (OM), scanning electron microscope (SEM) and

transmission electron microscope (TEM). Compositional and phase analysis of samples

were performed using X-ray diffraction and energy dispersive spectroscopy.

4.3.1 Bulk density and apparent porosity measurements

The bulk density of compacts was determined using the Archimedes method with

isopropanol as the immersing medium, which is applicable to almost all refractory shapes,

only if they have sufficient structural integrity to permit handling. In this method the

specimen was weighed in dry state (WD), then it was immersed in iso-propanol while its

weight was recorded in real time by a computer. The specimen was kept inside the liquid

since there were no more significant changes in the measured weight, which was assumed

as apparent immersed weight (WI). The density of the specimen (DS) was calculated using

these weights and the density of iso-propanol alcohol at experiment’s temperature (DL):

( ) ( 4-1)

Theoretical density was estimated using the rule-of-mixtures calculations that assumed the

nominal compositions of the powder specimens as specified, neglecting the limited

products of the interface reactions. The theoretical density of pure dense TiB2 fired bodies

was considered as 4.52 g/cm3

66. The theoretical density of Fe, Ti, and Al were also

considered as 7.87, 4.50, 2.70 g/cm3 respectively

67. Relative densities were calculated by

dividing the measured bulk density by the calculated theoretical density. Based on

48

replicated measurements on identical specimens, the uncertainty on relative density was

estimated to be less than ±1%. Hence, no error bar was included in the figures.

4.3.2 Mechanical properties evaluation

The evaluation of mechanical properties of the brittle ceramics is usually performed using

transverse bending test, in which a rod specimen (with either circular or rectangular cross

section) is bent under a three-point or four-point loading technique until fracture. In this

technique the bottom surface of specimen is placed under a tension stress, since the upper

surface is in the state of compressive stress. Considering that the compressive strength of

ceramics is higher than their tensile strength in the order of ten, the samples are always

fractured starting from their bottom surface; the flexure test is a reasonable substitute for

tensile strength. Since during bending the specimen is subjected to both compressive and

tensile stresses, the magnitude of its flexural strength is greater than its tensile fracture

strength 68

.

Figure 4–2- Schematic of three-point loading for measuring the flexural strength of

ceramics 68

.

The flexural strength of TiB2-based specimens in this project was measured using three-

point bending method with a 25.4 mm span at a deflection rate of 0.5mm/min following

49

ASTM C1161-02 standard method 69

. This test method covers the determination of flexural

strength of advanced ceramic materials at ambient temperature.

For three-point test, flexural strength (MPa) of rectangular specimens, S, was calculated

based on following equation:

( 4-2)

where:

P = breaking force (N),

L = outer (support) span (mm),

b = specimen width (mm),

d = specimen thickness (mm) 68, 69

.

4.3.3 Measurement of electrical resistivity

For the measurement of electrical resistivity at ambient temperature, a specimen was placed

between two copper plates, and pressure was applied (10 MPa) to obtain a good electrical

connectivity. The current density of about 1 A/cm2 was applied which is similar to the

current density of cathodes at cell condition. The resistance was measured and the electrical

resistivity (ρ) of the specimen was calculated using the following equation:

( 4-3)

where R is the electrical resistance, A is the contact surface area and l is the specimen’s

thickness.

22

3

bd

PLS

50

4.3.4 Wettability in liquid aluminum

The wettability of specimens by molten aluminum was investigated using sessile drop

technique. The surface of the Al pellet was polished prior to the test in order to reduce

surface oxides. The surface of specimens was also polished before the tests to remove

surface contamination, especially oxides. Polishing was carried out using a diamond

abrasive (6 µm) followed by cleaning with isopropanol in an ultrasonic bath. An aluminum

piece (90 mg) was placed on the surface of the specimen that was then heated up rapidly to

960°C under vacuum (10-3

Pa) and kept at this temperature. A light was fixed at one end of

the tube and the image of the Al drop over the specimen was recorded at the other end

during the experiment. The schematic of the setup used for this experiment is shown in

Figure 4–3. A software was used to evaluate the contact angle between the specimen

surface and the aluminum drop from the recorded images. The reported contact angle is the

average of left and right side angles. The time t=0 was set when the specimens’ temperature

reached 700°C approximately and a spherical liquid Al drop formed over the surface of the

sample. After the experiment, the furnace was cooled down at a rate of about 15°C/min

below the melting point of pure aluminum.

Figure 4–3- Schematic of setup used for sessile drop test and wettability investigation

4.3.5 Chemical stability and durability to liquid aluminum

Chemical stability and durability of specimens in liquid aluminum were also evaluated. The

specimen was glued to the tip of an alumina rod, using ceramic glue (Aremco Ceramabond

51

571), and it was then inserted into liquid aluminum under protecting argon flow at 960°C.

The surface of the specimen was covered by pure aluminum foil to protect the surface from

the oxidation during the insertion process. The specimen was removed after the test, cross

sectioned, polished and analyzed by SEM, EDS, EPMA, and TEM in order to examine the

microstructure and to evaluate their interaction with liquid aluminum.

4.3.6 Scratch test

The scratch test was conducted at room temperature by using a micro-tribometer test

system (UMT-2; CETR). The specimen surface was polished down to 6 µm diamond

abrasive paper prior to the test. The specimen was fixed to the lower holder, which was

automatically driven along a single horizontal axis while a conical diamond indenter

mounted in an upper holder sliding over the surface of the specimen and applying a vertical

force. The cone angle of the diamond indenter was 75o and its diameter tip was 400 µm.

The vertical component of the force (Fz) was increased gradually from 2 to 50 N over the

10 mm sliding distance. The horizontal component of the force (Fx), applied to the

specimen through the diamond indenter, was measured in a real time with a dynamometer

and reported as the test result.

4.3.7 Microstructural characterisation

Microstructure characterisation of the processed TiB2 specimens included optical

microscopy, scanning electron microscopy, and transmission electron microscopy. For OM

and SEM analysis, specimens were cut by automatic cutting machine (Struers) using a

Diamond Wafering Blade (Buehler) to obtain their cross section. Then, they were mounted

in epoxy resin and hardener (Struers) under vacuum. The cross sections were grounded and

polished to 0.1 µm surface finish using successively finer Diamond Grinding Discs

(Buehler) and diamond pastes. The final polishing was performed using a 0.05 µm alumina

suspension.

For the exposed specimens to aluminum, the infiltrated aluminum was removed by soaking

in a 0.3 N sodium hydroxide solution for 48 h at room temperature. Once the infiltrated

52

aluminum was removed, a porous structure, mainly composed of a TiB2 solid skeleton, was

obtained and characterized using SEM.

The SEM secondary electron images were used to observe the morphology and the back-

scattered electron imaging provided a distinctive image of different phases based on their

compositional differences.

TEM sample preparation was performed using focus ion beam (FIB; Hitachi FB2000A)

milling. TEM (Jeol JEM-2100F) observation enabled us to study the location and

composition of inter-particle bridges and grain boundaries as well as the nature of phases in

the microstructure.

4.3.8 Compositional and phase analysis

Phase analysis of the bulk of each sample was identified by X-ray diffraction (XRD;

Siemens D5000) using Cu Kα radiation at a scanning rate of 1.min-1

in the 2θ range of 25-

80. The detection limit of XRD apparatus was about 5 wt%. Energy dispersive

spectroscopy (EDS; PGT Avalon) was also used to identify the elemental composition of

the phases.

For precise element analysis of secondary phases, electron microprobe analysis (EMPA;

SX-100 CAMECA microprobe) was used. EMPA uses a high-energy focused beam of

electrons for detecting and measuring characteristic X-rays. Its electron beam current is

roughly 1000 times greater than that in a SEM. These higher beam currents produce more

X-rays from the sample and improve both the detection limits and accuracy of the resulting

analysis. Analysis locations are selected using a light optical microscope, which allows

accurate positioning to about 1 micron.

Chapter 5:

Pressureless sintering of TiB2–based composites using Ti

and Fe additives for development of wettable cathodes

Hamed Heidari1; Houshang Alamdari

1; Dominique Dubé

1; Robert Schulz

2

1 Department of Mining, Metallurgical and Materials Engineering, Université Laval, Québec (QC), Canada G1V 0A6

2 Hydro-Quebec Research Institute, 1800 Boul. Lionel Boulet, Varennes, QC, Canada J3X 1S1

This article was presented in TMS 2011 international conference and was published in the

journal: Light Metals 2011, P. 1111-1116.

54

5.1 Résumé

La diborure de titane est le matériau le plus prometteur pour la fabrication des cathodes

mouillables destinés à la production de l’aluminium. Il est considéré comme une alternative

au carbone afin de réduire la distance cathode-anode, ce qui entraine une meilleure

efficacité énergétique dans les cellules d'électrolyse. Dans ce travail, des spécimens à base

de TiB2 ont été consolidés en utilisant des additifs métalliques suivis par un frittage en

phase liquide sans pression. Différentes proportions de fer et de titane (≤ 10% en poids) ont

été étudiés pour la fabrication des additifs. Le frittage a été réalisé entre 1400-1650°C sous

atmosphère contrôlée. Les effets de la composition, de la température de frittage, du temps

de broyage ainsi que le pré-alliage des additifs sur la densification, la microstructure et les

propriétés mécaniques ont été étudiés. Il a été constaté que le pré-alliage et le temps de

broyage avaient une influence significative sur la densification, l'uniformité de la

microstructure et la résistance à la flexion. L’utilisation des additifs pré-alliées, broyés

pendant 30 min et frittés à 1650°C/1 h a permis d’obtenir une microstructure uniforme, et

sans fissures avec une distribution uniforme des pores ainsi qu’une densité relative

maximale de 91%. Une résistance à la flexion de 300 MPa a aussi été obtenue.

5.2 Abstract

Titanium diboride is the most promising candidate material for development of wettable

cathodes for aluminum smelting. It is considered as an alternative for carbon cathodes in

order to reduce the anode cathode distance resulting in higher energy efficiency in

electrolysis cells. In this work, TiB2-based ceramic specimens were consolidated using

metallic additives followed by pressureless sintering. Different proportions of iron and

titanium (≤ 10 wt%) were used as low melting point sintering additives. Sintering was

conducted at 1400–1650°C under controlled atmosphere. The effects of composition,

sintering temperature, milling time and pre-alloying of the additives on densification,

microstructure, and mechanical properties were investigated. It was found that pre-alloying

and milling time have significant influence on densification, microstructure uniformity and

bending strength. Uniform crack-free microstructure with even distribution of pores as well

55

as maximum relative density of 91% and bending strength of 300 MPa were obtained using

pre-alloyed additives, milling time of 30 min and sintering for 1 h at 1650°C.

56

5.3 Introduction

Liquid aluminum reacts with almost all materials with only a few having good stability 25

.

Most of the stable materials are very expensive metals or non-conductive ceramics, which

are major obstacles for their application as cathode material. Nevertheless there are some

electrically conductive ceramics such as graphite and TiB2. Graphite has been used for

more than a century in aluminum electrolysis cells 14

. However, liquid aluminum does not

wet graphite. Thus a relatively thick liquid metal pad is kept on top of the graphite cathode

to avoid the diffusion of electrolytes through the cathode blocks and to insure good

electrical current distribution within the cell. The magnetic fields present in smelting cells

apply significant Lorentz forces on the metal pad resulting in wave creation. In order to

avoid the short circuits between the metal pad and anode, the anode-cathode distance is

kept large (typically 4.5 cm) unnecessarily increasing the bath resistance 12, 13

.

TiB2 has very high melting point (about 3000°C), low density, excellent strength, high

hardness, and very good thermal and electrical conductivities 33, 70, 71

. It is chemically stable

in and well wetted by liquid aluminum 12, 13, 15

. It has been therefore the most promising in

the attempts to find an alternative material for carbon cathodes since the search began in the

1950’s 22

. Despite its extraordinary properties, strong covalent bonding and low diffusion

coefficient make sintering of TiB2-based ceramics quite difficult 33

.

Fully dense TiB2 is probably not necessary for cathode application. It has been shown that

using dense TiB2 cathodes results in early failure by cracking 30

. Liquid aluminum reacts

with impurities at grain boundaries and, after a period of time, results in the formation of

new phases, internal stress build up and crack formation 45

.

A number of techniques have been used to consolidate TiB2 49, 53, 60, 72

. Pressureless

sintering is a low cost technique to produce large and near net-shape components. However

to consolidate pure TiB2 with this technique, very high sintering temperatures are required

resulting in exaggerated grain growth and reduced mechanical properties. It has been

reported that at temperatures above 1700°C, the presence of titanium oxide at the surface of

57

particles increases both pore and grain size by increasing the surface diffusivity 46, 47

. In

industrial conditions, it is difficult to remove this oxide layer. Thus, it is preferred to sinter

TiB2-based ceramics below 1700°C. At these relatively low temperatures the use of

sintering additives are almost unavoidable to provide a liquid phase promoting

consolidation of TiB2-based composite. Under these conditions, the sintering of TiB2-based

composite requires an appreciable amount of liquid phase, wettability of TiB2 by the liquid

phase, and small solubility of the solid phase in the liquid 73

.

In this study, titanium and iron were used as additives to reduce the sintering temperature.

Titanium was chosen as the principal metallic additive. Upon infiltration of aluminum into

the porous cathode, it could react with the titanium additive to produce TiAl3 with a

melting point higher than that of aluminum. TiAl3 also shows a very good wettability with

respect to molten aluminum. Although TiAl3 is soluble in liquid Aluminum, it has been

shown to be stable when formed at the surface of TiB2 74, 75

. Iron, in turn, was chosen to

somewhat reduce the melting point of additives. Ti and Fe have a eutectic at 1078°C and

71.1 mol% Ti 76

. These metals wet the surface of TiB2 77

while TiB2 has a small solubility

in liquid Ti-Fe 78

.

The TiB2-Fe system is characterized also by a eutectic at 1340°C and 6.3 mol% TiB2 79

.

The reaction between TiB2 and Fe may accelerate densification and lead to the formation of

Fe2B. This phase could cause deterioration in the mechanical properties as well as the

resistance to liquid aluminum. The presence of titanium in the liquid mixture could prevent

the formation of Fe2B 80

. Fe can also react with liquid aluminum to form FeAl3 having a

melting at about 1160°C 81

.

In this work, the effects of different compositions, processing conditions, and sintering

temperature on density, microstructure and mechanical properties were investigated. A total

amount of 10 wt% Ti and Fe was selected as metallic additives to provide about 10 vol% of

liquid phase during sintering.

58

5.4 Materials and methods

Commercial TiB2, Ti and Fe powders (Atlantic Equipment Engineers Inc., Bergenfield, NJ)

were used as starting materials. The particle size of TiB2 powder was between 2 and 10 µm

with a mean size of 6 µm. Its purity was >99.7 % and the impurities were C, N, O and Al.

For Ti powder the particle size was <20 µm and for Fe powder, the particle size was

between 1 and 9 µm.

The TiB2 powder was mixed with selected proportions of Ti and Fe powders and then

milled in a high energy ball mill (SPEX 8000, Spex Industries, Inc., Edison, NJ) using

hardened steel vial and balls with a ball-to-powder weight ratio of 4:1. Ti and Fe powders

were added to TiB2 either separately or after being pre-alloyed.

Pre-alloying of Ti and Fe powders was performed, first by mixing titanium and iron in a

70:30 Ti:Fe weight ratio and then by pressing the powder mixture at 400 MPa using an

uniaxial steel die. The compacted specimens were then sintered at 1150°C for 1 hour. The

resulting pellets were subsequently crushed and milled using high-energy ball milling for 1

h to obtain the pre-alloyed additive in powder form. XRD analysis showed the presence of

α-Ti and the FeTi intermetallic compound in this powder.

The TiB2-Ti-Fe powder mixtures were pressed at 150 MPa in a uniaxial die to form pellets

of 16 mm in diameter and bars of 38 × 13 mm. The specimens were then heated in a tube

furnace (INP15-20, Norax Canada inc., QC) at a rate of 6°C/min from room temperature to

the specified sintering temperature. Sintering of pellets was performed under an Ar/5%H2

protective atmosphere for 1 hour. The various experimental conditions of this study are

given in Table 5–1.

59

Table 5–1- Experimental conditions used for consolidation of specimens

No. Code* Additives wt%** Sintering

Temperature (°C) Pre-alloying

Milling time

(min) Ti Fe

1 T8F2M10 8 2 1400-1600-1650 - 10 2 T7F3M10 7 3 1400-1600-1650 - 10

3 T8F2M30 8 2 1650 - 30 4 T7F3PM10 7 3 1600-1650 yes 10 5 T7F3PM30 7 3 1650 yes 30 6 T7F3PM60 7 3 1650 yes 60 7 T7F3PM120 7 3 1650 yes 120 8 T7F3PM240 7 3 1650 yes 240

*T: Titanium, F: Iron, P: Pre-alloying, M: milling time

**Unless noted, all compositions in this article are in weight percent

The green density of specimens was evaluated by measuring their weight and geometrical

dimensions. The bulk density of sintered specimens was determined using the Archimedes

method with isopropanol as the immersing medium. Theoretical density was estimated

using the rule-of-mixtures calculations that assumed the nominal compositions of the

powder specimens as specified. Relative densities were calculated by dividing the measured

bulk density by the calculated theoretical density. Based on replicated measurements on

identical specimens, the uncertainty on relative density was estimated to be less than ±1%.

Hence, no error bar was included in the figures.

The bending strength of the sintered specimens was measured using the three-point bending

test with a 26 mm span at a loading rate of 0.5 mm/min according to the ASTM C1161

standard 69

. The dimensions of the test specimens used for bending strength measurements

were 38 mm × 13 mm × 4 mm.

For microstructural investigations, the specimens were cut with a diamond saw and

polished to 0.1 µm surface finish using successively finer diamond abrasives. The

microstructure of specimens was investigated using a scanning electron microscope and the

chemical microanalysis was performed by energy dispersive X-ray spectroscopy (EDX;

PGT Avalon, Princeton, NJ). The crystalline structure was determined by X-ray diffraction

method (XRD; Siemens D5000) with a Cu Kα radiation at a scanning rate of 1° min−1

.

60

Wettability of sintered specimens was studied by placing a 90 mg aluminum piece at the

surface of sintered specimen surface and heating under high vacuum (1.2E-5

mbar) to

960°C. The wetting angle was monitored using an instant imaging system. From these

images, contact angles can be measured as a function of time.

5.5 Results and discussion

5.5.1 Effect of additive composition

The relative density was measured for specimens with two different compositions after

sintering at 1400, 1600 and 1650°C. As shown in Figure 5–1, at all sintering temperatures,

specimens containing 7 wt% Ti and 3 wt% Fe (T7F3M10) had higher density than that of

specimens containing 8% Ti and 2% Fe (T8F2M10). According to the Fe-Ti phase diagram

82 the additive with a mass ratio of 7Ti:3Fe is closer to eutectic. A lower melting point leads

to earlier formation of liquid phase during sintering, therefore promoting better

densification.

Figure 5–1- Comparison of the relative density as a function of sintering temperature

and composition of sintering additives (T8F2M10: TiB2+8%Ti+2%Fe; T7F3M10:

TiB2+7%Ti+3%Fe).

61

5.5.2 Effect of sintering temperature

As shown in Figure 5–1, for both compositions (T7F3M10 and T8F2M10), there is not

much densification after sintering at 1400°C. However, sintering at 1600 and 1650°C

resulted in an appreciable densification to approximately 70%. Between 1600 and 1650°C,

there is no further densification. An SEM backscattered (BS) micrograph of the T8F2M10

specimen, sintered at 1400°C, is presented in Figure 5–2. It shows that the additives formed

segregated phases. A temperature of 1400°C, well below the melting point of Fe and Ti, is

not high enough to provide the liquid phase required to promote densification. Moreover,

these additives were not locally mixed in the proper ratio and were not in intimate contact

with each other.

When Fe and Ti particles are in intimate contact, solid-state diffusion occurs at the interface

resulting in the formation of a thin liquid layer in between. Increasing the sintering

temperature to 1600°C resulted in the formation of significant amount of liquid and

consequently in higher densification of specimens.

Figure 5–2- Backscattered SEM micrograph of T8F2M10 (TiB2+8%Ti+2%Fe)

specimen, sintered at 1400°C for 1 h (The arrows show segregated phases containing the

additives).

62

5.5.3 Effect of pre-alloying additives

Besides the uniform distribution of additives, it is important to have Ti and Fe particles in

contact with each other to promote the formation of the liquid phase. Hence, the addition of

pre-alloyed additives, instead of adding Ti and Fe separately, was considered to achieve

this goal. The pre-alloyed additives were prepared by mixing, pressing, and sintering Ti and

Fe powders with a mass ratio of 7:3.

Table 5–2 compares the relative densities of specimens with pre-alloyed additives

(T7F3PM10) with those obtained by adding the additives separately (T7F3M10) after

sintering at two different temperatures. Under the same processing and sintering conditions,

pre-alloying of the additives resulted in better densification. The difference is significant at

1650°C.

Table 5–2- Relative density of specimens with separate (T7F3M10) and pre-alloyed

(T7F3PM10) additives

Specimen Relative density (%)

Green Density Sinter. 1600°C Sinter. 1650°C

T7F3M10 59 72 72

T7F3PM10 62 74 80

5.5.4 Effect of milling time

Milling was performed in order to achieve a uniform distribution of additives. However,

milling time should be as short as possible to reduce costs and to prevent oxidation of

powders. To investigate the effect of milling time on the densification process, powder

mixtures were milled for different times prior to sintering. Preliminary results showed that

by increasing the milling time from 10 to 30 min, density after sintering increased. These

preliminary experiments suggest that the milling time has an important influence on

densification. The effect of milling time was further investigated in a systematic way for

specimens containing 90 wt% TiB2, 7 wt% Ti, and 3 wt% Fe using five different milling

63

times (10, 30, 60, 120 and 240 minutes) followed by compaction and sintering at 1650°C.

(Specimens 4-8, Table 5–1).

The densities of specimens (green and after sintering) were plotted as a function of milling

time in Figure 5–3. No significant influence of milling time on green density was observed.

However specimens milled for 30 min showed a maximum density of 91% after sintering.

Further milling resulted in a slight decrease in density. Figure 5–4 shows that the three-

point bending strength of sintered specimens follows a similar trend: a maximum bending

strength of 300 MPa was achieved for specimens milled 30 min.

Figure 5–3- Influence of milling time on relative density of green and sintered specimens

(TiB2+7%Ti+3%Fe) using pre-alloyed additive and sintered at 1650°C for 1h.

64

Figure 5–4- Influence of milling time on bending strength of specimens

(TiB2+7%Ti+3%Fe) prepared using pre-alloyed additive and sintered at 1650°C for 1h.

In order to understand the influence of milling time on sintering, the particle size

distribution was determined and XRD analyses were performed on milled powders while

the microstructure of sintered specimens was investigated by SEM.

The effect of milling time on the particle size distribution of powders is shown in Figure 5–

5. After 10 min of milling, the powder mixture is mainly composed of particles with a

mean size of 6 micrometers similar to that of the starting TiB2 powder. However, a wider

particle size distribution was observed. This distribution widening as well as the appearance

of a shoulder at around 2 micrometers is most likely due to the TiB2 particle fracturing and

refining during milling. In addition, EDX analysis of the very large particles showed that

the peak at 200 micrometers is related to the Ti and Fe pre-alloyed particles. Milling of

mixed powders for 10 min reduced the particle size of additive powders, but some large

additive particles remained. Milling for 30 min, however resulted in a quite different

particle size distribution. The distribution is much wider and shifted toward the small

diameters. The quantity of particles smaller than 0.7 micrometer increased, and the peak

related to the metallic additive particles disappeared. This suggests that milling for 30

65

minutes results in a good refining and dispersion of metallic additives within the powder

mixture and provides partial refining of TiB2 particles (particles smaller than 0.7

micrometer).

Figure 5–5- Effect of milling time on the particle size distribution for powder mixtures

containing the 70%Ti and 30%Fe pre-alloyed additive.

By increasing the milling time to 60 min, a second peak appeared in the particle size

distribution at around 100 micrometers. Since the powder samples are deagglomerated

using an ultrasonic bath prior to analysis, the presence of this second peak suggests the

formation of strong agglomerates in the powder. By further increasing the milling time, the

quantity of these strong agglomerates increases but no significant increase is observed in

the quantity of small particles. EDX analysis of large agglomerates showed that they were

rich in Ti and Fe. They are usually formed by plastic deformation and cold welding of

smaller particles and are very difficult to deagglomerate even in an ultrasonic bath.

66

The typical shape of the large agglomerates observed after 240 min of milling is shown in

Figure 5–6. These large particles are basically composed of TiB2 particles welded together

by metallic additives.

Figure 5–6- SEM micrograph of a large agglomerate formed after 240 min milling in

T7F3PM240 powder (TiB2+7%Ti+3%Fe).

The XRD patterns of the TiB2 powders milled for 10 and 240 min are shown in Figure 5–7.

The peaks were slightly broadened after 240 min of milling while the intensities decreased

owing to the overall decrease of the crystal size of TiB2. At the resolution of these x-ray

scans, the minor phases corresponding to the pre-alloyed additives were not detected.

67

Figure 5–7- XRD analysis of powders containing pre-alloyed additives after 10 and 240

min milling. (Cu Kα).

SEM micrographs of polished sections of specimens containing pre-alloyed additives and

sintered at 1650°C are shown in Figure 5–8 to Figure 5–10. In Figure 5–8, the

microstructure of the specimen milled for 10 min revealed a highly porous structure. The

high level of porosity explains the low density (84%) and reduced bending strength of this

specimen (197 MPa). By increasing the milling time to 30 min (Figure 5–9), a more

uniform and denser microstructure was achieved after sintering. A significant increase of

density (91%, Figure 5–3) and bending strength (300 MPa, Figure 5–4) was observed for

this specimen compared with the previous one.

As shown in Figure 5–5, after 30 min of milling, there was a refinement of the particle size

and a broader distribution was observed leading to higher densification. During liquid

phase sintering, densification is mostly caused by rearrangement of particles upon

formation of the liquid phase 83

. Further densification is achieved by the solution-

precipitation process: small particles promote this stage due to their higher surface energy

and therefore higher solubility in the liquid phase 83

. A broader particle size distribution

increases the overall contact area between particles and eases the particle rearrangement in

68

the early stage of sintering. Moreover, the presence of small particles can help the solution-

precipitation stage. As a result, higher densification and bending strength could be achieved

after 30 min of milling.

By further increasing milling time, densification and bending strength decrease. As

revealed by the microstructure of T7F3PM120 specimens ball milled for 120 min (Figure

5–10), cracks were formed after sintering. These cracks explain the dramatic decrease of

bending strength (Figure 5–4). Crack formation was attributed to the presence of hard

agglomerates in the powder mixture. As shown in Figure 5–6, these large agglomerates are

formed by the cold welding of TiB2 particles with additive particles. In the early stage of

milling, the TiB2 particles are partially refined and stick to the additives. Upon further

milling, the TiB2 particles become embedded in additives and form large and dense

agglomerates. Upon sintering, these large agglomerates shrink initiating cracks around

them in the compact. This phenomenon has also been reported previously 73, 83

. These

cracks limit the densification of the sintered specimens and reduce their strength.


(TiB2+7%Ti+3%Fe) milled for 10 min and sintered 1 h at 1650°C.

69


(TiB2+7%Ti+3%Fe) milled for 30 min and sintered 1h at 1650°C.


(TiB2+7%Ti+3%Fe) milled for 120 min and sintered 1 h at 1650°C.

70

5.5.5 Wettability and stability in liquid aluminum

Since specimens milled 30 min with pre-alloyed additives (T7F3PM30) show the best

density, strength and uniform microstructure, their wettability by liquid aluminum was

investigated. Figure 5–11 shows images of the aluminum drop on the surface of T7F3PM30

specimen at 960°C during the wettability test. From these images, the contact angles were

measured as a function of time. The first image taken as the temperature reached 960°C

(t=0) shows that the liquid Al drop was almost spherical. The contact angle between liquid

aluminum and the specimen at this moment was 169°. After a while, the contact angle

started to decrease. After 60 min the contact angle was 96° and only 14° after 175 min.

After 185 min, liquid aluminum was completely spread over the surface, which indicates

that wetting occurs quite rapidly on this specimen and it has good wettability for liquid

aluminum.

Figure 5–11- Behavior of liquid Al drop over T7F3PM30 specimen (TiB2+7%Ti+3%Fe)

during the wettability test at different time. (The time from beginning of test are

reported in minutes)

The stability and reactivity of T7F3PM30 specimen in liquid aluminum have also been

investigated. The specimen keeps its integrity after being exposed to liquid aluminum

during 24h. Detailed results of these experiments will be published in an upcoming report.

71

5.6 Conclusions

TiB2-based composites with 10% of Ti and Fe additives were consolidated using

pressureless sintering. Specimens with 7%Ti+3%Fe additives showed better densification

due to the formation of liquid phase during sintering. Best results were obtained for a

sintering temperature of 1650°C. Pre-alloying of Fe and Ti before addition to TiB2 powder

significantly improved the densification. The milling time has also a marked influence on

densification and on the properties of the sintered TiB2 specimens. A maximum relative

bulk density of 91% and maximum bending strength of 300 MPa were achieved with

specimens milled for 30 min and sintered at 1650°C for 1h. The micrographs of specimens

milled for 30 min reveal a uniform crack free microstructure with an even distribution of

pores while those milled for longer times show the presence of numerous cracks in the

specimens. The sintered specimens showed some resistance in liquid aluminum although

more tests are needed. The resistance of specimens against aluminum infiltration and

erosion are under investigation.

5.7 Acknowledgement

The authors wish to acknowledge the kind collaboration of the technicians of the Dept.

Mining, Metallurgical and Materials Engineering of Laval University and of Sylvio Savoie

from Hydro-Quebec. The financial support of this project was provided by Hydro Quebec

and the Natural Sciences and Engineering Research Council of Canada (NSERC). The

research project was also partially financed by the “Fonds Québécois de la Recherche sur la

Nature et les Technologies (FQRNT)” via the Aluminum Research Centre – REGAL.

Chapter 6:

Investigating the potential of TiB2–based composites with

Ti and Fe additives as wettable cathode


1; Dominique Dubé

1; Robert Schulz

2



This article was presented in Thermec 2011 international conference and was published in

both the journal of the: Materials Science Forum, 2012, Vol. 706-709, P. 655-660; and the

journal of the: Advanced Materials Research, 2012, Vol. 409, P. 195-200.

73

6.1 Résumé

Dans ce travail, les matériaux céramiques poreux à base de TiB2 ont été frittés en phase

liquide à l'aide de Ti et Fe comme additifs métalliques afin d'effectuer le frittage aux

températures inférieures de 1700°C. Il a été montré que les paramètres de procédé, y

compris la durée du broyage avaient un effet considérable sur les propriétés finales des

spécimens frittés et leur comportement dans l'aluminium fondu. Les études de

microstructures ont été réalisées en utilisant le microscope optique, le MEB et la

microsonde électronique. Il a été trouvé que les spécimens avec une microstructure

uniforme et sans fissure pourraient être produits en utilisant des poudres pré-mélangées et

broyées pour une durée aussi courte que 30 min, avant le pressage et le frittage. Le test

mouillage avec la goutte a été effectué sur les échantillons broyés pendant 30 et 240

minutes. Leur interaction avec l'aluminium fondu a également été étudiée. Il a été trouvé

que le temps de broyage de 30 min permet à une meilleure conductivité électrique, ainsi

qu’une bonne mouillabilité et une bonne stabilité dans l'aluminium liquide.

6.2 Abstract

In this work, porous TiB2 ceramics were consolidated by pressureless sintering method

using metallic Ti and Fe as additives in order to perform sintering at temperatures lower

than 1700°C. It was shown that processing parameters including milling time of the starting

mixture had a considerable effect on final properties of sintered specimens and their

behaviour in molten aluminum. Microstructural studies were carried out using optical

microscope, SEM and EPMA. It was found that specimens with uniform and crack-free

microstructure could be produced using the pre-mixed powders milled for as low as 30 min

prior to compaction and sintering. Sessile drop test was performed on the specimens milled

for 30 and 240 minutes. Their interaction with molten aluminum was also studied. It was

found that 30 min milling time resulted in better electrical conductivity, wettability and

stability in liquid aluminum.

74

6.3 Introduction

TiB2-based ceramics have been studied since 1950’s as alternatives for carbon cathodes 21

.

Despite their widespread use for more than a century in aluminum smelting industry,

carbon cathodes are not ideal materials from the energy efficiency point of view 10

. The

issue with carbon cathodes is that they are not wettable by molten aluminum which results

in an increase in energy consumption for the production of primary aluminum. As they are

not wetted by aluminum, the gap between the cathode surface and aluminum pad increases

the electrical resistivity. In order to have a better contact, it is necessary to increase the

thickness of the pad. Moreover, a relatively large anode-cathode distance (ACD) must be

maintained to avoid electrical shortcut due to turbulence generated by Lorentz forces. The

gap and the superior ACD increase the resistivity and energy loss in the cell 10, 16

.

Improving the energy efficiency in energy-intensive industries such as aluminum

production has become critical due to the rising cost of energy. Wettable cathodes represent

therefore an interesting solution for saving energy 12

.

Numerous researches have been focused on the fabrication of wettable cathodes, and a wide

range of processing methods and materials including TiB2-based materials have been

proposed 15, 22

. However, none of the proposed solutions has yet been used in commercial

scale either because they are not economically feasible or because it is technically too

difficult to fabricate large cathodes. From the technical point of view, using highly dense

TiB2 components is not a viable solution because molten aluminum reacts with grain

boundaries after extended immersion time causing internal stresses at grain boundaries of

TiB2 due to volume expansion of reaction products. These stresses eventually result in

rapid crack propagation and failure of immersed components 30, 45, 84

. Considering these

issues, an economical process must be developed to fabricate large near net shape and

porous TiB2-based cathodes.

The pressureless sintering of TiB2-based composites for wettable cathode application was

previously investigated 1. The effect of composition, sintering temperature, pre-alloying of

additives and milling time on density, bending strength and microstructure were studied 1.

75

It was found that better density and bending strength could be achieved using pre-alloyed

Ti-Fe additive of near eutectic composition (70:30 Ti:Fe wt. ratio) and sintering

temperature of 1650°C. It also has been shown that the milling time of powder mixtures

had an important effect on the final properties of the specimens. Increasing the milling time

from 10 to 30 min increased the relative density from 84 to 91% and the bending strength

from 197 to 300 MPa. Further increase of milling time reduced these properties and a

milling time of 240 minutes resulted in relative density of 87% and bending strength of 192

MPa. In addition, shorter milling time reduces the production cost as well as the risk of

oxidation of starting powders which are important from industrial and technical points of

view.

In this work, the effect of milling time on microstructure, electrical conductivity, and

behavior in molten aluminum were studied considering their application as wettable

cathodes. Wettability of TiB2-based composites and its stability in molten aluminum were

investigated.


Titanium (Ti >99.8%) and iron (Fe >99.9%) powders (Atlantic Equip. Eng. Inc) were

mixed in 70:30 Ti:Fe wt. ratio, compacted and then sintered at 1150°C for 1 h. The

resulting pellets were subsequently crushed and milled for 1 h using high energy ball

milling in order to obtain the pre-alloyed 7Ti3Fe additive in powder form (particle size

distribution in Fig.1). Commercial TiB2 powder (99.7 % pure, Atlantic Equip. Eng. Inc)

with particle size ranging between 2 and 10 µm was mixed with 10% of the pre-alloyed

7Ti3Fe powder and then milled using high energy ball mill (SPEX 8000) in stainless still

vial and balls (balls/powder ratio: 4/1). The milling was performed for 10, 30, 60, 120 and

240 min to study the effect of milling time on properties of TiB2-based composites. Particle

size distribution of the milled powders was measured using laser scattering particle size

analyzer (LA-900, HORIBA).

76

The milled mixtures were compacted in a uniaxial die under 150 MPa pressure and then

sintered in a tube furnace for 1 h at 1650°C under Ar-5%H2 atmosphere (heating and

cooling ramp: 6°C/min). Specimens were cross sectioned with a diamond saw and polished

down to 0.1 µm surface finish with successively finer diamond abrasives.

Microstructural studies and chemical analysis were carried out using optical microscope,

scanning electron microscope equipped with energy dispersive X-ray spectroscopy (EDX;

PGT Avalon) and electron probe microanalysis (SX-100 CAMECA microprobe).

For measurement of electrical resistivity at ambient temperature, specimen was placed

between two copper plates, and pressure was applied (10 MPa) to obtain a good electrical

connectivity. The resistance was measured and the electrical resistivity (ρ) of the specimen

was calculated using the following equation:

( 6-1)

where R is the electrical resistance, A is the contact surface area and l is the specimen’s

thickness.

The wettability of specimens by molten aluminum was investigated using sessile drop

technique. An aluminum piece (90 mg) was placed on the surface of specimen that was

then heated at 960°C under vacuum (1.2x10-8

bar). The wetting angle was monitored using

an instant imaging system. The contact angle was measured from these images as a

function of time. Starting time (t=0) was considered once a spherical liquid aluminum drop

was formed over the surface.

The chemical stability of specimens in liquid Al was also evaluated. The specimen was

glued (Aremco Ceramabond 571) to the tip of an alumina rod and was then inserted into

liquid aluminum under protecting argon flow at 960°C. The exposure durations were 1h

and 24 h. The specimen was removed after the test, cross sectioned using a diamond saw,

77

polished according to abovementioned method and analyzed by EPMA in order to examine

the microstructure and evaluate their interaction with liquid aluminum.


Cumulative particle size distributions of the starting mixtures milled for different period of

time are shown in Figure 6–1. As shown in this diagram, pre-alloyed additive contains

large particles (>30 µm), but most of them were eliminated after 10 min of milling.

Increasing the milling time to 30 min completely eliminated these large particles while the

volume fraction of particles smaller than 4 µm increased. By further increasing the milling

time, large particles reappeared and the volume of the fine particles decreased due to the

formation of large agglomerates.

Figure 6–1- Cumulative particle size distribution diagram of pure TiB2, pre-alloyed

7Ti3Fe, and mixtures after different milling times (M10 e.g. means mixed powder

milled for 10 min)

78

Microstructures of the sintered specimens with different milling time of starting powders

are shown in Figure 6–2. After only 10 min of milling (Figure 6–2-a), large and unevenly

dispersed particles of additives remain visible in the microstructure. The liquid phase did

not infiltrate between TiB2 particles to enhance densification. As a result of these

phenomena, a highly porous microstructure with non-uniform distribution of additives was

formed.

The specimen milled for 30 min (Figure 6–2-b) had lower porosity and uniform

microstructure with even distribution of additives. Their particle size distribution resulted

in a better compaction and therefore sinterability of the powder. The microstructure of the

specimen milled for 240 min (Figure 6–2-c) showed a higher level of porosity compared to

M30, large cracks and segregated additives due to the presence of large agglomerates in the

powder.

The M30 and M240 specimens were then selected for further investigations. The

microstructure of M30 was quite uniform whereas that of M240 was cracked and more

porous. The nature of phases in microstructure of composites as determined by EDX

analysis is given in Figure 6–2 as follows: A: TiB2, B: TiFe, C: αTi, D: TiFe2.

Figure 6–2- Micrograph of specimens milled for a) 10 min, b) 30 min, c) 240 min and

sintered at 1650°C for 1h. (Phases in the microstructure are A: TiB2 ; B: TiFe ; C: α-Ti ;

D: TiFe2)

The electrical conductivity of specimens is an important parameter considering their

application as wettable cathodes. The measured electrical resistivity of M30 and M240

specimens were approximately 54 and 243 µΩ.cm, respectively. The electrical conductivity

79

of M30 was much higher than that of M240 which could be attributed to its lower porosity,

absence of crack as well as more uniform distribution of additives. Although the electrical

resistances of these two specimens were much different, they were well below that of

graphite which is about 1000 µΩ.cm 27

showing that this TiB2-based composite is more

conductive than the usual carbon cathodes.

Figure 6–3- Photo taken from liquid Al drop on the surface of M30 specimen.

The sessile drop technique was used to investigate the wettability of the specimens by

liquid aluminum. Figure 6–3 shows a typical image of a liquid aluminum drop formed over

the surface of M30 specimen during measurement. The wetting angle, θ, was measured

using these images. The wetting angle of M30 and M240 as a function of elapsed time is

shown in Figure 6–4. It shows that the wetting angle of both specimens decrease with time.

However, the curve for M30 presents continuous and smooth changes of wetting angles

with time, whereas for M240, the changes of wetting angle versus elapsed time exhibit a

step-like curve. The wetting behaviour of these specimens can be explained by differences

in their microstructure. For M30 specimen, during the sintering process, liquid phase was

formed upon melting of 7Ti3Fe additives. That liquid entered inside small pores and

applied capillary forces over pore walls resulting in particle rearrangement and

densification. However, for M240 specimen, strong agglomerates composed of intimate

mixture of TiB2 and Ti-Fe particle were formed. Inside these agglomerates, densification

occurred early during sintering leaving large cracks around the agglomerates. During the

sessile drop test of M240 specimen, the sudden reduction of the wetting angle corresponds

80

to a quick penetration of liquid aluminum into these cracks. Generally, the penetration of

liquid aluminum in cracks and pores is a sign of good wettability of the composite.

Figure 6–4- Wetting angle of liquid Al on M30 and M240 specimens as a function of

elapsed time

Another important parameter for wettable cathode materials is their stability during

immersion in molten aluminum. Figure 6–5 shows the EPMA micrographs of M30

specimen after being exposed to molten aluminum for 1 and 24 h. After 1 h, aluminum

wetted the surface as shown in Figure 6–5-a but started to penetrate inside the specimen

only in some areas. This is in accordance with the wetting curve of M30 specimen (Figure

6–4) which shows that complete wetting occurred after about 1 h. As it could be observed

from Figure 6–5-b and c, aluminum started to penetrate inside the specimen and Fe was

dissolved in aluminum and washed out from that area.

For M240 specimen, after 1 h of immersion, aluminum penetrated inside the specimen.

Figure 6–6-a and b shows the penetrated zone. EPMA investigation (Figure 6–6-c) revealed

that liquid aluminum dissolved iron and pushed it further inside the specimen. The

81

concentration of iron in front of the aluminum-penetrated interface is much higher than the

original concentration of iron in the specimen.

The microstructural investigation of cross sections of specimens after 24 h of aluminum

exposure revealed that liquid aluminum penetrated completely inside both specimens

(Figure 6–5-e and Figure 6–6-e). It is interesting to note that specimens maintained their

shape after aluminum penetration and no cracking or failure occurred. It confirms that TiB2

particles sintered together very well and their junctions are resistant to liquid aluminum.

Therefore, this porous TiB2 composite meets an important requirement for wettable cathode

which is resistance to aluminum penetration without cracking.

Figure 6–5- Micrograph of M30 specimens exposed to liquid Al at 960°C: for 1h: a)

BSE micrograph of contact area, b) Mapping of Al element, c) Mapping of Fe element;

and for 24h: d) BSE micrograph of contact area, e) Mapping of Al element, f) Mapping

of Fe element

82

Figure 6–6- Micrograph of M240 specimens exposed to liquid Al at 960°C: for 1h: a)

BSE micrograph of Al penetration zone, b) Mapping of Al element, c) Mapping of Fe

element; and for 24h: d) BSE micrograph of Al penetration zone, e) Mapping of Al

element, f) Mapping of Fe element

The mapping of elemental aluminum in Figure 6–5-e and Figure 6–6-e reveals that less

aluminum penetrated in M30 specimen than in M240 specimen. The lower aluminum

penetration in M30 is a result of the absence of crack in its microstructure. Figure 6–5-f and

Figure 6–6-f show that most of the iron was washed out after complete penetration of

aluminum and only a little trace of iron remained. These results suggest that 30 min of

milling had an important effect on the refinement of particle size and thereby a better

sinterability of TiB2 which led to higher electrical conductivity, enhanced wettability and

stability in liquid aluminum.

The detailed investigation on wettability and aluminum interaction of M30 specimen as

well as its stability in liquid aluminum for prolonged time is ongoing and the results will be

reported later.

83

6.6 Conclusions

TiB2-based porous composites were consolidated from starting powders milled for different

time. The particle size distribution and the microstructure of sintered parts were

significantly influenced by milling time. It was found that 30 min of milling eliminated

large particles and produced a higher volume of fine particles which resulted in a uniform

microstructure after sintering. The electrical conductivity of this specimen was higher than

that of graphite having the highest conductivity among the carbon cathodes. Liquid

aluminum completely wetted the produced TiB2 composites but M30 had a better

wettability due to its uniform and crack-free microstructure. The M30 was stable in liquid

aluminum after 1 h of exposure. Aluminum penetrated completely into M30 and M240

specimens after 24 h, however, the specimens maintained their forms and no sign of crack

propagation and failure was observed. M30 specimen showed better stability in liquid

aluminum due to the absence of cracks in its microstructure and hence lower liquid

aluminum penetration.

Chapter 7:

Interaction of molten aluminum with porous TiB2–based

ceramics containing Ti-Fe additives


1; Dominique Dubé

1; Robert Schulz

2



This article was published in the: Journal of the European Ceramic Society, 2012, Vol. 32,

Issue 4, P. 937-945.

85

7.1 Résumé

Dans cette étude, la mouillabilité et les interactions des composites poreux à base de TiB2

avec l'aluminium liquide ont été examinées. Les composites de TiB2 ont été frittés en phase

liquide avec des additifs Ti et Fe. Ces composites possèdent une bonne mouillabilité par

l'aluminium liquide. Lors de l'infiltration par l’aluminium liquide, des additifs de Fe et de

Ti sont dissous. Des composés intermétalliques contenant le Ti, Fe et Al sont formés à la

profondeur de pénétration. Puisque ces phases ont des points de fusion supérieure à la

température du bain liquide (960°C), la solidification isotherme a lieu au cours de la

pénétration de l'aluminium fondu. L’aluminium liquide n’attaqué pas le "squelette solide"

des spécimens de TiB2 et aucun signe de gonflement ou de fissuration n’a été détecté.

7.2 Abstract

In this study, the wettability and interaction of porous TiB2-based composites with liquid

aluminum has been investigated. TiB2 composites were consolidated with Ti and Fe

additives using pressureless sintering. The composites show good wettability with respect

to molten aluminum. During liquid infiltration, Ti and Fe additives are dissolved.

Intermetallic compounds containing Ti, Fe and Al are formed within the penetration depth.

Since these phases have melting points higher than the experiment’s temperature (960°C),

isothermal solidification takes place during the penetration of molten aluminum. Liquid

aluminum does not seem to attack the solid skeleton of the TiB2 specimens and no signs of

swelling or cracking were detected.

86

7.3 Introduction

Molten aluminum reacts with almost all materials except with some ceramics such as

borides which exhibit very good stability in liquid Al 25

. Due to its intrinsic stability, high

electrical conductivity and good wettability with respect to liquid aluminum, TiB2 has been

extensively investigated as an alternative for graphite cathodes in aluminum smelters. It has

been mentioned that by replacing graphite cathodes by TiB2 electrodes, more than 10% of

the energy required to produce aluminum could be saved 12, 13, 22

.

However, the fabrication of large and dense TiB2 parts is quite complicated 32, 71

due to the

poor sinterability of the material. Consolidation of pure TiB2 requires very high

temperatures. At temperatures higher than 1700°C however, exaggerated grain growth

occurs leading to materials with poor mechanical and thermal shock resistances 46, 47

. In

TiB2 ceramics consolidated with current methods, liquid Al is able to penetrate the grain

boundaries and this leads to cracking of components due to the low toughness of materials

22, 30, 85. Micro-cracks can result from the anisotropic thermal expansion of TiB2 or from the

formation of phases with higher molar volumes 45

when liquid Al reacts with grain

boundary impurities.

The addition of sintering aids is a way to lower the consolidation temperature and hence

prevent the exaggerated grain growth. The use of iron, chromium and nickel as sintering

aids for TiB2 has been reported 51, 52, 84

. In a previous work, 7wt % Ti and 3wt % Fe were

added to TiB2 particles to promote liquid phase sintering during consolidation 1, 2

. Sintering

of specimens at 1650°C for 1h resulted in a relative density of 91% and a bending strength

of 300 MPa. No attempt was made to achieve higher densities because it was expected that

the presence of uniformly distributed small porosities could help to prevent the propagation

of micro-cracks when samples are in contact with molten aluminum. However, porosities

ease melt penetration since liquid Al wets TiB2 quite well 86, 87

.

In this work, the reaction of porous TiB2 composites containing 7 wt% Ti and 3 wt% Fe

additives with molten aluminum was investigated. Microstructural characterizations were

87

performed in order to understand the reaction mechanisms between the composites and

liquid aluminum.


The TiB2 composites were prepared by mixing commercially pure TiB2 powder (2-10 µm)

with 10 wt% of pre-alloyed additives in powder form (7 wt% Ti and 3 wt% Fe). The

additives (a mixture of Ti and FeTi phases) were then milled with TiB2 for 30 min using a

high-energy ball mill. The powder was compacted with a uniaxial pressure and sintered for

1 h at 1650°C under an Ar-5%H2 atmosphere 1.

Sessile drop tests were used to investigate the wettability of the specimens with liquid

aluminum. An Al pellet (0.09 g) was used for the experiments. The surface of the Al pellet

was polished prior to the test in order to reduce surface oxides. The pellet was placed on

top of the specimen in a resistance tube furnace under high vacuum (10-3

Pa). The furnace

was heated up rapidly to 960oC (corresponding to the operating temperature of aluminum

electrolysis cells) and kept at this temperature. A light was fixed at one end of the tube and

the image of the Al drop over the specimen was recorded at the other end during the

experiment. A software was used to evaluate the contact angle between the specimen

surface and the aluminum drop from the recorded images. The reported contact angle is the

average of left and right side angles. The time t=0 was set when the specimens’ temperature

reached 700°C approximately and a spherical liquid Al drop formed over the surface of the

sample. After the experiment, the furnace was cooled down at a rate of about 15°C/min

below the melting point of pure aluminum.

The surface of some TiB2 samples was polished before the tests in order to remove surface

contamination, especially oxides. Polishing was carried out using a diamond abrasive (6

µm) followed by cleaning with isopropanol in an ultrasonic bath. Therefore, experiments

were performed on polished surfaces (identified as S1-P, P: Polished) as well as on as-

sintered surfaces (identified as S2-AS, AS: As Sintered).

88

For one specimen, the sessile test was performed using a bigger Al pellet (0.18 g). In this

case, the early stage of the liquid infiltration process was investigated by interrupting the

experiment after t=100 min. The sample was identified as S3-PP (PP: Partially Penetrated).

In a last experiment, a polished specimen was subjected to a complete sessile drop test then,

another aluminum piece was placed on the same surface and a second sessile drop

experiment was conducted (S4-2D, 2D: 2nd

Drop).

Reactions between liquid aluminum and ceramics were investigated by examining the

cross-sections of the specimens. The samples were mounted in epoxy resin, cut using a

diamond saw and polished down to 0.1 µm. The final polishing was performed using a 0.05

µm alumina suspension. The microstructure of specimens was investigated using a

scanning electron microscope (SEM) equipped with an energy dispersive X-ray detector

(EDX). Electron probe microanalysis (EPMA) was performed for elemental mapping of

aluminum, titanium, iron and oxygen. Focused ion beam (FIB) was used to prepare samples

for transmission electron microscopy (TEM) characterization. Phase identification was

carried out based on selected area electron diffraction (SAED) pattern analysis.


7.5.1 Sessile drop tests

Figure 7–1 shows images of a molten aluminum drop on the surface of a polished specimen

(S1-P) during a sessile drop test after different time intervals. The measured average

contact angles (average between the right and left contact angle) are shown below each

figure. During the experiment, the Al drop spreads over the surface and penetrates inside

the specimen. After 9 min, the specimen temperature was about 870°C and there was no

visible wetting. After 22 minutes, the temperature reached 940°C, wetting occurred and the

contact angle was about 85 degrees. The measured contact angles after 30 and 50 min were

29 and 6 degrees, respectively.

89

Figure 7–1- Photos from the contact between a liquid Al drop and a polished specimen’s

surface during a sessile drop test

Figure 7–2 shows the average contact angle as a function of the elapsed time for the

polished (S1-P) and the as-sintered specimen (S2-AS). Two steps can be distinguished in

the case of the S2-AS sample. During the first step (incubation period), the contact angle

decreases very slowly probably because of surface oxides and reaches 120° after about 140

min. In a second step, the drop starts spreading over the surface rapidly and penetrates the

specimen. The contact angle decreases at a much higher rate and reaches 5° after 180 min.

90

Figure 7–2- Average contact angle versus time during sessile drop tests for the as

sintered and polished specimens.

For the S2-AS specimen, the shape of the curve in the second step is similar to that of the

polished specimen for the same range of wetting angles. This suggests that the polishing

eliminates surface oxides and impurities and reduces significantly the incubation period.

7.5.2 Early Stage of Interaction

To study step one and the early stage of the interaction between liquid aluminum and the

as-sintered specimen, an experiment was carried out in which the test was interrupted after

100 min (specimen S3-PP). Figure 7–3 shows the backscattered electron (BSE) micrograph

from the cross section of the specimen-aluminum interface revealing the penetration of Al

inside the specimen.

91

Figure 7–3- BSE micrograph from the cross section of a specimen with partial

penetration of Al (S3-PP)

To investigate if counter-diffusion of elements from the TiB2 specimen towards liquid Al is

taken place, the cross section of the aluminum drop on top of the substrate was analyzed.

Figure 7–4-a shows small white particles found inside the Al matrix. Semi-quantitative

EDX analysis revealed that the composition of these particles corresponds to TiAl3 phase.

Particles of TiAl3 are usually found in aluminum either as a needle-like structure or in the

form of chunky equiaxed particles 29

. These results suggest that the Ti additives in TiB2

dissolve and diffuse in the Al droplet during the test and precipitate as small TiAl3 particles

during solidification 88-91

.

Chemical mapping was performed in selected areas within the Al matrix. A Fe map is

shown in Figure 7–4-b. Contrary to Ti, which was mostly found in the form of Ti-Al

particles, iron precipitated at grain boundaries during solidification.

These results indicate that the metallic Ti and Fe additives are dissolved in the liquid Al

during the sessile drop tests and precipitate in the form of Ti and Fe aluminum phases upon

solidification.

92

Figure 7–4- Partially penetrated test (S3-PP) a, b) BSE micrograph and EDX analysis of

TiAl3 particles formed inside Al drop, c,d) BSE micrograph and EPMA maps of Fe

element within Al drop

The interface region between the Al droplet and the S3-PP specimen’s surface was also

analyzed using EPMA. Figure 7–5-a) and b) show the elemental distribution of aluminum

and oxygen respectively in this region. The mapping of oxygen indicates a 20 µm thick

oxide-rich layer near the interface. Comparison of this map with that of Al reveals that

oxygen is present in the form of aluminum oxide. In the as-prepared specimen, oxygen is

most likely picked-up by the Ti additives during the sintering process. During the sessile

drop test, liquid Al reduces the Ti oxides to form aluminum oxide. A similar observation

was also reported by Pettersen 84

when hot pressed TiB2 specimens containing Ti additives

were put in contact with liquid aluminum. Ti reacts with the oxygen present on the surface

of TiB2 particles and with the oxygen from the atmosphere to form trigonal Ti2O3

93

predominantly. He also mentioned that the reduction of this titanium oxide with liquid

aluminum is thermodynamically feasible at temperatures around 1000°C. According to our

results, titanium oxide is present on the free surface of the particles and in the porosities.

The reaction of titanium oxide with liquid aluminum may result in some reaction products

with higher molar volume. However, the presence of about 9% of porosity in the specimen

provides enough space for these volume changes and thereby prevents the detachment of

TiB2 grains and specimen swelling.

The delay observed before the wetting takes place in the S2-AS case (Figure 7–2) can

therefore be explained by the presence of a thin oxide layer on the surface of sintered TiB2.

However, this surface oxide did not prevent full wetting of specimen ultimately.

Figure 7–5- Elemental distribution of aluminum and oxygen at the drop-specimen

interface after the partially penetrated test (S3-PP)

7.5.3 Later Stage of Interaction

By examining the cross section of specimens after complete sessile drop test, it was found

that the Al penetration resulted in the formation of four different zones. These zones are

identified as zones 1 to 4 in Figure 7–6. The zone 1 showed a high level of porosity with

almost no sign of metallic additives. In zone 2 and zone 3, the pores contain Ti and Fe

aluminum phases in addition of Al. These zones will be discussed in more details later in

94

this work. Zone 4 remained intact. It represents the microstructure of the specimen before

the sessile drop test.

Figure 7–6- SEM micrograph of the cross section of a S1-P specimen after the sessile

drop test

Figure 7–7 shows elemental line scans of Ti, Fe and Al through the cross section of the S1-

P specimen. The approximate boundaries between the different zones are indicated by

dotted-lines. The concentration of aluminum in zone 1 and 2 varies from point to point

depending on pore location and density. Its average concentration in zone 2 is considerably

higher than that in zone 1. It seems that the Al concentration decreases in zone 3 and falls

to zero in zone 4. There is a very low concentration of Fe in zone 1. Fe was detected at

some points in zone 2 and its concentration increases and reaches a maximum in zone 3

before falling to the bulk Fe additive concentration of 3 wt% in zone 4. The Ti distribution

in Figure 7–7 shows only small variations since the signal comes from both TiB2 particles

and the Ti metallic additives.

95

Figure 7–7- Elemental line scans carried out through the thickness of the specimen after

a sessile drop test at 960°C. (Depth=0 is corresponds to the aluminum-specimen

interface at the beginning of the test)

Figure 7–8 shows elemental maps of Ti and Al near the interface between zone 1 and 2.

The maps seem to indicate that the Ti concentration in the pores where Al is present is

lower in zone 1 (darker blue) than in zone 2 suggesting that the Ti additives have been

washed out from zone 1 during the liquid infiltration. The simultaneous presence of Ti and

Al in the pores of zone 2 suggests the presence of Ti-Al phases in this zone.

96

Figure 7–8- Mapping of aluminum and titanium showing the transition between zone 1

and zone 2

Figure 7–9 shows the mapping of Fe and Ti near the interface of zones 2 and 3. The

concentration of Fe in zone 3 is much higher than in zone 2 in agreement with Figure 7–7.

The Ti concentration in pores is lower (darker blue) in zone 3 than in zone 2 and in these

pores, Fe is present instead of Ti. The simultaneous presence of Fe and Al in the pores of

zone 3 suggests the presence of Fe-Al phases in zone 3.

Figure 7–9- Mapping of iron and titanium revealing the transition between zone 2 and

zone 3

97

Finally, Figure 7–10 shows elemental maps near the transition between zones 3 and 4. The

Al concentration decreases down to zero from zone 3 to zone 4 in agreement with the line

scan of Figure 7–7. The Ti content between the TiB2 grains or in the pores seems to

increase as we move from zone 3 to zone 4 in agreement with what as been said previously

regarding zone 3.

Figure 7–10- Mapping of aluminum and titanium between zone 3 and zone 4

Figure 7–11-a) and b) show SEM images of the specimen in zone 2 and 3 respectively. In

zone 2, the phase identified as P1 contains Ti and Al. Studies on the Ti-Al system have

shown that at temperatures between 700 to 1000°C, TiAl3 is formed prior to any other

titanium aluminide phases. It was also reported that this intermetallic compound can

dissolve 1.2 at.% Fe at 1000°C 81

. To verify the exact nature of this phase, TEM analysis

was performed. A TEM micrograph of such a phase is shown in Figure 7–12-a). EDX

analysis and the SAED pattern (Figure 7–12-b) taken from the [010] zone axis of P1

confirm that this phase is TiAl3. This titanium aluminide phase precipitates from Ti

saturated liquid Al in the pores between the TiB2 particles when the liquid phase enters

zone 2.

98

Figure 7–11- BSE micrographs of zone 2 and zone 3. Arrows show the presence of TiAl3

(P1) and Fe-Al compound (P2)

Figure 7–12- TiAl3 precipitate in zone 2, a) Transmission electron micrograph, b) SAED

pattern from the [010] zone axis of TiAl3

The phase identified as P2 in zone 3 (see Figure 7–11-b) contains Al and Fe with traces of

Ti. The phase P1 is also detected in zone 3. Studies have shown that Fe4Al13 and Fe2Al5

phases are more likely to precipitate in the Al rich side of the phase diagram 92

. Although

the growth rate of Fe2Al5 is higher, the formation of Fe4Al13 is dominant in Al-rich

interfaces. Therefore, it is highly probable that the P2 phase is Fe4Al13. The Ti solid

99

solubility limit in that phase is 6.5 at%. Ti can partially replace Fe in certain

crystallographic sites 93, 94

.

TEM analysis was also performed on P2 observed in zone 3 to confirm the nature of the

Fe4Al13 phase (see Figure 7–13-a) and b)). The SAED pattern shown in Figure 7–13 b) is

characteristic of the Fe4Al13 structure.

Figure 7–13- Fe4Al13 phase precipitated in zone 3, a) Transmission electron micrograph,

b) SAED pattern from the [010] zone axis of Fe4Al13

7.5.4 Reaction Mechanism

Based on these results, the reaction mechanism between the liquid Al drop and the

specimen could be described as follows. When liquid aluminum forms over the polished

surface of the specimen, it interacts with surface oxides and three other major

microstructural features: TiB2 particles, pores, and the metallic additives. After the

reduction of surface oxides by liquid aluminum and wetting of the TiB2 particles, the

metallic additives are dissolved and the liquid enters the pores. At the beginning of the test,

the liquid is pure aluminum. As it enters the porous structure, the concentration of Ti and

Fe in the liquid gradually increases. The liquid has a solubility limit for Fe and Ti above

which new phases will precipitate. Based on the binary phase diagrams of Al-Fe and Al-Ti,

100

the solubility limit of Fe and Ti in liquid aluminum at 960°C is about 17 and 2 wt%,

respectively. Figure 7–14 shows the ternary phase diagram of Al-Ti-Fe. In this work, the Ti

to Fe weight ratio was 7/3 and therefore, the concentration of Ti in the liquid should

increase more rapidly than that of Fe.

Since the Fe and Ti content in the liquid is below the solubility limit in zone 1, the additives

in this zone are simply dissolved while the aluminum is penetrating into the specimen.

Once the concentration of Ti reaches the solubility limit near 2 wt%, TiAl3 phase starts to

precipitate forming zone 2. In this zone, the concentration of Fe is still below its solubility

limit (17 wt%), thus Al continues to dissolve the additives and penetrates inside the

specimen until the solubility limit of Fe is reached. This marks the beginning of zone 3

where the Fe4Al13 phase is formed. At this point, the TiAl3 phase continues to precipitate

and this explains the presence of both P1 and P2 phases in zone 3. By the precipitation of

Fe4Al13 and TiAl3, the liquid aluminum is gradually consumed. In the final stage of the

reaction, the liquid aluminum acts as a “transient liquid phase” 95

which means that after

the diffusion of iron and titanium in the remaining liquid, its transforms gradually to solid

phases via isothermal solidification processes.

101

Figure 7–14- Isothermal section of the Al-Fe-Ti phase diagram at 1000°C 88

The formation of zone 3 where the liquid isothermally solidifies by the formation of

aluminide phases raises the question on whether this zone could act as a barrier against

further aluminum penetration. This hypothesis was put to test by performing an additional

sessile drop test on the specimen formerly subjected to this test. The cross section

micrograph of S4-2D specimen is shown in Figure 7–15. Comparing the cross section of

S4-2D with that of the S1-P specimen, we see that the penetration of the second aluminum

drop lead to an increase of the width of zone 1, zone 2, and zone 3.

102

Figure 7–15- SEM micrograph of cross section of specimen subjected to two subsequent

sessile drop tests

This observation suggests that the second drop follows the same reaction mechanism as the

first one resulting in the dissolution of solid phases and precipitation of new phases until

the solubility limits of Fe and Ti are reached. Zone 3 cannot act as a barrier to the

penetration of liquid aluminum.

However, it is important to note that the TiB2 specimens remained intact despite the

penetration of liquid aluminum and the dissolution of the metallic additives. This could be

attributed to the stability of the grain boundaries (bridges) between the TiB2 particles. The

boundaries formed between the TiB2 particles during sintering with Ti-Fe additives are

quite stable in liquid Al. The specimens did not show any swelling or crack in the

microstructure. The expansion of the structure of TiB2 sintered using transitional methods

has already been reported in the literature 96

. Further investigations concerning the stability

of these TiB2 composites in liquid Al will be presented in an upcoming publication.

7.6 Conclusion

The interaction of TiB2-base porous ceramics with liquid Al was investigated. The

reactions were occurring faster on polished surfaces because of the removal of the surface

103

oxide layer. The penetration of the liquid Al drop resulted in the formation of three distinct

zones. Only Al was found in the first zone, TiAl3 was found in the second zone and both

TiAl3 and Fe4Al13 phases were found in the third zone. When penetration of liquid Al into

the material occurs, the metallic additives are dissolved and their concentration increases

gradually. Once the Ti saturation limit is achieved, the TiAl3 phase starts to precipitate

inside the pores (Zone 2). The residual liquid continues to penetrate and dissolve the

additives until the saturation point of Fe is reached leading to the precipitation of Fe4Al13

(Zone 3). Dissolution and precipitation will continue up to the complete depletion of the

liquid phase (Isothermal solidification). Liquid aluminum did not seem to alter the solid

TiB2 skeleton of the specimen and no sign of swelling or cracking was detected.

7.7 Acknowledgements

The authors wish to acknowledge the kind contribution of the technicians of the Dept. of

Mining, Metallurgical and Materials Engineering of Laval University, of Sylvio Savoie

from Hydro-Quebec and of Jean-Philippe Masse from École Polytechnique de Montréal for

TEM analysis. The financial support of this project was provided by Hydro Quebec and the

Natural Sciences and Engineering Research Council of Canada (NSERC). The research

project was also partially financed by the “Fonds Québécois de la Recherche sur la Nature

et les Technologies (FQRNT)” via the Aluminum Research Centre – REGAL.

Chapter 8:

Pressureless sintering of TiB2-based ceramics with Ti-Fe

additives: sintering mechanism and stability in liquid aluminum


1; Dominique Dubé

1; Robert Schulz

2



This article was published in: Advanced Engineering Materials journal, 2012, Vol. 14, P.

802–880.

105

8.1 Résumé

Le développement d'un procédé approprié pour la fabrication de cathodes mouillables à

base de TiB2 pour l’électrolyse de l'aluminium demeure un défi depuis plus d'un demi

siècle. Dans ce travail, les céramiques à base de TiB2 ont été consolidées via le frittage sans

pression à l'aide des additifs de Ti, Fe et Ti-Fe. La microstructure, les propriétés physiques

et mécaniques ainsi que l'interaction et la stabilité du matériau dans aluminium liquide ont

été examinés. Il a été montré que les spécimens frittés avec l’additive Ti-Fe ont une

excellente stabilité dans l'aluminium liquide car le squelette solide de TiB2 a conservé son

intégrité et la résistance après 5 jours d’exposition dans l'aluminium liquide à 960°C.

L’analyse par le MET a révélé que la formation de ponts entre les particules de TiB2 pur est

la cause de cette bonne résistance dans l'aluminium fondu. Un mécanisme de frittage a été

proposé pour la consolidation de TiB2 avec l’additif de Ti-Fe. Le matériau céramique à

base de TiB2, fritté avec un alliage de Ti-Fe est donc suggéré comme un matériau

potentiellement fiable pour l'application en tant que la cathode mouillable pour électrolyse

de l'aluminium.

8.2 Abstract

The development of a proper processing method for the fabrication TiB2-based wettable

cathodes for aluminum electrolysis has been challenging for more than half a century. In

this work, TiB2-based ceramics were consolidated via pressureless sintering using Ti, Fe

and Ti-Fe additives. The microstructure, physical and mechanical properties as well as the

interaction and the stability of the material in liquid aluminum were investigated. It was

shown that specimens sintered with a Ti-Fe additive have excellent stability in liquid

aluminum as the solid TiB2 skeleton maintained its integrity and strength after 5 days of

exposure in liquid aluminum at 960C. TEM analysis revealed that the formation of inter-

particle bridges of pure TiB2 is the reason for the good resistance of the material in molten

aluminum. A sintering mechanism was proposed for the consolidation of TiB2 with a Ti-Fe

additive. TiB2-based ceramic sintered with a Ti-Fe alloy is suggested as a potentially

reliable material for application as wettable cathode for aluminum electrolysis.

106

8.3 Introduction

The idea of using a wettable cathode in Hall-Héroult aluminum electrolysis cells aiming at

reducing the energy consumption of primary aluminum production was first proposed in

1950’s 21

. Lots of efforts have been made thereafter to fabricate an operational wettable

cathode at reasonable cost while TiB2 has been always the most promising candidate for

this application 12, 22

. In addition to its excellent stability, pure TiB2 has a good wettability

for molten aluminum 29, 97

. Despite these advantages along with high electrical and thermal

conductivity, the use of TiB2-based wettable cathode for aluminum smelting has not yet

been established. This is mainly due to challenges related to the fabrication process and

difficulties of meeting the required properties of the consolidated parts. Considering the

technical and economical aspects, pressureless sintering could be an attractive fabrication

method to make large near-net-shape cathode parts at reasonable cost. However, TiB2 has a

very high melting point (3000C) and a low inter-diffusion coefficient which make

sintering using pressureless method quite difficult 32

.

Pressureless sintering of pure TiB2 is typically performed at temperatures higher than

2000C. However, above 1800C the oxide layer at the surface of TiB2 particles promotes

the growth of some TiB2 grains in preferential directions resulting in exaggerated grain

growth, lowering the mechanical properties and especially the thermal shock resistance of

the consolidated part 47

.

To address this problem, researches have been conducted toward using transition metals as

sintering additives 46, 52, 57, 72, 98-100

. These additives promote the densification of TiB2 parts

at much lower temperatures via liquid phase sintering and reduce the risk of abnormal grain

growth 101

. Although TiB2-based specimens consolidated using these additives are suitable

for wear resistant and structural applications 102

, they require sufficient chemical stability to

be used as cathode in liquid aluminum. Aluminum could penetrate and react with different

phases at the grain boundaries resulting in swelling and degradation of TiB2 parts 84

.

107

In previous works 1, 3, 4

, a near eutectic Ti-Fe sintering additive has been used with the aim

of preventing the formation of undesirable phases at grain boundaries and promoting liquid

phase sintering. The influence of processing parameters was studied because they were

found to have a great impact on the final properties after consolidation. Upon exposure of

sintered specimens to molten aluminum, no sign of expansion, swelling or cracking was

observed. These characteristics were found particularly interesting for wettable cathode

applications. In this work, specimens were further characterized before and after exposure

to molten aluminum. The effect of additives on final microstructure, physical and

mechanical properties as well as stability of specimens in molten aluminum for prolonged

exposure time have been investigated. Other specimens were consolidated using Fe only or

Ti only as additives and their mechanical properties and stability in liquid aluminum were

compared to those obtained for the specimens consolidated with Ti-Fe. The inter-particle

bridges in the specimen with the Ti-Fe additive were studied using transmission electron

microscopy (TEM) to understand the reasons of its good chemical stability with respect to

molten aluminum.


Commercial TiB2 powder (99.7 % pure), titanium (Ti >99.8% pure metal basis) and iron

(Fe >99.9% pure metal basis) were used as starting materials (Atlantic Equip. Eng. Inc.).

The particle size of TiB2 powder was ranging between 2 and 10 µm with a mean size of 6

µm. For Ti powder, the particle size was <20 µm and for Fe powder, the particle size was

between 1 and 9 µm. Specimens were consolidated with three different sintering additives:

Ti, Fe, and pre-alloyed Ti-Fe. To prepare the pre-alloyed Ti-Fe additive, Ti and Fe powders

were mixed in a 70-30 weight ratio respectively, compacted in the form of rectangular bars

and sintered at 1150˚C for 1 h. The resulting pellets were subsequently crushed and milled

for 1 h using high-energy ball milling (more details have been provided elsewhere 1).

The powder mixtures were then prepared by milling TiB2 powder with 10 wt.% of additives

in a high-energy ball mill (SPEX 8000) for 30 min using hardened steel vial and balls with

a ball to powder weight ratio of 4:1. The milled mixtures were then compacted in a uniaxial

108

die under 150 MPa pressure and sintered in a tube furnace at 1650˚C under a reducing Ar -

5%H2 atmosphere for 1 h.

The bulk density of the sintered specimens was determined with Archimedes method using

isopropanol as the immersion medium. Theoretical density was calculated using the rule-

of-mixtures assuming the nominal compositions of the starting powder mixtures. Relative

densities were calculated by dividing the measured bulk density by the calculated

theoretical density.

The three-point bending test was performed at room temperature following the ASTM

C1161 standard with a 25.4 mm span and a displacement rate of about 0.01 mm/s 69

. The

dimensions of the specimens used for bending strength measurements were 38 mm × 13

mm × 4 mm and the reported results are the average of measurements on at least five

representative specimens.

The interactions between the sintered specimens and molten aluminum were first studied

using the sessile drop test. A pure aluminum pellet (0.1 g) was placed on top of the

specimen in an electrical tube furnace under a high vacuum (10-3

Pa). The furnace was

rapidly heated up to 960oC (corresponding to the operation temperature of aluminum

electrolysis cells) and maintained at this temperature for 1 h. Prior to the drop test, the

surface of specimens was polished using 6 µm diamond abrasive followed by cleaning with

isopropanol in an ultrasonic bath. After the tests, the microstructures of specimens were

then examined. The dynamic wetting behavior of TiB2-based specimens by molten

aluminum has already been reported 3, 4

.

To evaluate the chemical stability of the specimens in liquid Al, they were covered with an

aluminum foil and then inserted into molten aluminum at 960˚C under a protective flow of

argon. Three different immersion times were used: 1h, 24 h and 5 days. Specimens were

removed after the tests and their microstructure were examined.

For some immersed specimens, the infiltrated aluminum was removed by soaking them in a

0.3 N sodium hydroxide solution for 48 h at room temperature. Once the infiltrated

109

aluminum was removed, a porous structure, mainly composed of a TiB2 solid skeleton, was

obtained and characterized using SEM.

Microstructural studies and chemical analysis were carried out using optical microscope

and scanning electron microscope (SEM) equipped with energy dispersive X-ray

spectroscopy (EDX; PGT Avalon). Electron probe microanalysis (EPMA; SX-100

CAMECA microprobe) equipped with the wavelength dispersive X-ray spectroscopy

(WDS) spectrometers was utilized to perform detailed multi-element compositional

mappings. Specimens were cross sectioned with a diamond saw, vacuum-mounted in epoxy

resin and polished down to 0.1 µm surface finish with successively finer diamond

abrasives. The final polishing was performed using a 0.05 µm alumina suspension. Focused

ion beam (FIB; Hitachi FB2000A) was used to cut out 100 nm thick samples for

examination in transmission electron microscope (TEM; Jeol JEM-2100F). Phase

identification was carried out based on selected area electron diffraction (SAED) pattern

analysis.

A scratch test was used to compare the bonding strength of particles at the surface of

specimens before and after immersion in molten aluminum. The scratch test was conducted

at room temperature by using a micro-tribometer test system (UMT-2; CETR). The

specimen surface was polished down to 6 µm diamond abrasive paper prior to the test. The

specimen was fixed to the lower holder which was automatically driven along a single

horizontal axis while a conical diamond indenter mounted in a upper holder was sliding

over the surface of the specimen applying a vertical force. The cone angle of the diamond

indenter was 75o and its diameter tip was 400 µm. The vertical component of the force (Fz)

was increased gradually from 2 to 50 N over the 10 mm sliding distance. The horizontal

component of the force (Fx), applied to the specimen through the diamond indenter, was

measured in a real time with a dynamometer. A hot-pressed TiB2 specimen obtained from a

commercial supplier (Ceradyne Inc., Costa Mesa, CA, USA) was used to compare the

results of scratch tests. Its reported relative density and flexural strength were 98% and 265

MPa, respectively.

110


8.5.1 Influence of additives on physical and metallurgical properties

Specimens with three different sintering additives were consolidated: Ti, Fe, and a pre-

alloyed Ti-Fe powder. The starting compositions of specimens as well as their as-sintered

properties are reported in Table 8–1.

Table 8–1- The starting composition of sintered specimens, their relative density and 3-

point bending strength.

Specimen Composition (wt%)* Relative

density**

(%)

Bending

strength

(MPa) TiB2 Ti Fe

10T 90 10 0 77 -- 10F 90 0 10 90 520 41

7T3F 1 90 7 3 91 300 32 * Unless otherwise indicated, percentages in this text are wt% ** The uncertainty on relative density was estimated to be less than 1%.

The back-scattered electron (BSE) micrographs from the polished cross-section of

specimens are shown in Figure 8–1. Figure 8–1-a) shows the microstructure of the 10T

specimen. Some discontinuities and the lack of integrity can be observed in the

microstructure of this specimen. Arrows indicate the presence of partially fused Ti-rich

phase between TiB2 particles. The relative density of this specimen, as reported in Table 8–

1, is about 77%. Its bending strength was low with widely dispersed measurements.

Usually, in pressureless sintering of ceramics, densification is achieved by formation of a

liquid phase at the early stage of sintering. This liquid phase enters inter-particle spacings

and applies a capillary force to the walls resulting in re-arrangement and densification 83,

103. Since the sintering temperature of 1650C used in this study was slightly below the

melting point of titanium (1660C), no liquid phase is expected to promote the re-

arrangement of particles. However, even a trace concentration of iron contamination

coming from the milling process can drastically reduce the melting point of the Ti resulting

in its partial fusion as shown by the arrows in Figure 8–1-a). The low relative density and

bending strength of the sample with Ti additive is due to the low sintering temperature and

111

its microstructure represents almost the original size and form of the starting TiB2 powder.

The shape of particles is angular and a lot of fine particles are present.

Figure 8–1-b) shows the microstructure of the 7T3F specimen in which a relatively dense

structure of TiB2 grains with a fine distribution of additives and pores is observed. This

specimen was sintered using a pre-alloyed Ti-Fe additive with a near eutectic composition

of 70%Ti-30%Fe (7T3F). In the Ti-Fe system, an eutectic reaction occurs at 1078C and

67%Ti 76

. Consequently, a liquid phase should form at sintering temperature enabling the

re-arrangement of particles and densification of specimen. Compared to 10Ti, the

discontinuities and gaps are not present in the microstructure of this specimen and most

small pores between grains disappeared. Aside from the small rounded pores, some larger

ones with longitudinal form were distributed uniformly in the microstructure. The size of

TiB2 grains in 7T3F specimen is slightly larger than that of 10Ti, but the grains are still

angular and the inter-particle distances are smaller. The relative density and bending

strength of this specimen are 91% and 300 MPa, respectively. The major phases found in

the microstructure after sintering the 7T3F specimen were TiB2, α-Ti, TiFe and TiFe2 3.

As shown in Table 8–1, specimens sintered with the Fe additive had a relative density of

90% and a bending strength of 520 MPa. Their typical microstructure (Figure 8–1-c)

reveals a uniform distribution of iron (the white components) suggesting that Fe is a

suitable sintering additive for pressureless sintering of TiB2. The very fine TiB2 particles

and sharp edges of grains were eliminated by dissolving in liquid phase and precipitation

on larger grains and surface with greater curvatures. In consequence the angular grains

transformed into larger and rounded shapes and left some large round-shape pores in the

microstructure. According to Shurin et al. 104

, in the pseudobinary Fe-TiB2 system, a

eutectic reaction is expected to occur at 1340C and 6.3 mol% TiB2. Jüngling et al. 80

found

that this eutectic provides a liquid phase suitable for sintering of TiB2. Einarsrud et al. 52

reported that abnormal grain growth could occur by sintering of TiB2+1.5%Fe specimen at

1700C which strongly reduces the mechanical properties. No abnormal grain growth was

observed in our specimens.

112

Figure 8–1- BSE micrographs from the microstructure of as-sintered specimens: a) 10T,

b) 7T3F, c) 10F. Arrows in (a) indicate the presence of metallic titanium between TiB2

particles.

113

In conclusion, the microstructure and physical properties of both 7T3F and 10F specimens

showed a similar degree of densification and relatively uniform microstructures. However,

the bending strength of the 10F specimens was 520 MPa, which is considerably higher than

that of 7T3F (300 MPa).

8.5.2 Interaction with a liquid aluminum drop

The interaction of specimens with a liquid aluminum drop was studied as a first step to

characterize their stability in liquid aluminum. Figure 8–2 shows the BSE micrograph of

10Ti, 7ti3fe and 10Fe specimens after interaction with a sessile aluminum drop. The 10T

specimen had about 23% porosity after sintering. During the experiment, aluminum

infiltrated the specimen immediately upon melting due to the high level of porosity and the

good wettability of TiB2 surfaces by liquid aluminum (Figure 8–2-a). Despite the rapid

infiltration of molten aluminum, it dissolved some of the titanium additive on its way inside

the specimen (zone 1). According to Al-Ti binary phase diagram 89

at 960°C, when the

titanium content of molten aluminum reaches about 2%, TiAl3 phase starts to precipitate

resulting in the complete depletion of aluminum and solidification of liquid phase (zone 2).

Quantitative EDX analysis of the phases in the Al-infiltrated area of 10Ti specimen also

confirmed the formation of TiAl3 phase (Phase A in Figure 8–3). The aluminum drop did

not infiltrate zone 3.

114

Figure 8–2- BSE micrograph of the cross section of specimens after reaction with

aluminum drop: a) 10T, b) 7T3F, c) 10F. Numbers on images refer to the different

zones formed as a result of aluminum infiltration.

115

Figure 8–3- BSE micrograph and EDX spectrum of phase A formed in the Al infiltrated

area (zone 2 in Figure 8–2-a) of 10Ti specimen, A: TiAl3 phase, B: Ti rich phase.

As shown in Figure 8–2-b 4, complete penetration of aluminum inside the 7T3F specimen

and its reaction with the Ti-Fe additive resulted in the formation of three new distinct zones

in the microstructure of specimens. Titanium and iron additives were dissolved as

aluminum infiltrated the specimen (zone 1) 91

and since the quantity of liquid aluminum is

limited in this experiment (0.1 g), Ti and Fe reached their solubility limits and Ti-Al-Fe

phases segregated (zone 2 and 3). The mechanism of the penetration of liquid Al and the

precipitation of Al-Ti-Fe phases was discussed in a previous work 4. Most of the Ti-Fe

additive was washed out from zone 1 (Figure 8–2-b). EDX analysis detected a Ti-Al phase

in zone 2, and it was confirmed later by SAED pattern that TiAl3 starts to precipitate in

zone 2 as Ti reached its solubility limit. By reaching the solubility limit of Fe, Fe4Al13

precipitated in zone 3 until the complete depletion of the liquid phase. The rest of specimen

(zone 4) was not infiltrated by liquid aluminum. In spite of the penetration of Al inside the

116

specimens and its reaction with the metallic additives, there was no sign of deformation or

swelling of the specimen due to the penetration of aluminum.

As shown in Figure 8–2-c, the 10F specimen was not resistant to liquid aluminum.

Aluminum penetrated and reacted with the phases at grain boundaries which resulted in the

separation of TiB2 particles and the expansion near the contact area (zone 1). Figure 8–4

shows the typical microstructure in the expanded volume of the 10F specimen. The TiB2

particles were separated after less than 1 h of exposure, showing a weak stability of grain

boundaries in liquid aluminum. EDX analysis showed the presence of metallic Al and Fe-

Al phases between TiB2 particles in the expanded volume. A similar behavior had been

reported for the TiB2 specimens consolidated using Ni as sintering additive. Finch and

Tennery 105

reported the disintegration of hot-pressed TiB2-10%Ni after being exposed to

liquid aluminum for 3 h. Weirauch et al. 96

also observed the expansion of the surface of

TiB2-Ni specimen due to the reaction of grain boundaries with aluminum drop during

wetting tests.

117

Figure 8–4- a) BSE micrograph from the expanded portion of 10F specimen (zone 1 in

Figure 8–2-c) after reaction with the liquid aluminum drop (A: TiB2, B: Al, C: Fe4Al13

phase, the black areas correspond to mounting resin), b) EDX spectrum of phase C.

In previous studies about the sintering of TiB2 with Fe additive 51, 106

, the presence of

elemental Ti was found at the triple junctions. In these studies, the only source of Ti in the

specimens was TiB2 grains; therefore, the presence of Ti at triple points confirms the

dissolution of TiB2 phase in liquid iron during a dissolution-precipitation stage. However,

as reported by Jüngling et al. 80

, during the precipitation process, Fe2B phase precipitates at

the grain boundaries and some Ti remains in the liquid phase. Iron borides are not stable in

liquid aluminum 107

and when the 10F specimen comes in contact with the aluminum drop,

118

the grain boundaries react with aluminum resulting in the separation of TiB2 particles and

swelling of the specimen.

8.5.3 Stability in liquid aluminum

To investigate the stability of the specimens in liquid aluminum, they were immersed in

liquid aluminum for 24 h. The 10F and 10T specimens were totally disintegrated in liquid

aluminum, as expected from the results of their interaction with the aluminum drop in the

previous experiment. The 7T3F specimens, however, maintained their integrity and keep

the same shape before and after immersion in liquid aluminum. SEM and EDX analysis

revealed that aluminum penetrated inside the specimen and dissolved most of the metallic

Ti-Fe additive. As reported in an earlier work 3, elemental mapping of the exposed

microstructure did not reveal any Fe within the specimen. Ti was not detected between

TiB2 grains, either. The solid skeleton of TiB2 particles appeared to be stable and grains

remained bonded together.

The stability of the 7T3F specimen in liquid aluminum was also studied during 5 days (120

h) immersion test. SEM investigation of the microstructure of this specimen showed that,

the solid skeleton of TiB2 grains maintained its structural integrity. To reveal the structure

of the TiB2 solid skeleton after prolonged Al exposure, the infiltrated aluminum was

dissolved in a NaOH solution. Figure 8–5 shows the solid TiB2 skeleton of this specimen

after removing the infiltrated aluminum. The TiB2 grains are connected to each other

through inter-particle bridges that are stable in liquid Al.

119

Figure 8–5- SEM micrograph of the 7T3F specimen after immersion into molten Al for

5 days at 960C revealing the solid TiB2 skeleton. The metallic phases were dissolved in

a NaOH solution.

8.5.4 Scratch test

A scratch test was performed in order to investigate the bonding strength between TiB2

particles in the 7T3F specimen immersed in molten aluminum for 5 days. The scratch test

results are shown in a diagram in Figure 8–6. An increasing vertical force was applied from

2 to 50 N over 10 mm length of sliding distance. The measured horizontal force (Fx)

reported here is the force applied to remove the particles from the surface and is ascribed as

scratch resistance. It is observed that the as-sintered 7T3F specimen shows higher

scratching resistance compared to the specimen exposed to liquid aluminum. The higher

scratch resistance of as-sintered specimen is attributed to the presence of FeTi phase which

acts as binder for the TiB2 matrix. The influence of the FeTi binder on the scratch

resistance can be evidenced by comparing with the scratch resistance of a hot-pressed TiB2

specimen provided by Ceradyne Inc.

As shown in Figure 5, despite the greater porosity of the specimen exposed to molten

aluminum, the strength of inter-particle bridges is slightly higher than that of the hot-

pressed specimen. These results suggest that the present sintered TiB2 ceramic possesses

acceptable mechanical strength even after exposure to liquid Al and the inter-particle

bridges have good stability in liquid aluminum.

120

Figure 8–6- Comparing the scratch resistance (Fx) between specimens (AS: as sintered,

5 days: 7T3F after 5 days of Al exposure, Ceradyne: Hot pressed TiB2 provided from

Ceradyne Inc.; The reported Fx is associated with 7T3F specimen after 5 days of Al

exposure.)

8.5.5 TEM analysis

The nature of inter-particle bridges and grain boundaries of the exposed 7T3F specimen to

liquid aluminum were studied using TEM. TiB2 exhibits hexagonal symmetry with

P6/mmm space group (a = b = 0.3028 nm, c = 0.3228 nm; α = β = 90°, γ = 120°) 28

. The

atoms are positioned at Ti(0,0,0), B(1/3,2/3,1/2) and B(2/3,1/3,1/2) in the unit cell 35

. Each

boron atom has three boron neighbors in a trigonal planar arrangement, forming a strong

covalently bonded hexagonal network structure 33

. Figure 8–7-a to c show the area selected

for TEM analysis and the interface between two contiguous grains identified as G1 and G2.

Figure 8–7-d shows the SAED pattern of G1 grain from zone axis and Figure 8–7-e

121

shows the SAED pattern of G2 grain from zone axis. It confirms that G1 and G2 are

distinct grains. EDX analysis detected the presence of aluminum in the unattached

interfaces of TiB2 grains. However, no sign of aluminum or other impurities was observed

in bridges developed at the interface of grains and only Ti and B were detected by EDX

analysis in these areas. The inter-particle bridge between G1 and G2 grains shown in Figure

8–7-f reveals that pure TiB2 crystallizes during sintering and binds the two TiB2 particles.

Since TiB2 has a good chemical stability in liquid aluminum, the stability of the specimen

is therefore attributed to the nature of the inter-particle bridges developed during sintering.

Figure 8–7- TEM micrographs from boundary between two TiB2 grains forming a TiB2

bridge: a) selected area for FIB, b-c) Interface of G1 and G2 grains, d) SAED of G1, e)

SAED of G2, f) Inter-particle bridge

Although both 7Ti3Fe and 10Fe specimens were consolidated through the liquid phase

sintering mechanism, unlike 7Ti3Fe, 10Fe specimen did not have chemical stability in

contact with liquid aluminum. During sintering of 10Fe, TiB2 dissolves partially in liquid

122

iron; however, as reported by Jüngling et al. 80

, Fe2B phase precipitates at the grain

boundaries instead of TiB2. The lack of stability of Fe2B in liquid aluminum is responsible

for the rapid degradation of 10Fe specimen in molten aluminum 80, 108

. According to

published thermodynamic data on the Fe-Ti-B system, a quasi-binary section can be found

along the TiB2-(Fe, 2 at% Ti) line characterized by a simple eutectic at about 1320C 78, 104,

108, 109. It has been proposed that below 2 at% Ti, Fe2B coexists with TiB2

100, 110. Therefore,

in order to avoid the precipitation of this undesirable boride, higher amounts of Ti must be

added to the starting composition.

During the sintering of 7T3F specimen, the excess amount of Ti in the liquid phase

promotes the precipitation of the TiB2 phase instead of Fe2B. The formation and growth of

new TiB2 crystalline planes result in the bonding of TiB2 particles and the consolidation of

specimens. Figure 8–7-f clearly shows that the inter-particle bridges between G1 and G2

TiB2 grains formed from the growth of new TiB2 crystalline planes during sintering. As

SAED pattern revealed, the growth of TiB2 mostly occurred epitaxially on (0001) basal

planes of G1 grain. Since TiB2 has hexagonal crystalline structure, the (0001) planes are

closely packed atomic layers and the growth on this orientation is energetically favorable

49. Therefore the presence of Ti not only prevents the formation of secondary boride phases

but also promotes the precipitation of TiB2 as inter-particle bridges.

The presence of oxygen on the surface of TiB2 particles increases the surface diffusivity

and promotes the abnormal grain growth. However, it has been suggested that the use of

strong reducing additives could reduce the adverse effect of oxygen 47, 111

. The excess

content of Ti in the liquid phase and the presence of hydrogen in the sintering atmosphere

used in this study likely removed the oxygen from the surface of TiB2 grains and thereby

prevented the exaggerated grain growth during sintering.

8.6 Conclusions

Pressureless sintering of TiB2 was performed using Ti, Fe and pre-alloyed Ti-Fe additives.

Fe or the pre-alloyed Ti-Fe promotes the sinterability of TiB2 by producing liquid phase

123

sintering. These specimens showed similar degree of densification and a uniform

microstructure after sintering at 1650C for 1h. Although the use of Fe resulted in higher

mechanical properties, the consolidated specimens did not resist to liquid aluminum and

their interactions with molten aluminum caused the disaggregation of TiB2 particles and the

swelling of the specimens. The lack of stability of these specimens is attributed to the

formation of Fe2B in the bonding regions which do not have stability in liquid aluminum.

In contrast, the specimens sintered with pre-alloyed Ti-Fe showed excellent stability in

liquid aluminum. The solid skeleton of TiB2 grains maintained its integrity after 5 days of

exposure in aluminum. Scratch tests confirmed that the strong bonds between grains are

comparable to those of a hot-pressed TiB2 reference specimen. TEM analyses revealed that

the TiB2 grains are bonded to each other with pure TiB2 phase. The excess concentration of

Ti promotes the precipitation of the TiB2 phase in the inter-particle bridges and this leads to

the stability of the specimens in liquid aluminum.

8.7 Acknowledgements

The authors wish to acknowledge the kind contribution and technical support of André

Ferland from Laval University, Sylvio Savoie from Hydro-Québec and of Jean-Philippe

Masse from École Polytechnique de Montréal for TEM analysis.

Chapter 9:

General discussion and conclusions

125

This thesis reports the works carried out as part of the project: Development of wettable

cathode for aluminum smelting. Developing new materials which are wettable by liquid

aluminum has been proposed as alternative for common carbon cathodes to increase the

energy efficiency of Hall Héroult process for production of primary aluminum. TiB2 has

been the most promising material in the efforts made for developing wettable cathodes due

to its excellent wettability for molten metals and electrical conductivity along with good

chemical stability to withstand the severe conditions of electrolysis bath for long durations.

However, the fabrication of large TiB2 cathode parts is encountered by several scientific

and technical issues, which made the development of wettable TiB2-based cathodes a

challenge for aluminum industry since 1950’s.

The objective of the project was to develop TiB2-based ceramic materials, which their

mechanical strength, electrical conductivity, and stability in molten aluminum are close or

superior to those of usual carbon cathodes, and also have good wettability by molten

aluminum. In addition to these properties, it is important to be able to fabricate large

cathode parts of the developed materials economically into desired shapes.

We proposed pressureless sintering method as a less expensive route for consolidation of

large near net-shaped parts. TiB2 has poor sinterability due to its high melting point

(3000°C) and low inter-diffusion coefficient and its consolidation requires very high

sintering temperatures (~2000°C), especially in the absence of applied pressure. However

sintering of TiB2 at temperatures higher than 1700°C results in exaggerated grain growth

which drastically decreases the mechanical properties and thermal shock resistance. To

avoid this issue, the use of low-melting-point sintering additives was proposed to provide

liquid phase at relatively lower temperatures and promotes the pressureless liquid-phase

sintering of TiB2-based materials. Titanium and Iron binary system forms a liquid phase

over a wide range of compositions at temperatures higher than 1450°C. Therefore the

mixtures of Ti and Fe were chosen as sintering additives.

TiB2-based composites with 10% of different proportions of Ti and Fe additives were

sintered at temperatures between 1400 and 1650°C. Specimens with 7%Ti+3%Fe additives

126

(7Ti3Fe), due to the formation of liquid phase at lower temperatures during sintering,

showed better densification compared to those with 8%Ti+2%Fe. Ti-Fe system have a

eutectic point at around 30 wt% Fe with a melting point of 1085°C. Besides the uniform

distribution of additives in powder mixture, it is important to have Ti and Fe particles in

contact with each other to promote the formation of the liquid phase during sintering. For

this purpose, instead of adding Ti and Fe separately, they were added in pre-alloyed form,

which significantly improved the densification. Maximum density was obtained for the

specimen with pre-alloyed additives sintered at 1650°C. Other investigated parameter in

this work was the effect of the milling time, which was found to have a marked influence

on the densification and properties of sintered TiB2 specimens. Powder mixtures were

milled for various milling periods from 10 to 240 min. A maximum relative bulk density of

91% and maximum bending strength of 300 MPa were achieved with specimens milled for

30 min (M30) and sintered at 1650°C for 1h. By increasing the milling time from 10 min to

30 min, a significant increase of density and bending strength was observed and a more

uniform and denser microstructure was achieved after sintering. While the specimen milled

for 30 min had a uniform crack free microstructure with an even distribution of pores, the

presence of numerous cracks was observed in the microstructures of specimens milled for

longer periods. Particles size distribution analyses revealed that the milling of the starting

powder mixture for 30 min results in grinding of large additive particles and produces a

higher volume of fine particles; thereby a uniform microstructure is obtained after sintering.

However, further increase of the milling time leads to the formation of some large strong

agglomerates through the cold welding of TiB2 particles with additive particles. Upon

sintering, these large agglomerates shrink initiating cracks around them in the compact,

which limits the densification and reduces the mechanical strength.

The measured electrical resistivity of specimens milled for 30 and 240 min were

approximately 54 and 243 µΩ.cm, respectively. The higher electrical conductivity of

specimen milled for 30 min was attributed to its lower porosity, absence of crack as well as

more uniform distribution of additives. The electrical resistances of developed TiB2-based

127

ceramics were well below that of graphite (~500 µΩ.cm) which has the highest

conductivity among the carbon cathodes.

The wettability behaviour of specimens for molten aluminum was studied as well. During

the sessile drop test, liquid aluminum completely wetted the surface of the specimens. The

diagram of wetting angle as a function of elapsed time for specimen milled for 30 min was

a continuous and smooth curve due to its uniform and crack-free microstructure.

Another important parameter for wettable cathode materials is their stability during

immersion in molten aluminum. The specimens milled for 30 and 240 min were immersed

in molten aluminum of 960°C for up to 24 h. The microstructural analyses of exposed

specimens revealed that aluminum penetrated completely all inside the specimens and

dissolved the metallic phases. However, the specimens maintained their forms and no sign

of crack propagation and failure was observed. Specimen milled for 30 min showed better

stability in liquid aluminum due to the absence of cracks in the microstructure and hence

the lower liquid aluminum penetration.

The interaction of the developed TiB2-base porous ceramic with molten aluminum drop

was also investigated. Results showed that the removal of the surface oxide layer by

polishing the surface of as-sintered specimen before the sessile drop test led to a faster

wetting by molten aluminum. The microstructural analyses and elemental mapping of the

cross section of the as-sintered specimen after the test revealed that molten aluminum

reacted with the thin titanium oxide layer formed on the surface and reduced it which made

a delay before wetting the surface.

Molten aluminum drop penetrated inside the specimen after wetting and its interaction with

the metallic phases resulted in the formation of three distinct zones in the cross section. In

the first zone (close to the surface), almost all metallic phases were washed out from the

spaces between TiB2 grains and only traces of Al was detected on the surface of TiB2

grains. TiAl3 phase was detected in the second zone and both TiAl3 and Fe4Al13 phases

were identified in the third zone. The penetration and interaction mechanism of aluminum

128

drop was proposed as follows: the metallic additives were dissolved in liquid Al during its

penetration into the material, and their concentration increased in aluminum gradually.

Once the Ti saturation limit was achieved, TiAl3 phase started to precipitate inside the

pores (Zone 2). The residual liquid (saturated on Ti) continued to penetrate and dissolve the

additives until the saturation point of Fe was reached leading to the precipitation of Fe4Al13

(Zone 3). Dissolution and precipitation were continued up to the complete depletion of the

liquid phase (Isothermal solidification). Liquid aluminum did not seem to alter the solid

TiB2 skeleton of the specimen and no sign of swelling or cracking was detected.

Pressureless sintering of TiB2 was also performed using Ti, Fe and pre-alloyed Ti-Fe

additives. Specimens with Fe and pre-alloyed Ti-Fe additives showed similar degree of

densification and a uniform microstructure after sintering at 1650C for 1h. These additives

promoted the sinterability of TiB2 by producing liquid phase at sintering temperature.

Although the use of Fe resulted in higher mechanical properties, the consolidated specimen

did not resist to liquid aluminum and its interaction with molten aluminum caused the

disaggregation of TiB2 particles and the swelling of the specimen. The lack of stability of

this specimen is attributed to the formation of Fe2B in the bonding regions, which do not

have stability in liquid aluminum. In contrast, the specimens sintered with pre-alloyed Ti-

Fe showed excellent stability in liquid aluminum. The solid skeleton of TiB2 grains

maintained its integrity even after 5 days of exposure in aluminum. Scratch tests confirmed

that the strong bonds between grains are comparable to those of a hot-pressed TiB2

reference specimen. TEM analyses revealed that the TiB2 grains are bonded to each other

with pure TiB2 phase which is the reason for the stability of the specimens in liquid

aluminum. The excess amount of Ti in the liquid phase promotes the precipitation of the

TiB2 phase, instead of Fe2B, in the inter-particle bridges. The formation and growth of new

TiB2 crystalline planes result in the bonding of TiB2 particles and the consolidation of the

specimen. In addition, The excess content of Ti in the liquid phase and the presence of

hydrogen in the sintering atmosphere used in this study likely removed the oxygen from the

surface of TiB2 grains and thereby prevented the exaggerated grain growth during sintering.

The results of this investigation provided us with more insight about the sintering

129

mechanism of TiB2 with Ti-Fe additive and the advantage of using this additive system

over the previously studied materials in terms of stability in liquid aluminum.

It is therefore concluded that the material developed by pressureless sintering of TiB2 using

a Ti-Fe additive meets the required properties and chemical stability in liquid aluminum

and is proposed as a reliable material for application as wettable cathodes in aluminum

smelting.

Chapter 10:

Perspectives of the project

131

The developed TiB2-based ceramic showed to be a promising candidate for wettable

cathode application. The use of Ti-Fe additive promotes pressureless sintering at relatively

lower temperature and most importantly, forms the inter-particle bridges with TiB2 nature

which make the solid skeleton of TiB2 with good stability in molten aluminum. Further

investigations would be suggested more evaluation of developed material or for fine-tuning

of the processing parameters:

The evaluation of the chemical stability of the developed material in molten cryolite

and its resistance for sodium penetration is proposed as an important property

requirement of wettable cathode.

Studying the stability of the developed material to liquid aluminum in dynamic

condition (exposed to the flow of molten aluminum on the surface and applying the

electrical current similar to the electrolysis bath conditions) and comparing the

corrosion rate with static condition. Preliminary results showed that the rate of

aluminum penetration in the dynamic condition is lower than the static condition.

Impregnate the pores or apply a coating on the developed TiB2 with a proper

material to protect it from oxidation and infiltration of the bath electrolyte during

the start-up period in the electrolysis cell; during the start-up cycle, electrolyte will

be in direct contact with the surface of the cathode. By increasing the cell

temperature and melting of cryolite and before the formation of aluminum pad, the

molten electrolyte might penetrate inside the pores and result in subsequent issues.

Investigation of the other compositions and processing parameters e.g. sintering

time, the percentage of additives, sintering cycle for consolidation of large parts,

possibility of using vibro-compactor to achieve more uniform densification during

the compaction of large parts.

132

Although the proposed material in this project is developed for wettable cathode

application, its potential for using in other applications such as armours, wear resistant parts

could also be investigated.

133

References:

[1] H. Heidari, H. Alamdari, D. Dubé, R. Schulz, Pressureless sintering of TiB2–based

composites using Ti and Fe additives for development of wettable cathodes, Light Metals

2011, (2011) 1111-1116.

[2] H. Heidari, H. Alamdari, D. Dubé, R. Schulz, Investigating the potential of TiB2-based

composites with Ti and Fe additives as wettable cathode, Mater Sci Forum, 706-709 (2012)

655-660.

[3] H. Heidari, H. Alamdari, D. Dubé, R. Schulz, Investigating the potential of TiB2-based

composites with Ti and Fe additives as wettable cathode, Adv Mater Res, 409 (2011) 195-

200.

[4] H. Heidari, H. Alamdari, D. Dubé, R. Schulz, Interaction of molten aluminum with

porous TiB2-based ceramics containing Ti-Fe additives, J Eur Ceram Soc, 32 (2012) 937-

945.

[5] H. Heidari, H. Alamdari, D. Dubé, R. Schulz, Pressureless sintering of TiB2-based

ceramics with Ti-Fe additive: sintering mechanism and stability in liquid aluminum, Adv

Eng Mater, 14 (2012) 802–880.

[6] J. Nyboer, K. Lutes, A review of energy consumption and related data: canadian

aluminium industries from 1990 to 2009, in, Canadian Industrial Energy End-use Data and

Analysis Centre, 2011, pp. 39.

[7] E.L. Bray, Aluminum, in: Mineral Commodity Summaries, U.S. Geological Survey,

2011.

[8] AAC, Primary aluminum in Québec : world-class production : regional leverage., The

aluminum asscociation of Canada, (2006).

[9] H. Kvande, W. Haupin, Cell voltage in aluminum electrolysis: A practical approach,

JOM Journal of the Minerals, Metals and Materials Society, 52 (2000) 31-37.

[10] W.T. Choate, J.A.S. Green, US energy requirements for aluminum production, in:

Energy efficient manufacturing processes symposium TMS, San Diego, CA, 2003, pp. 99-

113.

[11] M. Sørlie, H.A. Øye, The carbon lining, in: Cathodes in aluminium electrolysis,

Aluminium-Verlag, Düsseldorf, FRG, 1994, pp. 1-30.

[12] M. Sørlie, H.A. Øye, Inert and wettable cathodes, in: Cathodes in aluminium

electrolysis, Aluminium-Verlag, Düsseldorf, 1994, pp. 66-70.

[13] D.W. Richerson, Aluminum industry, in: D.W. Freitag, D.W. Richerson (Eds.)

Opportunities for advanced ceramics to meet the needs of the industries of the future, US

Office of Energy, 1998

[14] F. Habashi, Extractive metallurgy of aluminum, in: G.E. Totten, D.S. Mackenzie

(Eds.) Handbook of aluminum, CRC Press, 2003, pp. 1-46.

[15] H. Zhang, V. Nora, J.A. Sekhar, Non-carbon cathodes in the Hall-Heroult cells, in:

Materials used in the Hall-Héroult cell for aluminium production, TMS, San Diego, CA,

1994, pp. 37-54.

134

[16] W. Choate, U.S. energy requirements for aluminum production in: J.A.S. Green (Ed.)

Aluminum recycling and processing for energy conservation and sustainability, ASM

International, Materials park, OH, 2007, pp. 157-222.

[17] M. Imris, G. Soucy, M. Fafard, Carbon cathode resistance against sodium penetration

during aluminum electrolysis: An overview, Acta Metallurgica Slovaca, 11 (2005) 231-

243.

[18] M. Dell, Reactions in hall smelting cell potlining, Metallurgical and Materials

Transactions B, 15 (1984) 277-280.

[19] W.L. Worrell, Carbothermic reduction of alumina - A thermodynamic analysis, Can.

Metall. Q., 4 (1965) 87-95.

[20] M. Sørlie, H.A. Øye, Evaluation of cathode material properties relevant to the life of

Hall-Heroult cells, Journal of Applied Electrochemistry, 19 (1989) 580-588.

[21] C.E. Ransley, Electrolytic cells for the production of aluminum, in: B. Patent (Ed.),

British Aluminum Co. Ltd, British, 1959.

[22] K. Billehaug, H.A. Øye, Inert cathodes for aluminum electrolysis in Hall - Heroult

cells. Pt. I, Aluminum, 56 (1980) 642-648.

[23] A.L. Yurkov, Refractories and carbon cathode materials for the aluminum industry,

Refract. Ind. Ceram., 47 (2006) 11-13.

[24] X. Liao, H.A. Øye, Carbon Cathode Wear in Aluminum Electrolysis, TRIBOLOGIA,

Finnish Journal of Tribology, 17 (1998) 19-31.

[25] J.E. Hatch, Aluminum : properties and physical metallurgy, American Society for

Metals, Metals Park, Ohio, 1984.

[26] N.D. Spencer, J.H. Moore, Encyclopedia of chemical physics and physical chemistry,

Institute of Physics Pub., Bristol; Philadelphia, 2001.

[27] D. Munz, T. Fett, Physical properties, in: Ceramics: Mechanical properties, failure

behaviour, materials selection, Springer-Verlag, Berlin, 1999, pp. 9-18.

[28] P. Schwarzkopf, R. Kieffer, Borides of the forth group of the periodic table, in:

Refractory Hard Metals: Borides, Carbides, Nitrides, and Silicides, Macmillan, New York,

1953.

[29] G. Jones, J. Pearson, Factors affecting the grain-refinement of aluminum using

titanium and boron additives, Metall Mater Trans B, 7 (1976) 223-234.

[30] C.J. McMinn, Review of RHM cathode development, Light Metals 1992, (1992) 419-

425.

[31] O.A. Kashin, V.E. Ovcharenko, S.E. Salibekov, A.A. Zabolotskii, Reaction of carbon

filaments coated with titanium carbide and molten aluminum, Metal Science and Heat

Treatment, 22 (1980) 809-812.

[32] B. Basu, G.B. Raju, A.K. Suri, Processing and properties of monolithic TiB2 based

materials, Int Mater Rev, 51 (2006) 352-374.

[33] R.G. Munro, Material properties of titanium diboride, J. Res. Natl. Inst. Stan., 105

(2000) 709-720.

[34] D. Vallauri, I.C. Atías Adrián, A. Chrysanthou, TiC-TiB2 composites: A review of

phase relationships, processing and properties, J Eur Ceram Soc, 28 (2008) 1697-1713.

[35] D.-C. Tian, X.-B. Wang, Electronic structure and equation of state of TiB2, Journal of

Physics: Condensed Matter, 4 (1992) 8765-8772.

135

[36] I. Higashi, Y. Takahashi, T. Atoda, Crystal growth of borides and carbides of

transition metals from molten aluminum solutions, J. Cryst. Growth, 33 (1976) 207-211.

[37] P. Schwarzkopf, R. Kieffer, Preparation of borides, in: Refractory Hard Metals:

Borides, Carbides, Nitrides, and Silicides, Macmillan, New York, 1953, pp. 271-275.

[38] L.G. Boxall, A.V. Cooke, H.W. Hayden, Use of TiB2 cathode material application and

benefits in conventional VSS cells, in, Metallurgical Soc of AIME, Los Angeles, CA, USA,

1984, pp. 573-588.

[39] S. Devyatkin, G. Kaptay, J. Boutellion, J. Poignet, Electrochemical synthesis of

titanium diboride coatings from cryolite melts, Molten Sail Forum, 5-6 (1998) 331-334.

[40] G. L'Espérance, G. Botton, S. Dallaire, Analytical transmission electron

microscopycharacterization of plasma spray synthesized TiB2-Fe coatings, Thin Solid

Films, 193-194 (1990) 442-452.

[41] J. Sekhar, J. Liu, V. de Nora, A porous titanium diboride composite cathode coating

for Hall-Héroult cells: Part I. Thin coatings, Metallurgical and Materials Transactions B, 29

(1998) 59-67.

[42] L.G. Boxall, A.V. Cooke, Successful Application of Carbon/TiB2 Composite

Refractories to Conventional Aluminum Reduction, in, Inst of Mining & Metallurgy,

London, Engl, 1985, pp. 473-496.

[43] A.V. Cooke, W.M. Buchta, Use of TiB2 cathode material: demonstrated energy

conservation in VSS cells., in, Metallurgical Soc of AIME, New York, NY, USA, 1985, pp.

545-566.

[44] Y. Liu, X. Liao, F. Tang, Z. Chen, Observations on the operating of TiB2-coated

cathode reduction cells, in, Publ by Minerals, Metals & Materials Soc (TMS), San Diego,

CA, USA, 1991, pp. 427-429.

[45] M.S. Jensen, M. Pezzotta, Z.L. Zhang, M.A. Einarsrud, T. Grande, Degradation of

TiB2 ceramics in liquid aluminum, J Eur Ceram Soc, 28 (2008) 3155-3164.

[46] M.G. Barandika, J.J. Echeberría, J.M. Sánchez, F. Castro, Consolidation,

microstructure, and mechanical properties of a TiB2-Ni3Al composite, Mater Res Bull, 34

(1999) 53-61.

[47] S. Baik, P.F. Becher, Effect of oxygen contamination on densification of TiB2, J Am

Ceram Soc, 70 (1987) 527-530.

[48] J.-H. Park, Y.-H. Lee, Y.-H. Koh, H.-E. Kim, S. Baek, Effect of hot-Pressing

temperature on densification and mechanical properties of titanium diboride with silicon

nitride as a sintering aid, J Am Ceram Soc 83 (2000) 1542-1544.

[49] M.S. Jensen, M.-A. Einarsrud, T. Grande, Preferential grain orientation in hot pressed

TiB2, J Am Ceram Soc 90 (2007) 1339-1341.

[50] D. Mattia, M. Desmaison-Brut, D. Tétard, J. Desmaison, Wetting of HIP AlN-TiB2

ceramic composites by liquid metals and alloys, J Eur Ceram Soc, 25 (2005) 1797-1803.

[51] S.H. Kang, D.J. Kim, E.S. Kang, S.S. Baek, Pressureless sintering and properties of

titanium diboride ceramics containing chromium and iron, J Am Ceram Soc, 84 (2001)

893-895.

[52] M.-A. Einarsrud, E. Hagen, G. Pettersen, T. Grande, Pressureless sintering of titanium

diboride with nickel, nickel boride, and iron additives, J Am Ceram Soc, 80 (1997) 3013-

3020.

136

[53] A. Petukhov, I. Khobta, A. Ragulya, A. Derevyanko, A. Raichenko, L. Isaeva, A.

Koval’chenko, Reactive electric-discharge sintering of TiN-TiB2, Powder Metall. Met.

Ceram., 46 (2007) 525-532.

[54] J.S. Kim, Y.S. Kwon, O.I. Lomovsky, M.A. Korchagin, V.I. Mali, D.V. Dudina, A

synthetic route for metal-ceramic interpenetrating phase composites, Materials Letters, 60

(2006) 3723-3726.

[55] T.S.R.C. Murthy, B. Basu, R. Balasubramaniam, A.K. Suri, C.S.R.K. Fotedar,

Processing and properties of TiB2 with MoSi2 sinter-additive: a first report, J Am Ceram

Soc 89 (2006) 131-138.

[56] G.B. Raju, B. Basu, Densification, sintering reactions, and properties of titanium

diboride with titanium disilicide as a sintering aid, Journal of the American Ceramic

Society, 90 (2007) 3415-3423.

[57] I. Azkona, F. Castro, J. Sánchez, Abnormal growth of TiB2 crystals during sintering of

TiB2-Ni3(Al,Ti) cemented borides, Metall Mater Trans A, 36 (2005) 459-466.

[58] A.K. Shurin, V.E. Panarin, Phase diagrams for iron with interstitial phases as a basis

for developing wear-resistant and eutectic steels, Metal Science and Heat Treatment, 26

(1984) 166-169.

[59] R. Telle, S. Meyer, G. Petzow, E.D. Franz, Sintering behaviour and phase reactions of

TiB2 with ZrO2 additives, Mater. Sci. Eng. A-Struct. Mater. Prop. Microstruct. Process.,

105 (1988) 125-129.

[60] S. Torizuka, K. Sato, H. Nishio, T. Kishi, Effect of SiC on interfacial reaction and

sintering mechanism of TiB2, J Am Ceram Soc 78 (1995) 1606-1610.

[61] S. Torizuka, K. Sato, J. Harada, H. Yamamoto, H. Nishio, Microstructure and sintering

mechanism of TiB2-ZrO2-SiC composite, Journal of the Ceramic Society of Japan, 100

(1992) 392-397.

[62] H.-J. Kim, H.-J. Choi, J.-G. Lee, Mechanochemical synthesis and pressureless

sintering of TiB2-AlN composites, Journal of the American Ceramic Society, 85 (2002)

1022-1024.

[63] L.-H. Li, H.-E. Kim, E. Son Kang, Sintering and mechanical properties of titanium

diboride with aluminum nitride as a sintering aid, J Eur Ceram Soc, 22 (2002) 973-977.

[64] H. Baker, A.S.M.I.A.P.D. Committee, A.S.M.I.H. Committee, Alloy phase diagrams,

ASM International, Materials Park, Ohio, 1992.

[65] M.N. Rahaman, Ceramic Fabrication Processes, in: 2nd (Ed.) Ceramic Processing and

Sintering, Routledge, USA 2003

[66] M. Gasch, D. Ellerby, S. Johnson, Ultra high temperature ceramic composites, in: N.P.

Bansal (Ed.) Handbook of ceramic composites, Springer, NY, USA, 2005, pp. 197-224.

[67] J.F. Shackelford, W. Alexander, CRC Materials science and engineering handbook, 3

ed., CRC Press LLC, 2001.

[68] W.D. Callister, Mechanical properties, in: Fundamentals of materials science and

engineering, John Wiley & Sons, Inc., 2001, pp. 171-172.

[69] ASTM_Standard, C1161-02c Standard test method for flexural strength of advanced

ceramics at ambient temperature, in, ASTM International, West Conshohocken, PA, 2002.

137

[70] R. Telle, L.S. Sigl, K. Takagi, Boride-based hard materials, in: R. Riedel (Ed.)

Handbook of ceramic hard materials, Wiley-VCH, Weinheim; Germany, 2000, pp. 802-

945.

[71] R.A. Cutler, Engineering properties of borides, in: S.J. Schneider (Ed.) Engineering

Materials Handbook, Ceramic and Glasses, ASM International, Metals Park, OH, USA,

1991, pp. 787–803.

[72] T. Graziani, A. Bellosi, D.D. Fabbriche, Effects of some iron-group metals on

densification and characteristics of TiB2, Int J Refract Met Hard Mater, 11 (1992) 105-112.

[73] W.D. Kingery, H.K. Bowen, D.R. Uhlmann, Grain growth, sintering, and vitrification,

in: Introduction to ceramics, Wiley, New York, 1976, pp. 448-515.

[74] P.S. Mohanty, J.E. Gruzleski, Mechanism of grain refinement in aluminum, Acta

Metallurgica et Materialia, 43 (1995) 2001-2012.

[75] T.E. Quested, Understanding mechanisms of grain refinement of aluminium alloys by

inoculation, Materials Science & Technology, 20 (2004) 1357.

[76] H. Okamoto, Fe-Ti (iron-titanium), J Phase Equilib, 17 (1996) 369-369.

[77] V.N. Eremenko, Y.V. Niadich, I.A. Lavrinenko, Capillary phenomena and wettability

in liquid phase sintering process, in: Liquid phase sintering, Plenum Publishing

Corporation, 1970, pp. 16-21.

[78] L. Ottavi, C. Saint-Jours, N. Valignant, C.H. Allibert, Phase equilibria and

solidification of Fe-Ti-B alloys in the region close to Fe-TiB2, Zeitschrift für Metallkunde,

83 (1992) 80-83.

[79] V. Raghavan, B-Fe-Ti (Boron-Iron-Titanium), Journal of Phase Equilibria, 24 (2003)

455-456.

[80] T. Jüngling, R. Oberacker, F. Thummler, L.S. Sigi, K.A. Schmetz, Pressureless

sintering of TiB2-Fe-materials, Powder Metall Int, 23 (1991) 296-300.

[81] M. Palm, G. Inden, N. Thomas, The Fe-Al-Ti system, J Phase Equilib, 16 (1995) 209-

222.

[82] H. Baker, S.D. Henry, H. Okamoto, M.P.O.H.A.P.D.a.C. Asm International, ASM

handbook. Vol. 3 : Alloy phase diagrams, ASM International, Materials Park, OH, 1992.

[83] R.M. German, Liquid-phase sintering, in: Sintering theory and practice, John Wiley &

Sons Ltd., New York, 1996, pp. 225-313.

[84] G. Pettersen, Development of microstructure during sintering and aluminium exposure

of titanium diboride ceramics, in: Department of physics, NTNU, Norway, 1997, pp. 138.

[85] R.C. Dorward, Aluminum penetration and fracture of titanium diboride, J Am Ceram

Soc, 65 (1982) C-6-C-6.

[86] K. Nord-Varhaug, TEM investigation of impurity phases and the penetration of liquid

aluminum in hot isostatically pressed TiB2 compacts, J Am Ceram Soc, 79 (1996) 1147-

1154.

[87] S.K. Rhee, Wetting of ceramics by liquid aluminum, J Am Ceram Soc, 53 (1970) 386-

389.

[88] G. Ghosh, Al-Fe-Ti (Aluminium - Iron - Titanium), in: G. Effenberg, S. Ilyenko (Eds.)

Light metal ternary systems: phase diagrams, crystallographic and thermodynamic data,

2005.

138

[89] U. Kattner, J. Lin, Y. Chang, Thermodynamic Assessment and Calculation of the Ti-

Al System, Metall Mater Trans A, 23 (1992) 2081-2090.

[90] M. Palm, J. Lacaze, Assessment of the Al-Fe-Ti system, Intermetallics, 14 (2005)

1291-1303.

[91] V. Raghavan, Al-Fe-Ti (aluminum-iron-titanium), J Phase Equilib, 23 (2002) 367-374.

[92] H.R. Shahverdi, M.R. Ghomashchi, S. Shabestari, J. Hejazi, Microstructural analysis

of interfacial reaction between molten aluminium and solid iron, J Mater Process Tech, 124

(2002) 345-352.

[93] J. Grin, U. Burkhardt, M. Ellner, K. Peters, Refinement of the Fe4Al13 structure and its

relationship to the quasihomological homeotypical structures, Z Krist-New Cryst St, 209

(1994) 479-487.

[94] T.I. Yanson, N.B. Manyako, O.I. Bodak, R. Cêrny, R.E. Gladyshevskii, K. Yvon,

Crystal structure of Fe4Ti0.93Al12.07, a substitutional variant of the Fe4Al13 structure type, J

Alloys Compd, 219 (1995) 135-138.

[95] W. D. MacDonald, T.W. Eagar, Transient liquid phase bonding, Annu Rev Mater Sci,

22 (1992) 23-46.

[96] D. Weirauch, W. Krafick, G. Ackart, P. Ownby, The wettability of titanium diboride

by molten aluminum drops, J Mater Sci, 40 (2005) 2301-2306.

[97] N. Finch, The mutual solubilities of titanium and boron in pure aluminum, Metall

Mater Trans B, 3 (1972) 2709-2711.

[98] V.I. Matkovich, Boron and refractory borides, Springer-Verlag, Berlin; New York,

1977.

[99] M.K. Bannister, M.V. Swain, A preliminary investigation of the corrosion of a

TiB2/BN/AIN composite during aluminium evaporation, Ceram Int, 15 (1989) 375-382.

[100] M.G. Barandika, J.M. Sánchez, T. Rojo, R. Cortés, F. Castro, Fe-Ni-Ti binder phases

for TiB2-based cermets: a thermodynamic approach, Scr Mater, 39 (1998) 1395-1400.

[101] M.K. Ferber, P.F. Becher, C.B. Finch, Effect of microstructure on the properties of

TiB2 Ceramics, J Am Ceram Soc, 66 (1983) C2-C3.

[102] R. Telle, G. Petzow, Strengthening and toughening of boride and carbide hard

material composites, Mater Sci Eng, A, 105-106 (1988) 97-104.

[103] W.D. Kingery, M.D. Narasimhan, Densification during sintering in the presence of a

liquid phase. II. experimental, J Appl Phys, 30 (1959) 307-310.

[104] A.K. Shurin, V.E. Panarin, V.K. Sul'zhenko, G.Y. Kozyrskii, Properties of eutectic

alloys containing titanium diboride at high temperatures, Met Sci Heat Treat, 19 (1977)

696-698.

[105] C.B. Finch, V.J. Tannery, Crack formation and swelling of TiB2-Ni ceramics in

liquid aluminum, J Am Ceram Soc, 65 (1982) c100-c101.

[106] E.S. Kang, C.W. Jang, C.H. Lee, C.H. Kim, D.K. Kim, Effect of iron and boron

carbide on the densification and mechanical properties of titanium diboride ceramics, J Am

Ceram Soc, 72 (1989) 1868-1872.

[107] A. Kataev, K. Karimov, Y. Chernov, N. Kulik, V. Malkov, B. Antonov, E.

Vovkotrub, Y. Zaikov, Wetting of boride cathode coatings by low-melting-point cryolite

and liquid aluminum, Russ Metall (Metally), 2010 (2010) 689-701.

139

[108] B.S. Terry, O.S. Chinyamakobvu, Dispersion and reaction of TiB2 in liquid iron

alloys, Mater Sci Technol, 8 (1992) 491-499.

[109] K. Tanaka, T. Saito, Phase equilibria in TiB2-reinforced high modulus steel, J Phase

Equilib, 20 (1999) 207-214.

[110] J.M. Sánchez, M.G. Barandika, J. Gil-Sevillano, F. Castro, Consolidation,

microstructure and mechanical properties of newly developed TiB2-Based materials, Scr

Metall Mater, 26 (1992) 957-962.

[111] T. Watanabe, H. Yamamoto, K. Shobu, T. Sakamoto, Factors affecting the porosity

and bending strength of Ti(CN)-TiB2 materials, J Am Ceram Soc, 71 (1988) C202-C204.

Appendix 1: Ceramic fabrication

A 1.01 Introduction

Generally, ceramic fabrication involves creating fine particle size powders, forming

powders into compact and sintering to produce a cohesive body with the desired

microstructure and properties for a given application. Specific surface area of powders is

relatively high; therefore ceramic processing involves understanding and controlling the

physical chemistry of surface and interfaces. The finer the particles are, the higher the

specific surface area of powder and hence the surface energy of the system is. Surface

energy, which is also called surface tension, γ, is obviously very important in ceramic

powder processing and sintering, as it causes liquids to form spherical drops, and allow

solids to preferentially adsorb atoms to lower the free energy of the system. It also creates

pressure differences and chemical potential differences across curved surfaces that cause

matter to move.

The Laplace equation, which defines the pressure difference, ΔP, across a curved surface of

radius, r,

ΔP = γ (1/r1 + 1/r2) ( 1-1)

has been characterized as the fundamental equation of capillary 1. In ceramic processing,

the pressure associated with surface tension and capillary forces contribute to particle

clustering (i.e. agglomeration) and rearrangement, to the migration of liquids through pores

during mixing, shape forming, and drying, and to pores shrinkage during sintering.

The equilibrium vapour pressure, P, over a curved surface is defined by the Kelvin equation

ln P/P0 = 2γΩ/rkT ( 1-2)

where P0 is the equilibrium vapour pressure over a planar surface, Ω is the molecular

volume of condensed phase, k is Boltzmann’s constant, and T is the absolute temperature.

Because the chemical potential difference, Δμ, between a curved and flat surface is related

to the vapour pressure over those respective surfaces,

141

Δμ = kT ln P/P0 ( 1-3)

chemical potential is also related to surface curvature:

Δμ = μ – μ0 = 2γΩ/r ( 1-4)

The chemical potential of a curved surface is extremely critical in ceramic processing. It

determines reactivity, the solubility of a solid in a liquid, the rate of liquid evaporation from

solid surfaces, and material transport during sintering 1.

A 1.02 Powder processing

Advanced ceramics are typically made from refined raw materials or chemically

synthesized ceramic powders. The chemical and physical properties and characteristics of a

powder are modified within beneficiation processes. Advanced ceramics are sensitive to the

chemical and physical defects present in the starting raw materials, or those that are

introduced during manufacturing. Particles size reduction using mechanical energy may be

the most common process. Grinding or milling creates new surfaces by breaking down

aggregates and by fracturing particles 1.

Consolidating powder generally involves a forming pressure to produce the desired size or

shape ceramic component. Forming additives or processing aids are commonly added to a

ceramic powder to enhance processing. Organic additives adhere onto the surfaces of

ceramic particles to modify surface energy and particle-particle interactions. Two common

additives used in ceramic processing are binders and lubricants 1. The main purpose of a

binder, which also called coagulant or flocculent, is to provide strength to the powder

compact after shape forming. It may be necessary for subsequent handling or green

machining. Lubricant is added to lower interfacial frictional forces between individual

particles or between particles and forming die surfaces to improve compaction and ejection.

Coating with a film of low-viscosity lubricant can lubricate die surfaces.

Powder pressing is the most common method of forming ceramic components. The

powders can be compacted by applying mechanical pressure, which is most simply done

142

under uniaxial compression in steel dies (or with those with WC-Co lining). It is called die

pressing or cold pressing to distinguish it from hot pressing. Cold pressing, is an

economical, yet versatile technique for fabricating small, relatively simple-shape powder

compacts 1. Such pressing can be done with no or limited use of lubricant or binder. Dies

have a vertical cylindrical through hole whose shape and dimensions are based on desired

piece shape considering sintering shrinkage and machining (if applicable). In die cavity, the

pressure is applied onto the powder by opposing rams. The limit for cavity clearance is less

than 25 μm for fine micron-sized powders up to four time for coarser particles 2. An

important issue is the relative motion of the top versus the bottom ram. The uniformity of

the packing in the consolidated material is very important. In general, any non-uniformity

in the green body is exaggerated in the sintering process, leading to the development of

crack like voids or large pores between relatively dense regions 3.

Basically the die pressing consists of three stages: 1) die filling 2) powder compaction and

3) part ejection. Figure 1 shows the schematic of uniaxial die which is used for compaction

of powders. Friction between the powder and the die wall must be controlled, e.g. by using

lubricants, during forming to minimize pressing pressure gradients that can create defects in

the form of density gradients or cracking in a pressed powder compact 1. It may need

binders to enhance strength of green body that is required for part ejection from the die and

subsequent handling before sintering 2.

143

Figure 1- The schematic of uniaxial pressing of powders

A 1.03 Thermal processing

Generally, the last step in ceramic component manufacturing is thermal processing. It is the

stage where the weakly bounded particulate body produced during shape forming is heat

treated to produce a cohesive body with the desired properties for its end-use application.

Thermal consolidation, which is more commonly referred to as ‘firing’, typically involves

two steps, burnout and sintering. Generally both are accomplished in a single firing process

with burnout preceding sintering.

The burnout stage involves eliminating the organic processing aids and any residual

organic impurities or water prior to sintering. Most organic binders used in ceramic

forming can be burned out by heating to 500°C 1.

A 1.04 Sintering

Sintering involves the densification and microstructure development that transform the

loosely bond particles in a powder compact into a dense, cohesive body. The final

properties of a ceramic are largely dependent on the degree of densification achieved

144

during sintering, and on the microstructure produced; consequently, sintering is one of the

most critical steps in ceramic processing. Sintering is usually accomplished by heating a

powder compact to approximately two-thirds of its melting temperature for a given time 1.

Sintering can also occur by subjecting a powder compact to externally applied pressure, or

heat and pressure simultaneously (e.g. hot pressing and hot isostatic pressing). A ceramic

densifies during sintering as the porosity between the solid particles is reduced in size with

time. Concurrently, the cohesiveness of the body increases as the inter-particle contact (e.g.

grain boundary) area increases during sintering.

A 1.05 Driving force for sintering

Ceramic powder compacts are sintered as a result of the thermodynamic driving force to

minimize the Gibb’s free energy, G, of a system. This includes minimizing the volume,

interfacial, and surface energy in the system. In a powder compact, excess free energy is

present primarily in the form of surface or interfacial energy (i.e. liquid-vapour or solid-

vapour interfaces) associated with porosity 1. Under influence of elevated temperature

and/or pressure during sintering, atoms migrate to thermodynamically more stable positions

within a powder compact. Material transport is driven by the chemical potential difference

that exists between surfaces of dissimilar curvature within the system. Physically, in a

particulate system, atoms or ions move from higher energy convex particle surfaces to

lower energy concave particle surfaces to decrease the curvature and chemical potential

gradients in the system.

Material transport can occur by solid-state, liquid-phase, and vapour-phase mechanisms.

For polycrystalline ceramics, material transport commonly occurs as ions diffuse through

the volume, along grain boundaries and on particle surfaces. Additionally, ions can

vaporize from, and consequently condense onto, particle surfaces (evaporation-

condensation) 1. A powder compact will densify when material transport occurs in a

manner that allows particle centres to approach during sintering. Material transport by

volume and grain boundary diffusion can results in densification. Material transport that

145

changes the geometry of the system without densification is termed coarsening. Grain

growth is perhaps the most common form of coarsening during sintering. Coarsening can

occur when material is transported by volume diffusion, surface diffusion, or evaporation-

condensation.

Figure 2 shows the classic two-particle sintering model illustrating material transport and

neck growth at the particle contacts resulting in coarsening and densification during

sintering. Surface diffusion (a), evaporation-condensation (b), and volume diffusion (c)

contribute to coarsening, while volume diffusion (d), grain boundary diffusion (e), solution-

precipitation (f), and dislocation motion (g) contribute to densification.

Figure 2- The classic sintering model illustrating material transport and neck growth at

particle contacts resulting in coarsening (left) and densification (right) during sintering

1.

146

A 1.06 Densification and microstructure development

In a microstructure point of view, material transport during sintering shows itself as inter-

particle pore shrinkage, grain boundary formation, a decrease in the total volume of the

system (i.e. densification), and an increase in the average size of the particles that make up

the compact (i.e. grain growth) 1. Inter-particle contacts flatten, the curvature within the

system decreases, and the surface area and free energy of the system decrease during

sintering.

The ideal sintering process can be divided into three basic stages. Initially, material is

transported from convex particle surfaces to the pore-grain boundary intersection to form

necks between adjacent particles. As this occurs, grain boundaries grow to create a three-

dimensional array of approximately cylindrical, interconnected pore-channels at triple grain

junctions throughout the compact. These pore-channels shrink in diameter during

intermediate-stage sintering. Ultimately, the channels punch off to form approximately

spherical, closed pores at four grain junctions within the ceramic matrix. The radial

shrinkage of closed pores and the growth of larger grains at the expense of smaller ones

constitute final-stage sintering.

Factors like surface energy anisotropy and packing heterogeneities in real systems can

contribute to heterogeneous densification and microstructure development. To overcome

this problem, minor concentration of selected chemicals, which called sintering aids or

dopants, are commonly added prior to the sintering. These chemical impurities

preferentially segregate to high-energy crystallographic planes to decrease the crystalline

anisotropy in the system to provide improved control over microstructure development

during sintering (e.g. MgO-doped Al2O3) 1. Impurity segregation to high-energy grain

boundaries will also produce lower-energy interfaces that reduce the overall driving force

for material transport during sintering.

147

A 1.07 Solid-state sintering

Ceramics can be densified by solid-state, liquid-phase, and viscous sintering. Solid-state

sintering refers to the process whereby densification occurs by solid-state, diffusion-

controlled material transport. Densification occurs as higher-energy, solid-vapour (i.e. pore)

interfaces are replaced by lower energy, solid-solid (i.e. grain boundary) interfaces 1. The

change in free energy associated with the elimination of porosity, which drives

densification, can be approximated by:

dG = γssdAss - γsvdAsv ( 1-5)

Grain growth can further reduce the free energy of the system by reducing the amount of

high-energy, solid-solid interfacial area. The change in free energy associated with the

elimination of particle-particle interfaces, which drives grain growth, can be approximated

by:

dG = - γssdAss ( 1-6)

Because densification occurs via shrinkage of thermodynamically unstable pores,

densification and microstructure development can be assessed on the basis of the dihedral

angle, θ, formed as a result of the surface energy balance between the two solid-vapour and

one solid-solid interfaces at the pore-grain boundary intersection.

( 1-7)

where γss and γsv are the solid-solid and solid-vapour interfacial energies, respectively

(Figure 3) 1.

sv

ss

2cos2 1

148

Figure 3- Interfacial energies and dihedral angle, θ, 1

In the presence of a vapour phase, θ is larger than 120° because γsv is usually higher than

γss. In general, γsv is 2–3 times higher than γss and θ is around 150°. If the dihedral angle is

constant and the junction edges are randomly distributed in three dimensions, the measured

angle with the maximum frequency is the true dihedral angle 4.

The larger the dihedral angle, the larger the inter-granular pores that can be eliminated

during sintering and the greater the surface tension driving force for pore shrinkage.

Thermodynamics and/or kinetics limit the shrinkage of pores trapped within grains (i.e.

inter-granular porosity) and pores above critical size 1.

A 1.08 Liquid phase sintering

To promote faster densification at lower temperatures, relatively small concentrations of

chemical additives, called sintering aids, are commonly used to create a liquid-phase

sintering. The formation of liquid phase during sintering considerably increases the

diffusion rate of the components. It facilitates the displacement of solid particles with

respect to each other, which results in rapid filling of pores and capillaries and usually

increases the sintering rate. In liquid-phase sintering (LPS) the theoretical density of the

149

material may be attained in a short time 5. Kingery et al.

6 proposed that for having the

rapid densification, it is essential to have:

An appreciable solubility of the solid in the liquid,

An appreciable amount of liquid phase, and

Wetting of the solid by the liquid.

During sintering in the presence of liquid phase, a number of phenomena occur

simultaneously and it is significantly more complex process than the solid state process. In

LPS, there are more phases, interfaces, and material transport mechanisms to consider.

In general, densification will occur as long as it is energetically favourable to replace

liquid-vapour (lv), solid-solid (ss), and solid-vapour (sv) interfaces with solid-liquid (sl)

interfaces during sintering 1:

dG = γsldAsl – (γlvdAlv + γssdAss + γsvdAsv) ( 1-8)

where γsl and γlv are the solid-liquid and liquid-vapour interfacial energies, respectively.

With considering weak chemical reaction between the constituents; the surface tension

from the liquid phase is a significant factor in determining the sintering rate. In liquid phase

sintering, densification and microstructure development can be assessed on the basis of the

liquid contact or wetting angle, ϕ, formed as a result of the interfacial energy balance at the

solid-liquid-vapour intersection as defined by Young equation 1:

( 1-9)

A low contact angle favours liquid wetting of particle surfaces and densification during

LPS. Theoretically, ϕ must be less than 60° to achieve 100% of the theoretical density 1.

lv

slsv

1cos

150

Figure 4- Illustration of the wetting angle, ϕ 1

The liquid wets the solid particles and applies capillary force on the solid particles to

eliminate porosity and reduce interfacial area. It can be considered that the material is under

uniform hydrostatic pressure. Diffusion rates in liquid are relatively high and results in

faster bonding and densification compared to equivalent solid state sintering 5.

If the liquid do not wet the solid particles, it will not penetrate between each of the solid

particles and they will be joined only in contact points. In this case, sintering must occur

with the transfer of materials within the solid phase, in order for particle centers to move

closer and shrinkage take place in the specimen. Although, material transfer may take place

by solution-precipitation process in which the material is transferred through the liquid, it is

not essentially different from the single-phase process in which the materials are vaporized

and condense at the junction between particles 7.

Considering the solubility factors, four possible combinations of interactions and sintering

behaviours encountered in liquid-phase sintering are 5:

Low liquid solubility in the solid combined with a high solid solubility in the liquid

which result in LPS. It is applied to wide variety of systems that results in

densification, including: W-Ni-Fe, WC-Co, Si3N4-Y2O3, Al2O3-SiO2, TiC-Ni, and

Fe-Cu.

The opposite solubility situation, high liquid solubility in the solid combined with a

low solid solubility in the liquid, which gives swelling and a transient liquid. The

Fe-Al system has such behaviour.

151

The combination of low liquid and solid solubilities, requires high temperature to

induce significant liquid phase sintering and corresponds to a system such as W-Cu.

In this case, similar to the solid-state process, the materials transfer in an initially

porous compact under the force of surface energy and form pore-free structure. For

such a system solid state sintering of the skeletal structure can be more significant

than liquid-phase sintering.

Finally, the situation of high solubility proves to be the least predictable. This

combination exhibits both swelling and densification and is sensitive to many

processing variables such as sintering temperature, particle size and green density.

A conceptual summarization of the key solubility factors encountered in liquid-phase

sintering is shown in Figure 5 5.

Solid solubility in liquid

Low High

Liq

uid

so

lub

ilit

y in

soli

d

Lo

w Limited densification,

Rearrangement

Extensive densification,

LPS

Hig

h Swelling,

Transient liquid

Mixed effect,

Swelling & densification

Figure 5- Conceptual summarization of four possible interactions and sintering

behaviours considering the solubility factor 5.

When the surface energies are dominant, liquid-phase sintering densification occurs in

stages as sketched in Figure 6. Initially, mixed powders are heated to a temperature where

liquid forms. Based on phenomenological observations, Kingery 7, 8

considered that after

the formation of liquid three stages can appropriately be distinguished:

Rearrangement,

Solution-precipitation, and

Final-stage sintering or coalescence.

152

Figure 6- Schematic diagram of the classic liquid-phase sintering stages 5

Figure 7- Density dependence on the liquid content 5

Figure 7 summarize these stages by plotting the density versus the volume of liquid phase 5.

With good wetting the liquid infiltrates to the contacting sections of the solid particles,

153

which sharply reduces friction and wedging. The solid particles are rearranged, due to the

exertion of capillary force by the liquid; the voids and bridges between them disappear, and

thus the volume of the sintered material decreases 9. If there is a high liquid level, full

density can be achieved via rearrangement upon liquid formation. On the other hand, at low

liquid contents the solid skeleton inhibits densification, requiring the participation of

solution-precipitation events, where mass transport through the liquid controls

densification. Residual final porosity is eliminated by solid-state sintering of the rigid solid

skeleton 5.

With liquid formation, grains shrink as solid dissolves into the liquid and the solid grains

pack to a higher density, releasing liquid to fill pores between grains. For wetting liquids,

the surface energy of solid-liquid is lower than that of solid-vapour and densification results

in reducing system energy 5. During rearrangement the compacts exhibit viscous response

to the capillary action. The elimination of porosity increases the compact viscosity. As a

consequence, the densification rate decreases continuously. Full density is possible by

rearrangement if enough liquid is formed 5, 9

.

Another important condition for occurrence of the rearrangement is the penetration of

liquid between the grains. The extent to which the liquid enters the joints between particles

depends on the dihedral angle formed by the liquid phase at the boundary with two grains

of the solid phase 4. The dihedral angle is determined only by the interfacial energies and is

independent of the pressures in the phases. This means that the dihedral angle is constant,

irrespective of the pressure of liquid phase 4.

Whether viscous flow occurs or not at the beginning of liquid phase sintering depends

strongly on two parameters: the dihedral angle between the solid grains and the volume

fraction of liquid. When this dihedral angle is greater than 0° and a solid skeleton forms

during heating to the liquid phase sintering temperature, no viscous flow of particles and no

particle rearrangement are expected 4. Viscous flow and particle rearrangement may occur

only when the dihedral angle is 0°. If the liquid volume fraction is high, viscous flow can

occur. However, for a low liquid volume fraction, local rearrangement of particles must be

154

predominant 4. While a slight shrinkage can be achieved, larger pores are formed by

particle rearrangement, as in the case of solid state sintering. This result suggests that

particle rearrangement may not be beneficial for overall densification because the final

densification is governed by the elimination of the larger pores in the compact 4.

The second stage termed solution-precipitation. Although the various other events are

concurrent with rearrangement, the kinetics of rearrangement are initially so fast that the

other events are overshadowed 10

. As densification by rearrangement slows, solubility and

diffusivity effects dominate. Therefore, this stage followed by solution-precipitation. At

least a limited solubility of the solid in the liquid phase is necessary for this stage to take

place; otherwise this stage of densification is completely absent. Densification occurs more

slowly than in the rearrangement stage, because the transport of material must proceed by

means of dissolution and diffusion in the melt. Small grains with strongly convex

curvatures go into solution and the substance is precipitated on large particles

accompanying by grain growth and grain shape accommodation. The driving force for

material transport, results from the increased compressive stresses and hence from the

enhanced chemical potential and higher solubility in the contact zones 10

. The solubility of

grains in surrounding liquid varies inversely with their size; small ones, owning the smaller

radius of curvature, have higher energy and solubility compared to the large ones with

larger radius of curvature. The difference in solubility establishes a concentration gradient

in the liquid. Material is transported from the small grains to the large grains by diffusion

through the liquid. The net result is a progressive growth of the larger grains at the expense

of the smaller grains, giving fewer grains with a lager average size. Solution-precipitation

not only contributes to grain coarsening but also to densification via grain shape

accommodation, allowing a customized fitting together the growing grains to better fill

space. The amount of liquid determines the diffusion distance and the necessary degree of

grain shape accommodation 5, 9

. According to Cannon and Lenel 11

, the solution-

precipitation process cannot occur with less than 5 vol.% of liquid phase.

155

The final stage of liquid-phase sintering is coalescence and it is controlled by densification

of the solid structure. In the case of incomplete wetting or insufficient liquid phase, the

solid grains remain partly in contact without the interposition of the melt 10

. In their

sintering model, Cannon and Lenel 11

suggested that, after considerable densification by a

solution-precipitation process and the formation of grain boundaries, the contribution of

solution-precipitation became negligible and that final densification occurred by a sintering

process similar to solid state sintering. This type of sintering, however, is not believed to be

operative in liquid phase sintering because the densification kinetics of liquid phase

sintering are much faster than the estimated kinetics of solid state sintering 9.

Microstructural coarsening continues and the residual pores enlarge if they contain trapped

gas, giving contact swelling. In general, properties of most liquid-phase sintered materials

are degraded by prolong final-stage sintering. Hence, short sintering times are preferred in

practice 5. Figure 8 shows a hypothetical densification curve for these three stages of liquid-

phase sintering 9. The time scale in this diagram is logarithmic and it gives a sense of the

densification event.

Figure 8- Densification time scale for the three stages of liquid phase sintering 9

156

A 1.09 References

[1] N.D. Spencer, J.H. Moore, Encyclopedia of chemical physics and physical chemistry,

Institute of Physics Pub., Bristol; Philadelphia, 2001.

[2] R.W. Rice, Forming and pressurless sintering of powder-drived bodies, in: Ceramic

fabrication technology Marcel Dekker, Inc., New york, NY, 2003.

[3] M.N. Rahaman, Ceramic Fabrication Processes, in: 2nd (Ed.) Ceramic Processing and

Sintering, Routledge, USA 2003

[4] S.-J. L.Kang, Densification models and theories, in: Sintering: Densification, Grain

Growth, and Microstructure, Elsevier Butterworth-Heinemann, 2005, pp. 227-247.

[5] R.M. German, Liquid-phase sintering, in: Sintering theory and practice, John Wiley &

Sons Ltd., New York, 1996, pp. 225-313.

[6] W.D. Kingery, H.K. Bowen, D.R. Uhlmann, Introduction to ceramics, 2 ed., Wiley,

New York, 1976.

[7] W.D. Kingery, Densification during sintering in the presence of a liquid phase. I.

theory, Journal of Applied Physics, 30 (1959) 301-306.

[8] W.D. Kingery, M.D. Narasimhan, Densification during sintering in the presence of a

liquid phase. II. experimental, J Appl Phys, 30 (1959) 307-310.

[9] V.N. Eremenko, Y.V. Niadich, I.A. Lavrinenko, General principles of sintering in the

presence of a liquid metallic phase, in: Liquid phase sintering, Plenum Publishing

Corporation, 1970, pp. 1-15.

[10] F. Thummler, W. Thomma, The sintering process, Met. Rev., 12 (1967) 69-108.

[11] H.S. Cannon, F.V. Lenel, Some observations on the mechanism of liquid phase

sintering, in: Pulvermetallurgie, Plansee seminar de re metallica, Springer-Verlag, Vienna,

1953, pp. 106-110.

development of wettable cathode for aluminum …€¦ · developing new materials, as alternative...

Documents