Metal–Crucible Interactions

Metal–Crucible Interactions suggests solutions to a major challenge in high-temper-

ature materials processing. It offers a holistic presentation of the current knowledge
of metal–crucible interactions in a compact volume so that readers can take informed
decisions on materials selection. Presenting practical information, this book:

• Provides an extensive summary on the compatibility a huge variety of

metal–container combinations, assembles information about all known sig-
nificant interactions, and evaluates how they are managed
• Explains the underlying reasons for the occurrence and extent of incompati-
bility between metals and containment and presents some possible solutions
• Outlines analytical experimental techniques to quantify compatibility/
incompatibility
• Covers issues and resolution in interrelated solid–solid, solid–liquid, solid–
gas and solid–liquid–gas processes determining compatibility
• Discusses all the metals - ferrous, common non ferrous, reactive and refrac-
tory metals, rare earths, and the important alloys as well as compounds and
special compositions that tide over or remain prone to degradation due to
compatibility issues
• Highlights the value of addressing all interrelated issues in arriving at dura-
ble solutions to compatibility challenges

Aimed at readers in industries dealing with materials processing at high tempera-

tures, research scientists and engineers, and graduate students, this book addresses
a topic vital to stimulating immediate and long-term research and development ben-
efits, in ways not previously covered in other books.
Metal–Crucible Interactions

Nagaiyar Krishnamurthy
First edition published 2023
by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742

and by CRC Press

4 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN

CRC Press is an imprint of Taylor & Francis Group, LLC

© 2023 Taylor & Francis Group, LLC

Reasonable efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and publish-
ers have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
copyright holders if permission to publish in this form has not been obtained. If any copyright material has not
been acknowledged please write and let us know so we may rectify in any future reprint.

Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmit-
ted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying, microfilming, and recording, or in any information storage or retrieval system, without
written permission from the publishers.

For permission to photocopy or use material electronically from this work, access www​.copyright​.com or con-
tact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. For
works that are not available on CCC please contact mpkbookspermissions​@tandf​.co​​.uk

Trademark notice: Product or corporate names may be trademarks or registered trademarks and are used only
for identification and explanation without intent to infringe.

ISBN: 978-0-367-34809-0 (hbk)

ISBN: 978-1-032-30787-9 (pbk)
ISBN: 978-0-429-34556-2 (ebk)

DOI: 10.1201/9780429345562

Typeset in Times
by Deanta Global Publishing Services, Chennai, India
Sri Krishnarpanam
Contents
Preface.................................................................................................................... xiii
Acknowledgements.................................................................................................xvii
Author Biography.....................................................................................................xix

Chapter 1 Metal–Crucible Interactions: Evolution of Crucibles............................ 1

1.1 Introduction................................................................................ 1
1.2 History of Metals Processing.....................................................2
1.2.1 Stone Age...................................................................... 2
1.2.2 Native Metals: Copper, Gold, Silver............................. 2
1.2.3 Primitive Crucibles........................................................3
1.2.4 Chalcolithic Age............................................................ 5
1.2.5 Bronze Age.................................................................... 6
1.2.6 Iron Age......................................................................... 6
1.2.7 Middle Ages.................................................................. 7
1.2.8 Post-Medieval Period.................................................... 7
1.2.9 Industrial Revolution and Later..................................... 7
1.2.10 World War II and Later.................................................8
1.3 Non-Ferrous Metals.................................................................... 8
1.3.1 Copper........................................................................... 9
1.3.2 Brass............................................................................ 10
1.3.3 Tin............................................................................... 11
1.3.4 Zinc............................................................................. 12
1.4 Iron and Steel............................................................................ 13
1.4.1 Beginning of Iron Smelting......................................... 13
1.4.2 Blast Furnace............................................................... 15
1.4.3 Cast Iron...................................................................... 17
1.4.4 Steel............................................................................. 18
[Link] Cementation Steel........................................ 18
[Link] Crucible Steel Ancient Practice................... 18
[Link] Puddling Furnace......................................... 19
1.4.5 Bessemer, Open Hearth, Linz–Donawitz (L-D)
Processes..................................................................... 19
[Link] Open Hearth Furnace..................................20
[Link] Linz–Donawitz (L-D) Steel......................... 21
1.5 Refractories in the 20th Century.............................................. 21
1.5.1 Silicates....................................................................... 22
1.5.2 Dolomite...................................................................... 22
1.5.3 Chrome–Magnesite..................................................... 22
1.5.4 Graphite....................................................................... 23
1.5.5 Grain and Bond Phases...............................................24
1.5.6 Bricks and Monolithics...............................................24

vii
viii Contents

1.5.7 Refractory Bonds and Binders....................................24

1.6 Electrolytic Processes...............................................................25
1.6.1 Aluminium..................................................................25
1.7 Rare Metals.............................................................................. 27
1.7.1 Special Crucibles and Protected Environment............ 27
1.8 Metal Replacement (Metallothermic) Reactions......................28
1.8.1 Goldschmidt Process or Bomb Reduction..................28
1.9 Consolidation and Refining...................................................... 29
1.9.1 Vacuum Induction Melting.......................................... 30
1.9.2 Vacuum Arc Melting................................................... 31
1.9.3 Electron Beam Melting............................................... 32
1.10 Crystal Growing....................................................................... 32
1.10.1 Crystal Growth in a Container.................................... 32
1.10.2 Floating Zone Technique............................................. 33
1.11 Summary and Scope of Crucible–Metal Interactions.............. 33

Chapter 2 Basics of Interactions.......................................................................... 41

2.1 Introduction.............................................................................. 41
2.2 Thermodynamics...................................................................... 41
2.3 The Titanium–Oxygen System................................................. 47
2.4 The Vanadium–Oxygen System............................................... 53
2.5 Components of Interaction....................................................... 55
2.6 Wetting and Contact Angle...................................................... 57
2.7 Porosity..................................................................................... 61
2.8 Capillary Penetration................................................................ 61
2.9 Physical Properties of Refractories.......................................... 65
2.10 Thermal Expansion.................................................................. 65
2.11 Thermal Conductivity.............................................................. 67
2.12 Thermal Shock......................................................................... 68
2.13 Corrosion of Refractories......................................................... 69
2.14 Chemical Dissolution in Casting.............................................. 71
2.15 Corrosion Testing of Refractories............................................ 72
2.16 Summary.................................................................................. 73

Chapter 3 Crucible Materials............................................................................... 77

3.1 Introduction.............................................................................. 77
3.2 Ceramics................................................................................... 77
3.3 Refractories.............................................................................. 78
3.4 Silica-Alumina Group..............................................................80
3.4.1 Fireclays...................................................................... 81
3.4.2 Aluminosilicates.......................................................... 81
3.4.3 Mullite......................................................................... 82
3.4.4 Alumina....................................................................... 83
Contents ix

3.5 Magnesia–Lime Group............................................................. 83

3.5.1 Magnesia..................................................................... 83
[Link] Magnesite Bricks.........................................84
3.5.2 Dolomite......................................................................84
3.5.3 Magnesia–Chrome...................................................... 85
[Link] MgO–Al2O3................................................. 86
3.5.4 Magnesia–Carbon....................................................... 86
3.6 Carbon...................................................................................... 87
3.7 Beryllia..................................................................................... 88
3.8 Zircon and Zirconia.................................................................. 88
3.9 Yttria.........................................................................................90
3.10 Spinels......................................................................................90
3.11 Carbon-Containing Refractories.............................................. 91
3.12 Barium Zirconate..................................................................... 93
3.13 Calcium Zirconate....................................................................94
3.14 Silicon Carbide......................................................................... 95
3.15 Borides...................................................................................... 95
3.16 Boron Nitride............................................................................ 95
3.17 Silicon Nitride and Sialon........................................................96
3.18 Aluminium Nitride...................................................................97
3.19 Coatings on Crucibles............................................................... 98
3.20 Summary..................................................................................99

Chapter 4 Major Melt–Crucible Systems........................................................... 105

4.1 Introduction............................................................................ 105
4.2 Iron and Steel.......................................................................... 107
4.3 Blast Furnace Refractories..................................................... 108
4.4 Steel-Making Refractories...................................................... 109
4.4.1 Refractories for Secondary Steel............................... 111
4.4.2 Refractories for Casting............................................ 112
4.5 The Hall–Héroult Cell............................................................ 115
4.5.1 Interactions in the Cell.............................................. 116
4.5.2 Barrier Refractory..................................................... 117
4.5.3 Side Lining Materials................................................ 118
4.6 Post-Reduction Treatment of Aluminium.............................. 118
4.6.1 Refractories in Aluminium Melting and Holding
Furnaces.................................................................... 119
4.6.2 Alumina–Silica Refractories..................................... 120
4.6.3 Belly Band Zone........................................................ 122
4.7 Management of Aluminium Penetration................................ 123
4.7.1 Anti-Wetting Additions............................................. 123
4.7.2 Phosphate Bonding.................................................... 124
4.8 Refractory Inclusions.............................................................. 125
4.9 Refractories Resistant to Molten Aluminium......................... 125
x Contents

4.10 Copper.................................................................................... 126

4.10.1 Smelter....................................................................... 126
4.10.2 Converter................................................................... 127
4.10.3 The Mag–Chrome Refractory................................... 128
4.10.4 Calcia–Magnesia....................................................... 130
4.10.5 Olivine Slags............................................................. 131
4.11 Less Common Metals............................................................. 131
4.12 Metallothermy........................................................................ 132
4.13 Reduction of Uranium Tetrafluoride...................................... 134
4.14 Reduction of Thorium Tetrafluoride....................................... 136
4.15 Reduction of Oxides............................................................... 137
4.16 Reactions in the Absence of Metal Vapour............................ 140
4.17 Kroll Reductions..................................................................... 141
4.17.1 Reduction Plant......................................................... 141
4.17.2 Magnesium Reduction of Zirconium
Tetrachloride.............................................................. 142
4.17.3 Bimetal Reduction..................................................... 142
4.17.4 Crucible Contamination............................................ 143
4.17.5 Magnesium Reduction of Niobium and Tantalum
Pentachlorides........................................................... 143
4.18 Lithium Reduction of Rare Earth Trichlorides...................... 143
4.19 Ingot Reduction Processes...................................................... 145
4.19.1 Reduction in a Tantalum Crucible............................. 145
4.19.2 Reduction of Rare Earth Fluorides........................... 145
4.19.3 Calcium Reduction (Ames Process).......................... 146
4.19.4 Goldschmidt Process................................................. 146
4.20 Oxide Reduction Processes.................................................... 148
4.21 Fused Salt Electrolysis............................................................ 148
4.22 Oxide–Fluoride Electrolysis................................................... 149
4.22.1 Gray’s Cell................................................................. 149
4.22.2 Reno Cell Type 6....................................................... 149
4.22.3 Reno Cell Type 12..................................................... 149
4.22.4 High-Temperature Electrowinning Cell.................... 150
4.23 Crucible Contamination in Rare Earth Reduction................. 151
4.24 Pyrovacuum Treatments......................................................... 152
4.24.1 Rare Earth Metals..................................................... 152
[Link] Lanthanum, Cerium, Praseodymium
and Neodymium........................................ 152
[Link] Yttrium, Gadolinium, Terbium and
Lutetium..................................................... 152
[Link] Scandium, Dysprosium, Holmium,
Erbium and Lutetium................................. 153
[Link] Samarium, Europium, Thulium and
Ytterbium................................................... 153
4.24.2 Uranium..................................................................... 153
4.24.3 Vanadium.................................................................. 153
Contents xi

4.25 Titanium................................................................................. 157

4.25.1 Melting Techniques................................................... 157
4.25.2 Induction Melting...................................................... 158
4.25.3 Skull Melting............................................................. 158
[Link] Induction Skull Melting or Cold
Crucible Induction Melting (CCIM).......... 159
[Link] Vacuum Arc Melting (VAR)...................... 159
[Link] Electron Beam Melting (EBM)................. 159
4.25.4 Castings..................................................................... 160
4.25.5 Moulds....................................................................... 160
4.25.6 Crucibles for Titanium Metal.................................... 161
[Link] The Oxides: Al2O3, BeO, ThO2, CaO,
MgO........................................................... 162
[Link] Zirconia...................................................... 164
[Link] Rare Earth Oxides..................................... 165
[Link] Yttria.......................................................... 165
[Link] Carbon, Graphite and Carbide
Crucibles.................................................... 166
[Link] Borides....................................................... 166
[Link] Sulphides.................................................... 167
4.25.7 Crucibles for Titanium Alloys................................... 167
4.25.8 TiAl Alloys................................................................ 168
[Link] Zirconia (Stabilized) Crucible................... 168
[Link] Yttria.......................................................... 170
[Link] Calcia......................................................... 172
[Link] Graphite..................................................... 173
[Link] Nitrides: AlN and BN................................ 174
[Link] The Zirconates........................................... 176
[Link] CCIM......................................................... 179
[Link] EBM........................................................... 180
4.25.9 NiTi(-X)-Based Shape Memory Alloys (SMA)......... 180
[Link] Barium Zirconate (Calcia-Doped)
Crucibles.................................................... 184
[Link] VAR and EBM........................................... 184
4.25.10 Hydrogen Storage Alloys.......................................... 185
4.26 Summary................................................................................ 185
References............................................................................................................... 189
Index....................................................................................................................... 203
Preface
Metal–crucible interaction is a nearly universal problem, which every researcher and
materials processor faces, particularly when dealing with metals and materials near,
at or above their fusion temperatures. It is usually a deleterious interaction between
the container and the melt. It is not entirely preventable but invariably managed to
levels that are considered harmless in the given context. This has been the general
situation from the beginning of metallurgy thousands of years ago and is still the
case today. The know-how and the resources to contain molten materials have deter-
mined the manner and extent to which materials processing progressed in various
regions of the world and the range of materials to which various civilizations had
access. In modern times, too, this situation prevails somewhat implicitly.
Every beginning researcher in metallurgy wants and searches for a ready-to-use
table that lists various melts and the crucible materials that can be used with them.
Some are available, mostly in the manuals that come with expensive high-temper-
ature equipment, and some from colleagues and associates who have created their
own cheat sheets. While such collections are attractive and comforting, their use is
very limited, because each situation of metal–crucible interaction is unique. A sneak
preview into the real world of metal–crucible interactions should bring out the chal-
lenges in all their glory.
First of all, it is necessary to abandon the common perception that a crucible is
nothing but a cup-shaped vessel that is used to hold materials at high temperatures. A
crucible is much more than that. More inclusively, a crucible can be considered as the
material surface at work in contact with the melt and related fluids during material
processing; the work here being to contain the melt, resist and survive its attack, and
withstand the concurrent thermal and mechanical stresses the process entails, and
to do all these without causing any degradation of the melt. The crucible also needs
to maintain its integrity and functionality for durations that are useful in practice.
In the field of common metals, both ferrous and common non-ferrous, the crucibles
are usually refractory ceramic materials. In the case of less common metals, the
crucibles are more specialized and are often made of refractory ceramic compounds,
refractory metals and also some special alloy compositions.
Crucible materials include metals and alloys, oxide minerals, pure oxide compo-
sitions, and compounds like metal borides, carbides, nitrides, silicides, phosphides,
sulphides and even fluorides. Carbon is an exceedingly useful crucible material,
mostly as graphite and also as vitreous carbon. Many of the properties of these mate-
rials are available not only when they are used in monolithic form but also when they
are used as linings or coatings on other materials. An impressively versatile range of
applications is fulfilled by the class known as water-cooled crucibles, which gener-
ate a layer of solidified material (or skull) on the working surface and thus obviate
any chance of material contamination. Refractory materials for crucibles are broadly
grouped into siliceous, aluminous, aluminosilicates, magnesite, dolomite, chrome
ore, and a miscellany of materials known as special refractories. Each of these

xiii
xiv Preface

refractory types has its niche application areas. Every crucible material is subject to
attack by the melt it is holding. Melt penetration into the crucible and reaction of the
melt with the refractory occur. Generally, refractories have the ability to withstand
alteration by penetration, contamination and/or reaction in service and still function
as reliable engineering structures. Many factors control this interaction, and some of
these are manageable to a certain extent.
Metal–crucible interaction operates by chemical interaction between the mate-
rial of the crucible and the melt, augmented by concurrent physical, mechanical and
thermal processes. Chemical interaction, in turn, generates products that, because of
their own properties, occasionally end up protecting the crucible but usually result
in its degradation. The key process, therefore, is the chemical reaction, and chemical
reactions are governed by thermodynamics.
Even though the statement that at sufficiently high temperatures, everything
reacts with everything else, cannot be disputed, it may be tempered with many ifs
and buts. These conditions, which are extremely important in real life situations, are
physical, structural, mechanical and thermal. A crucible material should possess,
in addition to high thermodynamic (chemical) stability, high melting point, good
strength, appropriate thermal expansion and thermal conductivity and hence, ther-
mal shock resistance. Some of these properties are interrelated and can be tailored
by microstructure and porosity control to ensure that the chemical stability of the
crucible material is fully utilized in actual applications.
To interact, the melt and crucible have to be in close contact. This aspect is
decided by the wetting characteristics of the melt–crucible combination. The contact
angle is a measure of wetting and is determined by the interfacial tensions of the
pairs: liquid–solid, liquid–vapour and solid–vapour. The contact angle or wettability
is ultimately determined by the real-time chemical composition of the melt–solid
interface. Impurities in the melt or crucible play a role, and this is the basis of anti-
wetting agents. Anti-wetting additives act by controlling the interface chemistry and
raising the contact angle. They have been used extensively with aluminium melts
because of the temperatures involved. Nearly all anti-wetting additives decompose at
higher temperatures, and then, the protection will be lost (usually beyond 1000–1200
°C). For this reason, porosity control may turn out to be the more reliable handle
to obviate melt intrusion. However, modifying porosity may have other unintended
consequences.
Corrosive attack on the refractory of the crucible is often due to the combined
action of molten metal, slag and also process gas. The most significant and life-
limiting attack is by the slag and also by metal that acquires the characteristics of
slag. The attack proceeds in two ways. Direct dissolution is controlled by the rate of
chemical reaction at the slag–refractory interface. Indirect attack is controlled by the
rate of diffusive transport through the slag or through the newly formed solid phase.
Significantly, penetration and thus, corrosion are determined by local slag composi-
tion and also refractory composition and structure. These are also usable handles.
The melt properties, particularly viscosity, are an important factor in the corrosion
process. Many attack prevention strategies rely on controlling this property by tem-
perature and/or composition manipulations.
Preface xv

Throughout the history of metals processing, metal makers have faced the chal-
lenge of finding suitable crucibles for smelting as well as post-reduction melting and
casting operations. The challenges have been faced and circumvented with inge-
nuity, directed experimentation and sometimes plain luck, and there is now avail-
able a large and valuable body of knowledge on what works and sometimes, also
the underlying reasons for the success. The situation as regards some of the major
metal–crucible systems is dealt with not only to look back at the road travelled but
also to prepare for the road that lies ahead.
The purpose of this book is also to consolidate in one place all the different
metal–crucible combinations that have been tested and found to work, and in the
process, look for the common underlying reasons that make such compatibility a
reality. Materials interact or fail to interact because of certain well-defined reasons.
The reasons concern thermodynamics, on the one hand, and transport processes
at the interface, on the other. How these factors play out in each individual case is
unique, but as is generally stated, there is method in the madness. These methods are
rather implicit and are best revealed by describing the working combination. It was
therefore considered best to select certain major metal–crucible systems and look at
the key processes involving the crucibles from the point of view of the crucibles. The
systems considered in this book are iron and steel, aluminium, copper, reactive rare
metals like uranium, niobium and vanadium, rare earths, and last, and in consider-
able detail, titanium and its alloys.
Each of the few metal–crucible interactions selected for specific coverage in this
book is representative of the type of problems faced by many more metal–crucible
systems. The drivers for interactions and the remedial actions that can circumvent
or offset the damages are described. The objective is to relate what is known about
metal–crucible interactions at the basic level to the interactions playing out in real
processes. This will help the task ahead as regards managing new metal–crucible
combinations and suitable approaches that would improve the chances of success.
Most of the issues concerning any metal–crucible pair can usually be resolved by
analogy to those covered specifically here.
Systems with melts and containers in contact at high temperatures are encoun-
tered in two major areas of technology. One is in process metallurgy and post-reduc-
tion processing of materials, and the other is in liquid metal/alloy cooling loops.
Both are important and have been attracting considerable research effort all over
the world. There are, however, major differences in the problems faced in each of
these areas and the way solutions to the problems are developed. They have more
differences than commonalities. To maintain focus, this book covers essentially the
processing-oriented issues of metal–crucible interactions.
I began to mull over the various possibilities of creating a monograph dealing
principally with metal–crucible interactions soon after my book on rare earths
was completed in 2015. The break came in July 2017 when Florence Kizza, Editor,
Engineering & Environmental Sciences, Taylor & Francis enquired whether I would
be willing to author a title in their ‘Focus’ series. I was, and prepared a proposal for a
book on ‘Metal–Crucible Interactions’. My mentor at T&F, Allison Shatkin, received
the proposal, and the process began. I was obviously underestimating the matter that
xvi Preface

should be covered under this title when I originally aimed at fitting everything into
150 pages, the typical size of a ‘focus’ title. The highly experienced international
review panel who took a look at my proposal concluded otherwise and suggested a
bigger book with at least 250 pages. So, here it is.
During the preparation of the manuscript, which took considerably more time
than I intended, I was fortunate to receive cooperation and support from both the
publisher and my family. I have been constantly encouraged and guided all along
by Allison Shatkin. Gabrielle Vernachio has been extremely helpful. She was in
constant communication with me, answering my numerous queries concerning the
presentation of text and figures. Florence Kizza started the project, and Camilla
Michael, Ryan Farrar, Iris Fahrer and Lara S. Loes have at various stages provided
support My sincere thanks to all of them.
All through the course of this work I received wonderful cooperation and support
at home from my wife Kusuma and daughter Kavita. My son Kumar had always
urged me think of creating monographs that I would have liked to possess myself
during my formal working years. I would like to attribute all good things in the
book to these wonderful people. If any, errors and omissions, are entirely my doing.
If brought to my notice, I shall remedy them in future editions.

Nagaiyar Krishnamurthy
Acknowledgements
The role of an author in creating a monograph is aptly described by the phrase ‘stand-
ing on the shoulders of giants’, a metaphor coined by the twelfth-century French phi-
losopher Bernard of Chartres and made famous by none other than Sir Isaac Newton
in his letter to Robert Hooke in 1675. As an author, the extent of support I received
from people and institutions has been truly humbling.
Nearly all scientists and engineers whose work involves the processing of materi-
als at high temperatures have been confronted with issues of interaction between
melts and the containers. In their publications, they often highlight the issues and
also indicate if the interactions were circumvented or lived with. Original research
papers, reviews and compilations, and technical reports from government laborato-
ries, over the years, contain a wealth of information on this matter. Then there are
books and conference proceedings that both allude to or directly deal with technical
details on the interactions and also their ramifications in the overall processing. In
putting together the present volume, I have strongly relied on these sources of infor-
mation that have been made accessible to anyone wanting to read or use them by the
wonderful publishing houses and organizations. I have greatly benefitted from and
gratefully acknowledge such publications brought out by Elsevier, Springer, Wiley,
Taylor & Francis, Maney Publishing, Routledge, Vulkan-Verlag GmbH, The AISE
Steel Foundation, EDP Sciences, American Ceramic Society, American Chemical
Society, Quartz Business Media, The Minerals, Metals & Materials Society,
American Society for Metals, as well as the Institutions of the U.S. Government
and the Government of India. Wikipedia and the Encyclopaedia Britannica often
provided the linkage when the terrain got tough. The open access system in technical
publications is truly an exalted level in information dissemination, and every author
will remain deeply grateful to all those responsible for its coming into being.
In addition to all those wonderful people from Taylor & Francis mentioned in the
Preface, I gratefully acknowledge the contributions of Ms. Jayanthi Chander, Project
Manager, and her team at Deanta Global Publishing Services, Chennai, in bringing
out the book in its final form.

Nagaiyar Krishnamurthy

xvii
Author Biography
Dr. Nagaiyar Krishnamurthy was affiliated with the Materials Group, Bhabha
Atomic Research Centre, Mumbai, India for nearly four decades, conducting and
later also formulating and guiding research in the extraction and processing of rare
earths, reactive and refractory metals, and special less common materials. He earned
his BSc degree at the University of Madras in 1974, his MSc in 1980, and his PhD in
1992, both at the University of Bombay. His PhD dissertation was on the pyrometal-
lurgy of Group V refractory metals and their alloys.
He has been professor at the Homi Bhabha National Institute and has guided
many students to complete their PhD. Apart from regular classroom teaching of
postgraduate engineering and science students at the Bhabha Atomic Research
Centre Training School, Professor Krishnamurthy is a popular faculty member in
various quality improvement programmes for college and university teachers. He has
been active as a lecturer in popular science topics connected with materials and has
lectured to audiences ranging from high school and college students to members of
staff in universities and national institutions.
In addition to more than 100 original research papers published in peer-reviewed
international journals, Dr. Krishnamurthy has co-authored two editions of Extractive
Metallurgy of Rare Earths, published by CRC in 2004 and 2016. Earlier, he co-
authored the book Extractive Metallurgy of Vanadium, published by Elsevier in
1991, and a monograph, Binary Phase Diagrams of Tantalum, published in 1996 by
the Indian Institute of Metals and American Society for Metals. He also served as a
key reader for the journal Metallurgical and Materials Transactions.

xix
 1 Metal–Crucible
Interactions
Evolution of Crucibles

1.1 INTRODUCTION
The earliest historical reference to metals dates back to 10,000 BCE, which is
approximately the age of the oldest metal artefact found anywhere in the world. It
is made of copper. Between then and the beginning of the Common Era, 10 more
elements, of which 8 were metals, were added to the list of elements discovered. It is
amazing that so many great civilizations were sustained with such a small number
of metals. Furthermore, by the beginning of the industrial revolution (1720 CE), only
three more elements had been discovered. These were arsenic, antimony and phos-
phorus. The pace picked up by about 1720 CE, and one by one, all the remaining 72
naturally occurring elements, which include 59 metals, were discovered, and many
of them were put into use. The most prevalent processing method was melting and
casting. The facilities for melting a metal often determined whether the metal could
be used and also, how widespread its use would be. Suitable crucibles were needed
for melting metals, and the availability of a crucible was often the decisive factor in
metal processing and use.
The concept of the crucible evolved from the necessity to confine materials within
a physical boundary when processing them, especially by raising or lowering their
temperature. The process may be melting and casting to consolidate the smaller
pieces into a solid mass, a chemical reaction by bringing together the selected
reagents, or the controlled cooling of a melt to develop some properties in the solid.
The evolution of crucibles is, therefore, linked to the types and amounts of materi-
als they needed to hold, the manner in which heat was supplied to or withdrawn from
them, and the availability of substances out of which the crucibles could be made. A
large variety of materials, both non-metals and metals, have been found suitable to
function as crucibles. Chemical, physical and mechanical interactions always occur
between the crucibles and the materials they hold. The extent of these interactions
determines the quality of the products obtained at the end of processing, the integ-
rity and longevity of the crucible itself, and above all, the economics of the process.
Seldom were there perfect crucibles, but workable selections have always been made.
The range of sizes and shapes of crucibles used in materials production and pro-
cessing is vast. This can be fully understood if we consider a more inclusive defini-
tion of a crucible as a confining solid boundary within which processes involving

DOI: 10.1201/9780429345562-1 1
2 Metal–Crucible Interactions

chemical reactions and transport of heat and mass occur to yield a physically sepa-
rable metal or an alloy in a certain state of purity and crystalline perfection. It then
becomes possible to consider any metallurgical reactor as a crucible.
Crucibles have been used in one form or other for thousands of years. The mate-
rials and forms of crucibles have evolved in response to the needs of processing
more and newer materials throughout history. The elements that occur in nature have
always existed, and people in every time period would have realized the existence
of some of them. But, the ability to identify, access and use the metals was slow to
evolve. The pace of this evolution has been determined by the availability of cru-
cibles, among other things. The choice of crucible was a challenge at the very begin-
ning of metals processing thousands of years ago and surprisingly, still remains a
challenge today. Only the metals that pose the challenge have changed.

1.2 HISTORY OF METALS PROCESSING

The three-age system of describing human history by dividing it into the Stone Age
(ca. 2,500,000–3200 BCE), the Bronze Age (ca. 3200–1200 BCE) and the Iron Age
(ca. 1200 BCE–100 CE) was propounded by Christian Jürgensen Thomsen, the
Danish antiquarian, in 1836. He divided prehistory into discrete periods based on
archaeological evidence of the prevalence or the exclusive or predominant use of
certain metals.

1.2.1 Stone Age
The evolution of humankind has been closely linked with the use of tools. The vari-
ety and types of tools possible, in turn, depend on the range of materials available.
Stone tools, many types of them but nevertheless only stone tools, both naturally
existing and improvised, were used in the incredibly long Palaeolithic and Neolithic
eras (ca. 2,500,000–3200 BCE).
The Palaeolithic period began 2.5 million years ago, with the first evidence of
human technology (stone tools), and ended with some stabilization in human societ-
ies marked by the invention of agriculture and the domestication of animals. The
Neolithic period, which corresponds to the first farming societies, extended from
6000 to 3200 BCE. Earthenware pottery and some shaped and polished stone tools,
such as axes, began to be used. Metals, especially copper, gold, silver, lead and tin,
had already made their appearance sporadically in the form of ornaments, charms
and small tools from around 9500 BCE.

1.2.2 Native Metals: Copper, Gold, Silver

That the eras of civilization are named after metals highlights the exceptional influ-
ence of metals on human life from the start. Interestingly, the knowledge that there
are metals and that they are useful was revealed to our ancestors by nature present-
ing some of the most important and useful metals in native forms. A metal in the
un-combined or elemental state in nature is considered to be occurring in the native
Metal–Crucible Interactions 3

form. Native gold was relatively common and occurred as small flakes separated
from the original deposits by weathering and transportation. Native silver and copper
were less common. Copper in the native form was invariably found in small amounts
but in more widely distributed locations (Bunch and Hellemans 2004). Native iron
was found in locations where meteorites had deposited it on the surface of the earth
in large masses of iron or as an alloy of iron and nickel. No occurrence of native iron
could be linked to processes in the earth’s crust.
Nearly10,000 years ago, copper was already familiar to humans. The earliest
known metallic artefacts are small pendants, beads and pins of copper from the
period between 9000 and 7000 BCE. Native copper was shaped by cold hammering
in the period from 7200 to 6600 BCE. Along with beads of native copper, beads of
fused lead, dated to 6500 BCE, have also been found in the ancient Anatolian mound
of Catal Hüyük (Darling 1990).
For about five millennia, copper was probably the only metal known to man and
fulfilled all the applications that required a metal. Although lead was also known at
a very early date, copper was the first metal to be put to use. In the very early stages,
humans were first tasked with the processing of metals rather than preparing them.
Native copper was hammered directly into small articles of jewellery or good luck
charms.

1.2.3 Primitive Crucibles
In North America, implements and weapons were produced from native copper by
cold forging with frequent annealing at temperatures below 800°C (Tylecote 1992).
Interestingly, those craftsmen knew, even in 4000BCE, that copper could be hard-
ened by deformation and softened by annealing, but they could not work out a means
of melting copper. Indications of cast copper ingots were found much later in these
locations. Excavations at the copper smelting site at Batan Grande in Peru have
revealed a row of smelting furnaces. These furnaces were blown by blowpipes rather
than bellows.
Between the seventh and fourth millennia BCE, most of the larger copper arte-
facts produced in the Middle East have a microstructure characteristic of crystals
grown from the melt. In addition to hammer and anvil, the crucible became essential
processing equipment in the earliest days of metallurgy.
The burning of charcoal for heating and intensifying the heat using draught,
either natural or forced, were techniques known to humans well before any process-
ing of metal was attempted. Various configurations were developed to implement
these techniques. When native copper was first melted intentionally, it would have
been heated from above by charcoal fire. Beneath the fuel heap, the metal was placed
in a clay-lined saucer-shaped depression in the ground (Figure 1.1). The molten metal
would have run together, collecting and forming a convex ingot. This is probably
close to the first ever crucible and furnace arrangement used for processing any
metal. Provision of a forced draught, through a tuyere connected to bellows (Figure
1.2), and also a chimney would have completed the rudimentary furnace arrange-
ment to attain a higher temperature. The arrangement is exemplified in the Minoan
4 Metal–Crucible Interactions

FIGURE 1.1 Primitive metallurgical furnace – crucible and blowpipe.

FIGURE 1.2 Primitive metallurgical furnace – crucible and tuyere.

furnace for copper smelting and casting shown schematically in Figure 1.3. The
Minoan civilization flourished from ca. 3000 BCE to ca. 1450 BCE on the island of
Crete and other Aegean Islands (Darling 1990).
To produce objects such as flat axes or mace heads, which were cast directly to
size, crucible furnaces were used. These furnaces (Figure 1.4) had a vertical cylin-
drical clay shaft. Air entered freely at the lower end, providing the required draught.
A hemispherical clay crucible, about 10 cm in diameter, was supported about half-
way up the shaft by a bed of charcoal packed into the base of the furnace (Darling
1990).
Gold was known in Egypt well before the fifth millennium BCE. Gold was
refined by fire created by the charcoal–blow pipe combination. Charcoal served as
the crucible. Gold (m.p. 1064°C) was melted and held molten so that all other metals
present, except silver, were gradually oxidized and separated.
Thousands of years later, after it was discovered in native form, copper was also
the first metal to be smelted from an ore, by ca. 3500 BCE in Egypt and Mesopotamia.
Metal–Crucible Interactions 5

FIGURE 1.3 Minoan furnace for copper (From Tylecote, R.F. 1992).

FIGURE 1.4 Chalcolithic crucible furnace (From Darling, A.S. 1990).

The list of firsts associated with copper is rather impressive. It was the first metal
(native copper) to be cast into a shape in a mould, ca. 4000 BCE, i.e. centuries before
it was smelted. Copper was also the first metal to be alloyed on purpose with another
metal, tin, to make bronze, ca. 3500 BCE.

1.2.4 Chalcolithic Age
The beginning of the Copper Age or Chalcolithic Age is marked by the emergence of
smelting methods to extract copper from its ores. The ores were bright green (mala-
chite: copper carbonate hydroxide), blue (azurite) and metallic looking sulphide ores.
It is surmised that malachite, used in Egypt as a green eye pigment, accidentally
found its way to a wood or charcoal fire and was reduced to metallic copper. Once it
was observed that fire could transform certain earthy and stony materials into met-
als, rocks or stones that were heavy or strongly coloured would have been thrown
6 Metal–Crucible Interactions

into or treated in fire to find out if they would yield metal. Apparently, many did, and
pyrometallurgy was born in the most elementary form.
Archaeological finds in the 21st century have revised the probable first date of
copper smelting to 500 years earlier. Copper smelting was probably practised from
5000 BCE in present-day Serbia. The fire obtained by burning the most common
combustible materials – wood and charcoal – is not only a source of heat but also
provides a remarkably gradable (tunable) reducing environment to the materials
thrown into it. The combination of heat and reducing environment readily yielded
certain metals, and these are the ones that were produced and so extensively used in
early history.
Crucibles used for copper smelting during the Chalcolithic period were wide,
shallow vessels made from the local clay, also used for domestic pottery, with melt-
ing temperatures around 1100°C. The charge was heated from the top using blow
pipes (Rehren 2003). Thornton and Rehren (2009) also reported on an unusual met-
alworking crucible from the prehistoric site (dated to the middle of the fourth mil-
lennium BCE) of Tepe Hissar in northeast Iran. The crucible was made of steatite
(Mg3(Si4O10)(OH)2). It was apparently used for the melting and alloying of arsenical
copper with lead. Steatite was also used in the middle Minoan period (ca. 2000 BCE)
as mould material for casting copper (Tylecote 1992). Steatite is easily carved. In
those early days in history, people had the knowledge of refractory minerals and the
skill to selectively incorporate these materials into crucible making depending on
process need (Darling 1990; Bunch and Hellemans 2004).

1.2.5 Bronze Age
The Copper Age, which could have begun as early as 5000 BCE, ended by 3200
BCE with the rising popularity of bronze. In the nearly 2000-year period from 3200
to 1200 BCE, the use of bronze (an alloy of seven parts of copper to one part of tin)
was so widespread as the principal material in the manufacturing of weapons and
implements that the period is given the name Bronze Age. Bronze is not only harder,
stronger and more corrosion resistant than copper but also melts at a lower tempera-
ture (950°C). At the normal melting point of copper (1085°C), bronze is far more
fluid, a property very conducive to casting intricate shapes accurately. The amount
of bronze used in tools was dwarfed by the quantities used for making weapons. Tin
ore does not normally occur with or near copper ores.

1.2.6 Iron Age
The Iron Age, the successor to the Bronze Age, extends from 1200 BCE to the end
of the first century CE. The Roman civilization was largely concurrent with the Iron
Age and existed for 12 centuries in Italy, from the eighth century BCE to the fifth
century CE.
The earliest iron samples are all of meteoric origin and are actually iron–nickel
alloys. Iron was probably known for centuries in the Middle East before people liv-
ing in present-day Armenia chanced upon a technique to enhance its properties.
Metal–Crucible Interactions 7

They discovered that reheating and hammering iron produced a material greatly
superior to copper or bronze. The discovery had military consequences.

1.2.7 Middle Ages
The Middle Ages lasted nearly 1000 years. They began in 476 CE, at the end of the
reign of Romulus Augustule, the last Roman emperor of the Occident, and ended in
1453 with the taking of Constantinople by the Turks. The progress of metalmaking
during the Middle Ages was marked by technology diffusion rather than new inven-
tions. Technologies fairly established in the East were becoming known and adopted
in Europe. The route was through the Middle East. One was the making of steel for
edge weapons, particularly swords. The production of cast iron and zinc metal was
also becoming more familiar in the West during this period. Sporadic attempts were
also made to build and use blast furnaces for producing molten iron and casting it. The
Catalan forge was perfected for ironmaking by around the eighth century in Spain.

1.2.8 Post-Medieval Period
The post-medieval period in metallurgy refers to the time period from1500 CE to the
beginning of the industrial revolution, i.e. 1750 CE. At the beginning of this period,
the famous refractory crucible from Hesse, a region in Germany, became available.
At about the same time, another specialized crucible was also made in Southern
Germany. It was made of graphite. These had a very similar design to the triangular
crucibles from Hesse. Reverberatory furnaces were developed, first for melting and
eventually for smelting. There was a gradual increase in the size of blast furnaces,
with more mechanization of air blowing by using water wheels.

1.2.9 Industrial Revolution and Later

In the period of the industrial revolution (from 1750 CE onwards), there were certain
major changes in the way metals were prepared and processed. The steam engine,
arguably one of the greatest inventions of the industrial revolution, added new power
and reliability to the blowers and made the blast furnace, as it is known now, pos-
sible. Initially, it was the Newcomen engine and later, in 1775, the much improved
steam engine of James Watt. The next was the Bessemer process for mass production
of steel from molten pig iron. The process as introduced in 1856 had issues, which
were eventually resolved. The removal of impurities from the iron via oxidation as
air is blown through the molten metal turned out to be a more general technique ame-
nable to much improvisation and modification for wider applicability. The invention
of the electric battery in 1800 by Alessandro Volta led to the eventual development of
the field of electrochemistry. Humphrey Davy used electricity to reduce many of the
alkali and alkaline earth metals, Faraday came up with the laws of electrolysis, and
Hall and Héroult revolutionized metals extraction by inventing the process named after
them for electrolytically reducing alumina to aluminium. Amongst all these, a German
inventor, Joseph von Fraunhofer, invented the spectrometer in 1814, and this became a
8 Metal–Crucible Interactions

new tool to discover more metals. The discovery of metals by this and other classical
methods continued, and by 1939, all the naturally occurring metals, as well as other
elements, had been discovered. The majority of these metals, however, did not appear
to have any applications, and there was no incentive either to investigate them in detail
or to produce them in good quantities. But, that lasted only until World War II.

1.2.10 World War II and Later

The war and the concurrent development of atomic energy for military and civil pur-
poses changed the metals scenario totally. Many metals that had been discovered and
all but forgotten became relevant and important. The eventual advances in electron-
ics and aerospace technologies added more metals to the list of materials that must
be developed urgently. The period between 1940 and 1970 was a watershed moment
for metals and materials development. The metals originally produced and used for
centuries came to be called common metals, and the newly important metals came
to be known as less common metals or rare metals. In the process metallurgy of less
common metals, the existing collection of processing techniques and the associ-
ated hardware that was so effective with the common metals proved inadequate.
Generally, the less common metals cannot be processed while exposed to air. Either
reduced pressure or a protected atmosphere is necessary. The temperatures needed
are usually higher, and crucibles and other hardware should be chosen keeping their
pronounced reactivity in mind. Rare metals, in order to be useful for their special
functions, have to be in a state of very high material purity. Due to one or more of
these factors, new techniques of processing had to be developed in metallurgy in
the post-war era (Hampel 1961;Jamrack 1963; Winckler and Bakish 1971). Halide
metallurgy, controlled atmosphere processing and vacuum metallurgy became the
mainstream techniques. Considerable progress was achieved in all of these. For the
rare metals, it was not just the ore processing and obtaining the metal that was dif-
ferent and difficult. Equally challenging has been the conversion of the as-produced
metal into a usable form. That, incidentally, remains a work in progress to date.
The 20th century has been eventful in the evolution of materials and techniques,
not only as regards the final products but also with respect to the technology enabling
materials, a new name given to ceramics and refractories by Lee and Moore (1998).
As crucible materials, the development of ceramics in the last 50 years has been
remarkable, and their amenability to modification and enhancement has been the true
reason behind the hope and success in many difficult situations of metals processing.

1.3 NON-FERROUS METALS
The word smelting is probably the most used term in any extractive metallurgy text.
It represents the process by which a metal is obtained in the elemental form, or even
as a simple compound, from its ore. Smelting involves heating beyond the melting
point, usually in the presence of oxidizing agents, such as air, or reducing agents,
such as coke. As noted earlier, copper was the first metal to be smelted in ancient
times, followed by tin, lead and silver.
Metal–Crucible Interactions 9

Over thousands of years of metalmaking, certain devices have evolved, in one

form or another, as the key features of the smelting furnace. One is the tuyere, a
pipe through which air can be forced into the smouldering coke, then the bellows or
blower, an implement for forcing air into the tuyere, and the hearth, a place where
the burning fuel can be contained over or against the tuyere opening. Traditionally,
hearths have been built of mud-brick, fired brick, stone, or later, refractories of every
description.

1.3.1 Copper
The furnaces for producing non-ferrous metals and the small shaft furnaces used for
ironmaking in the Roman period are similar. These were of square cross section,
about 0.8 m on each side, and 2 m high externally. The main difference between the
furnaces for different metals was in the hearths, in the arrangements for separating
and tapping slag and metal (Tylecote 1992). A shaft furnace (Figure 1.5) is a general-
purpose fixed bed reactor with an upright working chamber and is used to roast or
smelt lumped materials. The heat required for the process is provided by the combus-
tion of fuel, which usually is also a reactant. The charge introduced at the top of the
furnace descends against the ascending stream of gases introduced in the bottom,
resulting in direct contact between the charge and the hot gases. The molten products
collect in the hearth at the bottom or flow out to a fore-hearth.
Pyritic (sulphide) copper ore was first roasted, in the open, over burning wood
in shallow cavities in the ground. The roasting usually went on for over a month
before completion. The oxide was then mixed with charcoal and fluxes and reduced
in shaft furnaces. The ore was charged with alternate layers of charcoal into a fur-
nace, and air was blown with bellows through the tuyeres. The furnaces were lined
with a clay–charcoal mixture (refractory lining) rammed into place at the bottom
and smeared on to the inner walls.

FIGURE 1.5 Shaft furnace.

10 Metal–Crucible Interactions

FIGURE 1.6 Reverberatory furnace.

The entire smelting process was rather elaborate, with repeated furnace treat-
ments for days on end. Iron oxide, which was inevitably in the charge, was also
reduced. Later, in 18th-century Europe, copper extraction from sulphide ores contin-
ued along the lines described (Tylecote 1992) as well as by the use of the reverbera-
tory furnace.
Chronologically, the reverberatory furnace was developed later than the shaft fur-
nace and the retort. In the shaft furnace, the fuel is blended with the charge to pro-
vide not only the heat but also reagents. In retorts, the charge is physically isolated
from the source of heat and the products of combustion. The reverberatory furnace,
represented schematically in Figure 1.6, comes midway between the two. A rever-
beratory furnace isolates the material being processed from contact with the fuel but
not from contact with combustion gases. The term reverberation is used here in the
sense of rebounding or reflecting of the flame and heat.
Reverberatory furnaces were first used in the medieval period to melt bronze for
casting bells. They were used for smelting only in the late 17th century. Starting in
about 1678, Sir Clement Clerke and his son Talbot built reverberatory furnaces for
smelting lead and a decade later, for smelting copper and also tin. These units were
1.7–1.8 m long by 1 m wide in the inside. The hearth was made of sand with some
clay, and the ore was fluxed with some lime.
In the second half of the 19th century, the major development in copper extraction
was the use of the Bessemer process for the rapid oxidation of copper mattes, which
had been made by concentrating and separating the sulphides in the blast furnace or
a reverberatory furnace. The electrolytic refining of a slab of crude copper from the
converter, invented by Elkington in 1865, was first successfully put into operation in
South Wales in 1869. The modern flowsheet for copper extraction was thus in place.

1.3.2 Brass
The term brass was used to denote brass as well as bronze in olden days. Even
Biblical references to brass are known to actually mean bronze. This confusion per-
sisted until recent times. Brass, an alloy of copper and zinc, was unknown in the
Metal–Crucible Interactions 11

Egyptian and Babylonian civilizations, as well as to the Greeks and the Romans
before the first century BC, while it was already in use in Palestine.
Starting from the Roman period, a process called cementation was used to make
brass. Solid copper metal, zinc oxide or carbonate, and charcoal were placed in a cru-
cible, which was heated to about 1000°C. Elemental zinc, which forms as a vapour
by the carbothermic reduction of zinc ore at 1000°C, diffuses into metallic copper
and converts it to brass. The ceramic crucible was kept closed with a lid to prevent
zinc vapour from escaping before it had a chance to combine with copper. The use
of porous ceramic vessels was important for the vessel to lose gas through the walls
and thus prevent the build-up of excessive pressure. Incidentally, the process for car-
rying out cementation for brass did not change markedly until the 19th century. The
ceramic used was a clay.
The crucible has an interesting function in the ancient process known as cupel-
lation. Native silver exists but is not commonly seen. It is usually found in nature
combined with other metals, e.g. in the lead minerals galena (lead sulphide) or cerus-
site (lead carbonate). Silver is produced primarily by smelting and then cupellation
of argentiferous lead ores. Smelting yields silver and lead combined. Lead melts at
327°C, lead oxide at 888°C and silver at 960°C. To separate the silver, the alloy is
melted in an oxidizing environment at 960°C to 1000°C. The lead oxidizes to lead
monoxide, known as litharge. The liquid lead oxide is removed or absorbed by capil-
lary action into the hearth linings. The base of the hearth was made in the form of
a saucepan and covered with an inert and porous material rich in calcium or mag-
nesium, such as shells, lime or bone ash (Bayley and Eckstein 2006). A calcareous
lining is essential because lead reacts with silica (clay) to form lead silicate. Lead
silicate is viscous and impedes the absorption of litharge. Lead does not react with
the calcareous materials. It is captured by the materials purely by the physical phe-
nomena of absorption. Some of the litharge may evaporate, but the rest is absorbed
by the porous earth lining to form ‘litharge cakes’. The litharge could be reduced
back to lead separately.
Cupellation was known and used since the early Bronze Age. However the vessels
(cupels) to carry out this process were first made only in the 16th century (Rehren
2003). The cupel is a porous container, usually made of bone ash or ceramic, and
resembles a small egg cup. By the Middle Ages and the Renaissance, cupellation
was not only the common process for refining precious metals but also a method for
assaying precious metals. The process had remained virtually unchanged since its
inception (Tylecote 1992). The cupellation furnace bottoms are separately smelted
to recover the litharge and any silver contained in them. Cupellation is still in use.

1.3.3 Tin
In England, the tin ores of Devon and Cornwall were smelted in shaft furnaces
(Tylecote 1992). In the 16th century, the furnaces for tin smelting were about 2.5 to 3
m high and 60 × 30 cm internally (Figure 1.5). They were constructed of large sand-
stone blocks and were lined with clay. The downward inclined tuyere was located
at the back opposite the taphole. A sloping sandstone slab was the hearth, and the
12 Metal–Crucible Interactions

taphole was always open. The metal and slag separated out in the large (0.3 m long)
fore-hearth, and the slag overflew from this. A charcoal cover was maintained over
the tin to protect it from oxidation. The tin metal, smelted with water power and local
fuel (probably peat), was re-smelted (or refined) and recast into ingots. The metal
was ladled out of the fore-hearth and refined by drossing in an iron ‘pot’, similar
to the 19th-century lead refining pot. After this, it was ladled into granite moulds.
Charcoal was the main fuel. Smelting in a shaft furnace with charcoal followed by a
limited amount of drossing to reduce the iron content yielded very pure (99.9%) tin.
The reverberatory was not introduced here until 1700 CE.

1.3.4 Zinc
Metallic zinc is not an easy metal to produce because (i) zinc oxide cannot be reduced
by charcoal at temperatures below 1000°C and (ii) zinc boils at 923°C, and the metal
is a vapour at the temperatures required for reduction. The reduction temperatures
of around 1000°C can be easily obtained by means of a bellows blast, but special
arrangements were necessary to capture zinc vapour and collect the zinc metal. Zinc
must be condensed to liquid metal before it comes into contact with enough air and is
oxidized. This was the reason for the delayed appearance of zinc metal in the histori-
cal narrative of early metallurgy. Besides, given the ease with which calamine brass
could be produced, there was not much incentive to use metallic zinc for this purpose.
In India, at Zawar near Udaipur in Rajasthan, zinc metal was produced in large
quantities, beginning in the 10th century. Large heaps of small retorts were found
there. They consisted of pointed elliptical masses of vitrified clay 25 cm long and
15 cm in diameter. These were all closed at one end but open at the other, with the
remains of 2.5 cm tubes inserted in them. The heaps at Zawar suggest the extraction
of about 100,000 t of metallic zinc (Darling 1990). A reconstruction of the produc-
tion facility indicates that as many as 36 of these retorts were placed in a square
furnace. Their openings projected through a square perforated plate such that the
zinc vapour condensed and ran down into saucer-like vessels below. The retorts were
charged with zinc mineral and charcoal.
In 1807, the Abbé Dony established a works at Liege, Belgium that used hori-
zontal retorts, shown schematically in Figure 1.7, in horizontal rows (Darling 1990).

FIGURE 1.7 Horizontal clay retort for zinc production (From Darling, A.S. 1990).
Metal–Crucible Interactions 13

A charge of ground calcined ore and coke was placed in the clay retorts. Zinc was
collected in clay tubes fitted to the ends of the retorts after charging. As late as the
mid-1950s, nearly half of the world’s total zinc was produced in horizontal retorts
of Dony’s type. In the 1930s, the vertical retort furnace was developed by the New
Jersey Zinc Company, and later, in the 1950s, the invention of the lead splash con-
denser made the blast furnace a practical and economic proposition for the recovery
of zinc (and lead).

1.4 IRON AND STEEL

In the chronology of metalmaking, iron entered the scene a little later than copper
and other major non-ferrous metals, but from then on, the metal and the techniques
and materials associated with its processing had an overwhelming influence on met-
allurgy. A technique or material validated for the processing of iron and eventually,
steel proved useful for other, non-ferrous metals as well with a few tweaks.

1.4.1 Beginning of Iron Smelting

When iron is smelted in the same way as copper or tin, the resulting metal is not in
a form that is remarkably stronger or harder. As in the case of copper, easily smelted
iron ores could have been the source of iron in the early days. The earliest iron pro-
duced was probably formed accidentally in fires burned while obtaining red ochre
for use as paint (Bunch and Hellemans 2004).
Around 1500 BCE, the Hittites began smelting iron in Anatolia using clay-lined
furnaces (Forrester 2016). Iron in malleable form was obtained by direct reduction of
its ores with charcoal. Two types of furnace were used for iron making (Figure 1.8,
Figure 1.9): bowl furnaces and shaft furnaces. Bowl furnaces were made by scoop-
ing out a hole in the ground and providing for air from bellows to be introduced
through a tuyere. Stone-built shaft furnaces also used tuyeres but generally relied on
natural draught. Originally, the process was carried out in pits lined with refractory

FIGURE 1.8 Bloomery.

14 Metal–Crucible Interactions

FIGURE 1.9 Roman bowl furnace (From Tylecote, R.F. 1992).

clay or on stone hearths. The bottom and sides of the hearth were lined with a thick
coating of charcoal dust. In these furnaces, smelting was carried out by laying out
a bed of red-hot charcoal on which iron ore mixed with more charcoal was added
from above to obtain alternate layers of iron ore and charcoal. In such rather primi-
tive arrangements, temperatures of only up to 1150°C could be reached. Reduction
of the ore occurred, but the elemental iron formed was low in carbon and remained
solid in the form of an incandescent bloom. Some slag and pieces of charcoal invari-
ably remained entrapped in the bloom. The bloom needed to be pulled out from the
hearth. By hammering this mass, the non-metallic constituents were squeezed out,
and the metal was consolidated. The furnace for making this iron bloom is known as
a bloomery. A bloom might have weighed about 8 kilograms during Roman times.
This process is the oldest iron making process and has been in use, more or less
uninterrupted, from as early as the second millennium BC until the early 20th cen-
tury. The making of iron artefacts involved heating blooms in a fire and hammering
the red-hot metal to produce tools and weapons. Iron obtained in this way is known
as wrought iron.
Over time, the bloomeries were improved by using bricks and natural refractory
stone for wall linings. A forced draught was supplied by bellows. Many bloomer-
ies used low shaft furnaces, i.e. blast furnaces with a short shaft. Early furnaces
were lined with natural sandstones or were even carved directly into stone ledges.
By about 1300, higher temperatures were reached by using manually operated bel-
lows to create the updraft in tall furnaces. By 1325, waterwheel-driven bellows had
enabled a more powerful and sustained blast, which led to greater heat generation.
With this improvement, bloom size could be increased. Blooms weighing over 100
kilograms were pulled out through the top of the shaft. The continuous operation
enabled by the application of water power to bellows eventually made the blast fur-
nace possible.
A popular version of the bloomery was the Catalan forge, an improved furnace
that was invented around 700 CE in Catalonia and remained in use in Spain until
the 19th century. The furnace, shown schematically in Figure 1.10, consisted of a
small square cavity or hearth with an inclined copper tuyere designed to direct the
blast onto the mixture of charcoal and ore. The nozzle of the device that provided
the blast opened into the tuyere, thus drawing much more air on the principle of
Metal–Crucible Interactions 15

FIGURE 1.10 Catalan furnace.

the injector. Another variant, the high bloomery furnace, had a taller shaft, e.g. the
3-metre high Stückofen. The blooms produced in this furnace were large and were
usually removed through a front opening in the furnace.

1.4.2 Blast Furnace
The blast furnace is a vertical shaft furnace in which air introduced under pressure
into the bottom of the furnace burns the fuel, usually coke, present in a mixture of
metallic ore, coke and flux fed into the top. The ensuing smelting reactions pro-
duce liquid metal and molten slag, which collect in the hearth at the bottom and are
recovered. Blast furnaces are used to produce pig iron from iron ore for subsequent
processing into steel, and they are also used for production of lead, copper and other
metals. Intense combustion is maintained by the current of air under pressure.
In Europe, the shaft furnace, the precursor to the iron blast furnace, developed
gradually over the centuries from the small furnaces of Roman times. Increase in
the height of the furnace, coupled with a continuous and strong flow of air sustained
by mechanical bellows, resulted in the much higher furnace temperatures needed to
produce the carbon-laden, lower-melting pig iron. The molten iron could be poured
into moulds or could be remelted and cast into any shape. Thus, the shaft furnace
of the bloomery metamorphosed into the blast furnace of the iron works. It was not
until the 14th century that smelting furnaces capable of melting iron were built in
Europe.
An incentive for the development of the blast furnace was the need for cast iron
to make cannons. A cast iron gun was not only cheaper but also better than a bronze
gun. Blast furnaces are necessary for producing cast iron. The first generation of
furnaces (Tylecote 1992) were square externally, built of stone and lined with sand-
stone. This type remained prevalent up to the 19th century. The higher tempera-
tures achieved by 17th-century blast furnaces led to the use of firebricks made from
kaolinite-rich clays.
16 Metal–Crucible Interactions

The main difference in operation between a bloomery and a blast furnace was in
the tapping of liquid iron in addition to slag from the bottom of the furnace. A 2-m
tall bloomery might be used to produce liquid iron by increasing the fuel/ore ratio to
make the atmosphere more reducing and carburize the iron, which would then melt
at about 1200°C instead of 1540°C. During this, the slag might become very viscous
as its iron content was reduced, and it would not run easily. The addition of crushed
limestone and maintaining a working temperature of at least 1300°C could result in
a more fluid limy slag. Typically, pig iron has a carbon content in the range 3.8–4.7%,
along with silica and other constituents of dross. As produced, it is very brittle and
has little use directly as a material. Charcoal was the only furnace fuel until the 17th
century, but its depletion and public concerns over the denuding of forests forced an
eventual shift to coal. This changeover from charcoal to coke and later, the advent of
the steam engine to power the air blowing cylinders of the blast furnace made much
larger blast furnaces possible.
Throughout the 18th century, there was a steady increase in the size of the blast
furnace. The maximum height increased from about 7 m in 1650 CE to 13.5 m in
1800 CE. By the middle of the 18th century, the shaft lining was usually brick,
although the bosh and the hearth were made of refractory sandstone.. Brick was used
for the stack in Maryport (1752 CE) and Low Mill (1761 CE). Moulded slag blocks
were used to build the shaft of a Swedish furnace at Soderfors.
The early development of metallurgical blowing engines was one of the reasons
why the Chinese gained early superiority in the cast iron tradition. The blast furnace
was being used in China long before it became commonplace in Europe. The blast
furnace was also in use in other countries of the East. The furnaces were made
of sun-baked fireclay, sometimes from bricks and sometimes monolithic, with the
better-grade material used near the hearth (zoning). The tuyere was of best fireclay.
The unroasted ore was smelted with charcoal and without flux. The slag was free
running and contained some iron.
Until World War II, the hearth walls were similar to a simple chimney-like
structure built of fireclay brick. Many refractory combinations were tested. Until
the 1960s, the hearth was composed of carbon wall and ceramic (fireclay) pads.
This was further improved by using carbon hearth pads in addition to carbon hearth
walls with a provision for cooling. In another arrangement, the combination was
low-conductivity ceramic in contact with the metal but cooled by a carbon backing.
The low-cost firebrick was eventually replaced with high-alumina brick.
The modern blast furnace is a tall, vertical shaft that consists of a steel shell with
a refractory lining of firebrick and graphite. A schematic is shown in Figure 1.11.
At the bottom is a parallel-sided hearth to collect liquid metal and slag. Above
the hearth is the bosh, an inverted truncated cone. Air is blown into the furnace
through water-cooled nozzles or tuyeres made of copper and located at the junc-
tion of the hearth with the bosh. The temperatures are the highest in the bosh and
hearth. Here, the lining is usually made of carbon bricks that are manufactured
by pressing and baking a mixture of coke, anthracite and pitch. Carbon is more
resistant to the corrosive action of molten iron and slag than the aluminosilicate
firebricks. For the remainder of the lining, aluminosilicate firebricks are used.
Metal–Crucible Interactions 17

FIGURE 1.11 Iron blast furnace.

In the bosh parallel, bricks containing 63% alumina are used, and 45% alumina
bricks are used for the stack.
During the 1960s and 1970s, the major changes in blast furnace linings were the
replacement of fire clays with more refractory (higher-alumina) materials for the
bosh and stack and a range of carbon materials for the hearth. Some of the new lin-
ings consisted of castables, silicon carbide and a thick ceramic layer for the upper
portion of the hearth pad. The use of staves, which are 2 m × 1 m × 250 mm cast iron
blocks embedded with steel pipes and ceramic inserts, also became common. In
some of the large blast furnaces, the cooling and lining design relies on the combina-
tion of machined copper plate coolers and high-conductivity graphite.

1.4.3 Cast Iron
Cast iron is the name for iron–carbon alloys containing carbon in the range from 1.8
to 4 wt.% and silicon at 1–3 wt.%. Depending on its composition, cast iron melts at
1127–1204°C. Its properties include low melting temperature and good fluidity and
castability. The oldest cast iron artefacts date to the fifth century BC, discovered in
the present-day Jiangsu in China. Certain factors naturally favoured China being
the earliest cast iron producer. The iron ore in China contained a high proportion of
phosphorus, and residual phosphorus lowers the melting temperature of iron. They
also had good supplies of refractory clay suitable to make kilns that could endure
high temperatures. Furthermore, they invented and used a form of piston bellow,
which provided a steady draught necessary for reaching and maintaining high kiln
temperatures.
Cast iron was recognized as a superior material for making cannons. In Europe,
cast iron production started only in the 13th century CE. The remelting of pig iron
from the blast furnace in a reverberatory or air furnace was the process used to
18 Metal–Crucible Interactions

produce high-quality cast iron for making cannons in Britain in the mid-18th century
(Tylecote 1992). Earlier, iron was cast direct from the blast furnace. The quality of
the cast metal was determined by the smelting regime itself, which was difficult to
control. When the raw pig iron from the blast furnaces was poor in quality and con-
taminated with slag, remelting in a reverberatory furnace enabled the slag to float out
because a higher temperature was available. The melting conditions being slightly
oxidizing, some of the carbon was also eliminated, giving a greyer and softer iron
with high fluidity. Cast iron was also made by melting pig iron in cupolas. The cupola
is another version of the shaft furnace, mainly used for remelting before casting.

1.4.4 Steel
Even before the start of the Iron Age, people living in present-day Armenia chanced
upon a technique to enhance the properties of iron. It was all meteoric iron at that
time. They discovered that reheating and hammering iron produced a weapon mate-
rial very much superior to copper or bronze. With this secret knowledge of making
iron weapons, they enjoyed unequal advantage in their military campaigns across
the Near East and the eastern Mediterranean (Bunch and Hellemans 2004). They
had actually made weapons of steel, probably without realizing it, just as in later
years, the makers of the famous Hessian crucibles used mullite as the material, prob-
ably without realizing it.

[Link] Cementation Steel
Steel is a mixture of iron and carbides of iron containing 0.5–1.5% carbon overall.
One characteristic of steel, that the simple operation of heating it to redness and
plunging it into water hardens it significantly, was a great reason to convert iron to
steel. Cementation steel is made by a process that converts low-carbon iron to steel
by letting carbon permeate into it. It may be recalled that in the context of brass,
cementation refers to permeation of zinc into copper. Steel was sometimes produced
in bloomeries when the temperature was high and charcoal was present in excess,
conditions conducive to picking up of some carbon by iron. More often, it is made
by heating iron in contact with carbonaceous mixtures. Even though exotic carbon
sources like horn dust were used in the 10th century CE, the process had already
been practised using more mundane materials many centuries earlier.

[Link] Crucible Steel Ancient Practice

Crucible steel was also made by melting together pig iron or cast iron, iron, some-
times steel, and often sand, glass, ashes and other fluxes in a crucible. In olden
days using charcoal or coal fires, temperatures high enough to melt steel or low-
carbon iron could not be reached, but temperatures high enough to melt pig iron
were reached. The process therefore consisted of melting pig iron, and by soaking
wrought iron or steel in the liquid pig iron long enough, the carbon content of the pig
iron was reduced as it slowly transferred into the wrought iron. Crucible steel of this
type was produced during the medieval era in south and central Asia. This steel was
then forged, filed or polished to make weapons.
Metal–Crucible Interactions 19

[Link] Puddling Furnace
In 1784, inventor Henry Cort unveiled his pudding furnace, which would convert pig
iron to malleable iron. In a coal-fired reverberatory furnace, Cort melted a charge of
pig iron, to which iron oxide was added as a slag former. The resultant ‘puddle’ of
metal was agitated to speed up the removal of the impurities, carbon, silicon, phos-
phorus and manganese, by oxidation. The ensuing change of composition raised the
melting point of the metal. While the slag remained fluid, the metal was rolled as a
solid mass, formed into balls and taken out of the furnace. The slag was squeezed
out of the ball by hammering. For a certain time period, puddling furnaces were
able to provide enough iron to meet the demands for machinery and construction.
Incidentally, the Eiffel Tower, which was constructed from 1887 to 1889, was made
using puddled iron, approximately 7300 tons of it.

1.4.5 Bessemer, Open Hearth, Linz–Donawitz (L-D) Processes

Henry Bessemer announced in 1856a process for the production of steel from molten
pig iron (Gale 1990). Another technique, the open hearth furnace for steelmaking,
came into the picture a decade later, in1865. In the Bessemer process, the removal
of impurities from the iron occurs by oxidation due to air blown through the molten
iron. The oxidation, being exothermic, also raises the temperature of the melt and
keeps it molten. An American inventor, William Kelly, had also asserted his inde-
pendent discovery of the ‘Bessemer process’ back in 1851(Tylecote 1992)
Oxygen in the air blown into the pig iron melt oxidizes carbon and also the impu-
rities silicon, manganese and phosphorus, to form the respective oxides. Oxides of
carbon escape as gas and the other oxides are removed as the slag. The slag is com-
posed of the oxidation products silica and manganese oxide, some iron oxide and
also a certain amount of the refractory lining. The process was first carried out in
the vessel shown schematically in Figure 1.12. The brick linings were made of pure

FIGURE 1.12 Bessemer’s fixed vessel (From Tylecote, R.F. 1992).

20 Metal–Crucible Interactions

sandstone or silica (SiO2), the only materials then available for service at 1600°C for
any length of time. Silica is chemically acidic and reacts on contact with the alkaline
materials lime and magnesia. Bessemer, like all other furnace builders, had used
siliceous materials for his linings, and these (acidic) linings would not combine with
the acidic phosphorus pentoxide (P2O5) that forms during the oxidation of phospho-
rus. Thus, phosphorus was not removed from the steel. This situation confined the
use of the Bessemer process to only non-phosphorus ores, i.e. those containing less
than about 0.05% P. The ores that accounted for most of the iron production of 19th-
century Britain had phosphorus content exceeding this limit and were unsuitable for
the Bessemer process. For solving the problem of phosphorus in steel, neutralization
of the acidic P2O5 with lime was considered. The basic slag so produced tended to
consume the siliceous lining also. The need for a basic lining was strongly indicated.
In 1877–1888, Sydney Gilchrist Thomas and Percy Gilchrist checked the effec-
tiveness of a basic lining using an 8-lb converter lined with limestone bonded with
sodium silicate (Lee and Moore 1998). With this lining and a slag made basic by add-
ing limestone, phosphorus was removed from the metal and poured off with the slag.
They then developed a lining of fired dolomite mixed with tar, which was operated
successfully in the larger converters at Middlesborough in 1879. The first Bessemer
furnace was built in the United Kingdom in 1884; it had a tarred dolomite hearth.
During operation, there is considerable movement of the entire vessel through the
positions shown schematically in Figure 1.13.

[Link] Open Hearth Furnace

The open hearth furnace (OHF) was developed by Carl Wilhelm Siemens and Pierre-
Émile Martin, and the process is also known as the Siemens–Martin process. The
open hearth process is another method to oxidize carbon and other impurities out of
pig iron to produce steel. The OHF itself consists of a shallow, rectangular hearth,
like the hearth of a reverberatory, to hold the charge, the liquid steel and the slag,

FIGURE 1.13 Bessemer converter positions.

Metal–Crucible Interactions 21

relying on external heating that uses the heat of combustion of gaseous or liquid fuels
by preheated combustion air. Typically, the OHF is first charged with light scrap,
and heated using burning gas. Once the charge has melted, heavy scrap is added,
together with pig iron from blast furnaces. Once all the charge has melted, limestone
and other slag-forming agents are added. Iron oxide and other impurities provide the
oxygen to burn away the carbon in the pig iron. To provide more oxygen in the heat,
iron ore can also be added.
This is a batch process, much slower compared with the Bessemer process and
thus, easier to control. It can be terminated when the target carbon content has been
achieved. Attempts were made to address the phosphorus problem using basic refrac-
tories. The basic OHP was first operated at Le Creusot, France, in 1880. However,
the OHP was not very effective at removing all the phosphorus.
Apart from the phosphorus problem, the brittleness caused by nitrogen in the
blowing air prevented Bessemer steel from being used for structural applications.
Nitrogen was not an issue in the steel produced by the OHP. Open hearth steel was
suitable for structural applications, and the process started gaining acceptance in
the year 1865. For about 100 years, the open hearth and Bessemer-based processes
jointly contributed to most of the steel that was made. Then came the L-D and elec-
tric arc furnaces, which represent the current steel making technologies.

[Link] Linz–Donawitz (L-D) Steel

Doing away with the bottom tuyeres, the L-D process uses a vertically positioned
lance, lowered from above, which is able to blow oxygen onto the surface of the mol-
ten metal. This process was developed in Austria by the Linz–Donawitz companies
following the initial demonstration of its working by Robert Durrer and Heinrich
Heilbrugge in 1948.
Basic oxygen steelmaking is more similar to the Bessemer process (decarburization
by blowing oxygen as gas into the melt) than to the open hearth (burning the excess
carbon away by adding oxygen-bearing substances into the charge). The L-D process
simultaneously addressed both the phosphorus and the nitrogen issues by having a
basic refractory lining and using oxygen blow instead of air blow. The L-D process
is often referred to by other names. Basic oxygen steel making (BOS), basic oxygen
process (BOP), basic oxygen furnace (BOF), oxygen steel making (OSM), oxygen con-
verter process are other names for the Linz–Donawitz (LD) steel making process.

1.5 REFRACTORIES IN THE 20TH CENTURY

Certain naturally occurring inorganic materials have been used as crucibles or in
lieu of crucibles for metals processing since the Bronze Age. Locally available clays,
stones, river sand, sandstones, granite, fire clays and fire bricks have all been used
at various stages, and these were all that were generally accessible for use up to the
post-medieval period. This rather ordinary-looking inventory of crucible hardware
was at times punctuated by the appearance of special refractories: a refractory con-
sisting of steatite (Mg3(Si4O10)(OH)2) in the Bronze Age, the Hessian crucibles made
of mullite and even graphite in the post-medieval period. Systematic and directed
22 Metal–Crucible Interactions

development of materials for crucible functions coincided loosely with the industrial
revolution, when furnaces that started operating at consistently higher temperatures
became common. The rapid developments of the 20th century in the production of
metals and many other types of materials have been closely interlinked with concur-
rent developments and advances in the refractories.

1.5.1 Silicates
Up to the end of the 19th century, there was a heavy reliance on natural materials,
such as rocks and fireclays, for use as refractories. Because of their variable com-
position and properties, the predictability of their somewhat modest performance
was also limited. Fireclay was probably the first fired refractory to be used. Being
plastic when mixed with water, it could be readily shaped into crucibles convenient
for small batch melting of metals at moderate temperatures. Clay-derived alumino-
silicate refractories are acidic in their chemical nature and therefore, were useful to
contain acidic slags and fluxes reasonably well. They proved mostly adequate during
the early industrial revolution. Silicate was the first lining material for the Bessemer
converter. In England, crushed Dinas rock from Wales was utilized from 1822 to
form nearly pure silica brick cemented together by the smallest possible amount of
lime as binding flux. These bricks, developed by William Weston Young, were the
most suitable refractory available for lining acid open hearth furnaces, where tem-
peratures were much higher than in Bessemer converters (Lee and Moore 1998). The
temperature capability of the refractories was stretched to the higher side by increas-
ing their alumina content. The reason was highlighted in 1924 when the binary
Al2O3–SiO2 phase diagram was determined by Bowen and Greig (1924).
The 1920s and 1930s can be regarded as the golden era of refractory science.
The development of X-ray diffraction (XRD) techniques for elucidating the crystal
structures of clays and oxide–silicate systems, and the determination of the impor-
tant binary and ternary phase diagrams, happened during this era. The publication
by Hall and Insley (1933) of a compilation of phase diagrams, apart from being an
invaluable resource, is also symbolic of this renaissance.

1.5.2 Dolomite
The first use of dolomite as a refractory was in 1878, when Sydney Gilchrist
Thomas lined his experimental Bessemer converter with tar-bonded, burned dolo-
mite. The production of burned (fired) dolomite for the steel industry began in
1901. Dolomite was used in the form of metal-cased magnesite for open hearth
and electric arc furnace (EAF) walls and as unfired, chemically bonded magnesite
bricks until about 1930.

1.5.3 Chrome–Magnesite
It was becoming known in the 1930s that bricks made from graded mixtures of
chrome ore (chromite, (Mg,Fe)(Cr,Al)2O4) and dead-burned magnesite showed
Metal–Crucible Interactions 23

synergistic improvement in properties. For instance, refractoriness under load of

chrome–magnesite was much higher than that of either material separately. Silica
roofs in all basic open hearth furnaces were eventually replaced by the superior
chrome–magnesite roofs.
Chrome refractories are chemically neutral. They were used in basic open hearth
furnaces later in the 19th century. In direct-bonded chrome refractories, the silicate
films, which surround the chrome ore particles and periclase grains in the conven-
tional brick, are replaced by direct periclase–spinel and periclase–periclase bonds.
In this form, they have much improved hot strength, more resistance to slag attack
and greater dimensional stability.
With the increase in the number of BOF installations, standard basic bricks con-
taining magnesia, dolomite and chromite aggregates and matrix powders containing
tars and pitches were developed. These were referred to as ‘tar-bonded’ and ‘pitch-
bonded’ basic brick. In early 1950s, pitch-bonded dolomite and later, magnesia-
enriched tar-bonded dolomite were developed for the BOFs. The original purpose of
tar in refractories was as a binder and a coating to protect the basic grain, which is
prone to hydration. It was soon realized that under the conditions of BOF steelmaking,
magnesia and dolomite linings lasted longer if they contained carbon. Unfired pitch-
and tar-bonded magnesia containing additions of graphite came into universal use.
The use of magnesia refractories in combination with carbon started in the early
1970s (Biswas and Sarkar 2020). A decade later, the development of resin-bonded
magnesia – graphite, first with higher carbon content and then with the addition of
antioxidant, further improved the life of BOF linings. Late in the 1970s, a product
containing phenolic resin as a cold bond for a mixture of periclase, graphite and
metallic antioxidants came to be used in Japan. Resin-bonded basic brick not only
has superior properties vis-à-vis the tar-bonded variety but also is more acceptable
environmentally. Fused magnesia with larger size crystals than sintered magnesia
was recently introduced to further enhance the corrosion resistance.
The oxide–carbon–metal composite bricks are used in EAFs and BOFs and also
in ladle slag lines. The oxide confers oxidation resistance, and graphite provides
increased thermal conductivity, non-wettability and slag resistance. The carbon
bond is developed in situ in these refractories on firing them in the furnace. The
resin or pitch decomposes, yielding a fine pyrolytic carbon bond.

1.5.4 Graphite
Natural graphite was first used as a refractory material in crucibles in the 15th cen-
tury in Bavaria. Gousse Bonnin obtained a patent in 1769 for a crucible made of
equal parts of clay and graphite (termed black lead or plumbago). The manufac-
ture of plumbago crucibles started in the United States in 1827 and in the United
Kingdom in 1856. Graphite use became more widespread in the 19th century with
the discovery of more deposits.
Edward G. Acheson synthesized graphite accidentally while he was experiment-
ing on carborundum, and he started commercial production of graphite in 1897.
Since 1918, graphite, 99 to 99.5%pure, has been produced using petroleum coke,
24 Metal–Crucible Interactions

small and imperfect graphite crystals, and organic compounds as the raw materials.
The use of graphite has since only increased both in variety and in quantity.

1.5.5 Grain and Bond Phases

During the past 100 years, in the usage of refractories, the trend in the selection
of raw materials has shifted from the natural sources used in the nearly as-mined
condition (clays and other aluminosilicates) toward more synthesized, higher-purity
materials (fused and tabular grain and complex bond phases).
Up to the 1940s, grain phases used in refractories consisted of crushed firebrick,
quartzite, pyrophyllite, calcined clays (grog), sintered magnesite, chrome ore, silli-
manite–kyanite–mullite and calcined bauxite. From the 1940s, tabular alumina and
fused grains of various oxides also came into use.
The production of refractories by fusion in an EAF or electrocasting began in
1902, when Carborundum Company produced alumina via this route. The Norton
Company produced a ceramic-bonded fused alumina crucible in 1906. Refractories
like magnesia and mullite are also made by fusion. They are also used as grain mate-
rial in the crushed form.
In 1934, tabular alumina was made in a shaft furnace by T. S. Curtis. The process
involved heating of 2-cm diameter pellets of calcined alumina at >1925°C, just below
the melting temperature (m.p. of alumina is 2072°C) until almost complete conversion
of the fine α-Al2O3 crystallites to large (50 to >900 mm), angular, elongated, tablet-
shaped crystals. Tabular alumina crystals are hard and dense, and have good thermal
conductivity and high crushing strength. The crystals contain substantial closed poros-
ity because of the large grain growth, leaving pores in the centre of grains where vol-
ume diffusion is slow. This is one of the reasons for the better thermal shock resistance
of tabular alumina than fused alumina. The closed pores act as crack arresters. Due to
properties like dimensional stability, corrosion and erosion resistance, and load-bear-
ing capacity, the use of tabular alumina in iron, steel, aluminium and other non-ferrous
metal production has seen a steady rise since the 1970s.

1.5.6 Bricks and Monolithics

Up to the 19th century, refractory brick was manufactured manually, using mainly
aluminosilicate materials. Between 1900 and 1920, machine pressing came into use.
Over the years, brick production was further automated, resulting in bricks more
uniform in shape, having a more isomorphous microstructure and better proper-
ties. In the 1970s, carbon-bonded bricks were developed. These are fired in situ.
Installation procedures were developed for monolithics for linings in the latter half
of the last century.

1.5.7 Refractory Bonds and Binders

The names of refractories often indicate the type of bond that holds the aggregate/
grain phases together, e.g. carbon bonded, silicate bonded, phosphate bonded,
Metal–Crucible Interactions 25

cement bonded and so on. For a long time, the fired integrity of refractories relied
on silicate or aluminosilicate bonds derived from vitrification reactions. During the
20th century, bonds were synthesized first through the use of fine additions, called
mineralizers, and later on by the introduction of separate cementitious and glassy
bond systems.
The significant development associated with manufacturing technology for lin-
ings in the second half of the 20th century has been the procedures for installa-
tion of monolithic or unshaped refractories that can be formed and shaped in situ.
Monolithic refractory materials consist of mixtures of aggregates and binders, ready
for use either as supplied or after the addition of suitable liquids. Since the 1960s,their
use has become widespread as castables, gunning mixes, and mouldables or plastics.
Unshaped refractories use similar aggregate materials as shaped brick, but the bind-
ers used to form the bonding phase are very different. The binders may consist of
aluminate, silicate and phosphate bond systems or their combinations. The develop-
ment of techniques for installation continued into the 1980s.

1.6 ELECTROLYTIC PROCESSES
At the very beginning of the nineteenth century, Sir Humphry Davy had stated
that even the most stable chemical compounds should be electrolytically reduc-
ible with the aid of the newly available voltaic cell, and supported his assertion
by obtaining sodium, potassium, barium, strontium and calcium metals using this
approach. However, he did not succeed in his attempts to obtain the element he
had only named ‘aluminium’. An impediment was the unavailability of ‘heavy’
electric current.

1.6.1 Aluminium
Generators capable of producing heavy electrical currents become generally avail-
able by the mid-1870s, and interest in electro-metallurgical possibilities revived. The
initial attempts in the United States to obtain aluminium by the electrolysis of fused
salts were made by Charles S. Bradley of Yonkers, NY. Even though his ideas and
approaches were remarkably similar to those of Hall and Héroult, Bradley’s patent
was granted only in 1892, by which time the Hall–Héroult process was already in
commercial operation in several countries (Darling 1990).
In 1885, Charles Martin Hall began to investigate various electrolytic approaches
to aluminium production in his private laboratory and concluded that fused salt baths
would be essential. On 10 February 1886, he also found that alumina readily dis-
solves in fused cryolite ‘like sugar in water’ and that the alumina/cryolite solution
thus obtained is a good electrical conductor. It was, however, already well known
that cryolite would dissolve alumina. By dissolving 15–20% of alumina in cryolite,
Hall obtained a bath whose melting point was between 900 and 1000°C. At this tem-
perature, its electrical conductivity was high enough to permit electrolysis. The cell
used by Hall is shown schematically in Figure 1.14 (Darling 1990). Hall found that
the bath rapidly dissolved silica also from the refractory materials used to contain
26 Metal–Crucible Interactions

FIGURE 1.14 Hall aluminium cell (From Darling, A.S. 1990).

it. He soon solved this problem by containing his melt in a graphite crucible and
obtained a number of small globules of aluminium, which formed close to the cru-
cible that also acted as the cathode.
On 23 April 1886, Paul Louis Toussaint Héroult applied for a French patent for an
invention which was identical to that of Hall. He described:

a method for the production of aluminium which consists of the electrolysis of alumina
dissolved in molten cryolite, into which the current is introduced through suitable elec-
trodes. The cryolite is not consumed, and to maintain a continuous deposition of metal
it is only necessary to replace the alumina consumed in the electrolysis.

Héroult did not realize initially that the bath required no separate external heating,
and corrected the lapse in a subsequent patent. His original cell arrangement, shown
in Figure 1.15 (Darling 1990), does show furnace heating of the cell.

FIGURE 1.15 Héroult aluminium cell (From Darling, A.S. 1990).

Metal–Crucible Interactions 27

1.7 RARE METALS
World War II affected major branches of technology in a profound but positive way,
accelerating progress in general and causing the initiation and intensification of
activities that would have, in normal times, remained peripheral. Developments were
powered by atomic energy and the aerospace and electronics industries. All three
of these were materials driven, and most of those materials were hitherto undevel-
oped or underdeveloped. Clifford Hampel, the author of the 1961 classic Rare Metals
Handbook (Hampel 1961), covers as many as 54 metals in his handbook, implying
their rarity before World War II and newfound relevance in technology in the post-
war period. That was the status and background at the beginning of the second half
of the 20th century. He mentions the story of uranium, noting that as late as in 1940,
only a few grams of good uranium were in existence and that the data on the metal
was so scanty that its melting point was listed only as <1850°C. The real value, as is
known now, is 1132°C.
Apart from the fact that they had been already discovered and also investigated as
scientific curiosities, all relevant details of their production in required purities and
sufficient quantities remained to be either developed or validated in the post-1940
period. These activities gathered momentum, and much was accomplished in the
first three decades of the second half of the 20th century.

1.7.1 Special Crucibles and Protected Environment

The rare metals are produced by techniques that are in many ways different from
those used for common non-ferrous metals and also iron and steel. More often than
not, the processing of rare metals has to be carried out in a protected atmosphere
and in inert crucibles. This is due to their pronounced reactivity with elements
like oxygen, nitrogen, hydrogen, carbon and silicon, especially at the high tem-
peratures needed for processing. Many of the rare metals are excellent getters and
avidly pick up these impurities from the surrounding atmosphere. The metals also
ingest them from any solid, such as the crucible, with which they are in contact
in molten condition during processing. It so happens that very often, this aspect
of rare metal processing is the decider that qualifies a process for production or
processing. The metal–crucible interactions play a crucial role in the metallurgy
of many rare metals.
Unlike the common metals, where the as-mined ore, after mineral processing
operations for separating it from gangue, concentration and roasting, is taken for
smelting, the rare metal minerals are put through very elaborate mineral process-
ing and separation operations, and a pure compound is taken for reduction. This is
because slag has literally no role in metal purification here, unlike with the common
metals. Impurities that enter the as-reduced rare metal, barring a few well-known
exceptions, cannot be removed from the metal sufficiently or easily. A certain simi-
larity can be perceived with aluminium extraction. Bauxite is put through the Bayer
process, and pure alumina is taken up for electrolytic reduction. Most of the metals
presently produced by electrolysis appear to be designed to be made that way only.
28 Metal–Crucible Interactions

Any attempts to produce them by alternate means have considerable technological

and economic penalties.
Electrolysis in a fused salt is a potentially useful route for the preparation of
metals that are not easy to prepare by a chemical reduction route. There are many
examples. Aluminium, magnesium, titanium, zirconium, and the rare earth metals
are some of the important cases. The success of molten salt electrolysis as a com-
mercial route to produce aluminium and magnesium encouraged the intensive search
for a similar process for zirconium and titanium. The challenge lies in the melting
points of the metals produced. Al melts at 660.5°C and magnesium at 650°C. Fused
salt electrolysis of these, carried out at 950–960°C for aluminium and at 700°C for
magnesium, yields metal in a liquid form, which is tapped or otherwise collected
conveniently without any problem of contamination. The melting points of zirco-
nium and titanium are 1855°C and 1670°C, respectively. At least at the time when
electrolysis was attempted for them, there was no intention of carrying out electroly-
sis at temperatures exceeding these values. The metal at the cathode will have to be
in the form of a solid.
Even though the temperatures needed for most of the electrolytic reduction pro-
cesses are lower than about 1100°C, there are often serious contamination issues
amongst the metal produced, electrolyte and cell materials.
As well as electrolytic processes, metal replacement reactions or metallother-
mic reductions have been widely useful for the preparation of the rare metals. Even
though some the processes were first demonstrated in the previous century, much of
the usable information on process details was generated in the 20th-century post-war
period.

1.8 METAL REPLACEMENT (METALLOTHERMIC) REACTIONS

Metal replacement reactions, also known as metallothermic reductions, are carried
out in many distinct ways, and there are important examples for each of them.

1.8.1 Goldschmidt Process or Bomb Reduction

Towards the end of the 19th century, Johannes Wilhelm Goldschmidt invented a
process whereby aluminium metal is oxidized by an oxide of another metal, which
is in turn converted to the metallic form. The oxide originally experimented with
was iron oxide, and the reaction occurs with an intense evolution of heat (highly exo-
thermic). It is known that Goldschmidt was originally interested in producing very
pure metals by avoiding the use of carbon in smelting. Events that led to the use of
his process in the 20th century indicate that he eminently succeeded in his objective.
The reaction is called the thermit reaction, and the process is called the Goldschmidt
process (Goldschmidt and Vautin 1898).
The Goldschmidt process raises the temperature of the reaction mixture to high
levels very quickly, and the reaction mixture would need to remain molten for some
time in order for the molten metal to separate cleanly from molten slag. These two
characteristics immediately point to an important role for any crucible in which the
Metal–Crucible Interactions 29

process is carried out. It is performed in a refractory-lined reactor of a suitable con-

figuration. In order to prevent pick-up of interstitial impurities like oxygen and nitro-
gen from the surrounding air by the metal when it is at a high temperature and in
molten form during the process, the thermit process is carried out under an inert gas
cover. A refractory-lined and closed reactor is therefore the standard equipment for
the Goldschmidt process.
As an example of direct application, pure, carbon-free niobium metal can be pro-
duced by aluminothermic reduction (Wilhelm et al. 1966) of niobium pentoxide.
The same applies to vanadium metal also (Carlson et al. 1966). Not only vanadium
metal but vanadium–aluminium master alloys have been produced routinely for use
in the manufacture of Ti6Al4V alloy, a commercial alloy needed in high volumes.
These processes were actively developed during the 1960s and 1970s. Wah Chang
corporation at Albany used the process industrially beginning in the 1970s (Wang
et al. 1970). A schematic of the closed bomb reduction reactor for commercial vana-
dium thermit (V-Al-O alloy) productionis given in Figure 1.16. Perfect (1967, 1981)
of Reading Alloys introduced the revolutionary concept of using a water-cooled cop-
per reactor, instead of a refractory-lined reactor, for conducting the thermit reaction.
Carlson et al. (1981) of Ames Laboratory showed that using water-cooled reactors for
aluminothermic reduction of vanadium can markedly improve the purity of metal
that is produced. Examples of processes that are variants of the Goldschmidt process
include the techniques for the production of uranium and plutonium metals. The dif-
ferences are that a fluoride, instead of an oxide, is used as the start material, and the
reducing agent is not aluminium but calcium or magnesium.

1.9 CONSOLIDATION AND REFINING

In the as-reduced condition, the rare metals are generally unusable. They need to
be consolidated and also usually refined before conversion to mill products for use
in fabrication. While the water-cooled copper reactor was first used in the metal
reduction stage as recently as in the 1980s, such arrangements were in use in metal

FIGURE 1.16 Closed bomb reactor for aluminothermic reduction (From Wang et al. 1970).
30 Metal–Crucible Interactions

refining systems much earlier. Water-cooled copper moulds for holding the melt are
key components in vacuum arc and electron beam furnaces. It has also been incorpo-
rated in induction melting, particularly in the induction skull melting configuration
(Winckler and Bakish 1971)
Metals like tungsten and molybdenum are best consolidated by powder metal-
lurgy techniques of compaction at high pressures and sintering at high temperatures.
Melting is not very attractive, not only because of their very high melting points and
absence of a suitable crucible if vacuum induction melting (VIM) is selected, but
also because these metals do not cast well. Most of the other metals can be melted
and generally cast well.
Melting to consolidate (with little or no refining) is always done in arc furnaces.
Some of the metals and alloys are best melted (for refining and consolidation) in an
Electron Beam Furnace (EBF) if there are no volatile components to be retained in
the melt stock, because in electron beam melting (EBM), refining by vaporization
also occurs along with consolidation. Except that the mechanism of heating is same,
the construction and function of arc furnaces for rare metals processing is very dif-
ferent from that used, for example, in steel making.
In all processing operations on rare metals, the common and significant feature
is the use of a protected atmosphere, either clean vacuum or an equally pure inert
atmosphere. Processing equipment is constructed accordingly.

1.9.1 Vacuum Induction Melting

Vacuum melting, or melting a substance while it is in vacuum, is a 20th-century
phenomenon (Winckler and Bakish 1971). The method of heating used here is only
electrical and confined to the use of resistance, induction, arc and electron beam.
Vacuum melting made the transition from the laboratory to the plant in 1917 when
Dr. Wilhelm Rohn melted nickel base alloys by resistance heating. Although resis-
tance-heated melting furnaces were used initially, the technique has never been pre-
ferred for production scale melting. This is due to several factors. Supplying heat
from an external source is thermally inefficient. The thermally insulated crucible
cannot be used, and the crucible is made to withstand temperatures in excess of that
of the bath throughout its wall thickness. Outer wall temperatures several hundred
degrees more than that of the bath are incurred when effective (practically useful)
heating rates are used with refractory oxide crucibles.
E.F. Northrup built the first prototype of a vacuum induction furnace in the United
States in 1920. Basically, in an arrangement for induction heating, a coil surrounds
a piece of metal. When an alternating electric current passes through the coil, eddy
currents are induced in the metal (charge). The resistance of the metal to the eddy
currents raises its temperature. In place of a solid metal it is often a metallic melt
suitably contained. By 1923, vacuum induction furnaces were commercially avail-
able from Heraus in Germany. They were used to produce thermocouple materials
and resistance heating alloys.
In the VIM process, virgin alloys or scrap are melted down under vacuum or
inert atmosphere to avoid oxidation and contamination from atmosphere. The molten
Metal–Crucible Interactions 31

charge is then cast, mostly under vacuum. With small furnaces up to 25 kg capacity,
it is standard practice to use pre-fired crucibles. For larger sizes, the economics of
rammed linings becomes more clear. An essential feature of VIM is the presence of
a crucible, which may never be completely inert to molten charge.
Commercial VIM developed rapidly with the advent of titanium in the United
States in the early 1950s and by the need for VIM of wrought nickel base alloys for
turbine blades. The next good thing in the development was the availability of much
improved vacuum pumps originating from the atomic energy industry.

1.9.2 Vacuum Arc Melting

The arc furnace is the first metallurgical furnace to work on electricity. The first arc
furnace was developed by Sir William Siemens in 1878–1879. In vacuum arc remelt-
ing (VAR), heat is generated by the application of a low-voltage high-current electric
arc to the surface of the molten metal from an electrode. Vacuum is in the range of
0.1–10 Pa. The electrode may be consumable or non-consumable. In order to assist
ionization and also to minimize vaporization of the electrode, melting is usually
carried out under a low pressure of inert gas and not in high vacuum. In the consum-
able electrode mode, as arcing proceeds, the electrode is consumed as it heats up,
melts and drips into the metal pool (Winckler and Bakish 1971). Titanium sponge is
consolidated to titanium ingot in this way. In the non-consumable mode, as the name
implies, the electrode remains intact, and metals fed into the arc melt and collect into
a water-cooled copper crucible.
In the standard consumable electrode arc furnace, the normal deep round crucible
used for ingot production may be replaced with a shallower water-cooled crucible
capable of being tilted through 90° so that its contents may be poured into conven-
tional moulds to make shaped castings (Clites and Calvert 1961).
The term skull melting implies the use of a water-cooled copper crucible and
denotes the fact that a thin shell, or skull, of metal remains behind in the crucible
after pouring due to the rapid chilling effect of the water-cooled crucible. After its
formation, the skull remains in the crucible, separating the melt and the crucible
inner surface. Depending upon the thermal constants of the metal being melted and
the power available, up to 90% of the crucible content can usually be poured. The
principle has obvious applications in the manufacture of castings in the reactive met-
als, which do not lend themselves to melting in ceramic crucibles. In the field of
titanium and titanium alloy castings, the method has been most widely adopted.
However, little superheat of the metal is possible using a consumable electrode tech-
nique, since generally speaking, higher melting power merely results in quicker
melting rates, and metal temperatures in excess of 30–40°C above the melting point
are difficult to achieve.
Compared with the conventional DC machines, the AC system offers a better heat
distribution and a shallower pool for a given melting current, since no current passes
through the crucible, and the danger of penetration is minimized. Large homoge-
neous ingots, 700 mm in diameter, of certain superalloys prone to segregation have
been made by Cooper et al. (1965).
32 Metal–Crucible Interactions

1.9.3 Electron Beam Melting

Electron beam melting (EBM) utilizes the kinetic energy of highly accelerated elec-
trons, which are made to impact the material to be heated or melted. Heating or
melting is a consequence of transfer of kinetic energy of the electrons to the lattice of
the material on impact, leading to local temperature rise. A greater number of faster
electrons will lead to higher temperatures.
The concept was first propounded during or about 1905, and its inventor was M.
von Pirani. However, the technique became a useful tool only half a century later
(Winckler and Bakish 1971).
Electron beam has truly outstanding advantages as a heat source for melting. It is
clean, does not have any real upper temperature limit and works with high vacuum
as its natural environment. The last advantage is sometimes cited as its limitation
also. In a different format, electron bombardment heating, where the beam is not
really focused but only directed, also has many applications as a heat source and is
amenable to excellent and close temperature control.
Electron beam melting and purification owes its success to the ability to fuse
any material known to undergo solid/liquid transition and to maintain the same in a
fused state for as long as is necessary to effect the purification required. The melt-
ing is carried out in a water-cooled copper hearth, the popular and reliable non-
contaminating crucible.

1.10 CRYSTAL GROWING
A melt, on normal cooling, solidifies to a polycrystalline solid. The nature and prop-
erties of the solid are in many ways determined or influenced by the details of these
crystals and their arrangement. The technique of ‘crystal growing’ deals with the
possibility of changing these details and arrangements on purpose.
Two major methods of crystal growing exist: techniques where the melt is con-
tained in a crucible and techniques in which the material is self-contained, i.e. by
surface tension.

1.10.1 Crystal Growth in a Container

The technique was actively investigated in the 1960s. Apart from the obvious role
of the container in the shaping of the crystal, a number of physical and chemical
properties of the container material are important. The container must be inert to
the metal or alloy being purified. Wetting of the container by the melt should be
minimized, because good wetting leads to sticking and eventual fracture of the con-
tainer. The coefficient of thermal expansion of the container and the charge should
be carefully matched. In order to minimize heat transfer problems, the thermal con-
ductance of the container material should be comparable to or less than that of the
charge. These general criteria allow a wide range of possible container materials.
The choice becomes more limited as the temperature and/or reactivity of the molten
charge increases.
Metal–Crucible Interactions 33

1.10.2 Floating Zone Technique

Chronologically, investigations of the floating zone technique started simultaneously
with the studies of the regular crucible techniques, in the 1950s and 1960s. In fact,
there was more research activity on the floating zone techniques in the 1950s than on
the other methods. The novelty value and the tremendous popularity of the reactive
and refractory metals probably drove these early investigators. In float zone crystal
growing, a molten zone of the material is held in place by its own surface tension
between two collinear solid rods of the same material. The technique was originally
developed by Keck and Golay (1953), Emeis (1954) and Theuerer (1956)to prepare
high-purity single crystals of silicon. This method avoids crucible contamination.
The material usually solidifies behind the molten zone to form a single crystal along
the entire length of the specimen rod. Common heating methods for this are induc-
tion and electron bombardment.
Induction heating makes it possible to work under high vacuum, in a reducing or
oxidizing atmosphere, or in a positive pressure of inert gas. Induction is particularly
attractive for the growth of crystals of the reactive metals. Electron beam heating
is possible only in a vacuum. It was first used successfully by Calverley and Lever
(1957). Control of the dimensions of the molten zone is easier than with induction
heating because the electrons can be focused on the specimen.

1.11 SUMMARY AND SCOPE OF CRUCIBLE–

METAL INTERACTIONS
In a metallurgical process with materials placed inside a crucible, when heat is sup-
plied to raise the temperature (or withdrawn to lower it), and substances necessary
for chemical reactions are introduced, the objective is to induce changes to occur.
Changes are of two types: those that are designed to occur, and those that are not
intended and usually not wanted either. In general, all of these occur, governed by
thermodynamic properties of the system and also, in reality, determined by the kinet-
ics of the processes. Among the most important of these reactions and processes are
those that involve the crucible material itself.
The general expectation is that the crucible material will remain inert through
all these changes. Apart from physically containing the reactants and products and
allowing or not allowing heat to pass through it, depending on the original pur-
pose of its choice, it should remain indifferent to its contents and also to processes
occurring within its confines. This is the perfect crucible. But, perfect crucibles
do not exist.
All crucibles interact with their contents. They interact chemically and also physi-
cally. Interactions cannot be avoided but can be managed. Interactions can be gener-
ally understood, and their negative effects can be circumvented so that the crucibles
remain available and useful for any processing requirement. The purpose of the pres-
ent volume is to address this area.
Even though metal–crucible interactions exist technically at all temperatures
and in all states of aggregation, they are important and warrant remedial action
34 Metal–Crucible Interactions

particularly when the temperatures are high and melts are involved. These are exactly
the conditions when thermodynamics expands the possibilities for interactions, and
kinetics causes them actually to occur. Systems with the melts and containers in
contact at high temperatures are encountered in two major areas of technology. One
is in-process metallurgy and post-reduction processing of materials, and the other is
in liquid metal/alloy cooling loops. Both are important areas of technology and have
attracted considerable research effort all over the world. There are, however, major
differences in the problems faced in each of these areas and the way solutions to the
problems are developed. They have more differences than commonalities. To main-
tain focus, this book covers only the processing-oriented issues of metal–crucible
interactions.
It would not be accurate to conclude that the size of the crucible does or does not
determine the number, type and extent of interactions, but there is a loose linkage.
A larger system tends to be more complex. It would therefore be unrealistic to iso-
late and consider only interaction between the metal and its container in such cases.
Attention needs to be paid to all major reactions and processes. Often, the compo-
nents of the melt attack the crucible synergistically. A great example is the corrosive
attack in the belly band zone of aluminium melting and holding furnaces. Generally,
it is the slag rather than the metal that attacks the refractory, so much so that an
oxidized metal melt becomes more corrosive then when it was purer. Sometimes,
a minor alloy addition to the metal or a binder in the crucible tends to become the
source of interaction.
In this chapter, the history of metal processing was considered from the viewpoint
of crucibles used at various time periods. The nature, type and role of a crucible
depend on the format of the overall equipment, and an attempt was made to pro-
vide a glimpse of the major equipment and processes that have been used in metal
processing over the years. The objective is to provide a framework for considering
the many types of possible metal–crucible interactions in the next three chapters. In
the following pages, an attempt has been made to present the various facets of this
interesting issue, which has challenged human ingenuity throughout known history
and continues to do so even today.
Each of the few metal–crucible interactions selected for specific coverage in this
book is representative of the type of problems faced by many more metal–crucible
systems. The drivers for interactions and the remedial actions that can circumvent or
offset the damages are described. Most of the issues concerning any metal–crucible
pair can be resolved by analogy to those covered specifically here.
A collection of metal–crucible pairs that have, at one time or other, been found
workable is listed in Table 1.1.
Appearing next is Table 1.2, which lists various crucibles that have been recom-
mended for use in the evaporation of metals.
The entries in this table have been compiled from published information from
equipment manufacturers. Evaporation does not necessarily involve melting, and as
the evaporation is carried out in the context of thin film applications, the duration
of contact between the metal and the crucible is expected to be short and the quan-
tity involved smaller. The tables are useful for identifying the members of the pair
Metal–Crucible Interactions 35

TABLE 1.1
Composition of Crucibles Used for Melting Metals
Metal Crucible Materials (Remarks) References
Alkali Refractory sulphides, Single crystal MgO (The Eastman et al. 1951, Brewer et
metals refractory sulphides include BaS, CeS, ThS, Ce3S4, al. 1948
Ce2S3, Th2S3, Th4S7 and ThS2)
Alkaline Refractory sulphides Eastman et al. 1951, Brewer et
earth al. 1948
metals
Aluminium Zircon, Zirconia, Graphite, Clay-graphite, Refractory Eastman et al. 1951, Brewer et
sulphides, Fire clay (dense), Aluminium nitride al. 1948, Knauft 1943, Long
(There is least contamination with zircon) and Foster 1959
Beryllium Beryllium oxide, Graphite, Refractory sulphides Kaufmann and Gordon 1947
(Melting must be conducted in a vacuum)
Bismuth Cerium sulphide, Thorium sulphide Eastman et al. 1951, Brewer
et al. 1948
Boron Boron nitride Finlay et al. 1952
Brass Magnesium oxide, Graphite, Clay-graphite Harvey 1945
High-grade fire clay (Charcoal or fluxing material
on surface of melt helps to reduce oxidation)
Cerium Titanium nitride, Zirconium nitride, Cerium Eastman et al. 1951, Brewer
sulphide, Barium sulphide (Cerium sulphide et al. 1948
crucibles resist attack better if the porositya is from
10 to 15%)
Chromium Aluminium oxide Gayler 1930
Copper Silica, Magnesium oxide, Aluminium oxide Harvey 1945, Booth 1940
Clay-graphite
Gold Magnesium oxide, Aluminium oxide Taylor et al. 1948
Halide Refractory sulphides Eastman et al. 1951, Brewer et
metals al.1948
Iridium Thorium oxide Taylor et al. 1948
Iron Magnesium oxide, Beryllium oxide, Aluminium Eastman et al. 1951, Brewer et
oxide, Zircon, Fire clay, Thorium oxide, Refractory al. 1948, Richardson 1935,
sulphides (For extremely pure iron, beryllia or Turner 1931, Thompson and
thoria is best) Cleaves 1939
Lead Iron, Chromite, Magnesium oxide, Fire clay (dense) Harvey 1945, Thews 1931
Magnesium Steel plate coated with Aluminium, Heat-coat Brewer et al. 1948, Taylor et al.
mixture of 10 to 60% ZrO2, 7 to 25% MgO, 30 to 1948, Field 1947
80% Al2O3, Cerium sulphide, Thorium sulphide
(The steel plate should be sand-blasted before
spraying and heat-treated after spraying)
Manganese Aluminium oxide, Magnesium oxide, Thorium oxide Gayler 1930
(The metal should be pure, since a manganese-rich
alloy will pick up silica)
(Continued )
36 Metal–Crucible Interactions

TABLE 1.1 (CONTINUED)

Composition of Crucibles Used for Melting Metals
Metal Crucible Materials (Remarks) References
Nickel Zircon, Magnesium oxide, Mullite, Fire clay (dense) Seybolt 1946, Jordan and
(For melting pure nickel, zircon or magnesia in a Swanger 1930
vacuum is best)
Platinum Thorium oxide, Zirconium oxide, Magnesium oxide, Taylor et al. 1948,
Zircon (Also for platinum-rich alloys) Baldwin1948
Rareearth Refractory sulphides, Tantalum, Tungsten Eastman et al. 1951, Brewer et
metals al. 1948, Dennison et al.
1966a, 1966b
Rhodium Zirconium oxide Taylor et al. 1948
Silicon Silica Gayler 1930
Silver Graphite, Magnesium oxide, Aluminium oxide Taylor et al. 1948
Steel Magnesium oxide, Silica (Basic steels are melted in Taylor et al. 1948
magnesia, and acidic steels are melted in silica)
Thorium Cerium sulphide, Thorium sulphide, Titanium Brewer et al. 1948, Richardson
nitride, Zirconium nitride (The metal sticks to the 1935
sulphide crucibles)
Tin Magnesium oxide, Cerium sulphide, Thorium Eastman et al. 1951, Brewer et
sulphide, Fire clay al. 1948
Titanium Cerium sulphide, Barium sulphide, Thorium Eastman et al. 1951, Brewer et
sulphide (Titanium sticks to the sulphide crucibles. al. 1948, Brace 1948
There is slight attack of the thoria, melting is
accomplished in either a vacuum or an argon
atmosphere)
Uranium Magnesium oxide, Calcium oxide, Cerium sulphide, Brewer et al. 1948, Richardson
Titanium nitride, Zirconium nitride, Thorium oxide, 1935
Thorium sulphide, Graphite, Nickel
Vanadium Beryllium oxide, Magnesium oxide, Calcium oxide,
Titanium nitride (These materials were calculated
to be suitable for containing molten vanadium)
Zinc Flint fire clay, Cerium sulphide, Thorium sulphide, Eastman et al. 1951
Clay-graphite
Zirconium Zirconium dioxide, Thorium oxide Richardson 1935

Source: Schwartz, M.A., White, G.D., and Curtis, C.E., Crucible Handbook: A Compilation of Data on
Crucibles Used for Calcining, Sintering, Melting, and Casting, United States Atomic Energy
Commission. ORNL-1354, Oak Ridge National Laboratory, Oak Ridge, TN, 1953.
a Porosity is defined as the volume of airspace expressed as percentage of total volume of the piece.

(metal–crucible) for further investigation. Like the Ellingham diagrams given in the
next chapter, the information provided by the table should at best be regarded as a
starting point for considering the combination, and good starting points are defi-
nitely important. As will be seen, fairly repeatedly, in the rest of the book, metal–
crucible interactions, both the extent and the rate, are determined by multiple factors:
Metal–Crucible Interactions 37

TABLE 1.2
Crucibles Suitable for Evaporation of Metals
Melting Temperature,
Metal point, °C °Ca Crucible materials (Comments)
Aluminium 660 972 AlN, TiB2-BN, ZrB2, BN, Vitreous carbon, W
Antimony 630 425 BN, C, Al2O3, Alumina-coated Mo and Ta
Arsenic 817 210 Al2O3, BeO, Vitreous carbon
Barium 729 462 W, Ta, Mo (Wets refractory metals without alloying.
Reacts with ceramics)
Beryllium 1289 987 BeO, Graphite, Vitreous carbon W, Ta (Wets W, Ta)
Bismuth 271 520 Al2O3, Vitreous carbon, W, Mo, Ta
Boron 2092 1797 Graphite, Vitreous carbon (Forms carbide with the
container)
Cadmium 321 180 Al2O3, Quartz, W, Mo, Ta
Calcium 842 459 Al2O3, Quartz, W
Carbon 3826 2107
Cerium 798 1380 Al2O3, BeO, Vitreous carbon, W, Ta
Cesium 28 75 Quartz, Stainless steel
Chromium 1863 1162 Vitreous carbon
Cobalt 1495 1262 Al2O3, BeO, W, Nb (Alloys with refractory metals)
Copper 1085 1017 Al2O3, Mo, Ta
Dysprosium 1412 900 Ta
Erbium 1529 930 W, Ta
Europium 822 460 Al2O3, W, Ta
Gadolinium 1313 1175 Al2O3, Ta
Gallium 30 907 AlN, Al2O3, BeO, Quartz (Alloys with refractory
metals, attacks crucibles above 1000°C)
Germanium 937 1142 Quartz, Al2O3, W, Ta, Carbon (Wets W, Ta, Mo)
Gold 1064 1132 Al2O3, BN, W, Vitreous carbon (Wets W)
Hafnium 2231 3090
Holmium 1474 950 W
Indium 157 742 Graphite, Al2O3, W, Mo (Wets tungsten and copper)
Iridium 2447 2380 ThO2
Iron 1538 1207 Al2O3, BeO, W (Attacks tungsten)
Lanthanum 921 1388 Al2O3, W, Ta
Lead 328 497 Al2O3, Quartz, W, Mo, Ta
Lithium 181 407 Al2O3, BeO, Ta, Stainless steel
Lutetium 1663 1300 Al2O3, Ta
Magnesium 649 327 Al2O3, Vitreous carbon, W, Mo, Ta, Nb
Manganese 1244 647 Al2O3, BeO, W, Ta, Mo (Wets refractory metals)
Mercury −39 −6
Molybdenum 2623 2117
Neodymium 1021 1062 Al2O3, Ta (Low solubility of tantalum)
(Continued )
38 Metal–Crucible Interactions

TABLE 1.2 (CONTINUED)

Crucibles Suitable for Evaporation of Metals
Melting Temperature,
Metal point, °C °Ca Crucible materials (Comments)
Nickel 1455 1262 Al2O3, BeO, Vitreous carbon, W (Alloys with
refractory metals)
Niobium 2468 2287 W
Osmium 3033 2760
Palladium 1554 1192 Al2O3, BeO, W (Alloys with refractory metals)
Platinum 1769 1747 Al2O3, ThO2, Graphite, W (Alloys with metals)
Plutonium 641 W
Polonium 254 244 Quartz
Potassium 63 125 Quartz, Mo
Radium 700 416
Rhenium 3186 2571
Rhodium 1963 1707 W, ThO2, Vitreous carbon
Rubidium 39 111 Quartz
Ruthenium 2334 2260 W
Samarium 1074 573 Al2O3, Ta
Scandium 1541 1100 Al2O3, BeO, W (Alloys with tantalum)
Selenium 221 170 Al2O3, Vitreous carbon, W, Mo
Silicon 1414 1337 BeO, Ta, Vitreous carbon, W (alloys with W)
Silver 962 832 Al2O3, W, Mo
Sodium 98 192 Quartz, Tantalum, Stainless steel
Strontium 769 403 W, Ta, Mo, Vitreous carbon (Wets refractory metals
without alloying)
Tantalum 3020 2590
Tellurium 452 277 Al2O3, Quartz, W, Ta (Wets refractory metals without
alloying)
Terbium 1356 1150 Al2O3, Ta
Thallium 304 470 Al2O3, Quartz, W, Ta
Thorium 1755 1925 W, Ta, Mo (Wets W)
Thulium 1545 680 Al2O3, Ta
Tin 232 997 Al2O3, Mo, Ta (Wets molybdenum)
Titanium 1670 1453 TiC, Vitreous carbon, W (Alloys with refractory
metals)
Tungsten 3422 2757
Uranium 1135 1582 Y2O3, Mo, W
Vanadium 1910 1547 W, Mo (Wets molybdenum. Alloys slightly with W)
Ytterbium 819 417 Ta
Yttrium 1522 1157 Al2O3, W, Ta (High tantalum solubility)
Zinc 420 250 Al2O3, Quartz, Mo, W, Ta (Wets refractory metals)
Zirconium 1855 1987 W
a Temperature in°C for 0.01 Pa vapour pressure.
Metal–Crucible Interactions 39

temperature for one, duration of contact, identity of the melt and its impurity content,
viscosity of the melt, its actual wetting characteristic, impurities in the refractory,
microstructure of the refractory, its porosity, and pore sizes and their distribution,
to name the most important. Even when we consider a metallic crucible, sometimes
whether it is a single crystal or polycrystalline determines whether it would actually
be useful as a crucible. The original papers that relate to the table entries usually
answer many of these. In the end, the value of an actual validation experiment can-
not be overemphasized.
 2 Basics of Interactions

2.1 INTRODUCTION
The tendency of materials to react with one another can be generally predicted by con-
sidering their thermodynamic properties. When the possibility of interactions at elevated
temperatures is considered, kinetics is fast enough to render thermodynamic predictions
closely reflect real situations. Graphical representation of the thermodynamics of the sys-
tem as Ellingham and phase diagrams is especially valuable to reveal possible interac-
tions. Even though in many instances, more than two components are involved in the
interaction, and multi-component phase diagrams must be consulted when accessible,
much useful information can still be obtained from the relevant binary phase diagrams
that may be more readily available. A variety of transport processes contribute to the
materialization of metal–crucible interaction. Apart from thermodynamic properties,
there are requirements placed on the crucible materials as regards their thermo-physical
properties, both intrinsic and vis-à-vis the material being processed. Fabricability of the
crucible in the required form is another key consideration.
Chemical analysis is, obviously, very important in assessing the interaction. This,
in addition to X-ray, microscopic and microprobe analyses of the interfaces, helps
to map the nature and extent of interaction more completely. These tools and tech-
niques are all relevant in the context of predicting, explaining and thus managing
metal–crucible interactions.
Thermodynamic considerations are the usual and also the right way to shortlist mate-
rials that could serve as crucibles for processing melts. Management of interactions
involves attention to several other factors, including properties and processes that act
together. Wetting and penetration of melt in refractory pores is a common issue. Volume
change during the process occurs due to formation of the product by chemical interaction
or due to temperature changes and the expansion characteristics of the various phases
present in the system. Volume change usually aggravates and occasionally prevents con-
tinuation of melt ingestion. Thermal shock and its consequences are a major issue caus-
ing or aggravating degradation. As emphasized in Chapter 1, the scope of metal–crucible
interactions covered in this work is fairly broad, relating principally to production and
processing of materials and not to usages like liquid metal loops in energy systems. The
term crucible is used to denote essentially all surfaces that contact the process liquids.
The origin of the liquid may be direct or collateral. Interactions of vapours and gases are
considered only to the extent that they support or suppress the liquid–solid interactions
and act concurrently.

2.2 THERMODYNAMICS
Materials in contact tend to interact with one another. The interaction is greater and
faster at high temperatures and when one of the materials is a fluid. Such interactions

DOI: 10.1201/9780429345562-2 41
42 Metal–Crucible Interactions

may occur on purpose, being a part of process being implemented, in which case
conditions are deliberately created to facilitate its occurrence. The other group of
interactions occur not on purpose and in spite of not being intended either, e.g. metal-
crucible interactions. Such interactions are seldom beneficial, and efforts are always
made to avoid them or at least, control them to the least harmful levels.
Metal–crucible interactions are driven by thermodynamics, whether it results in
the formation of a simple Raoultian solution between the melt and the crucible or
in an all-out chemical reaction resulting in products that have little resemblance to
the original melt or the crucible. Checking on the thermodynamic properties of the
system is a necessary and proper first step in assessing these interactions.
Basically, the question that needs to be answered here, as in every area of materi-
als processing, pertains to whether a compound MXn, where M is a metal and X is
a non-metal, will react with an element R, another metal, at a given temperature. A
special case is when the crucible and the melt are both elements, e.g. carbon and iron,
or tungsten and lanthanum. The situation is simpler here but a special case of the
more general scenario presently considered. In the context of metal–crucible interac-
tions, MXn may be considered a ceramic compound and R the metal to be contained
by the ceramic. The question is answered conclusively by thermodynamics, but the
answer is accessible only when the thermodynamic properties of all the entities in
the reaction are known or can be estimated. The present question may be represented
by the general reaction:

MX n + iR = M + iRX n / i (2.1)

where M is a metal; X is oxygen, carbon, nitrogen, boron, sulphur; and R may be a

metal such as titanium, vanadium, aluminium or any element. The reaction is possi-
ble only when at a chosen reaction temperature, the difference between the free ener-
gies of formation, ΔG, of the compound formed by the interacting element, RXn/I and
the starting compound, MXn, is negative: in other words, if the condition

iDG ( RX n / i ) - DG ( MX n ) < 0 (2.2)

holds good. Therefore, the free energies of formation of the compounds (oxides,
nitrides and so on) of all the elements and their dependence on temperature deter-
mine which element R will react with MXn to release the M. To make this informa-
tion easy to assess, the data on the free energies of formation of compounds were
presented graphically as straight line plots with temperature as abscissa and stan-
dard free energy of formation as ordinate, first by Ellingham (1944) and later by
Richardson (1948) and Richardson et al. (1950). The Ellingham diagrams of com-
pounds relevant to the selection of crucible materials are given in Figures 2.1 to 2.5
(oxides, carbides, nitrides, sulphides, fluorides). Data from the U.S. Bureau of Mines
Publications (Pankratz et al. 1984) have been used in preparing these diagrams.
The metals titanium, zirconium, vanadium and every one of the 16 naturally
occurring rare earths are extremely reactive in the sense that they tend to interact
strongly with all usual crucible materials and also with some unusual materials that
Basics of Interactions 43

FIGURE 2.1 Ellingham diagram of oxides.

FIGURE 2.2 Ellingham diagram of carbides.

44 Metal–Crucible Interactions

FIGURE 2.3 Ellingham diagram of nitrides.

FIGURE 2.4 Ellingham diagram of borides.

Basics of Interactions 45

FIGURE 2.5 Ellingham diagram of fluorides.

are considered to be rather stable. It seems instructive, therefore, to look in detail at the
process of selecting crucibles for them, e.g. for titanium. For instance, if it is planned
to use an oxide crucible, say alumina, the free energies of formation of titanium oxide
will be compared with the free energy of formation of alumina. Since titanium forms
many stable oxides, for comparison, the most stable oxide phase of titanium, TiO, will
have to be selected. The free energy of formation of Al2O3 will be compared with the
free energy of formation of TiO. The comparison must be carried out on an equivalent
basis; that is, the molar free energy of formation of alumina represents the free energy
of formation of alumina when it is formed with the consumption of 3 gram atoms or 1.5
gram moles of oxygen. The molar free energy of formation of alumina when multiplied
by the factor 2/3yields the value of free energy formation of alumina per gram mole of
oxygen. Similarly, the molar free energy of formation of TiO represents the free energy
of formation of TiO when it is formed with the consumption of 1 gram atom or 0.5
gram moles of oxygen. The molar free energy of formation of TiO when multiplied by
a factor of two yields the free energy of formation of TiO per gram mole of oxygen. So,
before comparing the values of free energy of formation of oxides, the values for free
energy of formation per mole of the oxide should be converted to the free energy of
formation per mole of oxygen by multiplying with the appropriate factor, and the free
energies of formation of oxides per mole of oxygen must be compared. If the formula
of the compound for which the molar free energy values are listed is MxNy, then the
factor is (2/y), y being the subscript of the anion. It may be oxygen, nitrogen, fluorine,
46 Metal–Crucible Interactions

carbon, silicon, sulphur or phosphorus. For gases, the mole or 1 gram mole equals 2
gram atoms, and for all other elements mentioned, 1 gram mole equals 1 gram atom.
This comparison of free energy formation must be done only in this way irrespective
of the compound that is considered, on an equivalent basis. For example, if we are
considering BaZrO3 as the crucible, then the factor is still 2/y (i.e. 2/3).
On comparison of the free energies of formation of the refractory compound with
that of the corresponding titanium compound, on an equivalent basis, as explained in
the previous paragraph, if it is found that the free energy of formation of the refrac-
tory compound has a significantly larger negative value than the free energy of for-
mation of the titanium compound, the refractory compound can be used as a crucible
for containing titanium as a first approximation. If the free energy of formation of
the refractory compound has a smaller negative value that that of the titanium com-
pound, the crucible will be useless for containing titanium. If the values are more or
less similar, further evaluations (calculations and experiments) will be necessary to
comment on the suitability of the crucible.
A plot of the standard free energy of formation of a compound as a function of
temperature is called an Ellingham line. In a single plot, several Ellingham lines can
be drawn, all for the same type of compound (an oxide or a carbide), for a number of
metals. The ordinate is the free energy formation per mole of oxygen or free energy
of formation per gram atom of carbon, and so on. The Ellingham diagrams for a
number of compounds are given in Figures 2.1 to 2.5. In the way they are plotted,
these diagrams directly depict the relative stability of compounds of the same type.
Elements with their Ellingham lines towards the bottom of the figure are more stable
than those whose lines appear towards the top of the figure. It is only necessary that
the Ellingham lines of the most stable compound of the metal to be melted and of
the compound for the crucible material are both plotted there. The bottom most line
is that of the compound that is the most stable among all the compounds appearing
in the figure. All it takes is a look at these figures to know which refractory can be a
potential crucible for a given metal. The larger the vertical gap between the lines, the
greater is the relative stability between the compounds concerned.
The free energy of formation mentioned so far is actually the standard free
energy of formation. It gives the free energy values corresponding to a process
where the reactants and products are in the standard states. The standard state
simply refers to the stable state at the given temperature and 1 atmosphere (ambi-
ent) pressure. Consider the standard free energy of formation of zirconium oxide,
ZrO2. The value refers to the free energy change when 1 gram atom of zirconium
at an ambient pressure of 1 atmosphere combines with 1 gram mole of oxygen
gas at 1 atmosphere pressure to yield 1 gram mole of ZrO2 at an ambient pressure
of 1 atmosphere, all at a fixed given temperature. The idealized relative stability
information provided by the Ellingham diagram should only be considered as the
starting point in the evaluation of relative stabilities of metals and compounds in
real-life situations.
If, for example, the group of oxides Al2O3, BeO, CaO, MgO, SiO2 and Y2O3 is
considered, then under standard conditions (i.e. 1 atmosphere pressure at the given
temperature), the order of stability, in descending order, among the oxides is (most
Basics of Interactions 47

stable) CaO > RE2O3 > MgO > Al2O3 >> SiO2 (least stable). It is implicit in the state-
ments made earlier that each of the solid phases is pure and the pressure of the gas
phase is 1 atmosphere. Similar plots can be made and relative stabilities presented
for any type of compound and for any number of metals in a group, provided ther-
modynamic data is available.
The picture under non-standard conditions that, more often than not, more closely
approximates the real process condition is somewhat different. Here, the reaction
mixture need not be composed of only pure constituents R, M and RXn/i, and the
ambient pressure may not be 1 atmosphere. In other words, the activity, a, of each of
the constituents, aR, etc., need not be kept equal to 1. Consider the following thermo-
dynamic expressions.

DG = DG 0 + RTlnQ (2.3)

Q = a ( M ) a i ( RX n / i ) / a ( MX n ) a i ( R ) (2.4)

Even when ∆G0 is positive, depending on the value of Q, ∆G can become negative,
and reactions can occur when none would be expected considering only standard
free energies. Usually, solutions form, and the activity of components appearing in
the numerator of the expression for Q becomes low, making Q less than 1. There is
a convenient way to obtain a small value for Q in reactions where the product is a
gas or a vapour. The activity term is replaced by partial pressure, and by decreasing
the partial pressure of the product (gas), a low value for Q is generated. Depending
on how small (less than 1) the value of Q is and how high the temperature is, the
term RTlnQ can become a large negative value, and a strong possibility of reaction
emerges. It will be seen later that this situation develops when calcia or baria is used
as a crucible at high temperatures and in vacuum.
The extent of reaction of metals with oxides is important for the problem of find-
ing suitable refractories for the melting of metals and alloys. It was emphasized by
Kubaschewski and Alcock (1979) that the higher the temperature, the less likely are
the kinetic checks, and the more significant are the positions of thermochemical
equilibria.

2.3 THE TITANIUM–OXYGEN SYSTEM

Even though Ellingham diagrams are available or can be created for any compound
type, the most frequently used is the one for the oxides. Refractory oxides are the
most widely used crucible materials, and oxygen contamination is among the most
prevalent and also the most critical of the consequences of metal–crucible interac-
tion. An examination of the Ellingham diagram of oxides shows up a fairly large
number of choices for crucible materials for titanium. The case of titanium is exem-
plary in many ways. The melting point of the metal, its other physical properties, its
chemical reactivity and alloying behaviour, taken together, bring to the fore most of
the problems that can come up in any metal–crucible interaction. It is a suitable case
to elaborate on.
48 Metal–Crucible Interactions

It is seen that CaO and Y2O3 are potential crucible materials for direct melting of
Ti because their free energies are more negative at all temperatures, whereas SiO2
and Al2O3 have a ΔG0f (kJ/mole of oxygen) that is less negative than that of TiO
(Lochbichlerand Friedrich 2007; Kostov and Friedrich 2006; Kostov and Friedrich
2005) at T ≥ 1600 K. The latter group of oxides cannot, therefore, be used as materi-
als for melting crucibles. The value of the free energy of formation ZrO2 is close to
the one for TiO, and an offhand comment would not be appropriate.
The simplified thermodynamic approach, which only considers the stability of the
crucible material vis-à-vis the most stable oxide of the metal to be melted (presently
TiO), overlooks certain key realities of the interaction of the melt with the crucible
material. The phase diagram, Figure 2.6, of the oxygen–titanium system (Massalski
1990b) reveals that oxygen has a large solubility in Ti, and the conclusions derived
from the Ellingham diagram refer to a titanium that is in fact a saturated solution of
oxygen in titanium (i.e. titanium loaded with up to approximately 30 atom percent
oxygen in solid solution). A similar situation prevails in many other systems also,
e.g. O–Zr (Massalski 1990b).
To address this and related issues where a simple comparison of free energies will
not present a complete or sufficiently accurate picture of possible interactions, there
are very useful software packages that perform detailed thermodynamic calculations
and arrive at the equilibrium composition of a given system by minimizing the total

FIGURE 2.6 Titanium–oxygen phase diagram (From Massalski, T.B. 1990b).

Basics of Interactions 49

free energy of the system under specified conditions. The conclusions are limited
only by the reliability and extent of input information on the various thermodynamic
data available for the system being assessed. Fact Sage has been a popular package.
In this context, an assessment of the interaction thermodynamics between TiAlV
melts and a number of binary and mixed oxides as candidate crucible materials was
carried outby Kostov and Friedrich (2005, 2006).
It was concluded that out of six binary oxides (CaO, Y2O3, ZrO2, MgO, Al2O3
and SiO2) and many of their combinations (oxide spinels), only CaO, Y2O3, ZrO2 and
Al2O3 have positive values of free energy, DG r0 , in respect to their interaction with
TiAlV melts at T = 1273–1973 K in the whole range of Ti concentrations. However,
for the alloys with high Ti concentration and high melting temperature, DG r0 with
Al2O3 is close to zero, which eliminates its use as a crucible.
While this approach is highly useful and practical for shortlisting potential cruci-
ble materials, it is also instructive to follow the mechanics of evaluation of potential
crucible materials as outlined by Lyon et al (1973) and later by Okabe et al. 1991.
The selection of a crucible for melting titanium remains, as of now, a work in prog-
ress. The reason for this state of affairs is more readily understood if the attempts
already made at a solution are revisited.
The free energies of formation of refractory compounds are compared with those
of the corresponding titanium phases (TiO, TiN and TiC). Among the oxide ceram-
ics, Al2O3, ZrO2, ThO2, MgO, CaO, BeO andY2O3 are good for possible experimen-
tal evaluation. This approach is, as indicated earlier, incomplete. It overlooks the
need to consider solution effects, i.e. the complications introduced due to titanium
metal accommodating considerable oxygen in solid solution. The picture become
clearer if the partial Gibbs energy diagram for the titanium–oxygen system given in
Kubaschewski and Alcock (1979) is considered. The partial Gibbs energy of disso-
ciation of 1 mole of oxygen over the entire composition range of the Ti–O binary is
given. The value, which was approximately 750 kJ in the TiO + α two-phase region,
rises rather steeply as the system enters the α solid solution range (at about 15 wt%
oxygen at 1000°C) and flattens (at 940 kJ) briefly in the β+α two-phase region (at
around 2 wt% oxygen) followed by another extremely steep rise in the β Ti–O solid
solution region to 1000 kJ. This information indicates that the least stable oxide
crucible that can still be used for melting titanium would be calcia. In going by the
simple Ellingham plots to assess the reducibility of oxides, the stability of the tita-
nium–oxygen system to be tackled was underestimated by nearly 250 kJ.
As titanium has not only great affinity but also large solubility for interstitials,
the possibility of occurrence of a reaction with a crucible needs to be evaluated by
considering the partial molar free energies of solution for interstitials in titanium.
For example, at 1400°C, the free energy of formation of ZrO2 is more negative than
any titanium oxide phase, but comparison of this with a free energy of formation of
titanium–10 atomic percent oxygen solution indicates existence of a driving force
for dissolution of the zirconia (Lyon et al. 1973; Gingerich 1968) into titanium. It
turns out that oxides such as Al2O3 and SiO2 are not compatible with (cannot be used
to melt) titanium, even at relatively low temperatures. For it to be considered as a
50 Metal–Crucible Interactions

TABLE 2.1
Activity of Titanium at the Solidus Temperatures of Various
Titanium Alloys
Alloy Activity of titanium Solidus temperature, °C
Ti 1.00 1668
Ti6Al4V(wt.%) 0.87 1686
Fe70Ti(wt.%) 0.56 1109
Ti44.5Al6Nb0.2C0.2B (at .%) 0.29 1582
Ti45Al7Nb1Mo0.2B (at .%) 0.28 1585
Ti47Al2Cr0.2Si (at .%) 0.27 1528
Ti50Al (at .%) 0.24 1517
Fe30Ti (wt .%) 0.07 1433

Source: Friedrich et al., 2016.

potential crucible material for melting titanium, the prospective oxide should be able
to coexist with a very dilute titanium–oxygen solution at temperatures near the melt-
ing point of titanium. This requirement eliminates most refractory oxides with the
exception of CaO, ThO2, MgO, the trivalent rare earths, and BeO among the binary
oxides. It is entirely likely that if the titanium alloys are considered, the partial molar
free energy of oxygen in the alloy at comparable concentrations is less negative than
the binary Ti–O solution, making the choice of crucibles somewhat wider. This is
because the activity of titanium is less than 1. That makes Q, described earlier, a
larger value, and ΔG becomes less negative. Friedrich et al. 2016 have given the val-
ues for the activity of titanium in various popular titanium alloys. The data presented
by them in the form of a figure is listed in Table 2.1.
In the TiAl group of alloys containing approximately 50 atom percent titanium,
the activity of titanium is much smaller than 0.5. It is approximately 0.3 to 0.25 and
does not change much with other alloying components such as niobium, molybde-
num, silicon and boron. The trend is the same with the FeTi alloys. The titanium
activity decreases to 0.56 with the mole fraction of titanium in the alloy coming
down to 0.73. There is a rather drastic decrease in titanium activity to 0.07 when the
mole fraction of titanium in the FeTi alloy is 0.33. Incidentally, all this is good news
when it comes to finding a crucible for melting titanium alloys, because much less
aggression can be expected from these alloys than from pure titanium.
There is considerable solubility of oxygen in titanium (Figure 2.6). Yttria has an
oxygen stoichiometry range and can exist as an oxygen-deficient compound (Figure
2.7). When evaluating the possibility of melting titanium in an yttria crucible, the
reaction considered by Lyon et al. (1973) was, therefore,

Ti + Y2O3 = Ti ( O in soln ) + Y2O3- X (2.5)

Basics of Interactions 51

FIGURE 2.7 Yttrium–oxygen phase diagram (From Massalski, T.B. 1990b).

They determined the quantities necessary for evaluation of the free energy change
for this equation in a series of mass spectrometric experiments performed on molten
titanium in contact with yttria crucible liners. The purpose of the experiments was
to determine the oxygen potential (partial pressure of oxygen over Ti–O solutions)
in molten titanium of known oxygen content and the oxygen and yttrium potential of
yttria of a known stoichiometry.
From the computed pressures over a range of oxygen concentrations, Lyon et al.
(1973) determined the Sievert’s law constant for oxygen in titanium at 1960 K as

X O = 107 p1O2/ 2 (2.6)

At 1960 K, the oxygen partial pressure was l0 −19 atm for oxygen-deficient yttria with
an O/M of 1.47 (i.e. for Y2O2.94). Based on the observed partial pressure of yttrium
over Y2O3−X, they determined that at 1960K, the yttria phase can exist at oxygen pres-
sures down to 10 −24 atm. At such low oxygen potential, the thermodynamic activity
of Y approaches unity, and the titanium melt will pick up yttrium more than at any
other composition. The yttrium content in titanium solution could go to possibly 20
weight percent when the activity of yttrium is about 0.25 in value. From mass spec-
trometric data, Lyon et al. computed the activity of yttrium to be 6 × 10 -5 in Y2O2.94
at 1960 K. With this information, Lyon et al. (1973) determined the thermodynamic
feasibility of producing an acceptable-quality titanium ingot in an yttria crucible.
52 Metal–Crucible Interactions

If the objective is titanium containing 1500 ppm oxygen by weight, i.e. 4.5 × 10 −3
atom fraction oxygen, the oxygen potential in molten titanium will need to be 2 ×
10 −19 atm pO2 at 1960 K. The value is obtained by substituting for XO in the Sievert’s
relation. Stoichiometric Y2O3 at this temperature has a potential near l0 −16 atm pO2;
hence, a reaction between stoichiometric Y2O3 and titanium is possible, and there
will be oxygen ingestion from the yttria crucible into the titanium melt. Were the
yttria crucible reduced to a composition of about Y2O2.94 prior to charging, then the
oxygen potential would be matched between the oxide phase (l0 −19 atm at 1960K)
and the titanium phase (2 × 10 −19 atm), and driving force for transport of oxygen
between them would be greatly decreased.
From an oxygen content standpoint, then, yttria with the lowest possible oxygen
content (stoichiometry) would be the best. However, yttria has a rather narrow range
of stoichiometry, and the O/M ratio cannot be any lower than 1.47 for yttria accord-
ing to the diagram given in (Massalski 1990b). Incidentally, this diagram is redrawn
from the one in Shunk (1969). However, the results of Lyon et al. (1973) show that
the yttria phase can exist at oxygen pressures down to 10 −24 atm at 1960 K, implying
a possible lower oxygen stoichiometry limit for Y2O3, i.e. a value more than 0.06 for
x in Y2O3−X.
Alluding to this possibility, Lyon et al. (1973) proceeded to observe that should
titanium containing 1500 ppm oxygen be melted in a crucible whose stoichiometry
is lower than Y2O2.94, then one would expect the titanium to give up some of its oxy-
gen to the crucible. As regards yttrium pickup by titanium melt, Lyon et al. (1973)
have given the following expression relating concentration and activity of yttrium in
titanium at 1960 K:

a = 1.5 ´10 -2 w (2.7)

where a is the thermodynamic activity of yttrium and w is the weight percent of

yttrium expected to be dissolved in titanium. At 1960 K, if 400 ppm by weight
is the maximum yttrium pick-up allowed, then the activity of yttrium must be
6×l0 −4 or less, and the oxygen pressure to which this yttrium activity corresponds
is obtained from

( a Y )( pO2 )
0.75
= 3 ´ 10 -19 (2.8)

where aY is the desired activity of yttrium and pO2 is the oxygen pressure in atmo-
spheres. Solving this for oxygen gives pO2 = 3 × l0 −21 atm. According to Eq. (2.6),
the oxygen content corresponding to this partial pressure would be 5 × l0 −4 atom
fraction in the titanium. This value corresponds to less than 200 ppm by weight
oxygen in titanium. It is therefore entirely feasible to expect good-quality titanium
products (~200 ppm oxygen) melted in yttria crucibles, provided the yttria has been
pre-treated to ensure that it is of the proper stoichiometry.
Using published (Kubaschewski and Dench 1954; Komarek and Silver 1962; Okabe
et al. 1991) values for the partial molar free energies of Ti–O solutions and their variation
Basics of Interactions 53

FIGURE 2.8 Oxygen potential of Ti–O solid solution and selected metal oxides (From
Zhang et al. 2020).

with temperature, plots like the Ellingham diagrams can be generated. Zhang et al.
(2020) presented in such a plot the temperature variation of equilibrium oxygen partial
pressures over titanium–oxygen solid solutions at oxygen concentrations between 10
and 10,000 mass ppm, and the free energies of formation certain metal oxides. The
plot given as Figure 2.8 effectively defines the limits of oxygen contamination that can
be reached in titanium metal when using the pure stoichiometric oxides as crucibles.
A magnesia crucible would result in oxygen contamination of at least 2 wt.%, a calcia
crucible >0.1 wt% and yttria, as already mentioned, in the range of 0.05 wt%.

2.4 THE VANADIUM–OXYGEN SYSTEM

A situation similar to that for titanium exists in the vanadium–oxygen system too.
There are multiple solid phases in the vanadium–oxygen system with varying stoi-
chiometry ranges. The free energy diagram of the system in the composition range
from V2O5 to pure vanadium is shown in Figures 2.9 and 2.10 (Allen et al. 1951).
The vanadium–oxygen solid solution is less extensive than those of Ti–O or Zr–O.
Vanadium metal dissolves a maximum of 0.25 wt.% oxygen at 1000°C and 5 wt% O
at the eutectic temperature 1667°C. The stability of solution increases steeply as the
oxygen content decreases; removal of oxygen (deoxidation) becomes more difficult
as the metal becomes purer with respect to oxygen. In other words, as the purity of
54 Metal–Crucible Interactions

FIGURE 2.9 Partial molar free energy of oxygen in vanadium–oxygen system (From Allen,
N.P., Kubaschewski, O., and Von Goldbeck, O. 1951).

FIGURE 2.10 Partial molar free energy of oxygen in vanadium–oxygen system (From
Allen et al. 1951).
Basics of Interactions 55

the metal melt increases, it becomes more reactive. At 1000°C, the partial molar free
energy of oxygen in vanadium–oxygen solid solution containing 0.1 wt% oxygen
is comparable to the partial molar free energy of oxygen in titanium–oxygen solid
solution containing 0.01 wt% oxygen (~1100 kJ/mole O2). Contamination of oxygen
from the oxide crucible is likely to be a more serious issue in vanadium than in tita-
nium. The melting point of vanadium (1910°C) is also considerably higher than that
of titanium (1668°C).

2.5 COMPONENTS OF INTERACTION
The overall metal–crucible interaction consists of the following main events (Fashu
et al. 2020): (i) wetting of the crucible by the melt, (ii) penetration of the melt into
the crucible to various depths, (iii) reaction occurring at the surfaces in contact,
including the interior of pores, and (iv) products of reaction influencing the physi-
cal integrity of the crucible and properties of the penetrating melt. These events
occur in a loosely sequenced manner and somewhat cyclically. It is the rate at which
interaction occurs vis-à-vis the duration of the main process that is carried out in the
crucible that finally determines whether the interaction can be lived with or should
be remedied.
On melting, the metal’s contact with the surface of the crucible is enhanced, and
the liquid metal can also penetrate the pores. The extent of such contact is deter-
mined by the wetting between the liquid and the crucible material, fluidity of the
melt, pore sizes and their interconnectivity, and the presence of other pathways for
melt entry. Another factor is the hydrostatic head at the zone of contact between
the liquid and the crucible. If the fluidity is high enough and if good wetting exists
between the melt and the crucible, the melt can penetrate into the crucible by capil-
lary forces, access and attack the grain boundaries (Figure 2.11) and weaken the
bond between the grain and the crucible body. As noted in the general description,
eventually the grain detaches from the crucible and is entrained into the melt. The

FIGURE 2.11 Melt penetration and physical erosion (schematic).

56 Metal–Crucible Interactions

process is repeated on the next grain, and so on. Once in the melt, the particles follow
the motion of the melt and are carried away from their initial location.
The removal of the particles results in the enlargement of the channel, giving
access to the liquid to enter into the crucible surface region and eventually, more and
more into the bulk. One thing leads to the next, and the degradation of the crucible
by the combination of physical and chemical actions continues. The process is per-
petuated by the continuous increase in the conductance of the channels in the cruci-
ble surface, the consequent continuous increase in contact area between the crucible
and the liquid, and easier transport of entrained particles into the bulk of the melt.
The eroded ceramic particles may either survive in the melt or be dissolved in it.
Generally, the interaction between a ceramic crucible and molten metal involves
chemical reactions on the crucible wall and physical erosion. In such a situation,
erosion will complement chemical attack by removing the reaction products from
the surface, in the process exposing fresh surface, and the process goes on. The
events are somewhat similar to the familiar case of atmospheric oxidation, wherein
the oxide layer is non-protective and falls off on its own. The physical erosion in
the present case is an event brought about by the conditions prevailing in the melt,
such as free or forced convection, liquid currents generated by inductive stirring
accompanying the heating process, or currents of liquid and bubbles generated by
degassing processes in the melt or even by the stirring action of mechanical stirrers.
A simple possibility is that the melt enters the pore till it reaches the freeze line
and solidifies. No reaction occurs, and as the pores are filled, further penetration of
the melt also stops. The consequence is more thermal than chemical. The thermal
conductivity of the infiltrated portion often increases, and this will affect the thermal
balance of the system. An additional problem will arise when the crucible is cooled
and the infiltrated melt solidifies completely. Depending on the thermal expansion
characteristics of the solidified melt vis-à-vis the original crucible material, the
integrity of the crucible will be affected.
After permeating the pores, if the melt interacts with the crucible material, many
possibilities emerge. The sequence is different when reaction follows penetration
into the pores of the crucible wall. The reaction weakens the crucible structure when
one or more of its constituents, usually the binder or matrix, is dissolved in the melt
or otherwise transformed. The binding becomes weaker, and particles become loose
and are released into the melt. The pore size increases, and the process escalates. The
reverse may happen if the dissolution of crucible material into the melt increases the
melt viscosity. The progression of the melt into the pores slows down, and eventually,
the permeation becomes sluggish or stops.
When the interaction between the melt and pore walls results in a new chemical
product, and the physical properties of the products are different, things begin to
happen. The property to look for first is the density or specific volume of the product.
If the density of the product is higher than that of the original crucible material, the
product occupies less volume than the material it consumed; the pore size increases,
and new pore-like pathways may also be created. There will be escalation of penetra-
tion. If the density of the product is lower than that of the original crucible material,
the product occupies more volume than the material it consumed. One beneficial
Basics of Interactions 57

thing that may happen is that if the density difference is just right, the formation of
the product will cause closure or packing of the pores due to the additional space
needed. Continued infiltration will stop. On the other hand, if the volume expansion
is greater than can be accommodated in the pores, the product will end up disin-
tegrating the crucible material in the infiltrated zone. Additional damage may be
caused by the product’s thermal expansion and thermal conductivity characteristics.
When the characteristics of the product of initial interaction are right, in the best
case, the product forms as a compatible continuous and dense layer on the crucible
material and effectively protects it from any further attack, just like the protective
oxide film that forms on certain metallic materials during oxidation.
It all begins with the melt contacting the crucible surface and proceeds with a
simple dissolution of one or more of its components or a chemical reaction with the
crucible material followed by dissolution. The beginning is always a contact, and a
close contact between the melt and the crucible is determined by how well the liquid
wets the solid.

2.6 WETTING AND CONTACT ANGLE

Wettability is the ability of a liquid to spread over a solid surface. As it describes
the extent of intimate contact between a liquid and a solid, it is an important char-
acteristic for metal–crucible interactions. Generally, greater wettability results when
the adhesive force between the liquid and the solid is more than the cohesive force
(surface tension) within the liquid.
Figure 2.12 shows two-dimensional cross sections of a droplet on a solid surface.
The left most droplet does not spread over the solid surface and is said to have a large
contact angle. The right most droplet has spread well and has a low contact angle.
The characteristics of the middle drop are in between. This spreading is known as
‘wetting’, and a droplet either ‘wets’ or ‘dewets’ when in contact with a solid surface.
The angle between the droplet outline and the solid surface is the contact
angle. The angle is taken through the droplet to the tangent drawn to the droplet

FIGURE 2.12 Wetting and contact angle.

58 Metal–Crucible Interactions

outline at the point where the droplet periphery contacts the solid surface. A
contact angle gives an indication of how well (or how poorly) a liquid will spread
over a surface. A contact angle can be large or small, depending on the physical
properties of the materials involved. An acute contact angle represents wetting,
and an obtuse one represents dewetting. The smallest acute contact angle indi-
cates maximum wetting.
There are three boundaries to consider when a droplet is in contact with a solid sur-
face: the solid, the liquid, and the vapour (usually air) surrounding them. Figure 2.12
shows a force diagram of the point at which a droplet edge meets a solid surface. The
three arrows represent the forces exerted by the surface tensions at three interfaces:
liquid–surface(solid), liquid–vapour and solid–vapour (shown as sl= TiC/Al, lv = Al/
gas, sv = TiC/gas,).
Each force is pulling away from the equilibrium point, so if the droplet is in equi-
librium, then the forces are balanced and can be described by the following equation:

g sv = g sl + g lvcosq (2.9)

Where cos θ gives the x-component of the liquid–vapour surface tension. This can
be re-arranged to give:

cosq = ( g sv - g sl ) g lv (2.10)

If γsv < γsl, then cos θ will be negative, and θ is therefore >90° and the droplet dewets.
This can occur with a low-surface-energy solid or a high-surface-tension liquid (such
as water).
If γsv > γls, cos θ is positive, and θ is <90° and the droplet wets. This can occur
with a high-surface-energy solid (such as a metal) or a low-surface-tension liquid.
(If γsl is larger than γsv, the contact angle θ is larger than 90°; if the reverse is true,
θ is acute. If γsv = γsl + γlv, the liquid will spread out over the solid surface.)

cos q = ( g sv - g sl ) g lv (2.11)

g sl = g sv - g lvcosq (2.12)

The interface energy can be determined from the contact angle and liquid surface
tension if the solid surface energy is known. The work of adhesion represents the
decrease in energy in bringing together a unit area of liquid surface and a unit area
of solid surface to form a unit area of interface. The work of adhesion can also be
calculated directly from experimental data:

Wad = ( g lv + g sv ) - g sl = g lv (1 + cosq) (2.13)

The work of adhesion is maximum when the wetting is perfect. In other words, there
is maximum decrease in energy when a unit area of liquid surface and a unit area
Basics of Interactions 59

of solid surface are brought together under conditions of perfect wetting (contact
angle = 0) to form a unit area of interface.
The wettability of a given solid/liquid couple can be assessed by measuring the
contact angle in a system consisting of a drop resting on a flat solid surface, in a given
atmosphere, by the sessile drop method. A general (feature) limitation of the infor-
mation thus obtained has been highlighted by Lopez and Kennedy (2006).
The wettability of TiC by molten aluminium has been investigated in various
temperature ranges by many groups. More often than not, the measured contact
angle changes with time even while experimental conditions are maintained the
same. The actual interface chemistry and surface-active impurities originally
present in the solid and melt phases tend to play a decisive role (Fashu et al. 2020).
The contact angle at any point of time is determined by the real-time composi-
tion of the contacting surfaces. This is also highlighted in the results obtained by
Durov et al. (2005).
Wetting of zirconia by metals was studied by Durov et al. (2005). Single crys-
tals of 3% yttria-stabilized zirconia and some polycrystalline ZrO2 ceramic samples
were the substrates. Reactive metals (aluminium and vanadium) wet zirconia; less
reactive metals (tin, copper, silver) do not wet it. But, Durov et al. (2005) noted that
relating wetting to affinity for oxygen is not straightforward; silicon does not wet the
zirconia, whereas germanium wets it. The interaction of germanium with zirconium
plays a role, because stable intermetallics exist in the Ge–Zr system. A similar situa-
tion exists for the interaction between germanium and rare earth oxides.
The wetting of zirconia by the alloys Cu–Ga and Cu–Ge and also ternaries with
titanium, zirconium and vanadium was studied by Durov et al. (2005). The addition
of titanium and zirconium improves the wetting. Vanadium, even at low concentra-
tions, steeply decreases the contact angle, and the effect is greater than that due to
Ti and Zr.
When the wetting of different types of zirconia ceramic by a metal melt is com-
pared, monocrystalline zirconia is wetted more (smaller contact angle) than poly-
crystalline zirconia. The structure of zirconia material has a larger influence on the
wetting than its composition.
Black zirconia (non-stoichiometric oxide) is wetted by melts of noble metals and
alloys (Cu, Sn, Ag, Au, Pd, Pt, Cu–Ga, Pd–Rh) somewhat better (contact angles are
smaller) than white zirconia (contact angles are larger). Remarkably, after contact with
gold, platinum and palladium, black samples regained their white colour. On examina-
tion of the distribution of elements on ZrO2/Pd and ZrO2/Pt interfaces, it was revealed
that the excess zirconium in oxygen-deficit zirconia passes into the noble metal melt,
resulting in removal of the oxygen deficit and reversion of the colour to white, which is
characteristic of stoichiometric zirconia. While noble metals restore the stoichiometry
of non-stoichiometric zirconia (turning black to white), active metals make the stoi-
chiometric zirconia non-stoichiometric (turning white to black).
Wetting, in turn, determines the ease with which the melt will penetrate the pores
(invariably present) in the crucible and set the stage for further interaction. Non-
wettability is a property much desired in crucibles. When the wettability is poor
between the ceramic crucible and the melt, there will be a decrease in infiltration
60 Metal–Crucible Interactions

and eventual degradation of the refractory by the melt. To run a preliminary check
on crucible–melt interaction severity and wettability, sessile drop tests may be per-
formed on potential crucible materials.
The wetting of graphite/alumina substrates by molten iron at 1600°C was studied
by Zhao and Sahajwalla (2003). Pure graphite substrate was wetted by liquid iron with
the contact angle varying from 64° at the beginning to 38° at the end. The presence
of alumina had a strong influence on contact angle for carbon–alumina substrates.
The contact angle showed a sharp change from wetting to almost non-wetting as the
alumina content was increased from about 17% to 23%. Alumina also retarded car-
bon dissolution. The influence of alumina content on the contact angle between liquid
iron and alumina–carbon mixtures is summarized in Figure 2.13. Mixtures of alumina
and petroleum coke in liquid iron displayed non-wetting behaviour. Orsten and Oeters
(1986) also reported slower dissolution of carbon in less ordered forms of carbon mate-
rial (like petroleum coke) as compared with ordered counterparts like graphite.
That the contact angle is so dependent on the real-time chemistry of the interface
is both good news and bad news in the design of crucible compositions. It is not good
news because the impurities, especially surface-active ones, in crucible composi-
tions can end up behaving in unpredictable ways with expensive consequences. It
is good news because the window is open for deliberate choice of impurities (now
called additives) to modify the contact angle by analogy and experimentation.

FIGURE 2.13 Contact angle of liquid iron with alumina–graphite (From Zhao, L. and
Sahajwalla, V. 2003).
Basics of Interactions 61

2.7 POROSITY
Almost all refractory materials are porous. The pores may occupy 1–80% of the
material. The pores may be open (connected with the atmosphere) or closed (isolated
from the atmosphere and each other). Refractory materials always contain some
number of closed pores.
The ratio of open to closed porosity in refractory ceramics is, to some extent,
determined by the sintering process. In bricks with porosity exceeding 15%, the pro-
portion of closed pores is normally small (<ca. 3%). The pores of the ceramic body,
fabricated from non-porous starting grains, remain open until the relative density
during sintering reaches 83–85%, after which closed pores will begin to appear. At
a relative density of 95%, almost all the pores are closed.
Generally, the porosity of traditional fireclay refractories is 22–25%, in modern
aluminosilicates it is 13–15% and in silicon carbide refractories it is usually 12–18%.
The porosity of low-cement and ultralow-cement castables is 12–15%, but the propor-
tion of permeable pores is lower. Cathode carbon refractories, used in Hall–Héroult
cells, have porosities of 15–22%, and almost all pores are open and permeable. In
carbon materials, for example, almost all the pores are open (Yurkov 2017).
Different refractory materials may have approximately equal values of porosity
but a different ratio of permeable pores, a different pore size distribution and a dif-
ferent pore structure.
Porosity and density are important and to some extent controllable characteristics
of refractory materials. The mechanical, thermal and other properties of refractories
depend on them.
Porosity influences the strength properties of refractories. A certain porosity is
usually necessary for the refractory to have the adequate thermal shock resistance,
while corrosion resistance or resistance to metal melt or molten slag attack becomes
better with a decrease in porosity. Both these conflicting requirements are to be met
simultaneously. Difficult, but it has to be done in many instances.

2.8 CAPILLARY PENETRATION
The primary condition for metal–crucible interaction is the close contact between
the melt and the crucible over a large interface area. Apart from a small contact
angle, penetration of the liquid metal into a porous crucible structure is a major fac-
tor. Porosity, pore size, pore size distribution, fluidity of the melt and the hydrostatic
head at the zone of contact between the liquid and the crucible are the key factors.
The earliest systematic theoretical description of liquid flow in cylindrical capil-
laries, taking into account capillary pressure, gravity and outside pressure, is attrib-
uted to Washburn (1921). A very elegant description of the whole process was given
by Kaptay et al. (2004).
The presumption in the classical capillary theory of penetration is that the refrac-
tory consists of a large number of vertical, cylindrical capillaries of identical internal
radii r (Figure 2.14). There are three pressure vectors acting on the liquid column in
those capillaries:
62 Metal–Crucible Interactions

FIGURE 2.14 The classical capillary penetration model (From Kaptay, G., Matsushita, T.,
Mukai, K., and Ohuchi, T. 2004).

1. If the liquid wets the inner wall of the capillary, contact angle, θ, is less
than 90, the capillary pressure (Pc), pushes the liquid up (capillary sucks in
the liquid). When the contact angle is more than 90, the capillary pressure
pushes the liquid out of the capillary

PC = (2g cos q) r (2.14)

where γ is the surface tension of the liquid (J/m2).
2. The pressure due to gravity (Pg), always acts to push the liquid out of the
capillary:

Pg = - r gh (2.15)

where
ρ = density of the liquid metal (kg/m3);
g =  9.81 m/s2, the acceleration due to gravity; and
h =  the penetration height, measured from the bottom of the refractory (m).

In experiments, it is usual to provide the outside pressure difference (ΔP0) so that the
liquid rises into the capillary. The liquid column in the capillary eventually comes to
equilibrium if the sum of these three pressures (PS) is zero

PS = PC + Pg + DP0 = 0 (2.16)
Basics of Interactions 63

When liquid column height is less than 1 mm, Pg is negligible compared with PC, and
the condition for penetration to begin becomes

PC + DP0 ³ 0 (2.17)

Substituting Eq. (2.14) into Eq. (2.17),

DP0 + (2 g cos q) / r ³ 0 (2.18)

From condition (2.18), two results follow.

1. Spontaneous penetration occurs (i.e. at ΔPo = 0) if the contact angle is acute.
This contact angle, known as the threshold contact angle of penetration, is usually
different from 90°. However, for capillaries with parallel vertical walls of any cross
section, it is always 90°
2. When the contact angle is larger than the threshold contact angle of penetra-
tion, for any penetration to occur, the outside pressure needs to be higher than the
threshold pressure. The threshold pressure can be expressed as

DPth = -(2g cos q) / r (2.19)

When ΔPo > ΔPth, penetration will start. As the liquid column height becomes
significant, Pg can no longer be neglected compared with Pc. Then, the following
condition of equilibrium can be obtained by substituting Eqs. (2.14) and (2.15) into
Eq. (2.16):

(2 g cos q) / r + DP0 - r gh = 0 (2.20)

From Eq. (2.20), the equilibrium (or maximum) penetration height (hmax) can be
expressed as

hmax = DP0 / r g + 2g cosq / r r g = (DP0 - DPTh ) / r g (2.21)

Zhang and Lee (1999, 2004) estimated the depth of slag penetration using the
equation:

L2 = (r cosq)gt / 2m (2.22)

where L = penetration depth, r = radius of the pores, θ= contact angle, γ = surface ten-
sion of liquid, t = interaction time and μ = viscosity of liquid slag.
It is apparent that a large pore size, wettable crucible surface, high melt surface
tension, less viscous melt and long interaction times cause deeper melt penetration
into the crucible. A decrease in pore radius and increase in viscosity of the melt can
effectively stop any penetration. These two are useful handles in liquid ingestion
control.
64 Metal–Crucible Interactions

The slag viscosity is a function of both temperature and chemical composition

of slag. The penetration depth can be decreased by using a refractory material of
low thermal conductivity, which creates a thermal gradient between the hot face
and the cold face of the lining. As the slag penetrates into the refractory, the viscos-
ity increases with depth of penetration (because the temperature decreases). After
some distance, the temperature drop and consequent viscosity increase stop further
penetration. The other factor to reduce the viscosity is the change of chemistry of the
slag during operation. The spinel structure, for example, is able to absorb Fe and Mn
ions from the slag. As a result, the slag becomes rich in SiO2, and viscosity increases.
Any other reaction that results in an increase in the concentration of silica in the slag
increases its viscosity. An increase in the concentration of FeO, MnO makes the slag
more fluid.
When discussing pore size dimensions in a refractory, it is necessary to men-
tion the surface tension of the melt. The penetration of the melt in the refractory
depends not only on the pore radii and viscosity but also on the surface tension of the
melt, which is in any case also a factor that determines the contact angle of wetting.
The typical threshold pore diameter above which penetration of various metal melts
becomes serious is given by Rouchka (2001) and is shown in Figure 2.15.

FIGURE 2.15 Critical pore diameter for infiltration at the melting point of metals (From
Routschka, G. (Ed.). 2001).
Basics of Interactions 65

2.9 PHYSICAL PROPERTIES OF REFRACTORIES

Apart from chemical behaviour, certain physical properties of both the refractories
and the melts are of key importance in determining the extent and rate of metal–cru-
cible interactions. Foremost among these are the melting point of refractories and
the freezing point of melts. Then, there are properties like thermal expansion coef-
ficient, thermal conductivity, high temperature strength, thermal shock resistance,
polymorphism and many more of contextual relevance. Data on melting, softening
points, shock resistance and Ellingham free energy of formation for selected poten-
tial refractory materials are listed in Table 2.2.

2.10 THERMAL EXPANSION
Like all solids, refractories expand on heating. Figure 2.16 shows the expansion
behaviour of some refractory types (Fruehan 1998). The shape of the expansion
curve and the magnitude of the expansion are distinctly different for silica. The
complex mineral makeup of silica brick confers on it a uniquely different expansion
behaviour. Other refractories show a constant linear behaviour, except that the slope
differs appreciably among the materials. In any refractory assembly, allowance is
always made for accommodating such expansions in service (Fruehan 1998).
Thermal expansion curves for magnesite–carbon bricks with various antioxidant
additives and plain magnesite are shown in Figure 2.17. Their shapes are more or
less similar. In the temperature range 370–540°C, the formation of a glassy carbon

TABLE 2.2
Properties of Refractory Ceramics for High-Temperature Applications
Coefficient of
thermal Melting Thermal
DG 02000 Thermal expansion (softening) shock
kJ/mole conductivity (×10−6/K), temperature, resistance
Ceramics O2 (W/m/K) 20–1000°C °C factor
SiO2 −550.7 1.2–1.4 0.5–0.8 1710 (1280) 93
Al2O3 −702.6 38 8.0 2072 (1540) 19–26
MgO −660.0 30–60 9–12 2852 (2100) 18
CaO −809.0 30 15.2 2572 (1950) –
ZrO2 −725.5 2.5 10.0 2715 (2010) 12–13
Y2O3 −891.3 8–12 8.1 2425 (1855) –
AlN −185.6* 280 4.5 2200 37
BeO −800.0 370 8 2507 –
BN −159.1a 1300 7.5 2200 –
Graphite – 140 0.6–5.2 3530 (2680) –

Source: Fashu et al., 2020.

a kJ/mole N
2
66 Metal–Crucible Interactions

FIGURE 2.16 Thermal expansion behaviour of refractories (From Fruehan, R.J. 1988,
Routschka, G. 2001).

FIGURE 2.17 Thermal expansion behaviour of magnesite–carbon bricks (From Fruehan,

R.J. 1988, Routschka, G. 2001).
Basics of Interactions 67

bond causes densification of the structure and shrinkage. In refractories that contain
metals, thermal expansion at high temperatures increases significantly with increase
in metal content.
Crystals with a cubic lattice (CaO, MgO) expand equally along all axes (isotro-
pic). The thermal expansion coefficients for such materials are 6–15 × 10 −6 К−1. The
coefficients of thermal expansion are different along different axes for anisotropic
crystals with low symmetry, but this difference becomes smaller with increase in
temperature. Materials with strong chemical bonds (silicon carbide, titanium dibo-
ride, diamond) have low values of linear coefficients of thermal expansion. Pores
do not affect the values of linear coefficients of thermal expansion if the continuous
media are solid particles. Thermal expansion is a key component in determining
thermal shock resistance.

2.11 THERMAL CONDUCTIVITY
As insulators, refractory materials have always been used to conserve heat for the
charge and protect the metal shell of reactor vessels from overheating. The resistance
to heat flow is the reason for their selection in many applications. Figures 2.18 and
2.19 show thermal conductivity curves for several refractories (Fruehan 1998).
The hot face refractories in a particular application must be able to withstand the
higher temperatures that will result when layers of insulating backup materials are

FIGURE 2.18 Thermal conductivity of various refractories I (From Fruehan, R.J. 1988,
Routschka, G. 2001).
68 Metal–Crucible Interactions

FIGURE 2.19 Thermal conductivity of various refractories II (From Fruehan, R.J. 1988,
Routschka, G. 2001).

added. Insulation increases the depth of penetration and chemical attack on the hot
face layer. Most insulating materials are prone to attack on direct exposure to metal
or slag, vaporized process components (alkali, sulphur compounds, acids) or their
condensates. In certain applications, high-conductivity refractories are used to facili-
tate cooling of the refractory lining and retard further refractory wear. For example,
in a bosh construction, graphite and semi-graphite materials with conductivities of
70–80 and 30–35 Wm–1K–1, respectively, at 1000°C are used. The heat conducted
is finally extracted through the copper coolers. Thermal conductivity, like thermal
expansion, is also a key factor in determining thermal shock resistance.

2.12 THERMAL SHOCK
Thermal shock resistance is a complex issue. Thermal shock or spalling is caused
by stresses that develop from uneven rates of expansion and contraction within the
refractory, caused by rapid temperature changes or high inherent temperature gradi-
ents in a refractory. A qualitative prediction of the resistance of materials to fracture
by thermal shock can be expressed by the factor (Fruehan 1998):

ks
aE
Basics of Interactions 69

Where k = thermal conductivity,

s  =  tensile strength,
a  =  coefficient of thermal expansion,
E  =  modulus of elasticity.

According to this factor, the material should have a high thermal shock resistance
if it has a low elastic modulus, a low value of the coefficient of linear thermal expan-
sion, and high strength as well as high thermal conductivity. The higher the value of
the factor in the expression, the higher the predicted thermal shock resistance of the
material. A substantial proportion of all refractory failures are traceable to thermal
shock issues. Refractories are damaged by poor thermal shock resistance at tempera-
tures much lower than their regular service temperatures. Thermal stresses may be
tensile, compressive or shear; due to thermal stresses, refractory materials may chip
and split off the linings. Thermal shock resistance is not a physical property of a
material; it depends on many physical characteristics of the material, but also on the
shape size of the refractory part and the cooling and heating conditions.

2.13 CORROSION OF REFRACTORIES
Corrosive attack on the crucible refractories in a metal processing unit may be
mounted by the molten metal, molten slag or the process gas. Often, it is the combi-
nation of all three, each contributing to the overall degradation. This is apart from,
and in addition to, the thermal and mechanical factors that cause weakening or fail-
ure of refractories. The most significant and life-limiting attack is by the slag. Even
the corrosion by the metal melt becomes more serious if it acquires characteristics
of a slag, such as when it is oxidized or laden with oxides. The attack can happen
in essentially two ways. The following description is fairly general and reflects the
essential features of refractory degradation, be it happening in the lining of a large
furnace or in a compact system comprising a crucible and melt in a controlled envi-
ronment (Lee and Zhang 2004).
Direct (also known as congruent or homogeneous) attack is controlled by the
reaction rate at the slag–refractory interface or the rate of transport of the reactant
through the liquid film adjacent to the interface. Indirect (incongruent or hetero-
geneous) attack is controlled by diffusive transport of species to the reaction site
through the slag or through a new solid phase, which forms at the original slag–
refractory interface. This may lead to passive corrosion. The composition and hence,
the viscosity of the local liquid slag adjacent to the solid refractory have a critical
influence on corrosion. As is true with most of the rate processes, it is seldom that
one process or other exclusively controls the rate. Usually, one predominates over
the other. Penetration and corrosion can be controlled either through the local slag
composition, via the refractory or the bulk slag, or by microstructural control of the
refractory, e.g. by internal generation of dense layers or external deposition/genera-
tion of passive coatings. Selective corrosion may also occur, in which only certain
phases in the solid are attacked. A good example of this is the decarburization of
carbon-containing refractories, which occurs by several mechanisms, including
70 Metal–Crucible Interactions

dissolution of carbon into molten steel. Once the carbon has been removed, the
refractory is open to be wetted by the slag, so that penetration and spalling of the
decarburized layer can occur.
In direct, congruent or homogeneous dissolution, atoms from the solid dissolve
directly into the liquid melt. Direct dissolution can be reaction or interface controlled
when the diffusivity of reaction products is faster than the rate of chemical reaction
at the interface. The dissolution process is then directly controlled by the reaction,
whose initial rate can be expressed by

J = K ( AC / AO ) Cm (2.23)

Where J is the dissolution rate (g/cm/s), K is the rate constant, AC is the actual area
of refractory (cm2), AO is the apparent area of refractory (cm2), and Cm is the con-
centration of reactant in the melt (g/cm3). Surface irregularities such as grooves and
porosity, which increase the ratio of Ac to Ao in Eq. (2.23), become important. For
this simple treatment, stirring of the melt has no apparent effect on dissolution rate.
The concentration of the reactant at the interface (considered to be the same as the
concentration in the bulk) and the area available for interaction are the only two
things that matter, and these are equally important.
For direct dissolution to continue, the products diffuse away from the interface at
a rate proportional to t1/2 as reactants are depleted and dissolved species build up in
the absence of liquid convection or stirring (Cooper and Kingery 1959). In a situation
here the rate of removal of reaction products by diffusion is slower than the rate of
chemical reaction, a solute-rich boundary layer builds up, whose interface with the
refractory is saturated with reaction products. The diffusion of reactants to, or the
products away from, the interface through the boundary layer then governs the dis-
solution process. If the boundary layer leads to formation of a solid interlayer, this is
termed indirect, incongruent or heterogeneous dissolution, and the rate of corrosion
can be expressed in terms of the Nernst equation:

J = D(CS / Cm ) / d (2.24)

Where D is the diffusion coefficient (in cm2/s), Cs is the saturation concentration of

refractory component in the melt (g/cm3), and δ is the effective boundary layer thick-
ness (cm), which is defined as

d = ( CS - Cm ) / (dc / dx ) (2.25)

Where dc/dx is the concentration gradient over the interface. For diffusion-con-
trolled direct dissolution, diffusion of the product through the melt boundary layer is
considered. However, once saturation occurs, and a solid interlayer has precipitated
between the melt and the refractory, this is the diffusion-controlled indirect process,
and solid-state diffusion of the reaction species from the melt through the solid inter-
layer must be considered.
Basics of Interactions 71

Stirring the melt or rotating the refractory sample enhances the rate of indirect
dissolution (or effectively converts it to direct dissolution) by reducing the thickness
of any liquid boundary layer (or breaking up any solid layer). Apart from dynamic
viscosity, the value of (Cs − Cm) has a great influence on the dissolution rate. If
the refractory oxide in the bulk slag has been saturated, then J = 0. Naturally, to
minimize the dissolution rate, (Cs − Cm) must be minimized. For example, increased
MgO content in slag decreases the corrosion of periclase, the primary phase in many
basic refractories. If Cm = 0, then (Cs − Cm) reaches a maximum value, and so does
the dissolution rate. Consequently, the solubility of refractory oxides in molten slag
and the saturation concentration at the interface between refractory and slag are very
important in corrosion of refractories.
Slag viscosity has significant effects on both the slag penetration and the refrac-
tory dissolution. A more fluid slag will be more penetrating and more likely to dis-
solve the solid refractory. Furthermore, if dissolving the refractory in the liquid leads
to increased viscosity, then mass transport through the melt layer next to it will be
slower, so that the melt layer becomes progressively saturated, giving rise to diffu-
sion control, i.e. indirect dissolution. On the other hand, if the viscosity of the melt
layer is decreased, then diffusion through it becomes more rapid, no saturated layer
forms, and reaction control or direct dissolution occurs. Raising the temperature to
give a more fluid slag can then lead to a change from indirect to direct dissolution.
Even in the ancient process of cupellation, the role of melt viscosity for process
success was highlighted. Basically, a melt containing lead oxide and silver at 960 to
1000°C had to be separated physically. Liquid lead oxide was to be absorbed into a
special porous crucible, leaving behind silver. The crucible had to be made of inert
and porous material rich in calcium or magnesium, such as shells, lime or bone ash.
Only a calcareous lining would work, because lead reacts with silica (clay) to form
lead silicate, and lead silicate is viscous and impedes the absorption of litharge. Lead
does not react with the calcareous materials, and fluid lead oxide melt is readily
captured by the crucible purely by the physical phenomena of absorption. Generally,
a silicate enrichment in an oxide slag makes it viscous, and so does FeO depletion.
This second aspect was in a way responsible for the delayed implementation of iron-
making with separate molten iron and slag tapping in shaft furnaces in the early days.

2.14 CHEMICAL DISSOLUTION IN CASTING

The principal difference between metal–crucible interaction during melting and dur-
ing casting is that in the latter case, the duration of contact between the melt and
crucible/mould is very limited, and the melt is relatively quiescent before it sets into
a solid. Immediately after its formation (i.e. melting in the crucible or teeming into
the mould), the melt tends to react with the crucible and dissolve the constituents of
the crucible. For example, for an Y2O3 crucible, the dissolution reaction is

Y2O3 = 2YL + 3O L (2.26)

Where L stands for melt, such as molten titanium or titanium alloy.

72 Metal–Crucible Interactions

In conditions conducive to mass transport, the dissolution will be at the maximum

rate, and the melt will soon attain a uniform concentration. The dimensions of the
crucible are also a factor. When the sample is allowed to solidify inside the crucible,
the interaction between the crucible and the melt develops somewhat differently.
With the progress of solidification of the alloy, in regions that have become solid,
mass transport can occur only by diffusion, and in portions that are still liquid, mix-
ing occurs but to a much lower degree, because stirring and similar mass transport
aids no longer operate. As a result, composition gradients develop between the sur-
face and the bulk and can be discerned in the solidified material.

2.15 CORROSION TESTING OF REFRACTORIES

There are various tests for single refractory properties. Because most refractories
wear by a complex combination of mechanisms, many simulated tests have been
developed to study refractory behaviour for specific applications (Fruehan 1998; Lee
and Zhang 2004). Tests have been used to create the environment a refractory is
exposed to during service. They are categorized as dynamic, where the fluid moves
relative to the refractory, and static, where there is no such motion.
In the button or sessile drop test, shaped slag is placed on a refractory substrate,
heated to temperature and held for a fixed time to allow the slag to wet and react
with the refractory. The technique is used to measure interface and surface energies
in systems where the fluid is liquid and does not react with the solid. In the dipping,
immersion or finger test, one or more cylindrical or square-pillar-shaped refractory
samples are held in the corrosive slag for a certain period in an electric or induction
furnace. In this method, the atmosphere is easily controlled, and by using a large vol-
ume of slag relative to the size of the samples, the composition variation of the slag,
associated with rapid saturation with reaction products, can be minimized. These
tests are isothermal, i.e. at the same temperature throughout their volume.
In the crucible, cavity, cup or brick test, a cored-out refractory brick is filled
with slag and exposed to high temperature to promote slag–refractory interaction.
In this simple method, many samples can be tested in a short time. However, it has
limitations associated with static tests, i.e. no temperature gradient, rapid saturation
of slag composition with reaction products (often, all slag is absorbed into the brick)
and no slag flow. In tests conducted in an induction furnace, a temperature gradient
can be established, and atmosphere and temperature are easily controlled. Rapid
and vigorous corrosion can occur at the melt/slag line, since the less dense slag
floats on the metal. An induction furnace test induces some motion (electromag-
netic stirring) upon heating and so has some dynamic aspects. The results of such
a test can better simulate the situation in a real furnace, except that flow is uncon-
trolled. Relative motion of liquid slag and solid refractory is particularly important
in systems where indirect dissolution occurs. The motion can remove the bound-
ary layer and enhance any active corrosion process. Dynamic tests are needed to
simulate such behaviour. In fact, this can be a signature test of whether dissolution
is indirect (in which case corrosion rate increases with motion) or direct (in which
case motion has no effect).
Basics of Interactions 73

The dipping test can be made dynamic by spinning the samples in the molten
slag, a rotating finger test. A popular corrosion test is the rotary slag test, in which
the slag is melted with a flame in a cylindrical refractory-lined drum rotating about a
horizontal axis. The slag is refreshed periodically during the test by tilting to remove
old slag, returning to level and adding new slag. In this method, many samples can
be compared in a single test; a temperature gradient can also be established, and the
composition and fluidity of the slag are partially controlled.
The chemical composition, hardness and microstructure at the metal–crucible
interface can be measured by secondary ion mass spectrometry, scanning electron
microscopy/energy-dispersive spectroscopy (SEM/EDS) and X-ray spectroscopy
(XRS). Chemical composition measurement can be performed by quantitative EDS
analysis with appropriate attachments using a scanning electron microscope. Overall
oxygen content may be measured using the inert gas fusion (IGF) technique, and
oxygen content variation from the surface to the inside of samples by secondary ion
mass spectroscopy (SIMS).
In SIMS, a 4 keV 200 nA beam of Ar+, for example, is focused in a 50 μm diam-
eter area at a time, and secondary ions produced by sputtering are mass analysed by
the attached quadrupole-type mass spectrometer. XRD identifies compounds pres-
ent on the microstructure. Microhardness may also be evaluated with a hardness
tester. Hardness is a popular indicator of material purity.
Used and or failed refractories are examined by chemical analyses and by micro-
scopic methods for clues. X-ray, SEM, or electron probe techniques are used as part
of postmortem examinations, and valuable insights are often gleaned, giving direc-
tion to effective corrective actions (Fruehan 1998).

2.16 SUMMARY
Metal–crucible interaction operates by chemical interaction between the material
of the crucible and the melt, augmented by concurrent physical, mechanical and
thermal processes. Chemical interaction, in turn, generates products that because
of their own properties, may cause changes that also result in degradation. The key
process, therefore, is the chemical reaction, and chemical reactions are governed by
thermodynamics.
Ellingham diagrams, essentially the plot of standard free energy of formation
of compounds as a function of temperature, do indicate whether a reaction may be
expected between a compound of one metal and another metal in its elemental form,
in the present case between the melt and the crucible material. In numerous cases,
the predictions have been true and complete, and processes have been founded on
that basis. In many other cases, the predictions do not reflect the real situation com-
pletely, because the data on the basis of which the diagrams have been prepared
are in themselves incomplete. This is especially true for reactive metals that form
extensive and stable solutions, and also oxides that exhibit non-stoichiometry. It is
therefore necessary to consider the relative stabilities of the oxides, including the
known hypostoichiometric oxides, and also that of the solutions the metal forms.
Proceeding thus, as a test case, it has been highlighted that in contact with the most
74 Metal–Crucible Interactions

stable oxide crucible, yttria, near the melting point of titanium, the residual oxygen
content in titanium cannot decrease below 0.05 wt.% oxygen. In contrast, a casual
consideration of Ellingham plots may point to the possibility of titanium metal (by
implication ‘pure’) being obtainable by reduction with many common reducing
agents (metals, of course). By extension, this also means the possibility of melting
titanium without contamination in many crucible materials. Experience with melting
procedures has shown that this assertion does not reflect reality.
Even though the statement that at sufficiently high temperatures, everything
reacts with everything else cannot be disputed, it can be tempered with many ifs
and buts. These conditions, which are extremely important in real-life situations, are
physical, structural, mechanical and thermal. A crucible material should possess,
in addition to high thermodynamic (chemical) stability, high melting point, good
strength, appropriate thermal expansion and thermal conductivity behaviour, and
thus, thermal shock resistance. Some of these properties are interrelated. Tailoring
of some of these properties by microstructure and porosity control has been resorted
to more or less routinely in order to ensure that the chemical stability of the crucible
material is fully utilized in actual application.
To interact, the melt and crucible have to be in close contact. This aspect is decided
by the wetting characteristics of the melt–crucible combination. The contact angle
is a measure of wetting and is determined by the interfacial tensions of the pairs:
liquid–solid, liquid–vapour and solid–vapour. The contact angle or wettability is
ultimately determined by the real-time chemical composition of the melt–solid inter-
face. Impurities in the melt or crucible play a role, and this is the basis of anti-wetting
agents. Anti-wetting additives act by controlling the interface chemistry and raising
the contact angle. They have been used extensively with aluminium melts because of
the temperatures involved. Nearly all anti-wetting additives decompose at high tem-
peratures, and then the protection is lost (usually beyond 1000 to 1200°C). For this
reason, porosity control may turn out to be the more reliable handle to obviate melt
intrusion. However, modifying porosity may have other unintended consequences.
Corrosive attack on the refractory of the crucible is often due to the combined
action of molten metal, slag and also process gas. The most significant and life-
limiting attack is by the slag and also by metal that acquires the characteristics of
slag. The attack proceeds in two ways. Direct dissolution is controlled by the rate of
chemical reaction at the slag–refractory interface. Indirect attack is controlled by the
rate of diffusive transport through the slag or through the newly formed solid phase.
Significantly, penetration and thus, corrosion are determined by local slag composi-
tion and also by refractory composition and structure. These are also usable handles.
The melt properties, particularly viscosity, are an important factor in the corrosion
process. Many attack prevention strategies rely on controlling this property by tem-
perature and/or composition manipulations.
Testing to quantify and possibly elucidate the nature of corrosion is an impor-
tant component of metal–crucible interaction management. In these tests, the extent
to which real process conditions are reproduced determines the reliability of the
results, which together with thermodynamic calculations can serve to describe the
overall process more accurately. An important aspect of testing is the post mortem
Basics of Interactions 75

examination of the failed metal–crucible (and also a successfully operating) system

and parts of it. More often, this information complements other tests or provides
new directions for evaluation. X-ray, microscopy and metallographic examinations,
as well as chemical analysis, are all brought in to complete the picture. In the matter
of metal–crucible interactions, focusing on the basics of interaction and using the
knowledge thus gained has been a highly effective route for managing the inter-
actions to save the equipment and the melt and to run the process in a cost- and
resource-effective manner.
 3 Crucible Materials

3.1 INTRODUCTION
Broadly, a crucible can be considered as the material surface at work in contact with the
melt and related fluids during material processing. This emphasizes the importance of
the popular concept of ‘working lining’ of process vessels. When a crucible is referred
to here, it means a vessel holding the melt and also the lining of a furnace system that
is in contact with the metal melt or slag. The work here is containing the melt, resist-
ing and surviving its attack, and withstanding the concurrent thermal and mechanical
stresses the process entails; and, doing all this without causing any degradation of the
melt. The crucible also needs to maintain its integrity and functionality for durations
that are useful in practice. In the field of common metals, both ferrous and common
non-ferrous, the crucibles are usually refractory ceramic materials. In the case of less
common metals, the crucibles are more specialized and are often made of refractory
ceramic compounds, refractory metals and also some iron and nickel alloys.
Crucibles can be made of metals and alloys, oxide minerals, pure oxide composi-
tions, or compounds such as non-oxide ceramics, like metal borides, carbides, nitrides,
silicides, phosphides, sulphides and also fluorides. Carbon is an exceedingly useful
crucible material, mostly as graphite and also as vitreous carbon. Many of the proper-
ties of these materials are available not only when they are used in monolithic form
but also when they are used as linings or coatings on other materials. An impressively
versatile range of applications is fulfilled by the class known as water-cooled crucibles.
Arc and electron beam melting furnaces in a variety of configurations are enabled by
water-cooled copper crucibles. While these crucibles impose certain limitations on
the processing and on the product, their role in materials processing is unique and
irreplaceable. An interesting and highly effective form of water-cooled ‘crucible’ in
induction heating systems is exemplified by skull melting or cold crucible induction
melting. This system is made possible by exploitation of the surface tension properties
of the melt. The same property is used in another technique that reaches a higher level
of sophistication–float zone melting, in both the induction-heated and electron beam-
heated modes. Interaction with the crucible is entirely avoided, at least during melting,
in the process known as levitation melting. The charge is heated and melted while sus-
pended out of contact with any material container. The amount of material handled by
this technique, as of now, is very small. In this chapter, materials that have been used as
crucibles are surveyed, focusing on the properties and characteristics that make them
relevant for applications and also features that limit their applications.

3.2 CERAMICS
The use of the term ceramic implies any of the numerous hard, brittle, heat- and
corrosion-resistant non-metallic materials. As a class, ceramics are the most widely

DOI: 10.1201/9780429345562-3 77
78 Metal–Crucible Interactions

used crucible materials and have been around for millennia. Early civilizations
found that certain clay minerals became plastic when sprinkled with water and could
be moulded into shapes. They could be hardened by drying in the sun and hardened
further in a wood or charcoal fire. Many of the raw materials used by the ancient
civilizations are still in use today and are generally known as ‘traditional ceramics’
(Richerson and Lee 2018).
It is only relatively recently, i.e. during the past few centuries, that naturally
occurring minerals have been refined and improved and also new compositions syn-
thesized to achieve enhanced and new properties in ceramics. These improved mate-
rials are often called ‘modern ceramics’. They are of controlled composition and
structure and are often engineered to the needs of applications that have demands
far beyond the capabilities of the traditional ceramics. Modern ceramics include
plain oxides (such as Al2O3, ZrO2, ThO2, BeO, MgO and Y2O3), oxidic compounds
(MgAl2O4, BaZrO3, etc.), borides, carbides and nitrides (such as TiB2, TiC, SiC, B4C,
Si3N4 and AlN) and many more.

3.3 REFRACTORIES
In day-to-day usage, ceramic structural materials that retain their integrity at tem-
peratures above about 1000°C and under the ‘harsh’ mechanical and chemical condi-
tions that often accompany such temperatures are known as refractories. A refractory
endures not only elevated temperatures but also a certain amount of mechanical load
concurrently (Phelps and Wachtman 2011).
The term high temperatures is somewhat unspecific but usually means about
1000°C and above, or temperatures at which, because of loss of strength, oxidation
or melting, neither the traditional ceramics nor metals can be used. The greatest use
of refractories is in the steel industry for construction of linings of equipment such as
blast furnaces, converters, transportation ladles and casting equipment. Similar uses
abound in non-ferrous metallurgical furnaces and process vessels.
The ‘refractoriness’ of a ceramicis, in one way or other, rated by the tempera-
ture up to which it can resist deformation under load and/or creep. A traditional
method is to measure its pyrometric cone equivalent (PCE), and refractory ceramics
are formally defined as non-metallic materials having a minimum PCE of 1500°C
(Fruehan 1998).
In the beginning, metal melting was the purpose that often started and generally
sustained the development of refractories. The refractories were mostly quartz-
containing clays. Apart from standing up to modest temperatures, they did not pos-
sess many of the properties that modern -day refractory products have by default.
Refractory technology was born at the beginning of the 19th century, when the
addition of pre-fired clay or grog material was found to result in bricks with better
dimensional integrity. The intrinsic properties of a refractory can be significantly
enhanced for a given application by suitable techniques for its fabrication into a
component.
Refractory materials may be broadly grouped into siliceous, aluminous,
aluminosilicates, magnesite, dolomite or chrome ore and inevitably, the
Crucible Materials 79

miscellaneous category of special refractories. Refractories are used in two basic

forms: (i) pre-shaped objects (bricks and shapes) and (ii) unformed compositions
or monolithics.
Specialties or monolithics are refractories that are applied to form a mono-
lithic, integral structure and cure to a stable dimensional form through hydraulic
or chemical setting post application. They are a combination of refractory oxides
that include Al2O3, SiO2, Fe2O3 and CaO, formulated for specific requirements
and capabilities for service at temperatures as high as 1800°C. They are available
as castable, mouldable and ramming mixes. The mortars used to install bricks
and shapes also belong to this group, and their compositions are chosen to match
the brick they bond together. The objective is to achieve a lining that is as close
as possible to a uniform refractory structure. Castables, or refractory concretes,
are dry, granular refractory mixes suitable for mixing on site with water prior to
application. They are used to form complete furnace linings and other non-stan-
dard shapes (Fruehan 1998). In hydraulic-bonded castables, the use of MgO is not
particularly widespread because of (i) excessive expansion due to hydration when
used with cement-containing hydraulic bond, (ii) rapid setting, (iii) high thermal
expansion of MgO at steel-making temperature and (iv) low strength at high tem-
perature. The use of micro-silica as binder prevents hydration and thus, may avoid
the associated problems.
As regards pre-shaped objects, hand moulding, once widely used, is now used
only for special shapes, and power pressing is common. Hot pressing and isostatic
pressing, both cold and hot, are used for difficult or special refractories. Slip casting
is also used for special refractory shapes. Post shaping, the refractories are usually
fired in kilns, but not the chemically bonded types. These are pressed and installed
without firing. The steel-clad refractories are encased in a metal sheath at the time
of pressing. During the heating after installation, iron in the sheath oxidizes and
interacts with the refractory and forms a bond between the individual bricks. The
common practice with clay refractories is to incorporate calcined clay or grog to
increase the density and to reduce the shrinkage on firing.
A high melting point, far above the intended process temperature, is a normal
requirement for a crucible refractory. Then, good high-temperature strength, low
coefficient of thermal expansion and good thermal shock resistance are all proper-
ties that are valuable in refractories. A low propensity to chemical reaction is very
important. This property is considered not just in the context of interaction with
the melt but also as regards convenience in storage and handling. For example,
the tendency of the magnesium oxide or calcia to hydrate is an issue to be circum-
vented. The possibility and also rate of corrosion of refractories by molten slags
and metallic melts is, of course, a key consideration. When the crucible refrac-
tory is porous, corrosion-dissolution often follows penetration into pores and in
all similar pathways including interconnected networks of voids. Both physical
penetration and chemical invasion are favoured by good melt/slag–solid wetting
and by low melt/slag viscosity. Corrosion may start with the dissolution of either
the matrix and/or the aggregates. The solid phase may just dissolve into the liquid
or may undergo a chemical reaction and cease to exist. Besides, process gas such
80 Metal–Crucible Interactions

as CO, SO2, Cl2, CH4, H2O, and volatile oxides and salts of metals may permeate
and take part in the reactions.
Generally, refractories have the ability, within certain limits, to withstand altera-
tion by penetration, contamination and/or reaction in service and still function as
reliable engineering structures. Many factors control this interaction. Some of these,
like porosity and wettability, are manageable to a certain extent. Reference to the
thermodynamic data and phase diagrams gives indications as to which combinations
of slag and refractory will react. Usually, tests are carried out to validate the indica-
tion. Some of the popular tests were mentioned in Chapter 2.

3.4 SILICA-ALUMINA GROUP
Silica (SiO2) is a major ingredient in many refractories (Fruehan 1998). The raw
materials used in the manufacture of conventional silica refractories contain 99%
or more SiO2 and low levels of impurities, particularly alumina and alkalis. Natural
silica is the mineral quartz, which is the primary constituent of sand, sandstone and
quartzite. Depending on the thermal history, silica refractories may contain propor-
tions of quartz, cristobalite and tridymite, which are the different crystalline forms
of silica (Figure 3.1).
Fused silica is produced by the actual fusion of very high-grade silica sands
in electric arc, electrical resistance or other furnaces. Crystalline raw material is

FIGURE 3.1 Phase diagram of silica.

Crucible Materials 81

converted into an amorphous glass known as fused silica. The key characteristics of
fused silica are very low thermal expansion, low thermal conductivity, high purity
and excellent resistance to thermal shock (Fruehan 1998).
The raw materials for silica brick lack a natural bond. Lime is the universally used
bonding additive. With additions of CaO to SiO2, the melting temperature remains
unchanged up to about 28% CaO. Silica brick has good tolerance for FeO. But, a
small amount (~1%) of Al2O3 can ruin this. Silica refractories account for about
one-sixth of total refractory production. They are made by firing crushed and ground
quartzite (ganister) with about 2% added lime to assist in bonding (Routschka and
Granitzki 2005). The outstanding characteristic of silica is its load-bearing ability
at high temperatures. It was therefore used in a very wide sprung-arch roof over an
open hearth.

3.4.1 Fireclays
In nature, fireclay is found as a normal clay, whiter and lighter in colour but with
higher alumina content than normal mud. Fireclays contain the aluminosilicate
mineral kaolinite. Fireclay refractories are made from clays containing kaolinite
(Al2[Si2O5][OH]4) with 15–45% alumina and 50–80% silica. Alkalis (~3.5%) and
iron oxides (< 3%) are the usual impurities. The impurities largely responsible for
lowering the refractoriness of fireclay refractories are CaO, MgO, FeO, Na2O and
K2O. The temperature at which the fireclay is usable (max~1600°C) depends upon
the alumina-to-silica ratio and also the impurity content. Fireclays are classified as
low-duty, medium-duty, high-duty and super-duty to indicate the intended tempera-
ture of use. Firebricks (fireclay bricks) have a relatively low thermal expansion coef-
ficient and thus, moderate thermal shock resistance. They are fairly inert in acidic
environments and become reactive in basic environments. A deleterious fluxing
agent for fireclay refractories is MnO, a relation similar to that of FeO with the
Al2O3–SiO2 system.

3.4.2 Aluminosilicates
Fireclay is the raw material from which the bulk (about 70%) of aluminosilicate
refractories are manufactured. High-alumina refractories are made from alumina-
rich bauxitic kaolins or bauxite clay combinations. These materials are mined,
blended and fired in rotary kilns to yield calcined aluminosilicates from about 50%
to over 70% alumina content (Fruehan 1998).
The refractoriness of alumina–silica refractories will be seriously affected by
very small (<1%) amounts of Na2O or K2O (Fruehan1998). Figure 3.2 shows the
Al2O3–SiO2 phase diagram. Impurities generally present in most commercial refrac-
tories in this system can have a pronounced effect on the usable service temperatures
vis-à-vis those indicated in the phase diagram. In the FeO–Al2O3–SiO2 ternary, the
liquid can form even below 1095°C. Iron oxide-bearing liquids are characteristically
very fluid and have high degradation potential.
82 Metal–Crucible Interactions

FIGURE 3.2 Alumina–silica phase diagram.

3.4.3 Mullite
Mullite rarely occurs in nature. It is produced commercially by firing various alumi-
nosilicates. Andalusite is considered the best precursor. The general composition is
3Al2O3.2SiO2. It has a good combination of properties for a refractory: high thermal
stability, low thermal expansion, high bending strength and creep resistance, and
excellent resistance to thermal shock. The properties of aluminosilicate refractories
are largely determined by the presence of two crystalline phases, mullite and corun-
dum, and the siliceous glassy phase. In the absence of glassy phase formation in the
grain boundary, mullite can retain >90% of its room-temperature strength at more
than 1600°C.
High-alumina bricks are produced using natural raw materials, high-grade (>45%
Al2O3) clay, bauxite and the sillimanite group of minerals, as well as synthetic mate-
rials (fused corundum, mullite, calcined alumina and sintered corundum). Bauxite is
primarily used for the manufacture of 76–85 wt% alumina-containing high-alumina
refractories. To develop the mullite matrix (to improve thermal shock resistance and
other high-temperature properties), materials from the sillimanite group with cal-
cined alumina are added. In aluminium melting furnaces, bauxite bricks are used in
the chemically bonded form.
Crucible Materials 83

3.4.4 Alumina
Aluminium oxide (Al2O3) occurs naturally as the mineral corundum. Alumina
powder, used to manufacture polycrystalline Al2O3 ceramics, is produced from the
mineral bauxite by the Bayer process. Bauxite is primarily colloidal aluminium
hydroxide intimately mixed with iron hydroxide and other impurities.
Many types of chemically and thermally processed alumina are used in refrac-
tories. These are calcined, tabular and fused alumina, all obtained from the alu-
mina made by the Bayer process. Calcined alumina is used to promote refractory
binding during manufacture or use, whereas tabular or fused products form very
stable aggregates. Tabular alumina is formed by calcination at 1925°C, whereas the
fused alumina, which is denser, is made by total melting and rapid solidification
(Fruehan1998).
Tabular alumina has excellent high-temperature properties and thermal shock
resistance. Its name is derived from the large (typically 100 to >400 μm), flat, table-
like α-alumina crystals in the microstructure. Tabular alumina is different from
white fused alumina. The presence of large numbers of closed pores and controlled
pore size makes tabular alumina superior in thermal shock resistance to white fused
alumina and other aluminosilicates.

3.5 MAGNESIA–LIME GROUP
The magnesia or magnesia–lime group encompasses all refractories made from
dolomite and synthetic magnesites. These are the principal refractories used for basic
steelmaking. Magnesia is noted for its compatibility with iron oxides and therefore
forms the base for all types of basic refractories, including those made from magne-
site, olivine, dead-burned dolomite, and magnesite and chrome ore. Calcia is more
reactive with iron oxide, forming low-melting calcium ferrites such as dicalcium
ferrite (2CaO·Fe2O3), which melts incongruently at~1440°C. Calcia is more prone to
hydration and disintegration.

3.5.1 Magnesia
Periclase (MgO) is an extremely rare mineral and does not occur in commer-
cial quantities. About three -fourths of the world’s magnesia production comes
from magnesium-rich minerals. Magnesite (MgCO3) is the most important, fol-
lowed by dolomite (CaMg (CO3)2). Deep well brines and sea water have also
been important sources of magnesia (Canterford 1985). Magnesium hydroxide,
Mg(OH)2 , is precipitated from brines (deep well type) and seawater by reac-
tion with calcined dolomite or limestone and treated in a rotary kiln to produce
refractory-grade magnesia.
Commercial magnesia is made in three basic forms: caustic-calcined, dead-
burned and fused. Caustic-calcined magnesia has a high surface with a small (1–20
µm) grain size and is reactive. Dead-burned magnesia (grain size 30–120 µm) is
formed at temperatures above 1200°C. The calcined magnesia is briquetted and
84 Metal–Crucible Interactions

fired in shaft kilns at temperatures up to 2000°C. The product is sintered, dense

refractory-grade magnesia. By melting a refractory-grade magnesia or other magne-
sia precursor in an electric arc furnace, fused magnesia (grain size 800 to 1000 µm)
is made. The molten mass from the furnace is cooled and broken up for use in refrac-
tories (Fruehan1998). For magnesia (MgO), the maximum operating temperature is
1800°C depending on grade. Magnesia ceramics usually contain around 90% MgO
together with Al2O3–SiO2 and CaO. The magnesia spinel variety contains around
90% MgO together with Al2O3.

[Link] Magnesite Bricks
Commercial-grade MgO brick contains atleast 80 wt% of MgO (Biswas and Sarkar
2020). Magnesite has poor sinterability, which is improved by adding bauxite.
Magnesia–alumina spinel (MgAl2O4) and forsterite (Mg2SiO4) form during firing.
In unfired bricks and bricks clad in steel casing, in situ sintering occurs during heat
up, and properties similar to those of fired magnesite bricks are developed.
In magnesite raw material, the impurities present include SiO2, Al2O3, Fe2O3 and
B2O4. Their level must be kept low to preserve the refractory properties. The proper-
ties of the magnesite bricks strongly depend on the CaO/SiO2 ratio. The coefficient
of thermal expansion of pure MgO refractory is very high. Consequently, the thermal
shock resistance is poor, but it can be improved using additives.

3.5.2 Dolomite
The double carbonate dolomite (CaCO3·MgCO3) occurs in nature. It can be con-
verted to refractory dolomite (CaO·MgO) by firing. Silica, iron oxide and alumina
are the persistent impurities. Manganese, sulphur and phosphorus are also present
(Fruehan 1998). Refractory-grade dolomites are usually enriched in magnesia, con-
taining a minimum of 20% MgO and a maximum of 2.5% impurities.
Dolomite bricks are made of sintered dolomite and fused dolomite. They con-
sist of around 60% CaO and 40% MgO. In dolomite, CaO and MgO remain as a
eutectic mixture of solid solutions, as shown in the phase diagram (Figure 3.3).
MgO grains are embedded in a fine matrix of CaO, and the matrix has a broad
composition range. Hence, oxides such as MgO, TiO2 and ZrO2can be readily
incorporated in the matrix. Fired direct-bonded dolomite bricks are made by
mixing graded dolomite with organic binders, compacting and firing at high
temperature. To improve the corrosion and hydration resistance, the fired bricks
are impregnated with tar or pitch.
Doloma brick shows a high degree of direct bonding. Direct-bonded doloma
bricks are susceptible to thermal shock, which is significantly improved through the
addition of small amounts of zirconia. Improvements to the slag resistance of both
direct-bonded and carbon-bonded doloma bricks have been made through the addi-
tion of high-purity magnesia. Tar-bonded dolomite has been in use for some time.
Thermosetting phenol–formaldehyde resin is presently used instead of tar/pitch. A
chemically bonded dolomite brick with 3–4% residual carbon is less wetted by slag
and has better thermal shock resistance.
Crucible Materials 85

FIGURE 3.3 Calcia–magnesia phase diagram.

3.5.3 Magnesia–Chrome
Mag–chrome or MgO–Cr2O3 refractories find predominant use in copper smelt-
ing, converting and refining furnaces (Schlesinger 1996). Widespread use of mag–
chrome began in the 1950s and 1960s and continues to date. Chromite ore contains
CrO and MgO and also Al2O3, Fe2O3 and SiO2 in significant amounts. The first mag–
chrome refractories produced, and which still remain in use, are the silicate-based
(“burned”) materials. They are obtained when the chromite ore is fired at tempera-
tures <1500°C with or without added magnesia. Their microstructure shows grains
of a spinel-based structure, (Mg,Fe)(Al,Fe,Cr)2O4, surrounded by a rim of fosterite
Mg2SiO4 (m.p. ~1900°C) and silicates. The silicate phases are low melting (~1200°C
for fayalite), and the usefulness of such refractories is thus limited.
Direct-bonded refractories were first produced in the 1950s and possess superior
high-temperature properties. Firing mixtures of chrome ore and magnesite at 1600–
1700°C results in the formation of a liquid silicate phase. On cooling, precipita-
tion of secondary spinel from the silicate phase facilitates bonding by interdiffusion
between the spinel grains themselves. To further enhance the properties of direct
bonded refractories beneficiated of chrome ores, that have lower silica content are
used as start materials
Fused cast mag–chrome refractories are produced by melting the entire mix-
ture in an electric furnace at about 2450°C and casting the melt. The fused grain
86 Metal–Crucible Interactions

materials still contain some silicate phase as minor discrete particles in contrast to
the continuous phase in the burned and direct-bonded bricks. Mag–chrome refrac-
tories produced from purer materials perform better. A higher calcia/silica ratio in
the mag–chrome refractory promotes the presence of more refractory silicate com-
pounds, and the refractory has higher temperature capability. A higher Cr2O3 content
relative to the combined amount of Al2O3 and Fe2O3has been associated with greater
resistance to silicate slag attack.
Magnesia and chrome ore synergism results in excellent properties in the refrac-
tory. One constituent tends to minimize the major weaknesses of the other constitu-
ent. The refractory is called chrome–magnesite or magnesite–chrome, denoting that
the first named is the dominant constituent. The properties of the bricks are a func-
tion of the magnesite-to-chrome ratio, among other things.
Rebonded magnesia–chrome brick is also made with synthetic fused magnesia,
shaped under high pressure and fired at very high temperature. The brick structure
then consists of large, fused aggregates embedded in a matrix of finer fused grains.
Another method was first to electrically fuse the chrome ore and the magnesia grains
at about 2450°C, pour the melt into moulds, cool, crush, add binder, press and fire at
1550°C to manufacture the bricks. Electrofused magnesio-chromite is a mixture of
chromite crystals inside a slag glass mainly composed of MgO (Bravo et al. 2017).
Chrome–alumina is another refractory, mainly used in induction crucibles. The typi-
cal composition of the ceramic is Al2O392.8%–SiO22.6%–Cr 4.0%+trace oxides.

[Link] MgO–Al2O3
The MgO–Al2O3 phase diagram is similar to that of MgO–Cr2O3 (Schlesinger 1996).
Refractories based on mixtures of MgO and Al2O3 have been considered for replace-
ment of mag–chrome. Periclase and alumina rarely occur combined in nature, and
MgO–Al2O3 refractories are made from purified raw materials. Castable and mono-
lithic magnesia–alumina materials are produced using a binder (calcium aluminate
cement) or by direct bonding. Fused cast materials are also made.

3.5.4 Magnesia–Carbon
The incorporation of carbon in magnesia refractories results in much improved per-
formance in terms of higher thermal conductivity, good thermal shock resistance
and resistance to slag penetration, and better high-temperature strength and refracto-
riness. Although the magnesia–carbon refractories are chemically quite simple, their
properties may be extensively changed by varying the carbon level (or C/MgO ratio),
type of carbon, type of magnesia and type of bond (Mikami and Martinet 1979;
Canterford 1985). Carbon is normally introduced by tar-bonding, pitch impregnation
and graphite additions.
The magnesia–carbon bricks are further grouped as tar-impregnated fired mag-
nesia bricks (2–4% C), carbon-containing impregnated magnesia bricks (7% C), and
resin-bonded MgO–C bricks (20–25%). Sintered and fused magnesia with larger
crystal size is used as the source of MgO. Graphite is normally the source of carbon;
other sources of carbon are carbon black, hard pitch and powder resins. Other things
Crucible Materials 87

being equal, purer MgO and graphite have improved physico-chemical properties
and corrosion resistance at high operating temperature. Antioxidants, such as Si,
Al, metal powders and B4C, are also added to the mixture to improve oxidation
resistance and high-temperature strength. The functioning of MgO–C refractories is
determined by the direct oxidation of carbon (and graphite) by air; the reduction of
magnesia (and other oxides) by carbon (graphite); and the role of antioxidants. These
are covered later in this chapter in the section on corrosion.
Magnesia–alumina–graphite refractories are Al2O3 – bricks that have MgO as
additive (Biswas and Sarkar 2020). When used in ladle operating conditions, the
spinel MgAl2O4 forms with an volume expansion of (thermal). This expansion
closes the joints of the bricks and ends up reducing metal penetration. With higher
magnesia content, the corrosion resistance is improved due to the formation of
spinel, but the larger thermal expansion here may end up cracking the bricks.
Al2O3 –MgO–C (AMC) emerged as an alternative working lining refractory to
MgO–C in the 1980s. The performance of AMC bricks depends on the alumina
aggregates used. In the ladle bottom, high-quality AMC bricks made of tabular
alumina are used.

3.6 CARBON
Carbon is the only elemental refractory that is also a non-metal. It is unmeltable and
is not wetted by metal melts, slags or salts under oxygen-free conditions. It dissolves
in pig iron melt and to a lesser but significant extent in certain reactive and refractory
metals. It has high thermal conductivity and low thermal expansion coefficient, and
this combination imparts to it high thermal shock resistance. The major limitation of
graphite is its susceptibility to attack by oxygen, steam and CO2 above 400°C.
Much of the graphite used in industry is synthesized. The raw materials (low ash
cokes, anthracite, and natural and artificial graphite) are blended together with tar
or pitch in heated mixers, and the hot mix is compacted by ramming or vibration.
The cold, hardened shapes are first fired at temperatures >400°C in the absence of
air and are eventually converted to graphite by self resistance heating at ca. 3000°C.
Depending on the quality and properties of the various raw materials used, graph-
ite with widely varying properties (thermal conductivity and oxidation, erosion and
alkali resistance) may be obtained.
In metallurgical processing, the principal use of graphite is as large carbon blocks
in the bottom and bosh of a blast furnace (iron), or as electrode cum lining for alu-
minium electrolysis cells (Phelps and Wachtman 2011).
Generally, graphite is used in refractories to reduce the wetting of the refractory
by the melt and/or slag and to increase the thermal conductivity, which will result
in better thermal shock resistance of the refractory. In oxide–carbon refractories,
the range of carbon content used is from 4% to 35%. With the increase in graphitic
content, the thermal conductivity of the refractory increases, but there is a decrease
in the density of the refractory, not only because graphite is much lighter but also
because the graphite flakes are not very conducive to close packing, unlike the gran-
ules (Fruehan 1998).
88 Metal–Crucible Interactions

Refractory bricks made of carbon can be classified into three groups: amorphous,
semi-graphite and graphite blocks. In recent years, two other superior-quality carbon
blocks have become available: micropore and super micropore. The average pore
diameter in these varieties is below 1 μm, and the thermal conductivity is also high.

3.7 BERYLLIA
Chemically, BeO is one of the most stable materials. In many of its properties – ther-
mal expansion, and electrical and mechanical properties – BeO is similar to alumina.
Thermal conductivity is relatively high, and for a material of good strength, this
gives better thermal shock resistance than Al2O3. Beryllia crucibles can be excellent
containers for liquid metals. One limitation is that it reacts with water vapour at high
temperatures to form a stable hydroxide species, probably gaseous Be(OH)2, which
causes it to evaporate significantly in a moist atmosphere (Brook et al. 1991).
Beryllia powder is commercially available with purities over 99% BeO. It can
be readily fabricated to high-density (<5% porosity) shapes by normal ceramics
fabrication methods such as pressing, extrusion, injection moulding or slip casting.
Sintering temperatures are in the range 1600–1800°C. Hot pressing in graphite dies
and hot isostatic pressing may also be used. Hot pressing, as well as slip casting and
sintering, has been used to achieve near theoretical density (<1% porosity) with the
purest powder.
Beryllia is toxic. This has led to increased costs of production and inhibited use.
Inhalation of fine beryllia dust results in respiratory disease in an acute form from
short, concentrated exposures or in a chronic form from long, continued exposure.
Special precautions are required when handling fine powder.

3.8 ZIRCON AND ZIRCONIA

Zircon, or zirconium silicate (ZrO2.SiO2), is a naturally occurring mineral having
excellent refractoriness (Fruehan1998). The main raw material for zircon brick is
zircon sand of 100–300 μm particle size. Zircon bricks have good wear resistance,
less wettability by molten metal and slag, and hence, good corrosion resistance.
Alumina is added to zircon to form zircon–mullite compositions, which are more
resistant to thermal shock and corrosion than plain zircon bricks. In steel making,
zircon refractory is used in the tundish nozzle due to its non-wettability by liquid
steel and good resistance to erosion and spalling.
Zirconia has a combination of high melting point (2715°C) and potential for high
strength with a maximum operating temperature of 2400°C depending on grade
(Fruehan 1998). Zirconium oxide (ZrO2) occurs naturally as the mineral baddeleyite,
commercially used zirconia is generally obtained by processing zircon. Combined
with high melting point, the superior resistance to corrosion and erosion makes zir-
conia an attractive refractory (Bullock et al. 1989). Zirconia is polymorphic. The
stable room-temperature phase is monoclinic. Around 1170°C, it transforms to the
tetragonal, which is stable up to 2370°C, when it becomes cubic. The tetragonal/
monoclinic transformation is associated with a large volume change (~4%). This is
Crucible Materials 89

FIGURE 3.4 Thermal expansion behaviour of zirconia (From Routschka, G. (Ed.). 2001).

the reason why pure zirconia wares break as their temperature changes through this
value. To counteract the deleterious effect of phase changes, zirconia may be con-
verted fully to the cubic phase by incorporating small amounts of calcia, magnesia
or yttria. The cubic phase then remains stable at all temperatures. Stability, thermal
shock resistance and hot load properties are all enhanced in the stabilized zirconia
(Fruehan 1998).
Fully stabilized zirconia refractories have very low thermal shock resistance
because of their high thermal expansion coefficient (Figure 3.4) and low thermal
conductivity. In many applications, however, good or even excellent thermal shock
resistance is required. With zirconia, this is achieved by optimizing the level of stabi-
lization (Johns and King 1970). Crucibles for induction melting of refractory metals
are made with zirconia compositions having a cubic zirconia content of about 40%.
In tundish nozzles in the casting of molten steel, a cubic zirconia content of 65%
is considered suitable. In the arrangement for partial stabilization, cubic materials
contain fine dispersions of tetragonal phase. Crack propagation is hindered when the
metastable precipitate transforms, resulting in significant improvements in tough-
ness and strength. Some toughening in zirconias can also result from the presence
of microcracks. This flexibility, combined with the capability of microstructural
90 Metal–Crucible Interactions

control through heat treatment, is the reason for the popularity of zirconia, as it can
be engineered to fit a wide variety of special refractory applications.
Zirconia is used in the natural, stabilized or fused state, the last often as a mixture
with alumina, silica or other compatible oxides (Fruehan 1998). The manufacture of
zirconia products starts with stabilized or partially stabilized zirconia raw materials,
which are ground and mixed, and the graded material is mixed with organic binder,
shaped in a press, dried and fired at >1700°C. Zircon with alumina can be processed
to form zirconia–mullite grains. This material has high thermal shock resistance
as well as good corrosion resistance. The well-known use of this material, with the
addition of carbon, is in manufacturing alumina–zirconia–carbon slide gate plate
refractory in steel ladles.

3.9 YTTRIA
The rare earth oxide yttria is one of the most stable refractory ceramic oxides. Its
availability, in high purity, improved with the expansion of the rare earths industry,
particularly after the late 1970s.
Yttria crucibles can be made by dry pressing followed by sintering. A crucible
with a somewhat porous structure is considered conducive to good thermal shock
resistance. Such porous crucibles are usually fabricated from a mixture of ceramic
raw material powders with varied diameters ranging from coarse to fine (Tetsui et al.
2012a). Other normal ceramic fabrication methods have also been used. The main
limitations of yttria are its cost and modest thermal shock resistance.

3.10 SPINELS
Spinel refers to a group of minerals of the common structure AB2O4, where A rep-
resents a divalent cation and B a trivalent cation. The common trivalent cations are
aluminium (Al+3), ferric iron (Fe+3) and chromium (Cr+3), and the divalent cations are
magnesium (Mg+2) and ferrous iron (Fe+2).
Magnesia–alumina spinel (MgO·Al2O3) and magnesia–chrome spinel are useful
refractories. The only compound in the binary system MgO–Al2O3 is magnesium–
aluminate spinel, with a formal stoichiometric ratio of 71.8% Al2O3–28.2% MgO.
MgAl2O4 (MA) spinel has a range of stoichiometry. It has a cubic crystal structure.
Magnesium–aluminate spinel is a highly refractory material, with a melting point of
2135°C, low thermal expansion, good thermal conductivity and high thermal shock
resistance. All types of spinel tend to form substitutional solid solution swith each
other and with other oxides. In MgO·Al2O3 spinel, both magnesium and aluminium
cations can be replaced by other cations of similar size. This results in the formation
of alumina-enriched spinel as well as magnesia-enriched spinel. Alumina-enriched
spinel can absorb FeO or MnO by substitution of MgO and consequently retard the
infiltration of slag into spinel-containing refractory. The picking up of Fe2+ and Mn2+
from the slag into the spinel leads to an expansion of the lattice and effectively pre-
vents further penetration of ladle slags. When the slag penetration of alumina, spinel
and aluminaspinel castables was compared, castables with 20% spinel displayed the
Crucible Materials 91

lowest penetration and corrosion, the reason being that alumina and spinel in the
castable react with CaO, FeO and MnO in the slag, locally increasing its melting
point and viscosity. This type of complex interaction, involving the pick-up of rapidly
diffusing cations from the slag into solid phases (such as Fe and Mn into spinel) and
dissolution of solid (initially fine, reactive matrix phases) into the slag, altering its
local composition, has also been highlighted by Goto et al. (1997) and Korgul et al.
(1997). The liquid product of this interaction is the phase in contact with the solid
phases of the refractory and determines any penetration and corrosion.
MgAl2O3 spinel can be introduced into a refractory composition by forming it in
situ or by the addition of pre-formed spinel. Spinel is formed in situ by adding mag-
nesia to high-alumina refractory or adding alumina to magnesia-enriched refractory.
During the firing of bricks or during preheating, MgO reacts with Al2O3 to form
the spinel. Spinel formation is accompanied by volume expansion. Pre-formed and
in situ spinel-containing refractory are very similar with respect to chemical com-
position and bulk density of the product, but they do show distinct differences in
other physical properties, such as thermal expansion, hot strength and slag corrosion
resistance.
Pre-reacted spinel does not present an expansion problem. This can be added in
larger quantity to high-alumina refractory. Expansion during in situ spinel formation
can result in cracking of the refractory. However, in situ spinel formation in refrac-
tory bricks and castables is advantageous for high-temperature strength and thermal
shock resistance (Biswas and Sarkar 2020). Hence, addition of MgO to high-alumina
refractory to form in situ spinel is restricted to about 3–5% MgO to form 12 to 18%
spinel, and pre-formed spinel is added as necessary to reach the desired spinel con-
tent for high corrosion resistance. Refractories containing MA spinel are potential
replacements for Cr-containing refractory linings in steelmaking (Lee and Zhang
2004) on account of their excellent thermal shock resistance and refractoriness
As an alternative to other carbon-free refractories, alumina spinel fired bricks
show superior performance in the highly corrosive environment in secondary steel
making (Biswas and Sarkar 2020). The alumina spinel bricks are made using high-
purity synthetic alumina aggregates, white tabular alumina and white fused alumina
and MgO–Al2O3 spinel, MgO and calcined alumina in finer fractions. High-alumina
bricks with spinel addition also have the advantages of lower thermal conductivity
and low thermal expansion coefficient compared with MgO–C or Al2O3–MgO–C
(AMC). The spinel-containing alumina being carbon free, dissolution of carbon in
ultra-low-carbon steel is avoided. This type of brick is widely used in producing
clean steel. Pre-reacted spinels have good thermal shock behaviour. This is mainly
due to differences in thermal expansion between alumina and spinel. The differential
expansion leads to microcracks in the matrix that act as crack arresters.

3.11 CARBON-CONTAINING REFRACTORIES
Carbon-containing refractories are widely used in the steel industry. Carbon is used
in two forms: pyrolytic carbon derived from pitch or resin binders, which ends up
in the matrix in fine and reactive form, or graphite flakes, which may be considered
92 Metal–Crucible Interactions

as part of the aggregates. Carbon in both forms is used with many refractory oxides
such as Al2O3, SiO2, ZrO2 and MgO. A typical unfired microstructure contains large
sintered and fused MgO grains bonded to graphite flakes by a nanoscale carbon
bond. MgO–C refractories are used in basic oxygen steel-making (BOS) and electric
arc furnace (EAF) steel-making furnace linings.
While carbon confers many of the good properties on oxide–C refractories, it is
also prone to oxidation. The weak nature of the C bond between the oxide and graph-
ite lowers the strength of the refractory. To inhibit oxidation and provide stronger
(ceramic) matrix bonding at high temperatures, metallic or carbide powder is added.
Several mechanisms for the good slag resistance of C-bonded doloma and MgO
refractories have been proposed (summarized by Lee and Rainforth 1994). These
include carbon’s non-wettability, so that liquid penetration is limited; reduction of
MgO by C to form magnesium vapour, which transports, re-oxidizes and deposits
as a dense layer of secondary MgO at the slag–brick interface, physically preventing
penetration; reduction of iron oxides by C to Fe metal, blunting their ability to attack
the oxides and form low-melting phases; and formation of CO vapour by oxidation
of C and/or Mg vapour (from reduction of MgO), which provides an overpressure
resisting slag/metal ingress. Which one of these mechanisms is the most significant
remains an unresolved issue (Lee and Zhang 2004).
The slag corrosion of oxide–C refractories follows the sequence: (i) formation of a
decarburized layer due to oxidation of graphite/pyrolytic C by oxygen in the furnace
atmosphere or FeO in the slag, (ii) infiltration of slag into the decarburized layer and
erosion of the oxide grain by slag penetration, aided by softening of any intragranu-
lar silicate bond, and (iii) reaction and dissolution of the oxide grain into slag and
molten steel at the prevailing high temperatures (Lee and Zhang 1999).
The protective role of a secondary MgO layer formed at the hot face of MgO–C
bricks has often been highlighted (Lee and Zhang 2004). In both dolomite and MgO
refractories, this arises from reduction of MgO by C to Mg vapour, which is trans-
ported to the hot face and re-oxidized to a fine-grained, dense MgO layer on the
hot face. This layer prevents slag/metal ingress into the refractory. This dense layer
does not always form. Specific conditions both inside and outside the refractory are
required to maintain it; a high CaO/SiO2 ratio in the slag is required, which main-
tains the delicate balance between dense layer dissolution and formation by limiting
the former. To improve the corrosion resistance of the refractory, MgO purity should
be higher, grain size larger and residual impurities SiO2 and B2O3 lower. Carbon in
MgO–C refractories is present as graphite or pyrolysed carbon from pitch and resin.
The advantage of one form over the other is yet to be established.
Two types of additives mainly used in oxide–C refractories are metals and alloys
(such as Al, Si, Mg and Al–Mg) and boron-based compounds (such as B, B4C, CaB6,
ZrB2, Mg3B2O6 and SiB6). These two types work in different ways. Metal/alloy
additives act by inhibiting carbon oxidation and forming high-temperature ceramic
bonds, thus improving hot strength. Metal additives facilitate dense layer formation
both at the slag–refractory interface and locally inside the brick, leading to early
protection in service (Lee and Zhang 2004). Slag composition also has a significant
effect on the corrosion resistance of MgO–C refractories. In general, corrosion rate
Crucible Materials 93

increases with increasing Cr2O3, iron oxide and MnO but decreases with increasing
MgO or basicity in the slag.
All said, Lee and Zhang (2004) identified two types of attack on refractories by
silicate liquids: direct, wherein atoms of the solid dissolve directly into the liquid,
and indirect, in which a boundary layer forms between the original solid and liquid.
Direct attack is usually undesirable, while the indirect mode may be beneficial if
the boundary layer turns protective. As often highlighted, the influence of the local
liquid at the point of penetration into the refractory has a critical effect on the attack.
Its composition is influenced by the bulk liquid and/or the refractory. Rapidly dif-
fusing species in the bulk liquid, such as Mn and Fe, migrate to the local liquid.
The refractory may take up species from the local liquid or release its own, thereby
altering the local composition and hence, viscosity. The phases in the refractory
may also dissolve in the local liquid, altering its composition compared with bulk
liquid. Slag attack can be controlled through the refractory, whose composition is
arranged to make the local liquid viscous by uptake or release of suitable species, or
by development of a tight texture and/or dense layers at the liquid/refractory inter-
face. Alternatively, the slag composition can be controlled to minimize attack on
the refractory by e.g. saturating it in the main refractory phases. In practice, passive,
protective coatings can be generated by alternative process techniques such as slag
splashing in BOF vessels.
The function of the slag in metal winning operations is to provide a vehicle for
impurity removal initially from the ores and later from the as-smelted molten metal.
Normally, a slag attacks the refractory. But, the attack is managed in many ways.
A novel new use of slag is as an in situ refractory. This is generally similar to the
formation and functioning of a frozen skull in water-cooled crucibles, common in
the handling of very reactive metals. In situ refractory is routinely created in the slag
splashing operations in BOS vessels. Lee and Moore (1998) outlined four types of in
situ refractories: (i) those arising due to reactions solely within the components of the
brick or monolith without any external contribution, e.g. formation of a MgO layer
initiated by reaction between carbon and magnesia in the refractory; (ii) those in
which reactions occur within the refractory but may be assisted by reaction with the
(liquid or vapour) furnace contents, e.g. magnetite formation in the upper regions of
flash smelting furnaces; (iii) refractories that react with the furnace contents, gener-
ating a protective interlayer between the refractory and furnace contents. Often, the
interlayer forms as a result of indirect corrosion of the refractory solid by penetrat-
ing liquid, e.g. the formation of a corundum layer on aluminosilicate refractories in
aluminium melting and holding furnaces; and (iv) those in which the slag from an
earlier batch is splashed or otherwise deposited on the interior of the furnace lining.
The next batch process fluids would have this slag deposit as the contact zone.

3.12 BARIUM ZIRCONATE
For the use of a ceramic as a crucible material, it is essential that the compound
shows no crystallographic phase transitions in the temperature region of interest.
This is in fact the case for BaZrO3. Barium zirconate has a cubic perovskite structure
94 Metal–Crucible Interactions

and a high density of 6.242 g/cm3 (Erb et al. 1995). It has a melting point of about
2600°C, a smaller thermal expansion coefficient than ZrO2 and therefore, better
thermal shock resistance.
In a binary ceramic such as barium zirconate, any deviation from the stoichio-
metric composition will lead to precipitation of the off-stoichiometric surplus at the
grain boundaries of the ceramic. This region will be prone to corrosion.
BaZrO3 powder is synthesized by heating BaCO3 and ZrO2 (purity >99%) in air
at 1200°C and fabricated by cold isostatic pressing (CIP). On CIP of the pre-reacted
powders (0.5–1 µm) at about 2000 bar, followed by sintering in air at about 1700°C
for 48 h, ceramic shapes with up to 98.5% theoretical density (TD) is obtained. A
Ca-doped BaZrO3 crucible could be made starting from BaCO3, ZrO2 and CaO in the
required mole ratios. The Ca-doped BaZrO3 refractory consists of Ba1−xCaxZrO3 and
CaO phases. Similarly, the yttria-containing variety is also obtained. However, doping
with Y2O3 decreases the sinterability of the BaZrO3 crucible (Chen et al. 2017).

3.13 CALCIUM ZIRCONATE
Calcium zirconate (CaZrO3) is a stoichiometric phase with a perovskite crystal struc-
ture in the pseudo binary CaO−ZrO2 system. It has a high melting point of 2368°C
and a chemical stability as good as CaO (Schaffoner et al. 2018).
The zirconate, unlike CaO, is inert to hydration. CaZrO3 can be prepared by the
solid-state reaction (between 700 and 1300°C) of equimolar CaO–ZrO2 mixtures
with subsequent CIP and sintering (Li et al. 2010) at 1750°C for 3 hours. The crucible
was used to melt NiTi and Ti6Al4V alloys at 1500 and 1750°C, respectively, without
contamination.
Coarse-grain CaZrO3 has also been prepared by slip casting of CaCO3 and
ZrO2, firing at 1500°C for 3 h and crushing. The product had 12.4% open porosity.
Schaffoner et al. (2013) used both solid-state synthesis and electric arc melting meth-
ods to prepare coarse-grained calcium zirconate. The fused calcium zirconate had
considerably lower porosity than solid-state synthesized material, but during electric
arc melting, the CaO/ZrO2 ratio decreased from 1 in the start material to less than 1,
resulting in the formation of cubic ZrO2 as a separate phase. After crushing and mill-
ing of fused CaZrO3, the preparation of large crucibles by CIP resulted in crucibles
with porosity of 15.9–16.5%.
Schaffoner et al. (2018) also prepared CaZrO3 refractory concrete (castable)
with calcium aluminate cement as a binder in the form of medium-sized crucibles.
Although the raw materials contained only a small amount of silica in bulk, a large
amount was accumulated in the calcium aluminate matrix, suggesting that the matrix
acted as a sink for the silicon oxide impurities of all raw materials.
Calcium aluminate binders were found to enable fast processing and sufficient
refractoriness for medium-sized melting crucibles. The properties of uniaxially
pressed and cast crucibles of CaZrO3 were generally similar. However, the matrix
containing calcium aluminates undergoes pronounced sintering shrinkage, resulting
in smaller apparent porosity and a lower refractoriness compared with uniaxially
pressed refractories.
Crucible Materials 95

The calcium aluminate cement consisted mainly of CaAl2O4, while it also con-
tained minor amounts of CaAl4O7 (Braulio et al. 2011). As in other components,
impurities should be avoided in calcium aluminate, since they impair not only the
refractoriness but also the corrosion resistance.

3.14 SILICON CARBIDE
Special refractories include many types of materials, and silicon carbide is one of
them. Silicon carbide, SiC, has good thermal and electrical conductivity, good ther-
mal shock resistance and strength at high temperatures. SiC has been found occur-
ring naturally only as small green hexagonal plates in meteoric iron. This same
hexagonal polymorph (α-SiC) has been synthesized commercially in large quantities
by the Acheson process (Richerson and Lee 2018). This simple process simultane-
ously produces lower-grade SiC for abrasives and high-grade SiC for electrical appli-
cations. The Acheson process consists of passing electric current through a large,
elongated mound of SiO2 sand mixture, thus heating it to about 2200°C. After cool-
ing, the mound is broken up and sorted to get various grades of SiC.
SiC can be prepared from almost any source of silicon and carbon. Silicon car-
bide bricks find increasing use in the blast furnace lining (in the shaft). They resist
corrosion by acid slags, alkali vapour, zinc vapour and oxidizing gases. The first
silicon carbide refractories were bonded with clay, and the refractory properties of
the bond placed the limit on the material performance. Self-bonded silicon carbide
has removed this limitation. The bonding phase can be SiC itself or silicon nitride
or sialon. Silicon carbide crucibles are used extensively for gold and silver melting.
Crucibles are of typical composition SiC 67.9%–Al2O3 23.4%–SiO2 8.3% with some
oxides in trace levels. They have excellent thermal shock resistance and can be used
up to 1500°C.

3.15 BORIDES
Diborides, especially TiB2 and ZrB2, are chemically stable in the environment of
a Hall–Héroult cell for aluminium production. They are not wetted by molten alu-
minium, and their electrical conductivity is high (Brook et al. 1991). In aluminium
electrolysis cells, replacing the conventional graphite cathode with diborides has
often been seriously considered.
Diborides can be used as crucibles for molten metals such as aluminium, copper,
silver, gold, zinc, cadmium, magnesium, germanium, lead, bismuth and chromium.
Many tonnes of TiB2 powder are consumed for the production of vacuum metalliz-
ing boats (TiB2, AlN, BN composite). They are resistance heated to about 1427°C
and used for the continuous evaporation of aluminium to produce thin metal films.

3.16 BORON NITRIDE
Boron nitride in its hexagonal, soft modification finds use up to very high (~2000°C)
temperatures. It is an electrical insulator. However, boron nitride may be blended
96 Metal–Crucible Interactions

with electrically conducting materials, such as titanium nitride or titanium diboride,

in order to modify high-temperature electrical conductance and make self resistance-
heated small boats and crucibles for the evaporation of aluminium. Billets of boron
nitride doped with calcium borate and hot pressed are readily machinable into metal
melting crucibles for use at temperatures up to 1500°C under vacuum or 1800°C
under nitrogen (Brook et al. 1991). The machining properties of boron nitride, like
those of graphite, are considered excellent.
Boron nitride is usable to 1250°C in air, provided the surface film of B2O3 is not
degraded by contact with other oxides. Boron nitride has only moderate strength
because of its low and anisotropic moduli of elasticity. Incidentally, its flexibility
is helpful in resisting thermal stress and shock. It has quite good thermal conduc-
tivity, which does not change much with temperature and is much less affected by
temperature than many other ceramics. Crucibles of pyrolytic boron nitride made
by chemical vapour deposition have been useful for high-temperature processing of
semiconductor materials. High pressures have been used to prepare cubic BN with
higher density and strength (Bullock et al. 1989).
Boron nitride, like graphite, undergoes crystal structure orientation during hot
pressing. When isostatically pressed, the solid dense material is anisotropic. Boron
nitride is inert and not wetted by a wide range of molten metals. It has low ther-
mal expansion, high thermal conductivity and excellent thermal shock resistance.
Its maximum operating temperature is 1600°C depending on grade (Wilson 2000).

3.17 SILICON NITRIDE AND SIALON

Silicon nitride (Si3N4) ceramic is stable only under conditions of moderate partial
pressure of nitrogen and very low partial pressure of oxygen. It may have applica-
tions as a crucible (Richerson and Lee 2018).
Silicon nitride does not occur naturally. It has been synthesized. Some of the
powder available commercially has been made by the reaction of silicon powder
with nitrogen at temperatures in the range of 1250–1400°C and consists of a mixture
of α-Si3N4 and β-Si3N4 polymorphs. The resulting powder is not of high purity but
contains impurities such as Fe, Ca and Al, which were originally present in silicon,
plus impurities picked up during crushing and grinding. Higher-purity Si3N4 powder
has been made by reduction of SiO2 with carbon in a nitrogen environment and reac-
tion of SiCl4 or silanes with ammonia. Both these methods produce powder of very
fine particle size.
The ceramic form of silicon nitride was first produced in 1955 (Xiao and Ralph
2009). This type of material was formed by nitriding silicon powder compacts and
came to be known as reaction-bonded silicon nitride (RBSN). During nitriding, the
silicon nitride grows within the pre-existing pores of the powder compact, and the
original dimensions of the silicon compact remain almost unchanged (linear shrink-
age <0.1%) at the end of nitriding. A complex shape can be machined to final size
either before nitriding or after partial nitriding.
Silicon nitride has high strength and high hardness due to strong covalent bonds.
These bonds also make sintering difficult. In 1961, Deeley et al. achieved high
Crucible Materials 97

densities by hot pressing previously formed silicon nitride powder using sinter-
ing additives. Using magnesium oxide, full density was reached by hot pressing at
1850°C under 23 MPa. The strength of this material was substantially better than
that of RBSN.
Sialon ceramics are solid solutions of AlN and Al2O3in β-Si3N4. They possess
excellent mechanical properties, good corrosion and thermal shock resistance, and
low wettability by aluminium alloy melt. This material may be a potential candidate
for long-term liquid aluminium containment. Sialon ceramics also show excellent
corrosion resistance to molten steel and slags.

3.18 ALUMINIUM NITRIDE
Aluminium nitride is a white to pale yellow solid nitride of aluminium, stable to high
temperatures in inert atmospheres, and melts at 2500°C. Very pure (100%) aluminium
nitride crystals are perfectly water white in colour. In a vacuum, AlN decomposes
at~1800°C. In air, surface oxidation begins above 700°C, and bulk oxidation becomes
significant above 1370°C. The material hydrolyses slowly in water. Aluminium nitride
resists attack by most molten salts, including chlorides and cryolite. The compound
was first synthesized in 1877 and can be made by carbothermic reduction under nitro-
gen or ammonia or by direct nitridation of aluminium. It is difficult to prepare in fully
dense form, and hot pressing is used with Y2O3 or CaO, as sintering aids, to produce
a material sufficiently dense for use. At a high nitrogen pressure, AlN melts without
decomposition. This property has been used to produce AlN crucibles without bind-
ers, plasticizers or other such additives. The process of ceramic production based on
such sintering with partial melting yields AlN crucibles with a high effective density
and low porosity (less than 5%). Long and Foster (1959)have given a detailed account
of aluminium nitride preparation, fabrication and use.
A unique material with low electrical but good thermal conductivity, AlN pos-
sesses, among the ceramics, a rare combination of chemical inertness along with
properties, such as thermal shock, linear expansion and erosion resistance, that com-
pare favourably with other materials. It can be used very well as a crucible material.
Aluminium nitride is inert to aluminium, copper, tin and calcium melts and does not
introduce foreign metal impurities into the notoriously reactive TiAl-based melts
either. It is an underexplored crucible material for melting reactive metals (Long and
Foster 1959).
High-density AlN crucibles are inert to titanium alloys, yielding non-contam-
inated castings with good ingot release after solidification of the alloys inside the
crucible (Kartavykh et al. 2008, 2009). They are expensive, and large crucible fab-
rication is difficult. Fabrication of AlN into usable small-size refractory ware by
hydrostatic pressing can be done with commercially available equipment. These are
best used as coatings on shock-resistant crucibles like Al2O3 or graphite. Slurry-
based coating methods like air spraying or painting cannot be used for AlN due to
its reactivity with water. Plasma spray coating is useful.
Long and Foster (1959) prepared aluminium nitride of highest purity by the pro-
cess of striking a dc arc between two high-purity aluminium electrodes in a nitrogen
98 Metal–Crucible Interactions

atmosphere. The aluminium vaporized in the arc, and the vapour reacted with the
nitrogen to form a hard lump between the two electrodes. The nitride lumps, crushed
to –100 mesh size, were heated in a graphite crucible in 1 atmosphere argon pres-
sure at 2000–2050°C for 1.5 hours. Ball-milled aluminium nitride compacted by
CIP yields green shapes with good handling strength. They are also machinable.
These shapes, on sintering in an induction-heated furnace in a graphite crucible,
yield products with excellent strength and low porosity even in 1 to 2 hours under
argon at 1950–2050°C.
Long and Foster (1959) heated high-purity aluminium metal to 2000°C in alu-
minium nitride crucibles in both argon and carbon monoxide atmospheres without
any noticeable attack on the refractory. Gallium wets the aluminium nitride, as it
does most other surfaces. Metallic gallium contained in an aluminium nitride cru-
cible was heated in vacuum for half an hour at 1300°C with no measurable pick-up
of metallic contaminants by the gallium.

3.19 COATINGS ON CRUCIBLES
There are rather few instances of a chemically inert crucible also having good ther-
mal shock resistance. In practice, the issue of getting a material with good chemical
inertness and high thermal shock resistance is circumvented by selecting a crucible
of shock-resistant material and coating it with a layer/layers of inert material. For
example, an Al2O3 crucible coated with Y2O3 (Zhang et al. 2012, Zhang et al. 2013a)
results in a crucible with good thermal shock resistance, reasonable chemical inert-
ness to the molten titanium and low cost.
As well as being inert, the coating must be adherent and resistant to abrasion and
shock. Y2O3 is probably the most investigated coating material for induction melting
crucibles (Barbosa et al. 2003, 2005, 2007; Gomes et al. 2011, 2013; Kartavykh et
al. 2009; Kuang et al. 2000). It can be applied by the simple process of slurry paint-
ing or more sophisticated plasma spraying. The issues in coating with yttria include
poor shelf life of the slurry, poor sinterability, low adhesion to many substrates,
and if the coating is thick, cracking due to thermal shock. Additives, e.g. SiO2, are
used to improve slurry life and sinterability of the coating, but such low-melting and
less chemically inert additives compromise the main property, the chemical inert-
ness, of the coating. For titanium melting crucibles, the thickness of yttria coating
is ~200–300 µm.
Porosity is a key parameter for both the crucible body and the coating (Camel
et al. 2017; Eastman et al. 1951; Gomes et al. 2013; Tetsui et al. 2010, 2012a). High
porosity improves shock resistance and facilitates adhesion of the coating to the sub-
strate. It also leads to higher surface wettability and melt infiltration. Functionally
graded coating, where the average particle size varies along the coating depth, has
been suggested as a means of fixing the issues of inertness and thermal shock resis-
tance of the coating (Gomes et al. 2011).
Any refractory that is good enough for melting crucibles generally qualifies for
making moulds. The melt–refractory reactions are less severe in moulds compared
with melting crucibles. There are, however, other issues that need to be addressed.
Crucible Materials 99

3.20 SUMMARY
In the beginning, a crucible was used not as a separate entity but as part of the
scooped-out earth mended with layers of clay or charcoal dust. The materials used
for pottery doubled as crucibles for metallic materials, and eventually, they were
made portable. The traditional ceramics, dominated by various clay minerals, and
modern ceramics, which included a range of materials from aluminosilicates to
many oxide and non-oxide compositions, came into use one by one, very gradually
till the 18th century, faster thereafter and rapidly during the past seven decades.
These materials became increasingly sophisticated in composition, purity and struc-
ture. They possessed superior properties, a wider usage spectrum and higher reli-
ability. Carbon also entered the picture as a crucible material a few hundred years
ago and remained unrivalled in superiority for usage in non-oxidizing environments.
Carbon, in its many forms and most notably as graphite, is by itself a complete cru-
cible material. Besides, as part of other crucible material compositions, it confers a
significant enhancement in properties. Yet, the development of crucible materials,
and of crucibles, remains, to date, a work in progress. The important types of cru-
cible materials that are currently relevant for immediate and future uses have been
covered in this chapter. The term crucible refers to the material entity in contact
with the melts being processed– both metallic and slag. As the need for attaining
and processing at higher temperatures developed, the term refractory came into use
to denote high-temperature capability. Refractory ceramics are invariably used as
materials for crucibles in metals processing.
Refractory materials for crucibles are broadly grouped into siliceous, aluminous,
aluminosilicates, magnesite, dolomite, chrome ore and a miscellany of materials
known as special refractories. These refractories are used in two basic forms: pre-
shaped objects and unformed compositions or monolithics. Each of these refractory
types and forms has its niche application areas. Every crucible material is subject to
attack by the melt it is holding. Melt penetration into the crucible and reaction of the
melt with the refractory both occur. Generally, refractories have the ability to with-
stand alteration by penetration, contamination and/or reaction in service and still
function as reliable engineering structures. Many factors control this interaction, and
some of these are manageable to a certain extent.
Silica (SiO2) is a major ingredient in refractories, and silica refractories account
for about one-sixth of total refractory production. Initially, naturally occurring fire-
clays and later on, aluminosilicates of a range of compositions and purity levels ful-
filled the refractory needs of many major metal processing facilities. The chemical
nature of aluminosilicate refractories is acidic, and they are unusable with basic slags
and in general environments that tend to be basic.
Mullite as a refractory is highly usable up to about 1600°C. High-alumina bricks
have found extensive use as a blast furnace refractory. Alumina is among the most
important refractory ceramics and ranks alongside zirconia and yttria in qualities
and usability range. Many types of chemically and thermally processed alumina are
used in refractories. Tabular alumina has excellent high-temperature properties and
thermal shock resistance.
100 Metal–Crucible Interactions

The magnesia or magnesia–lime group includes all refractories made from dolo-
mite and synthetic magnesites. These are the most dominant group of refractories for
the basic steel-making processes. Magnesia is noted for its compatibility with iron
oxides. Calcia is more reactive with iron oxide. It is also more prone to hydration and
disintegration.
For magnesia (MgO), the maximum operating temperature is 1800°C depending
on grade. The coefficient of thermal expansion of pure MgO refractory is very high.
Hence, the thermal shock resistance is poor. Other refractory raw materials may be
incorporated with magnesia in a refractory body to correct this.
The naturally occurring double carbonate dolomite (CaCO3·MgCO3) can be con-
verted to refractory dolomite (CaO·MgO) by firing. Dolomite bricks are made of
sintered dolomite and fused dolomite. To improve the corrosion and hydration resis-
tance, the fired bricks are impregnated with tar or pitch. Currently, the tar/pitch
binder is being replaced with a thermosetting phenol–formaldehyde resin. A chemi-
cally bonded dolomite brick with 3–4% residual carbon is less wetted by slag and has
better thermal shock resistance.
The refractories predominantly used in copper smelting, converting and refining
furnaces belong to the mag–chrome composition group. The properties of refracto-
ries made from magnesia and chrome ore combination are considered excellent, with
one constituent tending to minimize the major weaknesses of the other The proper-
ties of the bricks are a function of the magnesite-to-chrome ratio and also the type
of bonding. Refractories based on mixtures of MgO and Al2O3 have been considered
for replacement of mag–chrome due to perceived environmental threats of Cr6+.
The introduction of carbon into magnesia refractories by tar-bonding, pitch
impregnation and graphite additions results in much improved performance of the
refractory in terms of higher thermal conductivity, good thermal shock resistance and
resistance to slag penetration. The C/MgO ratio can be a handle to tailor properties.
Carbon is the only elemental refractory that is also a non-metal. It is unmeltable
and is not wetted by metal melts and slag. It dissolves in pig iron melt and to a lesser
but significant extent, in certain reactive and refractory metals. It has high thermal
conductivity and low thermal expansion coefficient, and this combination imparts to
it high thermal shock resistance. The major limitation of graphite is its susceptibil-
ity to attack by oxygen. Carbon is used in many forms, mainly as graphite, which
became widely available after Acheson developed a process to synthesize it in 1896.
In metallurgical processing, the principal uses of graphite are as large carbon blocks
in the bottom and bosh of a blast furnace and as electrode cum lining for aluminium
electrolysis cells. As an additive, carbon confers an important enhancement in prop-
erties on a variety of oxide refractories.
Chemically, BeO is one of the most stable materials. Potentially, beryllia cru-
cibles are excellent containers for liquid metals. Thermal conductivity is relatively
high, and for a material of good strength, this gives better thermal shock resistance
than Al2O3. But, beryllia is toxic. Zircon has excellent refractoriness, good wear
resistance, less wettability by molten metal and slag, and hence, good corrosion
resistance. Addition of alumina to form a zircon–mullite composition leads to fur-
ther enhancement in properties.
Crucible Materials 101

Zirconia has a combination of high melting point (2715°C) and the potential
for high strength, superior resistance to corrosion and erosion to reign among the
most important refractories. Zirconia is polymorphic. Around 1170°C, it undergoes
the tetragonal/monoclinic transformation, which is associated with a large volume
change (~4%). This is the reason why pure zirconia wares break up as their tem-
perature changes through this value. To counteract the deleterious effect of phase
changes, zirconia may be stabilized to the cubic phase with small amounts of calcia,
magnesia or yttria, with the result that stability, thermal shock resistance and hot
load properties are enhanced in the final product. It is also amenable to partial stabi-
lization; cubic materials contain fine dispersions of tetragonal phase. This flexibility,
combined with the capability of microstructural control through heat treatment, is
the reason for the popularity of zirconia, as it can be engineered to fit a wide variety
of special refractory applications.
Zirconia is used in the natural, stabilized or fused state, the last often as a mix-
ture with alumina, silica or other compatible oxides. Zircon with alumina can be
processed to form zirconia–mullite grains. This material has high thermal shock
resistance as well as good corrosion resistance. With the addition of carbon, the alu-
mina–zirconia–carbon material is used for slide gate plates in steel ladles.
The rare earth oxide yttria is one of the most stable refractory ceramic oxides. Its
availability, in high purity, improved with the expansion of the rare earths industry,
beginning in the 1970s. Yttria has outstanding chemical stability, but its thermal
shock resistance is not very good. This is partly circumvented by fabricating the
crucible with a somewhat porous structure. The material is expensive. This led to
exploration of its use as a coating on other oxides, and the effort has been rather
successful.
Magnesia–alumina spinel (MgO·Al2O3) and magnesia–chrome spinel are
widely used as refractories because of their high melting point of 2135°C, low ther-
mal expansion, good thermal conductivity and high thermal shock resistance. All
types of spinel tend to form substitutional solid solutions with each other and with
other oxides. In MgO·Al2O3 spinel, both magnesium and aluminium cations can be
replaced by other cations of similar size. This results in the formation of alumina-
enriched spinel as well as magnesia-enriched spinel. Alumina-enriched spinel can
absorb FeO and consequently, retard the infiltration of slag into spinel-containing
refractory. As an alternative to other carbon-free refractories, alumina-spinel fired
bricks show superior performance in highly corrosive metallurgical environments.
The spinel-containing alumina bricks do not contain carbon, and hence, dissolution
of carbon in ultra-low-carbon steel is eliminated. This type of brick is widely used
in producing clean steel.
Carbon-containing refractories are used widely in the steel industry. Benefits
of carbon include its non-wettability by slag and high thermal conductivity. While
carbon confers many of the desired properties on oxide–C refractories, it is also
prone to oxidation, and if that happens, the benefits carbon provides will be lost. To
overcome this problem, two main types of additive used in oxide–C refractories are
metals/alloys (such as Al, Si, Mg and Al/Mg) and boron-based compounds (such as
B, B4C, CaB6, ZrB2, Mg3B2O6 and SiB6).
102 Metal–Crucible Interactions

Refractories undergo corrosion when they come into contact with a molten slag
or a metallic melt. But, the attack is managed in many ways. A novel new use of slag
is as an in situ refractory. This is generally similar to the formation and functioning
of a frozen skull in water-cooled crucibles, common in the handling of very reactive
metals. In situ refractory is routinely created in the slag splashing operations in BOS
vessels. It is also possible to arrange reaction of refractories with the melt contents,
generating a protective interlayer between the refractory and the melt.
The composition of the liquid at the point of penetration into the refractory is
critical to the corrosion mechanism. A different local composition may arise for
several reasons: due to the components of the melt penetrating the refractory at dif-
ferent rates, due to refractories selectively reacting with or dissolving certain melt
components, and so on. The phases in the refractory may dissolve in the local liquid,
altering its composition compared with bulk liquid. The important consequence of
these processes is the resulting significant change in the viscosity of the local liquid,
which directly affects its ability to penetrate the refractory and degrade it.
For the use of a ceramic as a crucible material, it is essential that the compound
shows no crystallographic phase transitions in the temperature region of interest.
This is, in fact, the case for BaZrO3. Apart from a a melting point of about 2600°C,
BaZrO3 has a smaller thermal expansion coefficient than ZrO2 and therefore, better
thermal shock resistance. In a binary ceramic, any deviation from the stoichiometric
composition will lead to precipitation of the off-stoichiometric surplus at the grain
boundaries of the ceramic. This region will be prone to corrosion. This is remedied
by additives such as calcia and yttria, resulting in CaO-doped BaZrO3 and yttria-
doped BaZrO3. Calcium zirconate (CaZrO3) has a high melting point of 2368°C
and chemical stability as good as that of CaO. The zirconate, unlike CaO, is inert to
hydration. As in barium zirconate, the formation of off-stoichiometric compounds
was restored by incorporating CaO as additive. The crucible was used to melt NiTi
and Ti6Al4V alloys at 1500 and 1750°C, respectively, without contamination.
Special refractories include many materials. Silicon carbide, SiC, has good ther-
mal and electrical conductivity, good thermal shock resistance and strength at high
temperatures. Silicon carbide bricks find increasing use in the blast furnace lining.
They resist corrosion by acid slags, alkali vapour, zinc vapour and oxidizing gases.
Silicon carbide is also used in the side walls of Hall–Héroult cells on account of
its properties. The first silicon carbide refractories were bonded with clay, and the
refractory properties of the bond limited the usability of the material. Self-bonded
silicon carbide has removed this limitation. The bonding phase can be SiC itself,
silicon nitride or sialon. Silicon carbide crucibles are used extensively for gold and
silver melting. They have excellent thermal shock resistance and can be used up to
1500°C. Diborides, especially TiB2 and ZrB2, are chemically stable in the environ-
ment of a Hall–Héroult cell for aluminium production. They are not wetted by mol-
ten aluminium, and their electrical conductivity is high. In aluminium electrolysis
cells, replacing the conventional graphite cathode with the diborides has often been
seriously considered. Diborides can be used as crucibles for molten metals such as
aluminium, copper, silver, gold, zinc, cadmium, magnesium, germanium, lead, bis-
muth and chromium.
Crucible Materials 103

Boron nitride in its hexagonal, soft modification finds use up to very high
(~2000°C) temperatures. Billets of boron nitride doped with calcium borate and hot
pressed are readily machinable into metal melting crucibles for use at temperatures
up to 1500°C under vacuum or 1800°C under nitrogen. Sialon ceramics are based
on the solid solution of AlN and Al2O3in β-Si3N4. These materials possess excel-
lent mechanical properties at high temperatures, good corrosion and thermal shock
resistance, and low wettability by aluminium alloy melt. This material may be a can-
didate for long-term liquid aluminium contact. Sialon ceramics also show excellent
corrosion resistance to molten steel and slags.
A unique material with low electrical but good thermal conductivity, AlN pos-
sesses, among the ceramics, the rare combination of chemical inertness and good
shock resistance. It can be used very well as a crucible material. Aluminium nitride
is inert to aluminium, copper, tin and calcium melts and does not introduce foreign
metal impurities into TiAl-based melts either. It is an underexplored crucible mate-
rial useful for containing reactive melts such as molten aluminium and gallium arse-
nide. High-density AlN crucibles are not only inert to titanium alloys but also have
good shock and erosion resistance. AlN performed well, yielding non-contaminated
castings with good ingot release after solidification of titanium melts inside the cru-
cible. These materials are expensive, and large crucible fabrication is difficult. They
are best used as coatings on shock-resistant crucibles like Al2O3 or graphite.
There are rather few instances of a chemically inert crucible also having good
thermal shock resistance. In practice, the issue of getting a material with good chem-
ical inertness and high thermal shock resistance is circumvented by selecting a cru-
cible of shock-resistant material and coating it with a layer/layers of inert material.
For example, an Al2O3 crucible coated with Y2O3 results in a crucible with good
thermal shock resistance, reasonable chemical inertness to the molten titanium and
low cost. Only a beginning has been made on the use of coatings on crucibles for
circumventing metal–crucible interactions. There is a very long way to go.
 4 Major Melt—Crucible
Systems

4.1 INTRODUCTION
Throughout the history of metals processing, metal makers have faced the chal-
lenge of finding and using suitable crucibles for smelting as well as post-reduction
melting and casting operations. The challenges have been faced and circumvented
with ingenuity, directed experimentation and sometimes plain luck, and there is now
available a large and valuable body of knowledge on what works and sometimes, also
the underlying reasons for the success. In this chapter, the situation as regards some
of the major metal–crucible systems are dealt with not only to look back at the road
travelled but also to prepare for the road that lies ahead.
Originally, this chapter was titled ‘Metal–Crucible Combinations That Work’.
The intention was to consolidate in one place all the different metal–crucible com-
binations that have been tested and found to work and in the process, to look for the
common underlying reasons that make such compatibility a reality. It has become
clear in the matter presented so far that materials interact or fail to interact because
of certain well-defined reasons. The reasons concern thermodynamics, on the one
hand, and transport processes at the interface, on the other. How these factors play
out in each individual case is unique, but as is generally explained, there are methods
in the madness. These methods are rather implicit and are best revealed by describ-
ing the working combination. It was therefore considered best to select certain major
metal–crucible systems and look at the key processes involving the crucibles from
the point of view of the crucibles.
No metal has relied so much on the availability of suitable crucibles for successful
processing as iron and steel have. In fact, the bulk of all the refractories for metal
making belong to the sub-group of iron-making refractories or steel-making refrac-
tories. Not only has the metallurgy of iron and steel benefited immensely from the
development and availability of refractories, but also, the technology of refractories
was greatly enriched by the research and development focused on iron and steel. Iron
making in blast furnaces; primary steel making in Bessemer converters, open hearth
furnaces, Linz–Donawitz vessels and electric arc furnaces; secondary steel making
in ladles and degassers; and finally, tundishes, shrouds, slide gates and submerged
entry nozzles in the casting system have all relied heavily on the developments in
refractories.
It must have appeared that aluminium, with the relatively low melting point not
only of the metal itself but also of many of its commercial alloys, would go easy
on refractories in processing. This was not to be the case. Whether for primary

DOI: 10.1201/9780429345562-4 105

106 Metal–Crucible Interactions

aluminium production in Hall–Héroult cells or post-reduction treatments in anode

and refining furnaces, the containment problems were serious. Some of the solu-
tions were unique, and because of the temperatures involved, extensive efforts were
directed to decreasing the wetting of the refractory by the melt using anti-wetting
additives. For other issues, more general melt–refractory interaction management
techniques were used.
Copper is not particularly known for its reactivity, but Mark Schlesinger noted
that furnaces used for producing molten copper face exemplary challenges to the
refractory life in the form of highly aggressive slags, mechanical stresses, batch
operation and higher operating temperatures. Refractories from the mag–chrome
family fulfil nearly all the refractory needs in copper processing. However, two driv-
ers are forcing a change. One is the objection from environmental authorities regard-
ing chromium, and the other is the use of higher process temperatures, off gas richer
in SO2 and the use of calcium ferrite slag.
Beginning around 1940, activity started picking up in developing processes for
the production of many metals that had till then seen little life outside laboratories.
They are what eventually came to be known as less common metals and included
metals essential for emerging new technologies like nuclear energy, aerospace
and a little later, the electronics, telecommunication and biomedical industries.
At that point, the little-used technique of metal replacement reactions became an
important process for the ‘smelting’. In a popular version of this technique, the
reduced metal is produced in a molten form before allowing it to solidify as an
ingot. Thermal and reactivity management are key factors, and both were enabled
by the use of a suitable refractory lining of the reduction reactor, known popularly
as the ‘bomb reactor’. In an equally versatile format, the metallothermic reductions
were also carried out to yield the metal as sponge. Not that there was much choice
of obtaining these metals in any other form, because handling metals like tita-
nium, zirconium or hafnium in molten form at the reduction stage would have pre-
sented formidable crucible problems. The flagship process for obtaining reactive
metals in sponge form is the Kroll process, which is based on controlled reaction
between purified metal chloride and molten magnesium in a protected atmosphere
at temperatures in the range of 950°C. As long as sharp temperature excursions
are avoided, steel and Inconel reactors are fine for crucibles and have been widely
used. While the Kroll-type process can be used for the preparation of some of
the rare earth metals, a more robust method, relying on calcium reduction of rare
earth trifluorides, was eventually used for the preparation of rare earth metals.
The process was carried out in induction-heated tantalum crucibles in a protected
atmosphere. To minimize ingestion of tantalum into the metal, a relatively low-
temperature reduction based on lithium vapour as reducing agent was also used.
Tantalum, molybdenum and titanium wares were used as contact materials here.
A Pidgeon-type process based on reduction–distillation was also used for certain
rare earth metals. As well as all these, electrolytic methods were developed for
most of the rare earth metals, using graphite, molybdenum and tungsten as con-
struction materials. The high point is the production of yttrium in molten form in
an electrolysis cell using fluoride electrolytes operated at 1700°C.
Major Melt–Crucible Systems 107

Uranium is a metal that had to be processed frequently for a variety of purposes

in a molten form. There were also requirements to evaporate it from a crucible. All
this meant strong focus on the development of uranium melt-compatible crucibles.
Vanadium is well established as an alloy additive to steel, and not much was known
of its use as a metal or an alloy base. In fact, for three decades, there was just one
monograph on vanadium, written by William Rostoker in 1958, until another one
appeared in 1992. This was until interest in its possible role as a construction mate-
rial for liquid metal loops in fast (nuclear) reactors and thermonuclear fusion reactors
brought it back into focus. Vanadium is notoriously reactive and impossible to melt
for any length of time without contamination in a metal or ceramic crucible.
Somewhat less aggressive than vanadium, but still not easy on crucibles and
moulds, is titanium. It is exceedingly important, both technologically and finan-
cially, to find a crucible for routine melting and casting of titanium and its alloys. The
importance and urgency stem from its relevance as a premium aerospace, energy
conversion and biomedical material base, and in many cases, it is not replaceable in
its applications without technology and cost penalties. After nearly seven decades
of directed efforts, the interesting situation is that it may still be premature to assert
that we have succeeded.
What follows in this chapter is a crucible-centric exposition of the relevant pro-
cesses for all these metals. The objective is to relate what is known about metal–cru-
cible interactions at the basic level to the interactions playing out in real processes.
This will help the task ahead as regards managing new metal–crucible combinations
and suitable approaches that will improve the chances of success.

4.2 IRON AND STEEL

Few metals have been as closely dependent on the availability of crucibles for their
extraction as iron. Since iron was discovered and recognized as a highly usable
metal, for over two millennia, it was not prepared in molten form. All processes to
make products from iron were designed to work around the need for melting it. The
reports on the first iron castings being made in China also refer to their access to
fairly refractory clays available only in that country. Eventually, the development of
refractories for iron and steel became an important area of study and industry and
led the way for the development of the entire industry of metallurgical ceramics and
refractories.
Even though iron is the second most abundant metal in the earths’ crust (5.8%)
after aluminium (8.3%), iron is the most used metal on the planet, mostly as steel.
The iron-rich ores (haematite [Fe2O3], magnetite [Fe3O4]), are reduced using coke in
an iron blast furnace. The reduction process predisposes the as-produced iron, called
the hot metal or pig iron, to be liberally contaminated with a variety of impurities
derived from the ore, the reducing agent and the various additives and fluxes that
were put in to make the process efficient and smooth. There are also the inevitable
impurities that the iron picks up from the furnace structure. The impurities include
carbon, silicon, phosphorus, aluminium, manganese and sulphur, and many more in
minor quantities. Amazingly, the metal is still usable as such for a few applications
108 Metal–Crucible Interactions

(after a certain amount of metallurgical tending), but as the impurities are progres-
sively decreased both in number and in concentration, the variety of uses increases
greatly and the uses become more sophisticated.
Many elements are incorporated into iron and steel by design, for enhancement
of properties. The process metallurgy part of iron and steel is all about impurities
control, i.e. the removal of impurities from the metal and protection of the metal
from re-ingestion of impurities during further processing to a finished product. It is
amazing that steel is the only ‘metal’ that has the adjective ‘clean’ as part of its name.
‘Clean steel’, which has less than 50 ppm of residual impurities (PCSNHO), is the
cleanest metal produced routinely under industrial conditions in tonnage quantities
(Poirier 2015). The processes and materials used represent some of the finest and
most ingenious in process metallurgy. It is the scale of operation that is daunting.
Besides, the physical chemistry and metallurgical features of these operations are so
varied and so deeply challenging that developments in the materials and techniques
that constitute the making and treating of steel eventually became the core of the
evolving metallurgy practice and knowledge base. When confronted with challeng-
ing technical issues regarding materials and processes, it is not unusual for metal-
lurgists, irrespective of their specialization, to look for analogies in iron and steel
metallurgy. One such issue is the interaction of melts with the crucibles or contain-
ment and the consequent effect on properties.
Blast furnace iron making is the most popular process for extraction of iron from its
ores. Pure iron has a high melting point (1536°C), but when it dissolves carbon, the melt-
ing point decreases significantly, and the iron–carbon eutectic occurs at 1150°C and 4.2%
C. Steel, the alloy of iron and carbon with C content less than 4.2%, has the ability to take
many other elements, such as chromium, nickel and manganese, into its composition,
and the presence of these elements modifies and usually enhances one or more proper-
ties of the steel. There are hundreds of grades of steel characterized by compositions and
microstructures, and they are produced through secondary steel-making processes.
The primary steel-making process largely follows the blast furnace–LD pro-
cess/basic oxygen furnace (BF–LD/BOF) route and to a lesser extent, the directly
reduced iron–electric arc furnace (DRI–EAF) route. Secondary steel making cov-
ers the refining of primary steel with respect to composition and cleanliness before
casting. Secondary steel making involves deoxidation or killing with Al/Si additives,
vacuum degassing, desulphurization, homogenization, alloying, and control of inclu-
sion chemistry and content. These operations are commonly performed using ladle
furnaces (Fruehan 1998).

4.3 BLAST FURNACE REFRACTORIES

The functioningof a blast furnace depends on the performance of the lining refractory
(Biswas and Sarkar 2020). These refractories undergo exemplary physical, thermal
and chemical stresses, and the blast furnace has long been the test track for new and
improved refractories. The main sources of physical stresses are gas flow through the
stack, erosion by descending burden, and flow of liquid iron in the hearth. The ther-
mal stresses are due to the heat load inside the furnace. High-temperature reactions
Major Melt–Crucible Systems 109

of refractories with gases in the stack and belly area, corrosion reactions with slag at
the belly and bosh, and penetration of hot liquid metal in the hearth, in combination
with the physical and thermal stress factors, cause the degradation of refractories.
Alkali infiltration and reactions with the bonding phases, especially in the bosh and
lower stack, have the greatest effect on the lining. The alkalis come with the coke.
The blast furnace slag contains, as well as other components, FeO (+2, basic
oxide) and alkali vapour. These are the components that attack the refractory. The
various reactive species and the temperature range in which they are effective are
CO (on stack refractory: 450–850°C), alkali (800–950°C), zinc vapour (750–900°C)
and O2 (on C refractory: >400°C),as well as oxidation by CO2 and H2O (>700°C).
Hence, the lining refractory needs to be kept at temperatures lower than these tem-
peratures of reaction by external cooling. This ‘thermal solution’ is implemented by
cooling plates and staves, cigar coolers and spray coolers. All portions, including the
tuyere zones and the hearth, are cooled and the bias has been towards refractories
that have good thermal conductivity.
In the conventional refractory lining in the blast furnace stack and bosh, the lin-
ing was made of high-alumina bricks containing 62–85% Al2O3. Even in the 1960s
and 1970s, the refractory linings in blast furnace construction, including the hearth,
were made of different qualities of high-alumina bricks. Their performance and life
were limited due to poor alkali resistance, poor thermal shock resistance and also
very low thermal conductivity. This made the cooling of the lining ineffective. The
attack on high-alumina refractory by slag and alkalis resulted in the formation of
compounds that melt at <1400°C, resulting in severe corrosion of the refractory dur-
ing furnace operation.
The use of a carbon hearth in 1970 and of silicon carbide (SiC) refractory in
the bosh, belly and lower stack area came as major developments for modern blast
furnace lining. Since the turn of the century, there has been no brickwork in the
new-generation shaft and belly of blast furnaces in newly built or rebuilt furnaces.
Nitride- or oxy-nitride-bonded SiC bricks with coolers embedded in the stave are
used in the stack, and micropore carbon or graphite blocks are used in the belly and
bosh areas. For the hearth and bottom of blastfurnaces, semi-graphitized carbon
blocks and microporous carbon blocks are used. The carbon blocks used are all
of high heat conductivity. Superior-quality carbon blocks of micropore and super-
micropore quality (pore size <1 μm) and very high thermal conductivity are used.
Carbon hearth refractories are in direct contact with liquid iron. Under a high
ferro static load, the molten metal penetrates through the pores of carbon blocks and
embrittles the penetrated layer, which eventually dislodges due to turbulence in the
melt. This causes progressive thinning of carbon blocks. Penetration of hot metal and
eventual thinning may be avoided or minimized if the radius of open pores of carbon
blocks is small enough.

4.4 STEEL-MAKING REFRACTORIES
As mentioned in Chapter 1, the original Bessemer converter had an acid (siliceous)
lining. Its usage was restricted to a very low phosphorus- and sulphur-bearing pig
110 Metal–Crucible Interactions

iron, since these elements are not removed in the process. In the Thomas process,
also referred to as the basic Bessemer process, the vessel was lined with basic refrac-
tory (burned limestone) to handle the slag with a high basicity. Calcia was added as
flux during the air blow to form a basic slag with high CaO content, which was able
to remove phosphorus as calcium phosphate.
The hearth of Siemens’ furnace (open hearth furnace) was constructed of acid
bricks topped by a layer of sand. Later, the siliceous bottom of the acid furnace was
changed to permit the charging of limestone and use of a basic slag for the removal
of phosphorus. The hearth was constructed with a lining of magnesite brick, covered
with a layer of burned dolomite or magnesite.
A schematic cross section of an oxygen steel-making vessel is shown in Figure 4.1.
The refractories used in BOF linings are pitch-bonded magnesia or pitch-bonded
dolomite, which are made with resin bonds, metallics, graphites, and sintered and/or
fused 99% pure magnesia (Fruehan 1998).
All basic oxygen steel-making processes, top, bottom and combined blowing, rely
on oxidation by externally supplied oxygen gas. Oxygen is blown into the melt, and
dissolved elements, such as C, Si, P and some S, are oxidized to form CO, CO2,
SiO2 and P2O5. Manganese and iron are oxidized to MnO and FeO. The CO and
CO2 gases escape out, and other oxides collect to form an acidic slag. All these slag-
forming oxides can react with basic refractories and corrode them. CaO (burned
lime), MgO (burned dolomitic lime) and CaF2 (fluorspar) are added to the molten
bath to neutralize the acidic nature of slag, and form a basic slag, in the later part of
the blowing. The added lime is fluxed and dissolved by the siliceous slag, and the
resultant basic slag absorbs sulphur and phosphorus from the molten steel and also
protects the lining from attack by FeO and SiO2 (Biswas and Sarkar 2020). MgO
may be introduced as dolomitic lime to saturate the slag or even to supersaturate with
MgO. Supersaturation results in a MgO build-up on the vessel walls, which serves to
extend the life of the lining.

FIGURE 4.1 Schematic of a Linz–Donowitz vessel (From Fruehan, R.J. (Ed.). 1998).
Major Melt–Crucible Systems 111

FIGURE 4.2 Schematic of a Linz–Donawitz vessel – slag coated (From Fruehan, R.J. 1988).

The technology known as slag splashing evolved in the early 1990s. After tapping
a heat, it entails the use of the oxygen lance to blow nitrogen on the residual slag,
splashing it to coat the walls and cone of the converter (Figure 4.2). Slag splashing
seeks to utilize the technique of working off a slag coating instead of the working
lining brick or its gunned coating. (Fruehan 1998).
Until about 1976, the main refractory used in BOF was dolomite bricks, such as
tar-bonded dolomite or high-fired magnesia-enriched dolomite. During the 1980s,
the use of MgO–C bricks started due to their better corrosion and thermal shock
resistance. MgO–C bricks were used to line the whole vessel, and that practice has
continued to the present. As already mentioned, the durability of MgO–C bricks
is significantly improved if the dissolved MgO in BOF slag reaches or exceeds the
saturation point of MgO in slag at the operating temperature.
Carbon is used in refractories for its excellent thermal shock resistance and cor-
rosion resistance towards BOF slag and steel. But, it has an inherent drawback of
susceptibility to oxidation, and it also increases the thermal conductivity of the lin-
ing, which results in heat loss and increase in shell temperature. The development
of nano-carbon-containing MgO–C bricks with reduced total carbon content has
improved oxidation resistance and reduced thermal conductivity. The addition of
nano-carbon serves to lower the total graphite content of conventional magnesia car-
bon bricks without sacrificing the beneficial effects of carbon addition (Biswas and
Sarkar 2020).

4.4.1 Refractories for Secondary Steel

Dolomite and dolomite–magnesia bricks are the most commonly used refracto-
ries for the Argon Oxygen Decrburization (AOD) vessel. The very basic nature of
112 Metal–Crucible Interactions

dolomite refractories permits operation with higher-basicity slags, which enables

improved chrome recovery, and desulfurization. Magnesia–chrome and magnesia–
carbon bricks are also used in AOD. The same types and grades of refractories are
used in Vacuum Oxygen Decarburization (VOD) ladle applications, but with more
equal amounts of magnesia–chrome and dolomite–magnesia bricks. The choice of
brick composition, magnesia–carbon or fired magnesia–chrome, is based onthe car-
bon pick-up tolerated during processing.
In an RH (Ruhrshahl Heraeus) degasser, working lining refractories are exposed
to a variety of conditions depending on location. The linings are often zoned (Fruehan
1998). All locations are exposed to thermal (changes between 200 to 600°C) and
pressure (ambient to 70 Pa) cycling. Apart from liquid steel at 1650°C, snorkel and
lower vessel refractories face the erosive action of high-velocity, turbulent steel flow.
RH and RH-OB (oxygen below) degasser working linings are usually direct-bonded
magnesia–chromite bricks containing fused grain, or 50–60% alumina brick, or
90% alumina, or tabular alumina, or spinel castable. Castables with high-purity
materials are used for the snorkel exterior.

4.4.2 Refractories for Casting

In casting, it is desirable to preserve the purity and cleanliness of steel achieved
in secondary metallurgy operations. The machines and refractories are chosen
accordingly.
In 1940, the lowest level of total impurity (C, S, O, N and H) content possible in
steel was a few hundred ppm, and by the turn of the century, the value was less than
100 ppm. It dropped further to less than 30 ppm by 2015 (Poirier 2015). As well as
being present in solid solution, oxygen is often present as oxide inclusions of vari-
ous sizes in steel. They are both endogenous or micro-inclusions (originating from
within the steel) and exogenous or macro-inclusions (e.g. eroded and entrapped lin-
ing). The possible influences of non-metallic elements on various properties of the
metal have been summarized by Poirier (2015) and listed in Table 4.1.
Lower values (ppm)currently attainable for many elements are shown in the last
row of the table. For steel produced routinely in industrial conditions, the values are
somewhat higher. The overwhelming influence of oxygen on a range of steel proper-
ties cannot be ignored. Next comes sulphur.
As emphasized already, purity achieved in secondary metallurgy should not be
lost during continuous casting. The refractory materials directly determine both the
type and the amount of solute elements such as carbon, oxygen and the non-metallic
inclusions. The interactions between metal, slag, atmosphere and refractory products
are related to the chemical and mineral composition, microstructure and phase dis-
tribution of the refractory. The key reactions are direct dissolution of the refractory
with or without precipitation, dissociation, volatilization, redox reactions between an
oxide and a metallic element, and combination of the refractory and a non-dissolved
element present in the steel (inclusion).
When the steel melt is re-oxidized by the refractory lining, the oxygen reacts
with dissolved Al (aluminium) in steel to form Al2O3, which partly floats into slag
Major Melt–Crucible Systems 113

TABLE 4.1
Influence of Non-Metallic Elements on Steel Properties
Properties Hydrogen Carbon Nitrogen Oxygena Phosphorus Sulphur*
Internal soundness x x
Surface defects x
Fatigue x x
Bending x x
Deep drawing x
Anisotropy x x
Toughness x x x x
Weldability x x x
Electromagnetic x
properties
Lower limits of <1 5 10 5 10 5
residual elements
in steel making,
ppm

Source: Poirier, J. 2015.

a (Control of inclusions)

and partly remains in steel as inclusions. Very fine inclusions (the smallest size) may
not really float up for skimming or to be collected by the slag. Refractory materi-
als with the lowest oxygen potential result in lower aluminium losses and hence, a
decrease in inclusions by re-oxidation. Lime, corundum, magnesia and high-purity
synthetic materials such as alumina spinel, MgO–C and doloma–C are examples of
such materials. If present even in small amounts, SiO2, Cr2O3 and ZrO2, finally result
in a high inclusion content.
Shaped refractory parts (Figure 4.3) are used to guide and protect liquid metal
from the ladle to the mould where steel is solidified in continuous steel cast-
ing (Poirier 2015). Materials of the submerged nozzle have a role in clogging and
unclogging, which leads to metal contamination by alumina particles or clusters. The
tundish lining can also have a similar role. Contamination can also be caused by the
ladle shroud tube, where reactions similar to those in submerged nozzles can occur,
and by the sliding gate system, where the plates can undergo degradation.
Submerged nozzles are often alumina–graphite products. Clogging of submerged
nozzles by alumina build-up is a major cause of dysfunction in continuous casting
of aluminium-killed steel. The mechanism described by Poirier (2015) focuses on
the deposition of Al2O3 as a result of the thermochemical reduction of nozzle con-
stituents coupled with the oxidation of the aluminium in the steel. Carbon is the
key element. Alumina–graphite refractories contain secondary phases and impuri-
ties such as SiO2, Na2O and K2O, which can be reduced by the refractory carbon
at steel-making temperatures, and generate gaseous species. The carbon monoxide
so generated oxidizes aluminium in steel to alumina, which builds up as a solid
114 Metal–Crucible Interactions

FIGURE 4.3 Main refractories in relation to steel quality and metal cleanliness (secondary
steel making and casting) (From Poirier, J. 2015).

deposit. Avoiding carbon from the refractory should eliminate some of this alumina
deposition. Similar improvements can also be expected using highly pure A12O3–C
mixtures with as few silica and alkaline impurities as possible.
The tundish refractory is usually made of magnesia and forsterite (2MgO·SiO2)
raw materials. The lining may thus react with Al in steel, more so if the lining is
porous. At the steel/refractory interface lining, a layer composed of MgO–Al2O3
spinel forms by reaction between aluminium in steel and forsterite in the refractory.
Spalling due to the mismatch of properties between the spinel layer and the MgO–
forsterite refractory lining can lead to MgO–Al2O3 inclusions in the steel. The ladle
shroud is usually made of alumina–graphite. The same reactions as in the submerged
nozzle are also possible here.
The sliding gate system for steel ladles is used to control metal flow during teem-
ing, and consists of a mechanical assembly containing the refractory plates (Poirier
2015). Air leakage through the worn plates has adverse effects on the cleanliness of
the steel and the wear of the refractory by corrosion.
Al2O3–C slide plates have been widely used. They are produced using tabular
alumina, fused white alumina, and graphite with Si and Al metal powder addition.
However, these plates may not be ideal for producing low-carbon or ultra-low-carbon
clean steel because of the possible diffusion and dissolution of carbon into liquid
steel. The alternatives are SiO2-free Al2O3–C plates and spinel-containing plates
Major Melt–Crucible Systems 115

such as MgO·Al2O3–C and MgO–C. Al2O3–ZrO2–C plate is currently used to cast

Ca-treated steel.

4.5 THE HALL–HÉROULT CELL

The invention of an electrolytic process for aluminium extraction was outlined in
Chapter 1. Almost all the world’s aluminium is produced by electrolysis in Hall–
Héroult cells. During the electrolysis of alumina dissolved in molten cryolite, alu-
minium ions discharge at the carbon cathode and collect as molten metal at the
prevailing temperature (950–965°C) in the cell. The anions discharge at the carbon
anode and evolve as a mixture of carbon oxides: 20–40% CO and 60–80% CO2.
The melts of aluminium and of the electrolyte are non-miscible. In the cell, electro-
lyte (density = 2.1 g/cm3) remains above the molten aluminium (density = 2.3 g/cm3).
Typically, the depth of the molten aluminium layer is about 15–40 cm, and the depth
of the electrolyte layer is 15–25 cm (Yurkov 2017).
Pure cryolite, Na3AlF6 (3NaF·AlF3), melts at 1010°C. The melting temperature
of the electrolyte is brought down to 930–960°C either by adding fluorides (calcium
fluoride, magnesium fluoride) or by varying the ratio of sodium fluoride (NaF) to
aluminium fluoride (AlF3). The cell, shown schematically in Figure 4.4, is a shallow
bath tank (0.6–0.8 m deep), 10–15 m long and 3–5 m wide. Electrolysis is conducted
at 4.16–4.20 V, keeping the space between the anode and cathode at 40–60 cm. The

FIGURE 4.4 Hall–Héroult Cell (From Yurkov, A. 2017).

116 Metal–Crucible Interactions

bath is covered by a frozen electrolyte crust. The cells are lined with carbon cathode
bottom blocks and a side lining. Below the cathode block are the refractory layer and
the heat insulation layer.
The issues of metal–crucible interaction in the Hall–Héroult cell mainly concern
the interaction of the molten metal and the bath components with the cathode and the
refractory layer. Even though the temperatures involved are relatively low, interac-
tions are many, and intensity is high. Molten aluminium and electrolyte in operating
cells are in a state of constant movement, at a speed that may range from 1.5 to 30
cm/s, in the electromagnetic environment prevailing in the cell (Yurkov 2017).
During operation, the electrolysis current also heats the cell. A considerable
amount of excess heat is generated in electrolysis (i.e. more than what is required to
maintain the bath at 950–965°C), and is dissipated through the side walls of cells.
Approximately half of this heat is dissipated through the side wall, and the other
half goes away through the bottom and upper parts. Therefore, the side lining, made
of SiC blocks or carbon side-wall blocks, has minimal or no heat insulation, while
the bottom has good thermal insulation. The temperature profile is also maintained
such that a certain amount (a few centimetres thick) of electrolyte freezes on the side
lining and protects the lining materials from the molten cell contents. Freezing of
electrolyte is not permitted in the bottom within the cell but for a small portion near
the side wall. Insulation is therefore differently arranged in this portion. The thermal
balance will be disturbed if the electrolyte infiltration through the carbon cathode
(block) goes past the refractory layer and ends up in the heat insulation layer.
In the design and construction of the cell, the objective is to have the 800–850°C
isotherms within the main body of the upper, protective refractory layer (Siljan et al.
2002). The solidus of the advancing melt falls in this temperature window. The most
important purpose of the refractory layer, i.e. as a barrier to bath intrusion and to
prevent deterioration of the thermal insulation layer below, will then be served. Even
the thermal conductivity of barrier layer refractories, if infiltrated by electrolyte,
may change by a factor of 3–5.

4.5.1 Interactions in the Cell

Key interactions to be considered in the cell involve the electrolyte, molten alumin-
ium and sodium with the cell materials, particularly the cathode carbon block and
the refractory beneath it.
Sodium is generated and also transported through the cathode carbon due to reac-
tions given here (Siljan et al 2002):

In bath: 2Al + 6 NaF = 2AlF3 + 6 Na ( Al ) (4.1)

Metal pool/cathode: 6Na ( Al ) = 6Na ( C ) (4.2)

There are two main pathways for sodium to go past the cathode and reach the barrier
refractory layer of a reduction cell: sodium diffusion (intercalation) in the carbon
lattice and capillary flow of electrolyte through permeable pores of carbon cathode
Major Melt–Crucible Systems 117

blocks and carbon of ramming paste. The resulting sodium pressure at the lower
surface of the cathode block might reach as high as 0.02–0.03 atm.
The radius of a sodium atom being significantly larger than the distance between
the interlayers of carbon atoms, the intercalation of sodium atoms in the crystal lat-
tice causes swelling and eventually, cracking of carbon blocks. Cryolite (and also
aluminium) will penetrate through the cracks to the bottom of the cathode. More
than the dry penetration of sodium, it is due to ‘wet’ penetration, or the capillary
flow of electrolyte through permeable pores of carbon cathode blocks and rammed
carbon of ramming paste, that the deterioration of the refractory lining occurs.
Thus, as well as sodium and electrolyte, aluminium also ends up at the other side
of the cathode due to damage to the cathode. The pathways created by the damage
allow liquid aluminium to penetrate down to the top of the refractory lining. Then,
a number of reactions occur between the penetrated materials and the refractory,
which result in the formation of new phases and cause volume expansion and also
viscosity changes in the melt.

4.5.2 Barrier Refractory
Typically, the barrier refractory layer is made of aluminosilicate (usually fireclay)
bricks or unshaped dry barrier mixtures. The aluminosilicates are prone to infil-
tration and attack by molten electrolyte components. Sodium and molten fluorides
cause significant mineralogical transformation in the refractory.
Many chemical reactions are possible among cryolite, sodium fluoride, alumina,
silica and mullite. The reaction products are albite (NaAlSi3O8) and nepheline
(NaAlSiO4), which may coexist with cryolite. The volume effect of the reactions
is the real cause for concern. Depending on the conditions, the products of reaction
between the infiltrated electrolyte and the refractory barrier layer end up in the form
of a convex lens. This lens can grow up to almost a metre in thickness and press on
the carbon cathode blocks from below, causing bending strains and cracking. The
upper part of the lens (immediately below the cathode block) comprises albite, neph-
eline, cryolite and sodium fluoride. The middle part of the lens consists of albite,
nepheline, cryolite, sodium fluoride, quartz and mullite, and the lower part of the
lens consists of reacted material and of 30–50% of unreacted fireclay (Siljan et al.
2002).
One remedy suggested to diminish bath penetration through the cathode block to
the refractory level is to limit average pore size in carbon cathode blocks to 5–15 μm.
Bath penetration is facilitated if the pores in the carbon cathode blocks are above
25–30 μm. With bath components at the doorstep of the refractory layer, Siljan et al.
(2002) singled out the silica content of the aluminosilicates as the major handle that
can be used to stop the further progress of the reaction. By exposing samples of fire-
bricks to a fluoride melt at 950°C (‘cup test’) as well as to sodium vapour at 800°C,
Siljan et al. (2002) found that materials high in silica deteriorated less than materials
rich in alumina. The reason is that the silica-rich materials formed larger amounts of
a viscous glass phase. A fluoride melt, which has penetrated through the carbon cath-
ode, accumulates below the cathode carbon, while sodium aluminosilicate, which is
118 Metal–Crucible Interactions

denser than the fluoride, will settle below it and above the refractory lining. The con-
tinued contact and further reaction of the fluoride melt with the refractory is retarded
by the dense viscous layer of the sodium aluminosilicate-based melt. This has been
revealed in autopsies of shut-down cells as well as in laboratory-scale cup tests.
The composition of melt that results from the interaction of the infiltrated mate-
rials with the aluminosilicate refractory depends on the proportions of silica and
sodium fluoride in the reactants (Siljan et al. 2002). The key components are albite
and nepheline,which strongly determine melt viscosity, because melt viscosity
apparently plays a larger role than pore size of the refractory as regards eventual
melt intrusion into the refractory. The solute in the melt that increases the viscosity
is albite. The solubility of albite in cryolite is also high. It is good to have albite in
the melt. The viscosity of nepheline-rich melt isseveral orders of magnitude lower
compared with the viscosity of albite-rich melt. In practice, this (albite-rich melt) is
achieved by selecting a refractory with silica content above 60% and by a controlled
input of sodium compounds, e.g. with excess sodium fluoride.
There are two choices. To freeze out the fluoride melt at an isotherm located not
far below the cathode carbon, i.e. to ensure a low penetration depth into the refrac-
tory lining, a high solidus temperature is preferred. This is defined by the refractory
material and the penetrating bath. For this, higher alumina is needed in the refractory
composition. However, with a substantial amount of silica in the refractory lining,
the viscosity of the melt increases so much that penetration is in any case restricted
to a low depth. This aspect requires a silica-rich refractory (>53 wt.% SiO2). The
approach favoured by Siljan et al. 2002 is to focus on managing the viscosity of the
penetrating melt. They did also point out that the likely gain in solidus by increasing
alumina content in the refractory is not so substantial.
For the record, some of the chemically effective refractory barriers against the
penetration of cryolite are tin oxide, nickel oxide, compounds of nickel oxide, iron
oxide or zinc oxide (such as spinel NiFe2O4). These oxides almost do not react with
NaF and aluminium fluoride. But, their cost is impractically high.

4.5.3 Side Lining Materials

Anthracite-based carbon side blocks were used for many years before being replaced
with SiC in the 1980s. The advantages of SiC side lining compared with carbon side
lining are better corrosion resistance to electrolyte and aluminium, higher oxidation
resistance and higher thermal conductivity. Currently, nitride-bonded silicon carbide
is the only material used as side lining (Yurkov 2017).

4.6 POST-REDUCTION TREATMENT OF ALUMINIUM

As-reduced aluminium from the Hall–Héroult cell is 99.5 to 99.7% pure. It contains
non-metallic (electrolyte, alumina, aluminium carbide, carbon particles and oxides),
metallic (iron, silicon, titanium, sodium and calcium), and gaseous (hydrogen) impu-
rities. These impurities are removed in post-reduction treatments by fluxing (and
skimming) and degassing before casting the 99.8 to 99.85% Al into moulds.
Major Melt–Crucible Systems 119

The metallurgical equipment of cast houses at smelters includes ladles for Al

transportation, stationary or tilting holding furnaces, degassing units, induction fur-
naces and gas-fired furnaces. There is almost no difference in the constructions of
melting and holding furnaces. Usually, melting furnaces are heated by natural gas,
and holding furnaces are heated by natural gas or by electric heaters. The general
requirements for refractories in these facilities are porosity <15%, average pore size
<5 μm, and low wetting by Al.

4.6.1 Refractories in Aluminium Melting and Holding Furnaces

Schlesinger(2006)in his book on aluminium recycling observed that for its relatively
low melting point, aluminium places a rather disproportionate demand on materials
usable for its containment. He then listed the nine requirements that the refractories
in aluminium melting and holding furnaces need to fulfil: (i) lack of reactivity with
the molten aluminium, (ii) low solubility in molten flux, (iii) lack of reactivity with
the Al2O3in the dross, (iv) low thermal conductivity, (v) hydration resistance, (vi) oxi-
dation resistance, (vii) stability at temperatures up to 1200°C, (viii) good mechanical
strength and (ix) low cost.
As is common in such situations, no single refractory meets all the requirements.
The job is done by a combination of refractories. As an example, in the wall con-
struction in a reverberatory melting furnace for aluminium, a corrosion-resistant
refractory is used at the hot face and a low-density insulating material between it and
the steel shell. The purpose is to locate the ‘freeze plane’, below which penetrating
metal solidifies, well within (Dunsing 2002) the thickness of the corrosion-resistant
refractory lining. This is a rather universal requirement implemented in melt con-
tainment configurations. It was seen in the Hall–Héroult cell also. The insulating
brick is not sufficiently resistant to molten metal and is also porous, and will not
survive any contact with the melt. If the molten metal penetrates and ends up beyond
the thickness of the refractory facing, and freezes between the refractory and insu-
lating material, not only is the insulation arrangement ruined but the materials also
become unusable.
Such a combined refractory structure (Figure 4.5) works in the molten alu-
minium transportation ladle (Bonadia et al. 2006) as it does in the torpedo ladles
of molten iron transportation. This arrangement, in principle, works in crucible
induction furnaces also – a multilayer refractory lining comprising a chemically
stable working lining to contact the melt and a thermal insulation lining between
the working lining and the vessel walls (Figure 4.6). Phosphate-bonded high-
alumina bricks resist aluminium attack well due to their unique microstructure,
i.e. small pores, which significantly decreases metal penetration. This is a typi-
cal working lining. There is a thin layer of highly insulating microporous silica
between the steel shell and the intermediate brick layer, on the wall and at the bot-
tom, for thermal insulation. A semi-insulating castable coupled with ceramic–fibre
insulation is used for the ladle lid.
The crucible induction furnace shown schematically (Figure 4.6) is used for mak-
ing alloys and master alloys and for remelting scrap (Yurkov 2017). The lining of
120 Metal–Crucible Interactions

FIGURE 4.5 Composite refractory structure – aluminium transportation ladle (From

Schlesinger 2006, Bonadia et al. 2006, Yurkov 2017).

FIGURE 4.6 Composite refractory structure – induction furnace (From Schlesinger 2006,
Yurkov 2017).

induction furnaces is made from ramming mixes as well as plastic mixes. An anti-
wetting additive is normally used. Anti-wetting additives decompose, leaving a bulky
residue. A fine pore -size structure with a small pore dimension is a more reliable
option in such conditions. Phosphate-bonded bauxite bricks or low cement castable
(LCC) with filler from bauxite are used. Degassing units are used for the refining of
Al alloys from gases, fluxes and oxide films. The requirements for refractories for
the vanes of degassing units are the same as for melting and holding furnaces.

4.6.2 Alumina–Silica Refractories
The aluminosilicates are popular refractories for melting and holding furnaces.
High-alumina content (57 to 85% Al2O3) refractories, typically consisting of a mix-
ture of corundum (Al2O3) and mullite (Al6Si2O13), are used in areas that come into
Major Melt–Crucible Systems 121

contact with molten metal and dross. The refractories also contain ingredients like
CaO and free silica. The use of pure alumina does not result in improvements com-
mensurate with additional cost.
Silica in any form (crystalline, amorphous or combined) present in the refractory
material readily reacts with aluminium. Therefore, the use of aluminosilicate com-
pounds implies the use of remedial measures also to offset this vulnerability. The
effective remedial measures are the use of anti-wetting agents and restricting pore
sizes to <1 μm to hinder aluminium penetration and decrease the total contact area
available for reaction (Bonadia et al. 2006).
When aluminium reacts with the silica and silica-bearing constituents, corundum
and elemental silicon are formed.

4Al + 3SiO2 = 2Al 2O3 + 3Si (4.3)

8Al + 3Al6Si 2O13 = 13Al 2O3 + 6Si (4.4)

Brondyke (1953) noted that the reaction is accompanied by an increase in volume.

Cracks form as a result of this volume expansion. However, Nandy and Jogai (2012)
calculated a reduction in volume. Yurkov (2017) also noted that both reactions result
in a negative volume effect, which facilitates metal (Al) penetration. In any case, a
volume change accompanying a reaction causes stresses within the refractory and
degrades its integrity.
According to Brondyke (1953), the aluminium penetration of these refractories
proceeds at a decreasing rate because the penetrated material becomes an ever
increasing barrier to the diffusion of aluminium through it to the unreacted refrac-
tory. As the zone of penetration increases in thickness, the diffusion and hence, the
rate of reaction both decrease. Even though penetration of the alumina–silica refrac-
tories is the reaction of aluminium with silica and the silica-bearing constituents,
Brondyke (1953) discerned no apparent relationship between the percentage of silica
(free and combined) in the refractories and the extent of aluminium penetration. It
was also mentioned by Allaire (1992) and Afshar and Allaire (1996) that despite the
corrosion phenomenon being essentially characterized by silica reduction, no evident
correlation exists between metal penetration and the silica content of refractories.
The metallic silicon formed as reaction product ends up in the aluminium bath
(Brondyke 1953). But, the silicon pick-up by the aluminium tracks the silica content of
the refractory, with a definite decrease of silicon pick-up with increase in alumina content
(>50% alumina). Temperature has a significant effect. Penetration of the various refracto-
ries was negligible at 705°C but increased rapidly at 760°C and above. The porosity of the
refractories apparently had little effect on the degree of penetration, and the penetration
proceeded through the refractory grains by diffusion rather than by capillarity action
through pore channels. Commercial mullite refractories were penetrated to the same
degree and at about the same rate as other refractories having a similar chemical compo-
sition without all the silica and alumina being converted to mullite.
Among various investigations on the interaction of molten aluminium with alumi-
nosilicate refractories, the results obtained in apparently similar process conditions
122 Metal–Crucible Interactions

generally agree but often do not match closely with one another. This could be due to
the inevitable variations in the actual chemical composition of the melt, composition
and structure of the refractory.

4.6.3 Belly Band Zone

In the aluminium melting and holding furnaces, the zone most vulnerable to corro-
sion is the melt line, known as the bellyband (Schlesinger2006). It is shown sche-
matically in Figure 4.7 (McCollum 1989; Afshar and Allaire 1996; Nandi and Jogai
2012). Below the melt line, the reaction between molten aluminium and the silica in
the refractory to form corundum and elemental silicon stops because a surface layer
of corundum forms. The corundum layer prevents continued contact of the metal
with the refractory beneath the layer and stops the reaction. When zinc or magne-
sium levels are high in aluminium melt, the corundum layer is less adherent, and
consequently, refractory corrosion could continue.
The conditions are different at the bellyband. The molten metal penetrates into
the pores of the refractory and reacts directly with the oxygen present in the adjacent
furnace environment to form corundum. As the temperatures are higher at the belly
band, the reaction occurs faster, causing the metal to rise upward through the pores,
forming corundum at locations above the melt line, as shown in the figure. The
corundum layer is a composite with fine grains of corundum surrounded by an inter-
connected metallic network. The metallic aluminium network provides the channels
to supply fresh aluminium to the reaction interface during the corrosion process.
The corundum deposit is strongly adherent to the refractory. Besides, the deposit can
grow so large as to even reduce the furnace capacity.
Incidentally, constant or prolonged contact between the melt and the aluminosili-
cate refractory does not appear to be necessary for such product build-up. Brondyke
(1953) noted that build-up also appears on the upper side walls and roof (wherever
the aluminium melt would contact the refractory) due to splashing of the melt during
charging or gaseous fluxing of the metal.

FIGURE 4.7 Bellyband (From Nandi and Jogai 2012, Engel 2015, Schlesinger 2006).
Major Melt–Crucible Systems 123

At the metal line, corrosion of the refractory is fast and [Link] and
Allaire (1996) noted that the morphology of oxide layers depends upon the oxygen
in the environment. Under air, the oxide layer was porous, voluminousand non-pro-
tective, while under a nitrogen atmosphere, the oxide layer is thin and more com-
pact. When aluminium contains magnesium, the oxide formed is doped more or
less by MgO and is no longer protective. The use of higher-quality refractory and
eliminating free silica in the lining were measures recommended by Afshar and
Allaire (2001) for reducing corundum formation. A high-densityAl2O3 ceramic rod
displayed very good resistance to corrosion as well as to metal penetration in labora-
tory tests.

4.7 MANAGEMENT OF ALUMINIUM PENETRATION

The options available to counter the ability of the aluminium to wet and subsequently
penetrate and react with the refractory are: (i) the use of anti-wetting additions to
refractories, (ii) to use a phosphate bond, because it imparts good non-wetting prop-
erties to the refractory, and (iii) the use of sol–gel binders. The additional options
(Engel 2015) are increasing the ratio of alumina/silica in refractories, decreasing the
content of alkali oxides, magnesium and zinc, and lowering the pore size and the
porosity of refractories.
The dimensions of the pore through which the melt cannot penetrate due to the capil-
lary effects vary depending on the temperature, structure and composition of refracto-
ries, as well as on the type of melt. The critical pore dimension (below which the melt
cannot penetrate due to the capillary effect) for Al melt is <1–2 μm (Yurkov 2017).

4.7.1 Anti-Wetting Additions
The use of anti-wetting additions to refractories in contact with molten aluminium
is a popular approach, especially for castables containing calcium aluminate. Many
different materials have been used for this purpose, and the details remain propri-
etary. Hence, the exact mechanism(s) of their action remains uncertain (Engel 2015).
The anti-wetting additives are added to the refractory mix during manufacture
(Afshar and Allaire 2001). Barium sulphate is the most common additive. Aluminium
fluoride and calcium fluoride (AlF3, CaF2) are also used (Adabifiroozjaei et al. 2015),
as well ascertain boron- and phosphorus-containing materials. The decrease of
wetting due to anti-wetting additive is not an absolute protection. These additives
decompose at a certain temperature (depending on the additive) and once this is
exceeded, the refractory loses their protection. Yet, these anti-wetting additives do
add years of service life to the refractories.
Ba-based anti-wetting additives include BaSO4, BaO and BaCO3, Ca-based addi-
tives are CaO and CaF2, B-based additives are BN, B4C, B2O3 and 9Al2O3⋅2B2O3),
and P-based additives are AlPO4 and glass frits. In addition, compounds such as
Cr2O3, AlTiO5, cryolite, CeO2, AlF3 and ZrSiO4 have also been tried.
The anti-wetting additives act by increasing the interfacial energy between the
refractory and the aluminium melt through the formation of a new phase or phases
124 Metal–Crucible Interactions

(Adabifiroozjaei et al. 2015) by reactions between the additive and the refractory.
Increasing the interfacial energy lowers the tendency of melt to wet the surface of
the refractories. Even in non-wetting systems, aluminium and its alloying compo-
nents tend to diffuse through the interface to equilibrate. Here, anti-wetting mate-
rials may result in the formation of dense and rigid interfacial layers that retard
the diffusion.
Ba-containing compounds are effective additives. The effectiveness is attributed
to the formation of celsian, a monoclinic BaAl2Si2O8, in the matrix of the refrac-
tory. Celsian forms by reaction between the additive and the refractory components
(Al2O3 and SiO2) at temperatures exceeding 1000°C (Afshar and Allaire 1996, 2001).
According to Aguilar-Santillan (2008), BaSO4 additions also resulted in a substan-
tial decrease in the number of pores in the refractory.
Afshar and Allaire (1996) emphasize that corrosion of the aggregates was insen-
sitive to the non-wetting additive, and the beneficial effect of the non-wetting agent
was present only in the matrix. In some cases, the corrosion of aggregates promotes
corrosion in the surrounding matrix, even in the presence of a non-wetting agent
(Nandi and Jogai 2012). At temperatures below 1000°C, barium compounds react
with SiO2 and Al2O3 to form α-BaAl2Si2O8 (orthorhombic phase), which was cred-
ited with the protection observed (Schmutzler and Sandhage 1995). Above 1050°C,
the orthorhombic phase transforms to a monoclinic phase, and the protection is lost.
Incidentally, in their review, Adabifiroozjaei et al. (2015) attributed the non-wetting
behaviour to the monoclinic structure of BaAl2Si2O8. There are also conflicting
statements as to the effectiveness of celsian vis-à-vis barium sulphate. The current
information on the anti-wetting action of celsian vis-à-vis barium sulphate remains
contradictory. Adabifiroozjaei et al. (2015) consider monoclinic celsian to be confer-
ring superior anti-wetting properties on the refractory, while many other authors
consider barium sulphate to be superior.

4.7.2 Phosphate Bonding
A phosphate binder imparts good non-wetting properties to the refractory and does
not decompose until temperatures >1500°C (Engel 2015). In brick, it can be incor-
porated into the mix itself prior to firing. In castables, it can be the bonding agent
or supplement it, and in mouldables (plastics), it is the binder. Wollastonite (calcium
silicate, CaSiO3) is not wetted by Al and has also been considered as an anti-wetting
additive (Vikulin et al 2006) useful below 1125°C.
Generally, the joints in the refractory lining using bricks are particularly prone
to melt penetration and attack. Monolithic refractory linings have no joints but are
mechanically weaker. Both bricks and monolithic materials are used in furnaces
(Nandi and Jogai 2012). Phosphate-bonded materials are known for good resistance
to aluminium metal penetration and corundum growth. The addition of non-wetting
additives that tend to decompose at temperatures above 1000°C is useful in low-
temperature processed materials. Lower firing temperatures create smaller pore
sizes and improved non-wetting properties, which translate to higher penetration
resistance against alkali.
Major Melt–Crucible Systems 125

The use of sol–gel binders, in particular colloidal silica, results in small pores
that can hinder aluminium penetration (Engel 2015). In such a situation, a non-
wetting component may not be required or may be required in smaller quantities.
Certain special raw materials or binders that have low wettability with respect to
aluminium contain only lime and alumina, e.g. calcium hexaluminate or bonite.
Bonite is a new synthetic, dense refractory aggregate based on the mineral phase
calcium hexaluminate (CaO·6Al2O3 or CA6) (Buhr et al. 2016; Schnabel et al.
2011). Bonadia et al. (2006) noted that materials based on sintered CA6 or bonite
are inherently less wetted by aluminium, do not require anti-wetting additives and
keep their properties even at temperatures >1200°C.

4.8 REFRACTORY INCLUSIONS
Silicon oxide, iron oxide, titanium oxide and sodium oxide, which serve as the base
and constituents of refractories, are deoxidized (reduced) by Al. In the Al melt, there
are always impurities (sodium, potassium and gases) and alloying components (e.g.
magnesium, strontium, zirconium). The corrosion of aluminosilicate refractories (in
melting and holding furnaces) by molten aluminium generally leads to the formation
of an alumina deposit on the refractory. Such corrosion also promotes the formation
of inclusions in the molten metal by erosion. Inclusions can also originate from direct
oxidation of impurities in aluminium at the metal line (Allaire 2000).
The products of oxidation are a function of the alloy composition. At operating
conditions of aluminium melting furnaces, the presence of 0.3 to 18 wt.% magne-
sium in aluminium favours the formation of spinel (MgAl2O4). At less than about
0.3 wt.% Mg, alumina (Al2O3) is the most stable oxide in these conditions, while at
more than 18 wt.%, it is magnesia (MgO). These oxides may result from the action
of oxygen gas on the molten metal, which leads to the formation of a skim at the
metal line. This direct oxidation may also occur at the refractory/metal interface due
to the oxygen gas that migrates through the pores in the refractory. The oxidation of
aluminium by less stable oxides such as silica occurs in the more reactive regions of
the refractory, where vitreous phases surround more stable crystalline phases. The
latter are released to form inclusions in the metal.
Alkalis contained in the refractory (usually <2 wt.% Na2O) are highly suscep-
tible to reduction by aluminium, and sodium is released. AlF3in the refractory, as a
non-wetting agent, also contributes to the release of alkalis into the molten metal.
Sodium in the molten metal oxidizes, and Na2O convertsthe protective alumina layer
into sodium aluminate (NaAlO2), whose density (2.69 g/cm3) is lower than that of
alumina (3.96 g/cm3). Because of this difference in density, the protective alumina
layer cracks, and melt penetration is facilitated.

4.9 REFRACTORIES RESISTANT TO MOLTEN ALUMINIUM

Aluminium oxide, magnesium oxide, barium oxide and calcium oxide are refracto-
ries stable to molten Al. Magnesia bricks were used for the lining of furnaces in the
1960s, but the use did not continue due to their low thermal shock resistance. Yttrium
126 Metal–Crucible Interactions

oxide and zirconium oxide also may be resistant to molten Al. Aluminium nitride is
completely resistant to aluminium even at 2000°C.
Sialon was tested in molten aluminium. The corrosion test was performed in an
electric furnace at 800°C for 150 h in static air. In the experimental arrangement, the
sialon specimen was in contact with (i) air and molten aluminium simultaneously,
and (ii) only molten aluminium (no contact with air). Scanning electron microscopy
(SEM) studies showed no traces of Al in the pores of the crucible. There was no
infiltration. The material showed excellent resistance in contact with molten alu-
minium evenafter extended exposure (150 h) at 800°C. It has been suggested that
formation of AlN scale at the contact point of Al(l) and sialon is responsible for the
very slow corrosion attack of sialon and silicon nitride ceramics (Brook 1991), a ten-
able e­ xplanation in an oxygen-free atmosphere, as it occurred at the bottom of the
crucible, floating inside the Al melt.

4.10 COPPER
Mark Schlesinger (1996), in his review on the refractories for copper production, indi-
cated that the furnaces used for producing molten copper from concentrates and scrap
– flash smelters, converters, and anode and firerefining furnaces – face exemplary chal-
lenges to the refractory life in the form of highly aggressive slags, mechanical stresses,
batch operation and higher operating temperatures. When everything appeared to be
settling down, the use of ‘mag–chrome’ refractories began to be scrutinized and ques-
tioned as being a source of a possible environmental villain, hexavalent chromium. In
a way, the search for a ‘more’ suitable refractory begins all over again.
The copper production flowsheet consists of (i) matte smelting, in which sulphide-
based copper concentrates are melted and partially oxidized to forma silicate slag
and a molten matte. The matte is animpure mixture of sulphides consisting primar-
ily of copper and iron sulphides; (ii) converting,where oxygen-enriched air interacts
with the molten matte to produce molten blister copper and a slag containing most
of the iron; (iii) fire refining, where the blister copper is partially refined in ‘anode’
furnaces and cast into anodes, and (iv) electolytic refining, where relatively impure
copper in the anode is electro refined and recovered as pure copper cathode. There
are also refining furnaces for melting and removing impurities from scrap copper.

4.10.1 Smelter
For matte smelting, as well as reverberatory smelters, electric, submerged tuyere and
flash furnaces are in use. The sulphide mineral concentrates are melted down, and
part of the iron and sulphur in them is oxidized. The products of smelting, coming
from concentrate and from the burning of carbonaceous fuels, are: (i) a gas stream
containing SO2, N2, CO2, H2O and O2; (ii) fayalite slag, Fe2SiO4 (m.p.1200°C), formed
by the reaction of silica in the concentrate and added as a flux, and FeO, formed
on oxidation of iron sulphides. The slag also carries the gangue oxides, Al2O3 and
MgO, and also some (~2%) copper oxide generated during the oxidation reaction;
(iii) the matte – a molten solution of Cu2S, FeS and minor amounts of sulphides of
Major Melt–Crucible Systems 127

lead, nickel, tin, zinc, cadmium and other elements; and (iv) the dust-laden off gas.
Substantial amounts of copper, arsenic, iron, bismuth, zinc and antimony, besides a
few more elements, are carried off in this stream.
In the reverberatory and flash smelting furnaces, as well as in the Peirce–Smith
and top blown rotary converters, there is chemical, thermal and mechanical attack on
the refractory. FeO is an effective flux, and that makes the FeO-laden basic slag the
main factor. In flash smelters, slag lines the reaction shaft, and near the top, because
of prevailing high oxygen partial pressure, FeO oxidizes to magnetite, which is good
because the solid magnetite coating protects the lining underneath. Near the bottom,
such conditions do not exist, and no protective magnetite forms. The lining in this
portion is more vulnerable to slag corrosion. In the flash furnace, the falling droplets
of matte and slag are at a higher temperature and thus can dissolve and wash out the
refractory. In reverberatory furnaces also, the use of oxygen-enriched air pushes up
the temperatures in the roof and side walls and makes the environment difficult for
the refractories. The mixture of water vapour and sulphur dioxide in smelter off gas
is deleterious to the basic refractories, because they can be hydrated by water vapour,
and SO2 can convert them to fusible sulphates.

4.10.2 Converter
In the converter, matte is converted into blister copper and slag, and the conditions for
the refractories are harsher here. Peirce–Smith (PSC) units have been almost exclu-
sively used to convert all the primary copper produced. Enriched air blown through
the tuyeres in the side of the vessel,in the first stage of the blowing process, oxidizes
the iron sulphide in the matte, and the iron oxide formed reacts with the added silica
to form fayalite-based slag. The nearly pure copper sulphide remaining is oxidized in
the second stage to blister copper and slag. In the second-stage blow, a good amount
of Cu2O is also formed and collects in the slag, and this slag is reverted back to the
smelter. In the end, there is still about 0.2% S remaining in the blister copper.
Cuprite or Cu2O is a highly effective flux, and its presence makes the molten
slag very corrosive. Batch operation always implies deep temperature cycling and
mechanical impact due to periodic charging of the converter as well as punching
of tuyeres to unclog them. The refractories thus have to face mechanical damage
and thermal stress components. Refractory erosion by molten phases becomes more
intimidating due to bath agitation by submerged gas blowing (PSC) as well as by
complete rotation of the vessel (Top Blown Rotary Converter or TBRC). The infiltra-
tion into the furnace bottom and virtual freezing of molten copper is another hostile
occurrence special to converters.
The converter yields blister copper, which contains various metallic and non-
metallic impurities, and many of them are removed in a subsequent process known
as fire refining. In fire refining, less noble elements like arsenic, antimony and lead
are preferentially oxidized and removed from molten copper using a basic flux. Fire
refining is carried out in hearth and rotary furnaces. A high concentration of cuprite
also accumulates in the fire refining slag because of the highly oxidizing conditions
used. The melting of scrap copper is also carried out in such furnaces. The second
128 Metal–Crucible Interactions

part of the fire refining process and the final pyrometallurgical step in a typical cop-
per production flowsheet is the deoxidation and purification of molten metal using
hydrocarbon gases or occasionally, by the ancient practice of using wooden poles.
The slags in the refining furnaces are more basic than those in smelters and convert-
ers and have a much higher Cu2O loading.
Fire refining furnaces are operated on a semi-continuous basis and thus, have less
thermal shock issues. Secondary copper smelters, like converters, are batch oper-
ated with all the attendant load on refractories. These are usually rotary furnaces,
where the refractory life is shortened by erosion and mechanical stress as well as
­chemical factors. Converters that (i) operate continuously and (ii) do away with fur-
nace motion or rotation, which would ease thermal shock and mechanical stress fac-
tors on refractories, are still some distance away from routine deployment.

4.10.3 The Mag–Chrome Refractory

Flash smelters dominate the copper smelting facilities. Reaction shaft and slag line
are two areas prone to refractory wear and so, are lined with fused grain chrome–
magnesia bricks. Lower MgO compositions (less basic) are chosen because the SiO2
and/or water vapour in the smelter off gas would attack the MgO grains in higher-
basicity bricks. In areas less prone to slag attack, such as off gas shaft and side walls,
direct-bonded chrome–magnesia refractories are fine.
Silica and alumina refractories were still in use until the 1990s for roofs and side
walls of reverberatory furnaces. These areas do not come into contact with the slag
and matte, and these refractories can cope with the presence of SO2 and water vapour
in the environment. Later, when oxy-fuel burners were positioned in the roof, the roof
temperatures became higher. Because of this, and to deal with the corrosive impact
of dust and fume emanating from the smelting process, mag–chrome bricks, either
burned or directly bonded, began to be used. Beginning in the 1950s, converters have
been exclusively lined with mag–chrome. The best known (available) material is used
to line the tuyere area, where the highest refractory wear occurs. Originally, burned
brick was used. It is now direct-bonded refractory with fused grain (Rigby 2005).
The bricks face penetration by matte and fayalite slag initially, and later, by cop-
per oxide. The use of enriched oxygen leads to higher temperature at the tuyeres and
exacerbates thermal shock, consumes the protective magnetite layer and increases
the solubility of the refractory components in both first (iron blow) and second (cop-
per blow) slags (Schlesinger 1996). Any extra blowing to eliminate hard-to-oxidize
lead and antimony from the blister copper generates more copper oxide,which pen-
etrates the refractories and causes failure by slabbing.
When high-magnesia materials are used in the tuyere area, they react with the
fayalite slag generated during iron blow, forming a grain boundary solid solution of
fayalite and forsterite, which limits the subsequent penetration of matte and copper
oxide. A highly siliceous slag ends up dissolving this solid solution,exposing the spi-
nel grain, and thus promotes both penetration and refractory loss by erosion.
High-chrome compositions (it now becomes chrome–magnesia) better resist lower
temperature corrosion by molten silicates and are used throughout the converter. The
Major Melt–Crucible Systems 129

reaction between cuprite in the slag and chromic oxide in the refractory grain gen-
erates solid copper chromite (CuCrO2), which coats the refractory surface. Further
damage is then limited.
Chrome–magnesia bricks are also favoured in TBRCs due to their high strength
as well as other properties. Lower brick porosity reduces slag penetration. The
reduced thermal shock resistance associated with lower porosity is not critical away
from the tuyere zone.
In anode or fire refining and secondary smelting furnaces with high slag basicities,
the absence of siliceous slags points to the use of highly basic (magnesite) refracto-
ries. Even here, mag–chrome refractories are widely used. The most corrosive envi-
ronment in anode furnaces is the slag line. The combination of high cuprite slag and
air or oxygen here gives rise to the same melt penetration situation that impacts the
refractory life in the converters. Fused grain and fused cast mag–chrome bricks need
to be used in this region. At other places, direct-bonded bricks are good enough.
Oxidized copper penetrates the refractory easily. Thus, tar-impregnated magne-
site bricks are used in the floor. The reducing environment created by the presence
of tar reduces the dissolved oxygen content in the molten copper, which in turn,
reduces its ability to penetrate the refractory itself. Once infiltrated, the copper oxi-
dizes to form solid Cu2O and CuO; the brick grows and eventually fails by bursting
(spalling). In an oxidizing environment, the likely formation of a low-melting-point
eutectic in the Cu2O–MgO system precludes the use of magnesite refractories for
copper containment.
In a typical new-generation copper converter, which is stationary, the turbulence
created by the vertically injected oxygen is used to achieve the mixing action pro-
duced by the horizontal tuyere injection of PS vessels or the rocking rotary action of
TBRC. Mechanical stress on the refractories is thus less. Because the off gas gener-
ated in the new converters is richer in SO2, the use of basic refractories becomes
difficult. Impurities removal from high-grade mattes in newer converters will be
less effective, and the presence of impurities in blister copper can degrade refractory
performance. Besides, the greater reaction intensity in the newer converters coupled
with higher oxygen enrichment will result in higher temperature in the reaction zone.
Beginning in 1990, all solid wastes containing chromium oxide, including mag–
chrome bricks, came to be treated as hazardous material. The concern was over Cr3+
in these materials,which might be oxidized to Cr6+, and Cr6+ is water soluble. Cr6+
formation in refractories results from reactions in oxidizing environments with alka-
line components (Na2O, CaO, etc.) in slag or in the brick itself to form chromates. In
current practice, the chemical environment used in the smelters and converters does
not favour the formation of hexavalent chromium in copper production (Schlesinger
1996). The use of basic fluxes is generally not very prevalent. However, to improve
removal of arsenic and antimony, some refining practices do use sodium carbonate
or hydroxide. The use of calcium ferrite slags in the converting of high-grade mattes
may also result in the generation of Cr6+ in refractories.
The way out, then, is development of a chrome-free refractory suitable for
replacement of mag–chrome bricks in copper production furnaces. The clue is the
MgO–Al2O3 phase diagram, whichis similar to the MgO–Cr2O3 phase diagram.
130 Metal–Crucible Interactions

Refractories based on mixtures of MgO and Al2O3 could be suitable replacements

for mag–chrome (Schlesinger 1996).
Magnesia-alumina refractories may be as effective as mag–chrome, if not better,
in copper production furnaces. The solubility of MgAl2O4 in molten copper oxide
at 1400°C is slightly lower than that of MgCr2O4. Intergranular penetration of oxi-
dized copper into magnesite refractories can be equally well inhibited by the addi-
tion of Al2O3or Cr2O3 to the brick composition. In calcium ferrite slags, MgO with
20% added alumina corroded at about the same rate as magnesite with 20% added
chrome. Mag–alumina instead ofmag–chrome in almost every copper production
application is a possibility. In smelters, periclase–spinel brick might have better
resistance to the slag penetration. Alumina-rich compositions have better resistance
to sulphation and might be useful in roof and stack locations.
Steel ladle slags and flash smelter slags are similar. Alumina-rich compositions
working well for slag linings might be useful for flash smelters as well, but the
resistance of mag–alumina refractories to attack by high-cuprite slags needs to be
checked and validated. With increasing use of calcium ferrite slags in converting
operations, periclase–spinel or spinel–periclase might play a wider role.
Mitsubishi introduced the use of a calcium ferrite slag for converting. In com-
parison to fayalite, calcium ferrite slag has a lower viscosity and also a lower
solubility of Cu2O. Its higher basicity is conducive to the removal of impurities
that form acidic oxides. However, due to the low solubility of silica in this slag,
the use of this slag in smelting processes is restricted. Aided by the low viscosity
of calcium ferrite, magnesia–chrome bricks could undergo significant refrac-
tory penetration and wear (Fahey et al. 2004). The Ca-ferritic slag corrodes not
only the basic magnesia component but also the chromite and forms low-melting
Mg-Feoxide (magnesia–wuestite).

4.10.4 Calcia–Magnesia
In nature, calcia and magnesia occur as the mixed carbonate dolomite. On calcina-
tion, dolomite forms an oxide mixture of periclase and lime or doloma. It contains
about 60% CaO. The manufacture of refractories from this had to use additives like
iron oxide, silica or alumina to produce a lower-melting bonding phase. The tem-
peratures required for direct bonding are impractically high for any manufacturing
facility. The addition of binders is also needed to improve the density of the fired
product, since doloma hydrates easily. Tar or pitch can also be used as a binder
instead of acidic oxides. The tar has multiple roles: it not only minimizes hydration
during storage but also serves as a reducing agent in contact with molten metal. For
instance, the removal of dissolved oxygen from copper blunts its ability to penetrate
grain boundaries.
Oxidation of the matte generates SO2, which can permeate into the brick. The
SO2 subsequently oxidizes to SO3, which in turn, reacts with the basic oxides of the
chrome–magnesia brick to form the alkaline earth sulphates, MgSO4 and CaSO4, at
temperatures below 1050°C. These reactions are associated with volume increase,
which also causes filling up of the brick pores. The densification of the refractory
Major Melt–Crucible Systems 131

structure and the weakening of the brick bond eventually cause cracks in the bricks,
facilitating melt infiltration.
If metallic copper permeates into the refractory, there are two consequences. One
is an increase in thermal conductivity of the brick and as a consequence, substan-
tially higher temperatures at greater depths of brick. This will affect both corrosive
and thermal reactions. Incidentally, pure copper scarcely infiltrates the brick; oxygen
contamination facilitates penetration through its effect on the contact angle between
Cu and the refractory oxide.
Another possibility is mixing pure magnesite with doloma during refractory pro-
duction, generating a mag–doloma brick analysed at 60% total MgO. Mag–doloma
bricks hydrate less easily and resist slag corrosion better than pure magnesia.
Doloma and mag–doloma might be viable options for refining furnaces with their
highly basic slags and highly oxidized copper. The use of dolomitic refractories in
BOF linings subjected to similar temperatures suggests that these materials would
be competitive.
Malfliet et al. (2014) have summarized some consequences of melt infiltration
into refractories used for copper production. Melt infiltration occurs due to the pen-
etration of molten slag, copper sulphides and oxides, and the reaction products that
are formed. The acidic character and low viscosity of the slags, high temperatures
and the presence of additives such as soda aggravate the slag corrosion and make it
one of the main wear mechanisms of refractory in the copper industry. The slag dis-
solves specific phases in the bricks’ microstructure.
Slags in the primary smelting and converting processes are of the ferrous silicate
type. On contact with fayalite slag, refractory corrosion occurs through the dissolu-
tion of MgO, more so when the silica content of the slag is higher, the reason being
that periclase (MgO) is more basic than chromite and therefore, more susceptible to
acidic corrosion. The chromite grains do not react with the slag but are washed away
once they are unbound. Refractories that are higher in chrome ore and are direct-
bonded with a spinel bond rich in Cr2O3 are more resistant to chemical corrosion.
A high SiO2 supply caused by changes in the processing and/or the uncontrolled
addition of silica sand results in the copious formation of forsterite (Mg2SiO4) fol-
lowing contact with periclase, which due to the associated volume expansion, causes
forsterite bursting and deterioration of the brick structure.

4.10.5 Olivine Slags
A slag with properties midway between those of fayalite and calcium ferrite slags
is the ferrous calcium silicate or olivine slag used in the Ausmelt C3 converting
process. Yazawa (2001) proposed this slag to reduce the copper loss, increase the
absorption of both basic and acid oxides, and limit refractory problems.

4.11 LESS COMMON METALS

The unique technology benefits of a large group of metals known collectively as
rare metals or less common metals could be accessed only when methods were
132 Metal–Crucible Interactions

developed for their production in sufficient purity and for processing in good quan-
tities. Reflecting their unique nature or potential application areas, groups among
them have been referred to as reactive and refractory metals, nuclear metals and rare
earth metals. These metals were little more than laboratory curiosities but emerged
into prominence in the post-World War II period, and their development was driven
by the nuclear, aerospace, electronics and biomedical industries. Interestingly, even
today, these metals remain less available and less familiar to metal users at large.
They are notoriously reactive and need elevated temperatures for their production
and processing, and the task of handling almost every member of the less com-
mon metals group has been filled with challenges. The process for the production of
many of these metals is metallothermic reduction, also known as metal replacement
reaction. This is dictated by sound science and technology reasons as well as cost
considerations.

4.12 METALLOTHERMY
In contrast to the ferrous as well as the common non-ferrous metals, the less com-
mon metals are prepared not from beneficiated ores or concentrates but from inter-
process intermediates, which are invariably a pure compound usually obtained after
elaborate separation procedures. The objectives of the metallothermic reduction
have been summarized by Herget (1985) as follows: (i) the reaction should occur
quickly and give product metal in high yields, (ii) the reaction products must be
obtained in compact forms – metal as an ingot and the slag as a well-separated layer,
(iii) the reaction should be self-sustaining and once initiated, should proceed without
the need for additional external heating, (iv) the product metal must be of high purity,
(v) the reaction proceeds in an open atmosphere, (vi) the reaction should proceed
safely without risk, and (vii) the reaction should be amenable to being carried out
in commercially available reactors using readily available refractories as contain-
ers. Metallothermic reductions are generally exothermic reactions, and by a suitable
choice of reactants, fluxes and fluidizers, and process conditions, it has been possible
to realize many of the listed characteristics to a great extent and obtain the metal
from its compound. To repeat, refractories have played a key role in the implementa-
tion of metallothermic reductions.
For a host of technical reasons, the compound chosen is either an oxide or a
fluoride or a chloride. Other things (such as reducibility, purity of as-reduced metal
and possibility of post-reduction purification) being equal, an oxide is a convenient
choice and invariably preferred. Between the fluoride and chloride, the chloride is
the more versatile intermediate but is difficult to handle because of its volatility,
hygroscopicity and corrosive nature. Fluorides undergo a neat reduction but do lack
certain advantages special to chlorides. All these compounds, with very occasional
exceptions, are reducible under technologically useful conditions by metals such as
aluminium, calcium, magnesium, sodium, potassium and lithium. Reduction using
a metal invariably needs a high temperature, attained either by external heating or
by the exothermicity of the reaction or by a combination of both. Thermal manage-
ment of the process is an integral part of using metal as the reducing agent. Besides,
Major Melt–Crucible Systems 133

with very few exceptions, all these reductions have to be carried out under a pro-
tected atmosphere (vacuum or inert gas) with inert containment arrangements to
protect the compound intermediate, the reducing agent and the as-reduced metal
from contamination.
Metallothermic reduction has been used to obtain metals in a variety of physical
forms, from well-consolidated ingots to a fine powder. If the heat generated during
the reaction is sufficient to raise the temperature of both slag and metal above their
melting points, and if they remain molten for a sufficient length of time, so that
the molten metal (denser) settles by gravity with the immiscible slag layer (lighter)
remaining on top, a metal ingot topped by a solidified slag results on cooling. Slag
and metal are then separated mechanically. When the heat of reaction is insufficient
to result in an all-liquid reaction mixture, other types of products can form. If the
reduction occurs, but neither the metal nor the slag melts, the metal is formed as a
powder dispersed in the slag matrix. The slag in this case is leached away to sepa-
rate the metal powder. Metal in the form of powder can result even if the slag melts
when the melting point of the metal is very high and the metal is not sintered to
any extent at the temperatures involved. When the reduction to metal occurs with
only the slag melting, but not the metal, and the melting point of the metal is not
very high, the metal particles may coalesce and undergo partial consolidation into a
sponge. Here, the metal and sponge are separated either by aqueous leaching or by
vacuum distillation.
By a suitable choice of reductants, reactors and reaction conditions, the metal
can usually be obtained in the chosen form. The form of the product obtained in
the reduction determines how difficult it is to select a suitable container or crucible.
For example, the notoriously reactive metals titanium, zirconium and vanadium are
obtained in the form of sponge at the primary reduction stage when their chlorides
are reduced with magnesium at temperatures lower than 1000°C. The crucibles
usable for reduction are mild steel or Inconel. If the reduction process were to involve
handling these metals in a molten form, choosing a crucible would become a com-
plicated and almost unsolvable problem. The seriousness of a reactive melt increases
first of all with its reactivity and also when the melt forms at a higher temperature,
in larger quantities, and persists for longer durations. The formation of the metallic
melt and its proper containment are invariably necessary when the metal in the as-
reduced condition is obtained in the form of an ingot.
The properties of reactants and products, temperatures reached during reduction,
scale of operation and impurities tolerated in the as-reduced metal – all these deter-
mine the materials of construction of the reactor and the choice of melt facing refrac-
tories, be it a metal or a compound. In the as-reduced condition, less common metals
are obtained in the form of an ingot in two types of reactors. One is the bomb reactor,
which is usually a refractory-lined steel vessel that is closed during the course of a
reduction after loading the charge. Calciothermic and aluminothermic reduction of
oxides and calciothermic and magnesiothermic reduction of fluorides are carried out
in bomb reactors.
In many metallothermic reductions, an adiabatic environment is usually arranged
for a certain duration when the objective is to obtain the product in the form of an
134 Metal–Crucible Interactions

ingot. This is achieved by selecting a reaction that occurs rapidly and providing good
thermal insulation of the system. The total heat input within the reaction volume
through exothermicity of the main reaction, exothermicity of the booster reaction
and external heating (usually as a means of initiating or triggering the reaction) is
carefully conserved and utilized in three ways. The first is the choice of a reactor that
is thermally insulated, such that the heat once accumulated during the reaction is
not lost from the reactor volume before slag–metal separation is complete. A booster
reaction is a reaction of high exothermicity (high kJ per gram of reaction mixture)
that occurs alongside the principal reaction, independently but in the same reaction
mixture. It shares the reducing agent, and the product should flux with the product
of the main reaction to form a fusible and/or fluid slag. Heat is best conserved in
closed bomb reactors, which are essentially steel vessels lined on the inside with
calcia, magnesia or alumina refractory The refractory lining is principally to contain
the melt while it forms and separates into two layers – metallic melt as the bottom
layer and slag as the top layer. The lining resists and is inert to the attack by the melt
constituents, and that is why it is there in the first place. It also stops any rapid loss of
heat by conduction because it is not a good thermal conductor. Second, metallother-
mic reductions, particularly the thermit types, occur very fast. Once triggered, they
take a minute or two to go to completion even for a large batch size. Sufficient heat
for an intended metal–slag separation is generated fast and uniformly throughout
the reaction mass. With a fairly well-insulated system, almost all the heat generated
persists there for molten slag and molten metal to clearly separate into two layers.
Third, the density difference between the metal and the slag is usually large and
conducive to a rapid separation. This is further facilitated by ensuring that the slag
is sufficiently fluid by itself or made sufficiently fluid by using slag fluidizers. These
are the general features of a metallothermic reduction process for the production of
less common metals in the form of an ingot.
Calciothermic reduction of fluorides yielding the reduced metal in a consolidated
form is also carried out by reducing many of the rare earth trifluorides in an exter-
nally heated tantalum crucible under an inert gas cover. In these cases, the heat of
reaction is not a major factor in meeting the thermal requirements of the process.

4.13 REDUCTION OF URANIUM TETRAFLUORIDE

Much of the systematic research on metallothermic reduction for ingot metal produc-
tion was done on uranium, and the results obtained proved to be highly useful leads
for various other metals as well (Wilhelm 1955, 1960; Wilhelm and Peterson 1961).
Uranium is a highly reactive metal and forms stable compounds with a num-
ber of non-metals. The flowsheet of uranium does not normally involve extensive
post-reduction purification, beyond vacuum melting. It is therefore necessary that the
metal in the as-reduced condition is pure.
A lining of dolomite was originally used, but a thick lining of calcium or mag-
nesium fluoride can be used as the refractory lining if it will not collapse through
melting when in contact with molten slag of the same composition. Calcium fluoride,
generated as a product of the calciothermic reduction of uranium tetrafluoride, was
Major Melt–Crucible Systems 135

routinely used as a lining in uranium metal reduction reactors. The lining prevents
the contact between uranium melt and the steel crucible, and using the slag itself
as lining eliminates the possibility of ingestion of impurities into metal from other
lining compounds. This may seem similar to the popular process of slag coating of
BOF vessels. The idea ofusing the slag for furnace lining appears to be much older.
Centuries ago, moulded slag blocks were used to build the shaft of Swedish iron-
making furnaces at Soderfors.
After placing the charge (UF4 + 10% excess Ca) into the lined reactor, a reaction
is initiated from cold by electrically heating a small coil of molybdenum or tungsten
embedded in the charge or trigger mixture (Mg + Na2O2) placed in the charge. An
arrangement is shown in Figure 4.8. The run, which is over in minutes, yields a
clean ingot underneath a solidified layer of slag. It is recovered without necessarily
destroying the lining. The CaF2 slag obtained in the process is crushed, ground,
screened and reused for lining the reactors in the next heat. The reactor may be
cooled by circulating water through the cooling coils brazed on the outer wall of the
reactor vessel.
Calcium reduction may also be carried out in an open bomb and the charge ignited
at room temperature. A protective cover of calcium vapour backed by a CaF2 layer
prevents the product metal from contacting the air for oxidation. Uranium ingots
thus obtained are usually remelted in a graphite crucible in a vacuum induction fur-
nace and cast into cleaner ingots.
Both in operation and in the quality of the metal produced from it, calcium reduc-
tionis entirely satisfactory. However, it has been largely replaced by the lower-cost
magnesium reduction process, which gives an equally good product.
In the technique of magnesium reduction of uranium tetrafluoride developed in
USAEC (United States Atomic Enegy Commission) cylindrical reactors lined (25

FIGURE 4.8 Reactor for uranium tetrafluoride reduction with calcium.

136 Metal–Crucible Interactions

FIGURE 4.9 Closed bomb reactor for uranium production (From Wilhelm, H.A. 1960).

mm thick) with electrically fused dolomite are used (Jamrack 1963). (Burned lime
and MgF2 are also suitable lining materials.) The reaction is initiated by heating the
entire assembly in a gas-fired furnace at 365°C for about 3 h. Unlike calcium reduc-
tion, the option of carrying out the reduction in an open bomb does not exist for mag-
nesium, because apart from limited exothermicity, the boiling point of magnesium
(1091°C is lower than the melting point of uranium (1132°C). It has to be a closed
reactor. A schematic is shown in Figure 4.9.
Graphite is a suitable lining for a magnesium reduction process. Graphite tends to
react with calcium, and the product ends up with the carbon impurity. Problems due
to high thermal conductivity and specific heat are circumvented by using only a thin
graphite lining, separated from the reactor wall by a gas space or a loose packing
of insulating powder (reminiscent of the composite lining – barrier refractory and
insulation layer – used in steel and aluminium metallurgy). The process is executed
like a typical magnesium reduction. The ingot, slag and graphite come out separated
fairly cleanly. No significant metal–crucible interaction happens, and the graphite
assembly is largely reused.

4.14 REDUCTION OF THORIUM TETRAFLUORIDE

Thorium melts at 1750°C, approximately 620°C above the melting point of uranium,
and is the highest -melting component of the reaction mixture. Calcium fluoride
melts at 1418°C. Sulphur is used as the booster with a large excess of calcium. The
reaction between calcium and sulphur is highly exothermic, and the heat evolved is
available for meeting the thermal requirements of the main reduction process. The
bomb-type reactor is lined with graphite. The reactants are compacted together to
form pellets and stacked inside the graphite liner. The reaction is initiated by means
of an electrically heated molybdenum coil embedded in a trigger mixture of a few
grams of calcium and sulphur. Well-formed thorium billets are obtained in over 90%
yield. This was the UK process. Average impurity contents of billets produced were
Major Melt–Crucible Systems 137

as follows: oxygen 0.26%, nitrogen 0.038%, carbon 0.24%, sulphur 1.0%, fluorine
0.33%, calcium 0.21%.
In the American process, the reaction takes place in a closed steel vessel, which is
lined with electrically fused dolomite or lime. The charge mixture (ThF4, ZnCl2, Ca)
is designed to yield Th–6%Zn alloy, which melts at 1200°C. Post reduction, the zinc
is distilled off from the Th–Zn alloy under vacuum between 1000°C and 1100°C
in graphite pots. The development of the process, focused on keeping the reaction
temperature as low as possible, concurrently eliminated crucible or lining reactivity
with thorium from developing into an issue.

4.15 REDUCTION OF OXIDES
The thermit process, invented by Hans Goldschmidt over a century ago (Goldschmidt
and Vautin 1898), has been the framework for all future commercial processes for
the production of vanadium and niobium metals and their alloys as well as numerous
low-carbon ferroalloys.
The first ingot of vanadium metal was prepared by a process involving calcium
reduction of vanadium oxides. No calciothermic reaction for ingot production aims
to melt CaO (m.p. 2570°C) unaided. Iodine and sulphur are highly suitable for boost-
ing calcium reduction of vanadium oxides. Not only does the heat thus released add
to the heat released by the oxide reduction, but the product, calcium iodide or cal-
cium sulphide, fluxes the main CaO slag, there by reducing the temperature needed
for a molten slag.
In 1954, Beard and Crooks carried out the calcium reduction process on a
kilogram scale in a reactor similar to that shown schematically in Figure 4.9. A
tapered magnesia crucible was placed inside the steel crucible to serve as the liner.
Vanadium ingot produced by this method was about 99.7% pure, and the yield
of metal was 84%. Oxygen and nitrogen contents were about 0.03% and 0.01%,
respectively. In a later report, Chindgren et al. (1963) noted that magnesia crucibles
tended to contaminate the vanadium metal obtained with iron and silicon. The ves-
sel lining was changed to one formed in place by tamping magnesia powder, rela-
tively iron and silica free, around a tapered wooden mandrel that served as a form.
Such a lining also served as insulation against excessive heat loss and protected
the reactor vessel (steel) from attack by molten vanadium and slag. This method of
lining the bomb reactor was generally the standard in all subsequent bomb reduc-
tion experiments everywhere.
Sodium is not generally used as the reducing agent due to its rather low boil-
ing point (880°C), and potassium is too expensive. Magnesium and calcium are
favoured. Magnesium is the cheaper of the two, but reactions with magnesium also
form an extremely high-melting oxide, MgO (m.p. 2852°C), as slag. The melting
point of calcia is also high at 2572°C. Aluminium melts at a relatively low tempera-
ture (660°C), and the boiling point is fairly high (2470°C). The oxide formed during
reduction, Al2O3, melts at a relatively low temperature for a refractory oxide (m.p.
2072°C). Besides, alumina can be readily fluxed with CaO and CaF2, and the molten
slag can be made very fluid.
138 Metal–Crucible Interactions

Reactors for metallothermic reduction have usually taken simple geometrical

shapes and are made from or lined with a small selection of refractory materials:
lime, magnesia, electrically fused dolomite, calcium fluoride and graphite. On
another level, a water-cooled copper vessel has also been used as a reactor for metal-
lothermic production of ingot metal/alloy. The third category of reactors is simple
crucibles, usually of refractory metals, used in a controlled environment. They are
mainly, or almost exclusively, used for the preparation of rare earth metals.
Even though early investigations on the preparation of consolidated vanadium
relied on calciothermic reduction of oxides, eventually, aluminothermic reduction
turned out to be more convenient not only in the reduction stage but also for post-
reduction purification. The seminal work on aluminothermic reduction of Group V
metal oxides was published from Ames Lab in the same year, 1966, for both vana-
dium and niobium (Carlson et al. 1966; Wilhelm et al. 1966).
In the experiment described by Wilhelm et al. (1966), the Al2O3 slag was dis-
carded; however, it could very well be used in subsequent bomb reductions as
crucible liner material. As in the case of aluminothermic reduction of vanadium
pentoxide, no booster or slag fluidizer was used in the case of niobium reduction
either. The reactor was essentially the same as that used for vanadium – steel shell,
alumina lining and closed (Figure 4.10).
The simplest working arrangement in aluminothermic reduction is exemplified by
the process known as open aluminothermic reduction for the production of niobium
metal. A schematic of the reactor appears in Figure 4.11 (Kamat and Gupta 1971)
The reaction was triggered, with the entire charged reactor at room tempera-
ture, by electrically heating a nichrome coil embedded in the centre of the charge.
The reduction is very intense, with good evolution of white fumes, and is over in

FIGURE 4.10 Bomb reactor for vanadium and niobium (From Carlson, O.N., Schmidt,
F.A., and Krupp, W.E. 1966).
Major Melt–Crucible Systems 139

FIGURE 4.11 Open reactor for aluminothermic reduction (From Kamat, G.R. and Gupta,
C.K. 1971).

minutes. It was usually possible, immediately (say 30 seconds to a minute) after

triggering the reaction, to shovel more charge (Nb2O5–Al mixture) into the reactor
to continue the reduction and more fully utilize the reactor volume. After overnight
cooling, the thermite (Nb–4Al–0.3O in wt.%) topped by an alumina slag is easily
recovered from the reactor (invert and lightly tap).
It may be recalled that the great exothermicity of the UF4 + Ca reaction permitted
the production process to be carried out in an open reactor also. The evolving fumes
and slag cover protected the molten metal from any atmospheric contamination. The
exothermicity could afford the additional loss of heat that happened due to the open
reactor and still keep enough heat for slag–metal separation. The situation in open
aluminothermy is somewhat similar. The exothermicity of Nb2O5–Al is good enough
for the reduction reaction in an open reactor. Adequate slag–metal separation was
still occurring. The oxygen contamination in the thermit was also not excessive.
Metal yield was also good at 98%. The only significant difference is that use of 15%
excess aluminium in the charge resulted in only 4% aluminium loading in the ther-
mit. In a closed bomb reduction, the corresponding figures were 5% and 2%. Some
of the aluminium undergoes atmospheric oxidation and is lost in the open reduction.
This loss is avoided in the closed bomb.
Vanadium can also be produced smoothly by open aluminothermic reduction.
Mukherjee and Gupta (1972) ventured to use this process in a sequence that would
end up with pure vanadium metal. An open-top, cylindrical, mild steel vessel (150 mm
diameter by 300 mm high) lined with alumina (50 mm thick) was the reduction
reactor.
Oxygen is an unwanted, unavoidable but removable impurity in vanadium.
Nitrogen is an also unwanted, avoidable but not easily removable impurity in
vanadium. Any open (to air) process will result in nitrogen ingestion, and open
140 Metal–Crucible Interactions

aluminothermic reduction is not an exception. Nitrogen does not leave (degas from)
vanadium during electron beam melting. Only fused salt electrorefining works. If
the electrolysis process does not integrate well into the process flow sheet, it is best
to avoid nitrogen pick-up in vanadium, which is possible. This is the reason for all
commercial processes for vanadium using a closed bomb aluminothermic reduction.
(Typical nitrogen pick-up: closed refractory-lined bomb reduction 0.006 wt.%, open
refractory-lined bomb reduction 0.076 wt.%)
A significant development in carrying out aluminothermic reductions involves the
use of a water-cooled copper crucible instead of a refractory-lined steel bomb, a tech-
nique pioneered by Perfect (1967, 1981), as already mentioned in Chapter 1. Such
reactors enable reductions to be conducted under high-purity conditions (Carlson
et al. 1981), because impurity pick-up from the refractory liner is avoided, and evacu-
ation of the reactor volume and backfilling with pure inert gas is easier and effective
to produce vanadium thermit under sterile conditions starting from very high-purity
raw materials. The key impurities in thermit were (in wt.%): carbon 0.003, iron
0.006, nitrogen 0.002 and silicon 0.002. It has thus been possible to maintain a close
control of the purity of the environment (crucible+cover gas) during these highly
energetic reduction processes in a water-cooled copper reactor.
Many of the high-melting rare metals are produced relatively easily in the form
of powder, even on an industrial scale, using an oxide or halide reduction by cal-
cium, sodium, magnesium or hydrogen. These are relatively low-temperature pro-
cesses and involve no liquid metal contact. The occasional use of refractory barriers
between the reacting mass and the reactor material suffices to circumvent potential
crucible problems (Jamrack 1963).

4.16 REACTIONS IN THE ABSENCE OF METAL VAPOUR

The reactions are carried out at relatively low temperatures, and no melt phases are
involved. Crucible problems are usually not serious. A variety of refractory metals
have been directly obtained in powder form (Hampel 1961; Jamrack 1963). For the
oxide–calcium reaction for metal powder production, the reactants, oxide powder
and calcium shreds are pressed together into pellets several inches in diameter and
H/D =1. The pellets are stacked one on top of another, on a plate of sintered alumina
or a thin layer of powdered lime inside a simple mild steel reactor, and reacted at
1000°C for long enough under argon.
For the production of thorium powder, kilogram batches of calcium mixed with tho-
rium oxide loaded in nickel trays are passed through a furnace at 1000°C under argon
flow. A steel crucible lined with molybdenum sheet can also be used for the reaction.
Sodium reduction of fluorides has been used to produce metal powders in unlined
simple mild steel or stainless steel reactors. The powder product did not have any
significant iron contamination. The charge, alternate layers of K 2TaF7 crystals and
sodium cubes, is stacked between deep layers of NaCl flux in a stainless steel bomb,
which is heated externally to about 450°C to initiate the reaction.
An alternative method has also been used on a pilot scale for the production
of capacitor-grade tantalum powder by sodium reduction of potassium tantalum
Major Melt–Crucible Systems 141

fluoride (Jain et al. 1971; Sundaram et al. 1992), in Inconel-lined reactors at 925°C.
Sodium is metered into a stirred melt of K2TaF7 at 925°C. From the reacted mass,
tantalum powder is eventually separated.
Silica ware has been widely used in processes involving contact with metal chlo-
rides as well as hydrogen chlorides and chlorine. Molybdenum trays loaded with thin
layers of the relatively inert black solid NbCl3 are heated to 800–1000°C in silica
tube furnaces in hydrogen. In production facilities, multiple tubes each 3.5 m long
by 150 mmdiameter, filled with molybdenum trays, are used to produce quintals of
niobium metal powder. Molybdenum is very suitable for use in a chlorine environ-
ment as long as oxygen and water vapour are kept out.

4.17 KROLL REDUCTIONS
As preparation methods, heating a compound precursor in the vapour of a reducing
agent or melting the metallic reducing agent in the vapour of the compound precur-
sor often make their appearance, not only in exploratory experiments but also in
industrial processes. Molten magnesium (b.p. 1091°C) reduces zirconium tetrachlo-
ride vapour in the well-known Kroll process. Lithium (b.p. 1330°C) and sodium (b.p.
883°C) are also used in chloride reductions (Jamrack 1963; Hampel 1961).
A metal is obtained in the form of a sponge when its compound (e.g. chloride)
reacts with a molten metal reductant (e.g. magnesium or sodium) at a temperature
(e.g. 850°C) sufficient to melt the resulting slag but not the metal itself. The reduced
metal, initially formed in a finely divided form, coalesces in the molten slag and
partially sinters into a sponge-textured material.
Magnesium chloride and sodium chloride are suitable slags due to their low melt-
ing points (712°C and 804°C, respectively) and low viscosity. Slag removal from the
metal sponge is relatively easy – draining, distillation or leaching.
This method of metal making was originally developed by Kroll (1940, 1956, 1959)
for titanium using titanium tetrachloride and magnesium. The technique has since
been applied extensively to zirconium (Kroll et al. 1946, 1947, 1948, 1950)and also
experimentally to vanadium and niobium (Hampel 1961; Jamrack 1963). This type of
metallothermic reduction, the Kroll process, is perhaps the most important and also
the most widely used. Metal–crucible interaction is generally not a major issue in Kroll
reductions if temperatures are controlled, because molten metal–crucible contact does
not happen. However, the reactive nature of chlorides results in the formation of the
chlorides of certain impurity elements present in the crucible or reactor construction
materials; these chlorides are co-reduced, and the impurity finds its way into the as-
reduced metal sponge. It is impurity ingestion from the crucible to the reduced metal
but happening through the vapour phase – more of a chemical transport process.

4.17.1 Reduction Plant
The metal chloride reduction is always carried out in a closed reactor (Figure 4.12) in
an argon atmosphere to keep out oxygen (and nitrogen). The reaction is exothermic,
and it is necessary to limit the rate of reaction so that the temperature does not rise
142 Metal–Crucible Interactions

FIGURE 4.12 Titanium reduction furnace (From Darling, A.S. 1990).

to 1100°C (> the boiling point of magnesium) or fall too low, i.e. below the melting
point of magnesium chloride (712°C). When a crust of MgCl2forms, it blocks the
chloride vapour from accessing magnesium. The temperature is, in practice, limited
to between 720 and 920°C. Besides, alloys are formed between the zirconium or
titanium (sponge) and the steel (reactor) at about 1000 to 1100°C, resulting in a fairly
rapid penetration of the reactor or at least contamination of the sponge with iron.
The reaction vessel is usually constructed of mild steel, 25/20 nickel/chromium steel
or stainless steel except in the case of niobium production, because niobium penta-
chloride is reduced by metallic iron even at quite low temperatures. A molybdenum
liner is sometimes used as a protection from possible interaction (due to temperature
excursions) between the steel reactor and the metal produced.

4.17.2 Magnesium Reduction of Zirconium Tetrachloride

Magnesium reduction of zirconium tetrachloride was developed at the U.S. Bureau
of Mines under the direction of W. J. Kroll. Compared with titanium reduction, the
Kroll process for zirconium is procedurally elaborate. Zirconium chloride is sub-
limed (it is a white solid that sublimes at 331°C) into molten magnesium, whereas
liquid titanium tetrachloride was metered into molten magnesium. A suitable
arrangement is incorporated for subliming zirconium tetrachloride in a controlled
manner. Three heating zones are provided for in such furnaces.

4.17.3 Bimetal Reduction
Reduction of ZrCl4 using a mixture of sodium and magnesium metal is more advan-
tageous than using only magnesium. The reduction is carried out initially with
sodium at 600°C (10% of the ZrCl4 is reduced with sodium) and then with Mg
(for the remainder of ZrCl4) at 800°C. Bimetal reduced sponge separated by pyro
Major Melt–Crucible Systems 143

vacuum distillation is purest; the impurity concentrations (Cr,Al, Fe, Mg) are lower.
The reasons are the gettering ability of sodium, lower solubility of Fe in Na (Fe from
the reactor/crucible), better fluxing ability and hence, scavenging action of NaCl with
impurity chlorides (FeCl3, AlCl3) and better separation of MgCl2–NaCl slag (more
fluid than MgCl2 alone). NaCl is a better conductor of heat, and temperature excur-
sions in the reactor are reduced, resulting in purer sponge.

4.17.4 Crucible Contamination
The purity of sponge is also dependent on its location in the reduction crucible. Ti
sponge at the bottom of the crucible is less pure compared with that in the middle
of the crucible. The metal in the top periphery is fine and pyrophoric. The reasons
for variations in purity and form are variations in local reaction conditions and also
extent of contact with the crucible.
Vanadium sponge was produced by the Magnesium Electron Company, United
Kingdom, at a 20 kg batch size by magnesium reduction of vanadium trichloride in a
mildsteel reactor. After reduction and distillation, vanadium was obtained as a single
piece of fairly soft sponge. Vanadium from near the crucible wall has a higher iron
content (Jamrack 1963).

4.17.5 Magnesium Reduction of Niobium and Tantalum Pentachlorides

While the steel reactors worked well enough for titanium, zirconium and vanadium
in implementing the Kroll-type process for metal sponge production by chloride
reduction, vapour-phase corrosion of the steel reactor derailed the process in the case
of niobium and tantalum (Jamrack 1963).
Niobium pentachloride vapour attacked the metallic iron of the reactor, forming
niobium trichloride on the reactor wall and releasing ferric chloride into the vapour
phase. On contacting the molten magnesium, ferric chloride was reduced to iron,
which entered the product. Nickel-plating of the steel reactor did reduce the iron
content in niobium, but a reactor made entirely of nickel would solve the problem of
iron contamination. The results of a separate investigation on tantalum pentachloride
were similar.

4.18 LITHIUM REDUCTION OF RARE EARTH TRICHLORIDES

The rare earth metals are, as a group, extremely reactive and have the reputation of
being difficult to prepare. The reactivity of rare earth metals really shows up at high
temperatures and when they are molten.
Nolting et al. (1960) and Moriarty (1968) used reduction of rare earth ­trichloride
by lithium vapour to prepare yttrium and other high-melting rare earth metals.
Nolting et al. (1960) carried out the reductions at 950°C in a stainless steel reac-
tor lined with molybdenum foil (Figure 4.13). The molten rare earth chloride was
contained in a tantalum crucible and held at 950°C for 2 h in the reactor filled with
lithium vapour. The vapour reduced yttrium trichloride to yield yttrium crystals and
144 Metal–Crucible Interactions

FIGURE 4.13 Reduction bomb assembly for lithium vapour reduction (From Nolting, H.J.,
Simmons, C.R., and Klingenberg, J.J. 1960).

lithium chloride slag, which were eventually separated. Yttrium was 99.8% pure
with 0.16% oxygen as the major residual impurity. The as-reduced metal was in con-
tact at any time only with tantalum, and there was no contamination from tantalum.
Metal–crucible interaction is generally circumvented by preparing the rare earth
metal without melting it.
Lithium reduction of rare earth trichlorides was carried outon a commercial scale
by Moriarty (1968). Reduction was carried out at by holding molten rare earth chlo-
ride in titanium crucibles at 800–1000°C in a stainless steel chamber filled with
lithium vapour. The crucible arrangement was in principle the same as that used by
Nolting et al. (1960). The lithium reacted with the molten rare earth chloride, reduc-
ing it to rare earth metal and forming lithium chloride slag. Moriarty (1968) reported
the preparation of all the rare earth metals except promethium, samarium, euro-
pium and ytterbium. Apart from stating that the contamination of the reduced rare
earth metal crystal with the crucible material was very low, no values were given by
Moriarty. The major impurity in these metals was oxygen, which ranged from 225
ppm for holmium to 1600 ppm for terbium.
In a significant effort to reduce residual oxygen in the as-reduced rare earth metal,
Croat (1969) used lithium–5% calcium as reductant in place of pure lithium. Any
oxygen in the lithium vapour was gettered by calcium. The reduction was entirely
carried out in tantalum crucibles inside a tantalum retort. The as-reduced metals were
of 99.94 wt.% or 99.2 atomic% purity. The tantalum contentswere below 30 ppm.
The investigation by Croat (1969) actually served to define the limits to which purity
levels of as-reduced rare earth metals can be raised in a real process.
In the same year as Nolting et al. (1960) published their work on lithium vapour
reduction of yttrium trichloride, Block et al. (1960) reported on the reduction of
yttrium trichloride, but with sodium instead of lithium. The arrangement and pro-
cedure were very similar to the Kroll process for zirconium. Sodium and yttrium
chloride were loaded into two separate molybdenum containers in the stainless steel
retort. Sodium was vaporized and the sodium vapour reduced yttrium chloride. The
Major Melt–Crucible Systems 145

reduction temperature was 850°C. Molybdenum contamination in the yttrium was

lower than 100 ppm in both cases. However iron, chromium and nickel contents were
not insignificant, some probably being derived from the reactor.

4.19 INGOT REDUCTION PROCESSES

The classic closed bomb metallothermic reduction method that was used for pro-
ducing ingots of uranium, thorium and vanadium metals was also used for the rare
earths. The reactors, charge composition and general reduction procedures were
similar. A combination of rare earth chlorides and calcium was used. Rare earth
chloride reductions are not known for exothermicity. A booster reaction, usually
Ca–I, was always used. The oxide liner, sintered lime or dolomitic oxide, was suf-
ficiently inert to molten rare earth metals Ce, La, Pr and Nd only up to about 1020°C
(Spedding and Daane 1961;. Beaudry and Gschneidner Jr 1978).

4.19.1 Reduction in a Tantalum Crucible

Two major problems confronted the attempts made to prepare high-melting rare
earth metals by chloride reduction. At high temperatures, the calcium chloride
frothed and prevented clean slag–metal separation, and the oxide liners were
attacked by the metal and impurities were introduced. Subsequently, Spedding
and Daane (1952) attempted the reduction in a tantalum crucible. The procedure
was as follows.
In a typical run, rare earth chloride and calcium were mixed and either jolt packed
into a tantalum crucible or compacted in a powder press into cylinders that were then
placed in the tantalum crucible. The loaded crucible was covered with a perforated
tantalum lid and heated in an induction furnace under argon to 1000°C for lantha-
num, cerium, praseodymium and neodymium and 1350°C for gadolinium. The melt
was held for 13 min to permit complete coalescence of the product metal. After
cooling to room temperature, metal and slag were separated. The use of a booster
reaction was not necessary, because external heating was used. Tantalum was more
resistant to molten rare earth metal attack than any refractory material.
The method of producing pure rare earth metals by the chloride reduction route,
though feasible for most of the metals, is usually an elaborate procedure requir-
inggreat procedural care and experimental facilities, particularly on account of their
hygroscopicity and volatility. These two limitations are, to a large extent, circum-
vented by the use of fluorides in place of chlorides.

4.19.2 Reduction of Rare Earth Fluorides

The process was developed at Ames. Calcium was used as the reducing agent. The
lower vapour pressure of calcium fluoride permitted reduction temperatures as high
as 1700°C, higher than the melting point of any rare earth metal. It was thus pos-
sible to prepare all the rare earth metals (except those exhibiting stable divalency)
irrespective of their melting point.
146 Metal–Crucible Interactions

FIGURE 4.14 Tantalum crucible for rare earth metal production (From Daane, A.H. 1961a).

4.19.3 Calcium Reduction (Ames Process)

In the Ames process (Daane and Spedding 1953), the reductions were carried out
by heating the mixture of calcium and rare earth fluorides in a tantalum crucible in
argon in an induction furnace (Figure 4.14. The reaction initiated at 800–1000°C and
was not violent. Heating was continued to a temperature where both the products,
slag and metal, were molten, i.e. 50°C above the melting point of calcium fluoride
or 50°C above the melting point of the metal being prepared, whichever was higher.
Once the furnace had cooled to room temperature, the crucible was removed from
the furnace in air, and the slag layer that had formed on top of the metal ingot was
knocked out of the crucible, leaving a clean ingot of rare earth metal in 97–99% yield
with 0.1–2% calcium as the principal impurity. Sufficient thickness of the tantalum
crucible permits its repeated use (Daane 1961b). Yttrium is the highest-melting rare
earth. Spedding et al. (1970) gave a nice picture, reproduced here as Figure 4.15,
showing the cross section of a typical reduction crucible in calciothermic reduc-
tion of rare earth trifluorides. The sketch clearly shows the distinctly concave top of
solidified rare earth metal and the convex top of the solidified CaF2 layer. Molten
rare earth metals wet tantalum, and molten fluorite does not wet tantalum.

4.19.4 Goldschmidt Process
In the industrial practice described by Herget (1985) for calcium reduction of rare
earth fluoride, the fluoride was mixed with granular calcium, and the mixture was
placed in a tablet form or as rammed material in a steel reactor. The reactor was lined
on the inside with tantalum, which was coated with a layer of calcium fluoride. After
the reactor was closed and evacuated, the metallothermic reduction was initiated by
a spark discharge. The reduction proceeded to completion within minutes.

2REF3 + 3Ca = 2RE + 3CaF2 (4.5)

Major Melt–Crucible Systems 147

FIGURE 4.15 Crucible cross section after calciothermic reduction of rare earth fluoride
(From Spedding et al. 1970).

The rare earth metal collected as an ingot in the bottom part of the reactor, topped
by solidified calcium fluoride slag.
The preparation of rare earth metals by reduction of their trifluorides with calcium
in tantalum crucibles leads to the introduction of up to 0.5% tantalum as impurity.
This happens in particular in higher-melting rare earths, because higher tempera-
tures are reached in their production, and the solubility of tantalum in rare earths
increases with temperature (Dennison et al. 1966a, 1966b). In the case of scandium,
the tantalum content goes up to 2–5%.
The temperature necessary for fluoride reduction was lowered, and thus, the
contamination of crucible material in the resulting rare earth metal was decreased
accordingly through variants of the basic process. The process now proceeds through
the formation of an intermediate low-melting alloy with zinc (Spedding et al. 1960).

2ScF3 + 3Ca + 3Zn + LiF ® Sc – 60% Zn alloy

+ 3CaF2 - LiF ( eutectic composition ) (4.6)

The alloying of scandium (m.p. 1541°C) and formation of the low-melting slag made
it possible to carry out the reduction and separation of the products into molten metal
and slag layers at 1100°C. The reaction is carried out in a tantalum crucible. After
reduction, the product:scandium–zinc alloy, which also contained 1–2% Ca, was
crushed and vacuum distilled to remove zinc, leaving behind pure scandium as a
porous sponge containing only traces of tantalum. This process is similar to the cal-
cium reduction process for the preparation of thorium, wherein a thorium–zinc alloy
is prepared and eventually converted to pure thorium metal (by distilling off zinc).
148 Metal–Crucible Interactions

4.20 OXIDE REDUCTION PROCESSES

Rare earths form highly stable oxides. They cannot be reduced to metal by an oxide
reduction process in the classical sense. However, not all rare earth metals can be
produced by halide reduction either. Samarium, europium and ytterbium exhibit sta-
ble divalency in addition to trivalency. On reduction, the trivalent halide is reduced
to the divalent halide, and the process stops there.
In the 1950s, Daane, Dennison and Spedding (1953) at Ames devised the reduc-
tion–distillation process. This method, which is similar to the Pidgeon process
for magnesium (Pidgeon 1944), has been remarkably useful for the preparation of
samarium, europium and ytterbium.
Lanthanum is the least volatile of the rare earth metals; dysprosium has a vapour
pressure nearly 300 times that of lanthanum at the same temperature. The metals
samarium and europium are similarly very volatile. Besides, lanthanum oxide has
the most negative heat of formation among the rare earth oxides. These observa-
tions led Daane et al. (1953) to devise a method for reacting the oxides of samarium,
europium or ytterbium with lanthanum metal and driving this reaction to completion
by distilling away in vacuum the volatile metals formed as product. The free energy
of the reaction was made negative by strongly influencing the value of Q (Chapter
2). The method has since been called lanthanothermic reduction or lanthanothermy.
An important feature of the reduction–distillation process is that it is forced to
quantitatively yield the product required by removal (by vaporization) of the rare
earth metal formed from the reaction mixture. The reactor and crucible material
used is tantalum. Maximum reduction temperatures range from 1450 to 1600°C, and
the metals obtained are usually very pure with respect to lanthanum and tantalum.
Interaction with the crucible has not developed into an issue. Significantly, no melts
are involved. In fact, a Goldschmidt-type reduction of rare earth oxides to yield rare
earth metal ingot has strong heat of reaction and slag melting barriers.

4.21 FUSED SALT ELECTROLYSIS

Rare metal electrolytic processes are invariably carried out in an inert atmosphere.
The rare earth metals that have been produced by fused salt electrolysis of chlorides
on a commercial scale (Morrice and Knickerbocker 1961; Hirschhorn 1968) are lan-
thanum, cerium and didymium (Nd + Pr). The cells had an iron-, carbon-, graphite-
or refractory-lined steel vessel to contain the molten bath. The cathode was the bath
container itself or an iron or carbon block at the bottom of the container. The anode
was a set of graphite rods lowered into the cell through its top. Besides RECl3, the
electrolyte contained NaCl, KCl and CaCl2, and the bath was molten even at 650°C.
Iron contamination was greater for cerium (0.19%) than for lanthanum (0.11%)
or didymium (0.05%), even though the electrolysis temperature was the lowest for
cerium. The maximum operating temperature of the chloride cell is 1100°C, and
neodymium (m.p. 1021°C) is the highest-melting rare earth metal that can be pro-
duced by chloride electrolysis, albeit in low yield. At high temperatures in a chloride
electrolyte bath, the volatility of bath constituents, the solubility of metals in their
Major Melt–Crucible Systems 149

chlorides, and the attack of cell wall materials by the rare earth metals become
severe. Considerably higher-temperature operation, however, is possible in cells
using fluoride electrolyte baths.

4.22 OXIDE–FLUORIDE ELECTROLYSIS
4.22.1 Gray’s Cell
The most successful among the early experiments on oxide–fluoride electrolysis was
reported by Gray (1951). CeO2 was dissolved in a CeF3–LiF–BaF2 electrolyte con-
tained in an externally heated carbon container with 57 mm inside diameter. The
anode was of high-purity graphite rod, and the cathode was a molybdenum rod; both
were suspended from the top and immersed in the melt. A molybdenum crucible was
kept below the cathode to collect the product. Unless it is oxidized, molybdenum does
not dissolve in cerium. The bottom of the cell was lined with a thin sheet of molybde-
num to prevent any accidentally spilled cerium reacting with graphite to form cerium
carbide. Molten cerium metal dropped off the cathode and collected in the molyb-
denum cup kept below. The overall purity of the metal was 99.7–99.8% cerium. The
product contained 0.1–0.6% calcium and up to 0.3% magnesium as major impurities.
Many types of electrolytic cells were designed and used at the Reno facility of the
US Bureau of Mines for carrying out oxide–fluoride electrolysis (Morrice and Wong
1979). The first was a 150-mm (6-inch) diameter electrowinning cell, known as cell
Type 6 (Morrice and Knickerbocker 1961; Morrice et al 1961).

4.22.2 Reno Cell Type 6

The cell consisted of a 150-mm (6-inch) diameter bath container made of graphite.
Three 10-mm diameter molybdenum cathodes and three 25-mm diameter carbon
anodes extended down into the container through its top. This electrode arrangement
was first used to melt the bath with alternating current and later used for electrolysis
with direct current.
The cell was arranged to operate fully under an argon atmosphere. It was inter-
nally heated both for raising the temperature from cold and later, for sustaining the
temperature sufficient for electrolysis, 810–830°C. One of the important features of
the cell was the formation of a layer of frozen bath or skull on the interior surface of
the graphite container bottom. The skull prevents the molten electrowon rare earth
metals from contacting the graphite crucible. The thickness of the insulation on the
side walls and bottom of the graphite cell, and the geometry of the electrode with
respect to the side wall and bottom of the cell, were chosen so as to maintain this
electrolyte skull.

4.22.3 Reno Cell Type 12

These were the electrolytic cells in which the electrolyte was contained in a ~300-
mm (12-inch) diameter graphite crucible. The key feature of the cell was that the
150 Metal–Crucible Interactions

bottom and sides of the graphite crucible were lined with a molybdenum sheet and a
tapping pipe was attached to the molybdenum hearth so that the molten metal could
be tapped. The tapping pipe extended into an inert atmosphere casting chamber.
A typical continuous electrowinning run extended over 3 days, during which
metal was electrowon and several tappings of homogeneous cerium metal were
made. The metal purity was 99.8%. Molybdenum figured as a major impurity. It
was picked up by cerium during the extended contact with the molybdenum hearth.
Tungsten contamination under similar circumstances would be much less.

4.22.4 High-Temperature Electrowinning Cell

A high-temperature electrowinning cell, also from Reno (Henrie and Morris 1966),
operating in the temperature range 1370–1700°C and used for the preparation of
liquid gadolinium, dysprosium and yttrium, is shown in Figure 4.16.
The cell was heated internally by DC, and the electrolysis current was supple-
mented by AC applied between the graphite anodes, which supplied approximately
75% of the total power. The cell bottom could be cooled by passing helium or air
through the copper coil placed below the graphite crucible to create an electrolyte
skull for collecting the metal. The cell, like all the others from Reno, was enclosed
in an inert atmosphere (controlled atmosphere temperature and pressure) chamber to
prevent reaction of the cell components and chemicals with the atmosphere.
In the construction of the high-temperature cell, the materials were taxed to the
extreme, and Morrice et al. (1968) used a combination of materials, as shown, to
design the cell. The electrolyte for gadolinium electrowinning consisted of an equi-
molar mixture of GdF3 and LiF. Gadolinium oxide was fed to the molten electrolyte
maintained at 1370°C. The metal was deposited as liquid at the cathode and dropped
down to collect as nodules in the frozen electrolyte skull at the bottom of the graphite
crucible. The temperature here was maintained at 810°C by cooling the bottom of

FIGURE 4.16 High-temperature electrolysis cell for rare earth metals (From Morrice et al.
1968).
Major Melt–Crucible Systems 151

the cell. Such a difference between the temperatures of deposition and collection was
essential to achieve high recovery and to ensure high purity of the metal product. A
typical gadolinium nodule contained 0.05% carbon, 0.01% oxygen and 0.05% tung-
sten as major impurities. Liquid yttrium and dysprosium were also electrowon from
their oxides dissolved in melts composed of 50 mole% LiF and 50 mole% YF3 or
DyF3. The metals were deposited on the cathode at 100 to 200°C above their melt-
ing points (m.p. of Dy 1412°C, m.p. of Y 1522°C) and collected at temperatures 400
to 500°C lower. Yttrium and dysprosium nodules thus recovered contained as little
as 0.08% carbon and 0.10% oxygen. The highest temperature at which the cell was
operated was 1700°C, for electrowinning yttrium.

4.23 CRUCIBLE CONTAMINATION IN RARE EARTH REDUCTION

A reduction carried out in a tantalum crucible results in an average of 0.05% tanta-
lum in the as-reduced metal (Huffine and Williams 1961). This concentration can be
much higher if reduction is carried out at higher temperatures or for longer periods
of time. A tantalum crucible was used by Croat (1969) in the preparation of Dy, Ho
and Er. These metals were prepared by the reduction of their anhydrous chlorides
with lithium. The temperature of reduction was 900°C, which was well below the
melting point of these metals. The tantalum contamination of as-reduced Dy, Ho and
Er was quite low, i.e. <30 ppm. This can be contrasted with the tantalum content of
0.1–0.2% in the as-reduced form of Dy, Ho and Er when they are obtained by cal-
cium reduction of their fluorides in tantalum crucibles. These reductions are carried
out at high temperatures, above the melting point of the rare earth metals.
Daane (1961a) stated that because the solubility of tantalum in liquid rare earth
metals increases measurably with increasing temperature, the lowest possible tem-
perature is used in the processes wherein the rare earth metal is processed in contact
with tantalum in order to keep the tantalum impurity as low as possible. In the reduc-
tion of the light rare earth metals (lanthanum through neodymium), the amount of
tantalum in the metal is approximately 200–300 ppm. For higher-melting heavy rare
earths and yttrium, this figure increases to about 1000–5000 ppm. Among the rare
earth metals, tantalum is the most soluble in scandium. As much as 3% tantalum is
present in scandium melted in tantalum. Tungsten is better as a crucible material. In
the liquid rare earth metals, less tungsten than tantalum dissolves at any temperature
(Dennison et al. 1966a, 1966b). The refractory metals molybdenum, niobium, tita-
nium and zirconium are distinctly more soluble in molten rare earth metals than are
tantalum and tungsten.
In reduction procedures, where an alloying element is used to lower the melting
point of the as-reduced metal (for example magnesium in yttrium reduction), titanium
or zirconium may be used as crucibles. When yttrium–magnesium alloy was reduced
in titanium crucibles, the alloy picked up an average of 0.15% titanium, and when
zirconium was used, the pick-up was 0.58% zirconium (Huffine and Williams 1961).
In the electrolytic preparation of the rare earth metals, cell corrosion can occur
extensively, and the corrosion products can end up in the metal produced. Carbon
impurity in the electrolytically produced metals usually comes from graphite anode
152 Metal–Crucible Interactions

and/or cell walls (Beaudry and Gschneidner 1978). Besides carbon, depending on
the materials used for cell construction, the metals may be contaminated with iron,
silicon and other metallic impurities. The use of Mo as a liner or crucible in the cell
leads to pick-up of the element in the electrolytic metal. Electrolytic cerium metal
obtained using Mo as cell material showed a Mo level of >750 ppm. Tantalum was
not picked up significantly (Shedd et al. 1964) when used in place of molybdenum.

4.24 PYROVACUUM TREATMENTS
4.24.1 Rare Earth Metals
The rare earth metals can be divided into four groups primarily on the basis of their
volatility. In the post-reduction vacuum treatment, each group is processed some-
what differently (Beaudry and Gschneidner 1978).

[Link] Lanthanum, Cerium, Praseodymium and Neodymium

These four light rare earths have relatively low melting points, very high boil-
ing points and hence, extremely low vapour pressures at the melting point. High-
temperature vacuum melting is the only process by which they are purified.
As-reduced La, Ce, Pr and Nd from the fluoride process just separated from the
CaF2 slag contain Ca, CaF2 and F, and H. All these impurities are more volatile than
these rare earth metals and are removed by induction melting the metals in a vacuum
for 10 min. Melting of La and Ce was carried out at 1850°C; for Pr, the temperature
was 1750°C, and for Nd, it was 1650°C (Beaudry and Gschneidner 1978). A tantalum
crucible was used. At the high temperatures used for vacuum melting, the rare earth
metals dissolve a considerable amount of tantalum (Dennison et al. 1966a, 1966b).
However, the solubility of tantalum in these metals decreases with temperature, and
it is very low at the metals’ melting point. This information was utilized in restricting
the amount of tantalum impurity in the rare earth metal.
After vacuum melting the metals at high temperature, the furnace was slowly
cooled to about 10°C above the melting point of the metal to allow the Ta to precipi-
tate out of the solution and settle to the bottom of the crucible. Later, when the ingot
was removed from the furnace, the thin tantalum crucible and the dendrites that had
settled to the bottom of the crucible were machined off.

[Link] Yttrium, Gadolinium, Terbium and Lutetium

The complete pyrovacuum purification sequence for these metals consists of vacuum
melting followed by distillation. As in the case of the first group of metals (La, Ce,
Pr and Nd), the use of a tantalum crucible in a vacuum melting at ~1800°C leads to
the dissolution of a certain amount of tantalum in Y, Gd, Tb and Lu. Unlike in the
four light rare earths, however, the tantalum solubility in Y, Gd, Tb and Lu does not
decrease much on reducing the temperature to their melting points, and therefore,
the tantalum in the solution does not precipitate out by cooling. The only way to
remove tantalum is by distilling off the rare earth metals, leaving tantalum in the
residue. If tungsten crucibles were used instead of tantalum in reduction and vacuum
Major Melt–Crucible Systems 153

melting, tungsten would dissolve in the rare earth. But after vacuum melting in these
metals, tungsten could be precipitated out in exactly the same way as Ta was in La,
Ce, Pr and Nd. This alternative is available if the 0.012 at .% W that will persist in the
Gd after precipitation can be tolerated. Similarly, at the end of precipitation, 0.03 at
.% W will remain in Tb and 0.07 at.% W in yttrium (Dennison et al. 1966a; Beaudry
and Gschneidner 1978).
The distillation of Y, Gd, Tb and Lu was carried out at Ames in tungsten cruci-
bles, because refluxing of some of the distilled metals eroded the tantalum crucibles,
causing them to leak. A tungsten-lined condenser was used, since tantalum diffuses
into the hot condensate more readily than tungsten.

[Link] Scandium, Dysprosium, Holmium, Erbium and Lutetium

These metals have appreciable vapour pressure at their melting points. The vacuum
melting, therefore, had to be relatively brief, or much of the metal would be lost by
vaporization. The other part of pyrovacuum treatment in the refining of Sc, Dy, Ho
and Er is sublimation. The high volatility of these metals is an advantage here, and
purification of metals by sublimation is comparatively easy. Tantalum introduced
into the metals during the vacuum melting step was removed by subliming off the
metals, leaving tantalum in the residue. Lutetium has been purified by distillation
as well as by sublimation (Spedding et al. 1968). When distillation was carried out
at 1850°C in tantalum crucibles, tantalum dissolved in the refluxing liquid lutetium,
and the liquid metal often eroded a hole through the tantalum crucible. However, the
use of a tungsten crucible and tungsten-lined condenser enabled distillation of lute-
tium without extensive crucible attack (Beaudry and Gschneidner 1978).

[Link] Samarium, Europium, Thulium and Ytterbium

These four metals are prepared by the lanthanothermic or reduction–distillation
method. Tantalum does not become the major impurity with the metals Sm, Eu, Tm
and Yb, because they do not undergo the high-temperature vacuum melting step.
Hence, there is no crucible interaction issue.

4.24.2 Uranium
The results of uranium melting in graphite, tantalum, yttria and yttria-coated cruci-
bles are summarized in Table 4.2. Yttria coating by plasma spray (Chakravarthy et al.
2015) not only effectively isolates the crucible proper from molten uranium but also
does not itself contaminate uranium or become degraded by it.

4.24.3 Vanadium
Vanadium is a relatively difficult metal to prepare (Gupta and Krishnamurthy 1992).
It is likewise a difficult metal to purify. Vanadium canbe produced by widely differ-
ent methods from a variety of its compounds (Rostoker 1958). In every case, there
are issues relating to containment of vanadium, especially when a vanadium melt
contact with the container is involved. There are examples of problems and solutions
154 Metal–Crucible Interactions

TABLE 4.2
Crucibles for Uranium Melting
Melting conditions:
temperature,°C
Crucible (coating if (time, min),
any) atmosphere Results Reference
Graphite (yttria 1300–1500 (30) No carbon pick-up Koger et al. 1976
coating)
Yttria – 1400 (12,000) Uranium picked up oxygen Tournier et al.
polycrystalline, from yttria, UO2·Y2O3 1998
sintered formed at the melt–crucible
interface Solubility of Y in
U is 0.5 at .% at 1300°C
Tantalum Molten uranium Crucible punctured by Kuznietz et al.
intergranular attack 1988
Tantalum (yttria 1300 (4800) No attack or impurity pick-up Alangi et al. 2011
coating 80–100 µm or interdiffusion of U into
by thermal spray) yttria
Graphite (yttria 1200 (120) Yttria effectively isolated Chaktavarty et al.
coating 300 µm by uranium melt from graphite 2015
atmospheric plasma
spray)
Yttria 1600 (15) U-10 wt.% Zr alloy was Kim et al. 2013
melted without interaction

when vanadium is produced by metallothermic reduction of oxides and also when

the metal is obtained by metallothermic reduction of its chloride.
As-reduced vanadium is usually obtained in 80–99%purity. It contains many
metallic and non-metallic impurities. Among the metallic impurities, iron is usually
present along with aluminium, calcium, chromium, copperand occasionally magne-
sium. The non-metallic impurities invariably present are carbon, oxygen, nitrogen
and silicon. Certain impurities in the metal can be avoided or decreased by a suitable
choice of reduction conditions and by using high-purity start materials. However, it
has not been possible to obtain, in the reduction stage itself, vanadium metal pure
enough with respect to all the impurities. Post-reduction purification is an unavoid-
able step in pure vanadium production.
Vacuum induction melting (VIM) is not only great as a refining process but also
integrates very well with many good fabrication techniques. The major problem
with the VIM method for consolidation and purification of vanadium concerns the
container material. Molten vanadium attacks most, if not all, the usual refractory
materials and becomes impure because it picks up oxygen, carbon or other impuri-
ties from the crucible. The crucible is also rendered useless because liquid vanadium
seeps through it. Rostoker (1958) has summarized the results of melting vanadium
in some of the refractory crucibles as in Table 4.3. Interestingly, the list could not be
Major Melt–Crucible Systems 155

TABLE 4.3
Results on Induction Melting of Vanadium in Various Crucibles
Hardness of Remarks
Time held vanadium as cast, Impurity pickup
Crucible molten, Weight Vickers Hardness Crucible and product
material min melted, g Number attack condition
Recrystallized 0.75 20 410 Slight
alumina
Al2O3 569 1.52% O pick-up,
0.20% N pick-up
Alundum 0.5 26.6 Very severe No button
BeO 0.5 20.3 221 Very slight
BeO 2 35 251 Slight
BeO 564 1.15% O pick-up,
0.21% N pick-up
CaO 5 30 423 Very severe
CaO 545 1.13% O pick-up,
0.13% N pick-up
CeS 182 0.04% O pick-up,
<0.04% N
pick-up, 0.15% S
pick-up
Graphite (grade 4 50 325 Slight 2.98% C pick-up
ATZ)
Graphite (grade 0.75 50 315 Slight 3.91% C pick-up
AUC)
Graphite (grade 8 200 280 Slight 3.29% C pick-up
AUC)
Graphite (grade 8 500 298 Slight 2.08% C pick-up
AUC)
MgO 40.2 Very severe No button
recovered
Magnorite 0.5 26.6 564 Very severe
Carbon-bonded 30 Very severe No button
SiC recovered
ThO2 1 27 247 Slight
TiO2 Very severe
ZrO2 (stabilized) 1 39 438 Severe

Source: Rostoker 1958, Merrill 1958.

General:
Atmosphere: helium
Hardness of vanadium as arc melted: 217 VPN
Analysis: 0.05% O and <0.04% N
156 Metal–Crucible Interactions

updated with new additions in the last six decades, because there have been none.
Fabrication of vanadium alloys have been by the routes involving vacuum arc melt-
ing or electron beam melting (Gupta and Krishnamurthy 1992).
Among the oxide crucibles used, beryllia and thoria withstood the attack of molten
vanadium. This result is applicable only for very short melting times (0.5 to 2 minutes).
The pick-up of oxygen by molten vanadium from the oxide crucibles is very high and
sufficient to render the metal useless for many applications. Liquid vanadium only
slightly attacks graphite crucible for upto about 8 minutes of melting time but picks
up between 2% and 4% carbon in the process. The amount of carbon picked up by
the molten vanadium strongly depends on the quantity of vanadium melted in a given
crucible. This points to the possibility that melt–crucible contact area is a major fac-
tor determining the ingestion of carbon, more even than the duration of melting. The
carbon pick-up from graphite by different quantities of vanadium melted shows this.
A similar situation will be encountered later in the melting of NiTi alloys in graphite
(Otubo et al. 2006), where the batch size of the melt plays a decisive role in contamina-
tion. This strong carbon contamination is not reflected in the hardness values of as-cast
vanadium, because carbon solubility in vanadium is limited, and compared with other
interstitials, the contribution of carbon to the increase in the hardness of vanadium is
less. Among the other refractories tested, cerium sulphide (CeS) was found to be good
in that it did not contaminate liquid vanadium with oxygen. However, some amount of
sulphur (0.15%) got into the metal (Rostoker 1958).
Regarding the refractory metals tantalum, molybdenum and tungsten, tantalum
and tungsten cannot be used to hold molten vanadium, because even before vana-
dium melts, these metals form a liquid phase with vanadium due to the occurrence of
a minima in the liquidus at a composition near pure vanadium. Even though no such
minima occurs in the Mo–V system, these metals form a continuous series of solid
and liquid solutions over the entire composition range. It therefore appears that no
crucible – metallic or ceramic – can contain liquid vanadium without being attacked
and/or contaminating vanadium. To hold molten vanadium for a very short time
(2 min), beryllia and thoria crucibles appear satisfactory. Rostoker (1958) has indi-
cated the possibility of using beryllia- or thoria-coated crucibles to melt vanadium.
However, none of the crucibles described earlier can be used to melt vanadium for
refining the metal, because it will be necessary not only to superheat the liquid metal
but also to contain the vanadium melt for a longer time to facilitate complete refin-
ing. Thus,VIM does not appear to be a workable melt refining process for vanadium.
According to the Ellingham diagram, the oxides MgO, CaO, ThO2 and BeO are
all far more stable than the lowest (and most stable) vanadium oxide, VO(s), at all
temperatures. That beryllia and thoria are resistant to attack while magnesia and
calcia are strongly attacked by molten vanadium suggests that factors other than
simple thermodynamic stability of oxides may be at play. The literature information
available so far on molten vanadium–crucible interaction does not provide details on
the wetting behaviour and contact angle data, density and porosity conditions of the
crucibles tested, or the interface between the melt and the crucible. One or more of
these may help in reconciling observations with known thermodynamic and kinetic
behaviour and provide clues to more resistant crucibles for use in VIM.
Major Melt–Crucible Systems 157

4.25 TITANIUM
The industrial importance and the quantity of research effort that titanium has attracted
in the past 80 years among the less common metals is comparable to those of iron and
steel among the common metals. Titanium is a very abundant element in the earth’s
crust, and its major ores are easily accessible in every part of the world. The question
of titanium metal production in sufficient purity and decent quantities was presented
with a workable solution when William Justin Kroll developed a magnesium reduction
process for titanium tetrachloride, the ‘Kroll process’, in the 1940s (Kroll 1940, 1959).
Certain titanium alloys have outstanding combinations of properties that make them
highly attractive for applications as structural materials in the aerospace and automotive
industries, for chemical, petrochemical and marine applications, and in the manufacture
of biomedical components and surgical instruments. Some other alloys of titanium have
a combination of properties – shape memory and hydrogen storage – that make them
very useful as functional materials for special applications in power generation, energy
storage, robotics, biomedical and telecommunications, among others.
TiAl-based alloys are difficult to prepare and have low ductility at room tempera-
ture and low formability. Precision casting is the way to go for these alloys (Lapin
2009; Freuh et al. 1996, 1997). The critical process is alloy melting. The crucible–
melt interactions in a VIM operation depend on the titanium content of the alloy
as well as the temperature needed for its melting. In practice, TiAl(γ) alloys can be
melted at temperatures between 1550 and 1750°C, while Ti6Al4V alloys require
1660–1800°C due to the high Ti content (Fashu et al. 2020). Until now, no refractory
material has been found to be absolutely inert against these alloys, and some inter-
actions between melt and crucible and even mould materials have always occurred
during melting and casting (Fashu et al. 2020; Freuh et al. 1996, 1997; Uwanyuze
et al. 2021). These melt–ceramic material interactions lead to contamination and
degradation of properties (Tetsui et al. 2012a, 2012b). If titanium and its alloys can
be melted and cast without significantly degrading the crucible and/or the metal
itself, in sufficient volumes and inexpensively, it will be considered, and remain for
eternity, a major milestone in the annals of materials processing.

4.25.1 Melting Techniques
Titanium and its alloys tend to dissolve substantial quantities of impurity elements.
Hydrogen can be removed by vacuum treatment, but oxygen and nitrogen, once dis-
solved, cannot be readily removed by any simple or straight forward method. The
focus is therefore on not allowing the metal to ingest the impurities rather than trying
to remove them once ingested.
There are different techniques to manufacture titanium alloy parts. Investment
casting is one of them and probably, the most significant. Investment casting is an
established process for manufacturing near-net-shape intricate parts. The technique
is known and has been practised for centuries as lost wax casting. It is carried out in
ceramic shell moulds and is preceded by melting of the alloy. However, due to the low
fluidity of titanium alloys, the melt needs to be taken to a high superheat. Presently used
158 Metal–Crucible Interactions

furnaces for melting titanium alloys typically have water-cooled copper crucibles and
an inert gas cover. Temperatures sufficient for melting, however high they may be, are
attained easily using arc, electron beam and also induction heating. But after melting
occurs, raising the temperature of the melt much further is either not possible (vacuum
arc remelting [VAR]) or beset with cost problems (induction skull melting [ISM], elec-
tron beam melting [EBM]). These melting techniques and equipment – ISM (Harding
et al. 2011; Harding and Wickins 2003), VAR and EBM (Ou et al. 2015; Sakamoto et al.
1992) are specifically made for alloys that are very reactive in the molten state. They
are, by nature, exceedingly energy intensive, accommodate only a small metal volume
and are only able to achieve a small superheating temperature (Harding and Wickins
2003). VIM in a ceramic crucible is significantly more energy efficient and capable
of much higher superheating and thus, is more conducive to obtaining a heat-laden
melt pool suitable for larger castings (Fadeev et al. 2015; Kamyshnykova and Lapin
2018; Gomes et al. 2011). Thus, the key requirements for a durable investment casting
procedure for titanium alloys are (i) an energy-efficient method for melting titanium
alloys under a controlled atmosphere that can handle high melt volumes, provide good
homogeneity of alloys and enable a high superheat, and (ii) a non-reactive material
that is conducive to making a viable mould. A material that corrodes during melting
may still be usable as a mould material because of the shorter contact time and lower
temperature during casting compared with melting (Freuh et al. 1996). Current melt-
ing practices of titanium have evolved to circumvent issues like temperature needed,
reactivity of titanium, its sensitivity to interstitial elements, and the unavailability of a
sufficiently inert crucible and mould.

4.25.2 Induction Melting
Induction melting is considered attractive, among other reasons, because the mixing
it creates is good for producing homogeneous ingots. The crucible, if also used as
susceptor, is at or lower than the melt temperature. Induction melting in an uncooled
ceramic crucible provides accurate metal temperature control, melt thermal homo-
geneity and any desired molten metal superheating. A few of the titanium alloys are
commercially melted by this method. But across the board, the use of this process is
possible only when the crucible problem is solved. The problem remains one of the
most actively investigated areas of titanium metallurgy.
Experimental results have often indicated that it is not thermodynamically possible
to completely predict the inertness of a crucible material to attack by molten titanium.
The general limits of theory in choosing the right crucible for melting titanium were
described in Chapter 2. For checking on the possible reactions between potential cru-
cible materials and molten titanium, numerous melting experiments are required.

4.25.3 Skull Melting
A good workable interim method for melting titanium without contamination is by
the use of a water-cooled copper container to hold the melt. This arrangement is
Major Melt–Crucible Systems 159

used in three major melting methods: vacuum arc melting, EBM, and ISM, or cold
crucible induction melting (CCIM).
The real crucible is generally a water-cooled copper container. As the titanium
is melted in this container, the melt immediately in contact with the water-cooled
crucible wall freezes. From then on, the layer of titanium that is frozen acts in prac-
tice as the crucible inside wall in contact with the molten metal. The frozen layer is
called the skull, and the melting style is known as skull melting. Skull melting can be
implemented using a variety of heat input modes: by arc, induction or even electron
beam, as noted earlier.

[Link] Induction Skull Melting or Cold Crucible Induction Melting (CCIM)

Cold crucible induction melting uses a segmented water-cooled copper crucible, and
the melt is directly coupled to the high-frequency power supply. ISM of titanium
scores well in alloy composition uniformity and is free from any crucible contami-
nation. It is good enough for the production of small castings, because the technique
is severely restricted due to lack of superheat control and the batch size limitation.
Much of the applied energy to achieve optimal melt superheating is lost to the cool-
ing water of the segmented crucible.

[Link] Vacuum Arc Melting (VAR)

Schuyler et al. (1976) summarized the status of titanium alloy melting and cast-
ing as regards various crucibles and moulds in the mid-1970s. Titanium casting
producers used skull melting in a water-cooled copper crucible by the consum-
able and/or non-consumable electrode method, and for moulds, the choices were
ceramic investment mould, using a non-reactive face coat, and rammed graphite
moulds. In Russia (Glasunov 1970), consumable electrode melting into a graph-
ite-lined, water-cooled copper jacket and casting in moulds fabricated from steel,
graphite, alumina or sillimanite (Al 2SiO5) lined with a graphite dispersoid was
practised.
Most of the titanium produced today is arc melted in water-cooled copper cru-
cibles under an inert atmosphere. Either non-consumable electrodes of graphite
or tungsten or consumable electrodes of titanium are used. Consumable elec-
trode arc melting in water-cooled moulds is a working alternative, but only a
fraction of the metal is molten at any time, and the superheat margin is narrow.
This condition is not conducive to alloy homogeneity, and multiple remelting is
usually done to obtain homogeneous ingots. Even though arc melting methods
must be considered as expedient, induction melting using an uncooled crucible
remains the goal.

[Link] Electron Beam Melting (EBM)

EBM has been known since the 1950s and extensively used for refining refractory
and reactive metals. More recently, this process has been used to produce alloys
such as Ti6Al4V and very clean superalloys (Winckler and Bakish 1971) as well as
titanium alloys (Otubo et al. 2003a, 2003b)
160 Metal–Crucible Interactions

4.25.4 Castings
Titanium alloys are not easy to machine. Casting is the way to go. At present, the
consumable electrode vacuum arc method is used. Once the titanium electrode has
been consumed, there is no more heat input to the system, and melt cooling is fast and
intense. The metal must, therefore, be poured into the moulds very quickly before
solidification in the crucible sets in. Under these conditions, a centrifugal casting
arrangement is used to move this metal into the moulds (Bridges and Hauzeur 1991).
Although titanium is cast under vacuum and in a mould made of an appropri-
ate non-reactive material that current technology can provide, it is not yet possible
to avoid contamination of the casting surface by oxygen, carbonaceous material or
first-coat oxides from the mould surface picked up by the molten metal at the time
of pouring. The resulting impurity-laden layer, known as alphacase, is very hard
and can range in thickness from zero upto 0.6 mm. The alpha case (α-case) is the
subsurface layer that forms on a titanium casting in the area that is in contact with
the mould. It is a thin, impurity-laden layer that develops due to mass transfer of
elements such as oxygen, carbon, nitrogen, aluminium or silicon from the ceramic
mould into the alloy as it solidifies (Saha and Jacob 1986). There is always a concen-
tration gradient of elements across the thickness of the α-case, which has a break at
the point where this layer ends. The α-case significantly decreases ductility, fracture
toughness and fatigue life of the casting. There are various methods for removing
it, including grinding, laser irradiation, water jetting, cathodic de-oxygenation and
chemical milling in mixed acids. Yet, the better option would be to prevent or limit
it rather than treat it. The best case is no case (layer). A mould material resulting in a
thinner α-case is better than one that gives rise to a thicker α-case.

4.25.5 Moulds
The methods for making moulds are: ceramic shells made by the lost wax process for
investment casting to make complex shapes with a good surface finish, and precision
sand and graphite mould-rammed sand, generally used for simpler shapes and with
a surface finish slightly inferior to that obtained on investment castings. Standard
investment casting moulds in general are not acceptable for casting titanium alloys.
Zircon, alumina and silica are better avoided in the melt-facing surface due to their
reactivity with molten titanium.
Shortlisting of materials for investment moulding can be done from crucible data.
In any case, a mould is in contact with the reactive metal for only a short period,
unlike a melting crucible, which contains a melt for a considerably longer duration.
However, materials that may undergo dimension changes in firing is still acceptable
for crucibles but not for moulds (Freuh et al. 1996, 1997; Kuang et al. 2000, 2001).
Crucible investigations have identified several ceramics that possess low reactivities
with the titanium alloys. These ceramics have been considered for mould applica-
tions also.
Ti-48Al was induction melted in a yttria-coated YSZ crucible at 1600°C and cen-
trifugally poured into cylindrical moulds with face coats of zirconia, yttria and silica
Major Melt–Crucible Systems 161

– all under argon (Barbosa et al. 2006). Contamination occurs only at the casting
surface, to a depth depending upon the mould material and solidification time. This
is in addition to any α-case. The main contaminants for the moulds listed are oxy-
gen and the element present in the oxide used as mould material: Si, Zr, Y. The best
results were achieved in a mould made of zirconia with an inside coat of yttria. The
α-case was only 25 µm, and there was no apparent metal–mould reaction. It is pos-
sible to cast titanium aluminides with very low contamination and good surface fin-
ish if yttria is used as face coat in the moulds. Plain zirconia moulds resulted in the
formation of a ~15-µm thick Al2O3 layer on the sample surface (the result of a fairly
complex sequence of reactions as described by Barbosa et al. 2006). The α-case was
75 µm thick. The silica mould resulted in the formation of a complex oxide layer,
20–50 µm thick, and the α-case was 250 µm in this situation.
Earlier, Schuyler et al. (1976) had observed that Ti-2.5Be, induction melted and
cast in the Y2O3·15Ti crucible, contained oxygen levels (0.2–0.4%) that did not
severely degrade cast alloy ductility. Yttria second-phase particles were observed in
these castings. Castings made with the experimental Y2O3/K2SiO3 face coat mould
and the Howmet Monograf moulds showed acceptably low surface contamination.
The possibility of significant preheating (1100°C) of ceramic moulds in the con-
ventional vacuum investment process decreases the temperature differential between
the mould and the molten alloy and enhances casting quality and reliability. This
procedure is not yet feasible in a routine manner for commercial titanium alloys. The
issue is materials. New materials are only just coming into view and are yet to be
accessed and satisfactorily tested.

4.25.6 Crucibles for Titanium Metal

Kroll himself started the search for a crucible to melt titanium when he investigated
CaO as the crucible material in the year 1940. Investigations on ceramic crucibles
suitable for titanium melting and reviews of various workable options started appear-
ing regularly in publications beginning in the late 1940s. The first was an elegant
paper by Brace (1948), who examined alumina, beryllia and thoria as possible can-
didates for a crucible. Among the three, thoria was judged to be the best. About
a decade later, after considering many alternatives, Weber et al. (1957) suggested
oxygen-deficient ZrO2as the optimal crucible material.
Beginning then, and during the seven decades that have since elapsed, the crucible
materials that have been investigated at one time or other and for one reason or other
have included refractory compounds of every description: nitrides, borides and car-
bides (Chapin and Friske 1954a, 1954b, 1955; Samsonov et al. 1960; Weimer 1997;
Kartavykh et al. 2009, 2010; Sadrnezhaad and Raz 2005); sulphides (Eastman et al.
1951); oxides (Kuang et al. 2000; Sakamoto et al. 1992; Tetsui et al. 2012a, 2012b;
Zhang et al. 2012); fluorides, oxyfluorides, graphite and refractory metals (Savitskii
and Burkhanov 1970; Zhang et al. 2006; Frenzel et al. 2004; Nayan et al. 2007); as well
as certain perovskites. However, the issue of crucible materials for induction melting of
titanium and its alloys is still not fully solved, and no solution that has been put forward
has evoked sustained confidence for significant commercial trials (Fashu et al. 2020).
162 Metal–Crucible Interactions

[Link] The Oxides: Al2O3, BeO, ThO2, CaO, MgO

When titanium was melted in contact with Al2O3, BeO and ThO2 in vacuum, molten
titanium reacted vigorously with A12O3, less so with BeO, and slightly with ThO2
(Brace 1948). Among these three oxides, only thorium oxide showed any promise as
a crucible material for melting titanium or titanium-base alloys.
The most spectacular case was that of Al2O3. The interaction was so rapid that a
visible increase of temperature occurred (something that could be observed in real
time in the pyrometer used for sighting the charge and measuring its temperature)
as soon as the titanium melted, and the titanium became so highly alloyed that
the product bore no resemblance to either of the metals – titanium or aluminium.
A titanium–aluminium intermetallic of sufficient stability could easily upset the
stability order suggested by simple Ellingham plots. Later, Economos and Kingery
(1953) noted that titanium melt could penetrate along the grain boundaries of Al2O3
at 1800°C, leading to the alternation of these oxides. The specimen melted in con-
tact with BeO was also embrittled, indicating that some transfer of Be to titanium
occurred. With thorium oxide, there appeared to be no significant interaction.
Although it cannot be regarded as inert to molten titanium, it was potentially a
suitable crucible material. Al2O3 and BeO would be useless as crucibles for melting
titanium metal (Brace 1948).
Kroll (1940, 1956, 1959) stated that melting titanium on ThO2 may be useful for
commercial purposes if the presence of small traces of oxygen due to ThO2 reduction
is not objectionable. Titanium melted very quietly on a thoria base. When melted on
an alumina, zirconia or yttria base, titanium readily reacted to form alloys. The men-
tion of yttria along with alumina and zirconia is surprising, considering that many
later investigations proved yttria to be largely inert to titanium. It needs to be taken
into account that in 1940, suitable techniques for the separation of rare earths from
one another were still in the early stages of development, and pure rare earth oxides,
including pure yttria, were not readily available.
Pure ThO2 crucibles fired at temperatures as high as 1860°C were also wetted
by the titanium melt (displayed a concave meniscus; Weber et al. 1957). The Ti melt
in the ThO2 crucible had a dull, grey surface and a duplex microstructure resulting
from solution of the crucible material in the melt and reprecipitation.
Schuyler et al. (1976) evaluated many crucible materials, and the results are listed
in Table 4.4. They used a standardized method of evaluation. A 10-gram charge
of the Ti–2.7Be alloy was melted in vacuum (0.1 to 0.01 Pa) in the crucible and
held for 10 minutes at 1577°C (i.e. 150°C above the alloy melting point or at 150°C
superheat).
Lyon et al. (1973) noted that hypostoichiometric yttria as a crucible resulted in
less oxygen contamination in the melt compared with a fully stoichiometric oxide
crucible. A similar suggestion was earlier made by Weber et al. (1957) regarding
zirconia. To promote low oxygen stoichiometry in the crucibles, ThO2 crucibles
were coated on the inside surface with a carbon–acetone slurry, dried, and then
re-fired at temperatures greater than 1615°C (Schuyler et al. 1976). When the cru-
cible material was a substoichimetric ThO2 (i.e. oxygen-deficient thoria of ThO2−x
Major Melt–Crucible Systems 163

TABLE 4.4
Performance of Various Crucibles in the Vacuum Melting of Ti-2.7Be Alloys
Increase in oxygen content, Increase in other element content,
Crucible wt.% wt.%
CaO·15Ti 1.06 Ca: 0.004
ThO2 0.57 Th: 0.63
CaO (in inert gas) 0.59 Ca: 0.37
MgO (in inert gas) 3.41 Mg: 0.004
Y2O3 0.09 Y: 0.19
Y2O3·15Ti 0.12 Y: 0.19
SiC Decrease of 0.02 Si: 7.51C: 5.85
CeS 0.03 Ce: 0.56S:0.19

Source: Schuyler et al. (1976).

Temperature: 1577°C (150°C more than the alloy solidus), holding time: 10 minutes, vacuum: 0.01 to 1 Pa.

or hypostoichiometric thoria) instead of fully stoichiometric ThO2, the oxygen con-

tamination of melts was 50% less.
Kroll (1940) noted that CaO is a satisfactory refractory for melting titanium but
recommended that it be first fired near its melting point in a tungsten crucible in a
good vacuum so as to crystallize it. Even this refined, crystallized CaO resists the
action of molten titanium metal only at the metal’s melting point. At temperatures
slightly above this, a violent reaction occurs between titanium and calcia, resulting in
the evolution of calcium vapour. Argon was used as cover gas for melting Ti alloys in
CaO and MgO crucibles, because these crucibles show excessive metal boiling and
spitting during melting in vacuum.
In regard to calcia crucibles and titanium melt, Schaffoner (2020) noted that cal-
cium oxide, which decomposes to oxygen and calcium, shows fast dissolution with
commercially pure Ti and Ti6Al4V due to the high vapour pressure of calcium.
Calcia crucibles are good for titanium aluminide melts generally due to a much lower
melting temperature of only 1460°C, calcium aluminate formation and possibly a
lower titanium activity in the alloy (Morscheiser et al. 2008).
To assess CaO as a refractory material for titanium melts, its decomposition reac-
tion has to be considered. At higher temperatures, calcium oxide decomposes to
calcium vapour and oxygen, CaO → Ca(g) +0.5O2(g), a process noted by Kroll him-
self. A strong evaporation of calcium during the melting of titanium alloys was often
reported, while the oxygen is picked up by the melt. As expected, the pick-up of
calcium in titanium is very low, because most of it evaporates and does not remain
there to be picked up. Calcia, therefore, lacks corrosion stability (Schaffoner 2020)
with many low-alloyed titanium melts. Economos and Kingery (1953) observed that
titanium melt could penetrate along the grain boundaries of MgO at 1800°C, leading
164 Metal–Crucible Interactions

to the alternation of these oxides. The situation as regards BaO is similar to that of
CaO. There is also evaporation of BaO itself.
In their review, Weber et al. (1957) mentioned the results of titanium melting in
some unusual crucible materials – titanium oxide, certain oxyfluorides and interme-
tallics. None of these could serve as crucible material for melting titanium.

[Link] Zirconia
The possibility of using a zirconia crucible for melting titanium was checked early
on by Eastwood and Craighead (1950), who reported on many other crucible mate-
rials as well. These materials were refractory metals (W, Mo), carbon, carbides,
oxides, nitrides, sulphides, borides and silicides. They considered that all materials
other than zirconia were quite soluble in the titanium melt, and stabilized zirconia
was the only refractory not wetted by the melt. The meniscus was convex. Oxygen
was absorbed by the melt, but the Zr content of the ingot was generally low, hinting
at the possibility that ZrO2 was partly reduced and that lower oxides of zirconium
might perform better as crucibles. However, the stabilized zirconia crucible failed as
the ingot cooled. Economos and Kingery (1953) observed that titanium melt could
penetrate along the grain boundaries of zirconia.
Zirconia stabilized with 15.5 mole% MgO and fired in an oxidizing atmosphere
to maturity at 1860°C was used to melt titanium (Weber et al. 1957). A significant
reaction occurred, and the melt stuck to the crucible. MgO-stabilized ZrO2 crucibles
blackened (indication of stoichiometry change of the oxide) during the melting oper-
ation, and the melt appeared to have diffused through the crucible wall.
Avoiding the alkaline earth oxides to stabilize zirconia, Weber et al. (1957)
investigated ZrO2 made oxygen deficient by reacting ZrO2 with Ti metal pow-
der. Titanium amounting to 15 atomic%was incorporated into ZrO2 to make the
crucible more resistant to attack by titanium melt. According to Ruh (1963),
titanium could react with zirconia, and up to 10 at .% of Zr and O were taken
into the solid solution of Ti, while the zirconia transformed into oxygen-deficient
zirconia (Ruh and Garrett 1964). Zirconium entered the titanium lattice substitu-
tionally, and oxygen went to interstitial positions. Lyon et al. (1973) pointed out
that the free energy of formation of a Ti–10 at.% O solution at 1400°C was more
negative than that of ZrO2. The stoichiometry change in zirconia should not be
surprising. Saha et al. (1990) also stated that ZrO2−x and α-Ti(O) were formed
when Ti reacted with ZrO2.
When titanium was melted in 5 mol.% CaO–ZrO2 or 3 mol.% Y2O3–ZrO2 at
1750°C, the events were similar (Lin and Lin 1998, 1999b). Interfacial reactions
between zirconia and titanium were preceded by the infiltration of liquid titanium
through the open pores of zirconia. Zirconia was reduced to oxygen-deficient zir-
conia (ZrO2−x), and the oxygen released was dissolved in titanium. The stabilizing
oxide was expelled from zirconia. The Ca content increased in the solid solution of
ZrO2−x, leading to the precipitation of CaZrO3. This may be contrasted with the fate
of calcium when titanium was melted in pure calcia crucible. The yttria stabilizer
was expelled as oxygen-deficient yttria. Saha et al. (1990) had made similar observa-
tions earlier.
Major Melt–Crucible Systems 165

[Link] Rare Earth Oxides

The possible use of rare earth oxides as crucible materials for melting titanium was
investigated by Lyon et al. (1973) and Schuyler et al. (1976). The rare earth oxides
were screened for their possible interaction with titanium in the solid state itself at
1000°C for 64 h under vacuum by setting up diffusion couples. Among the oxides,
Y2O3, Er2O3, Ho2O3, Gd2O3 and Dy2O3 were the least reactive. The reactivity studies
in molten condition were carried out in an arc melting furnace, melting the metal
placed over the ceramic disc in the copper hearth. Induction melts were made at
1760°C under 3 mPa. Slight reaction in the form of some dissolution and precipita-
tion of small dendrites close to the interface was shown by Pr2O3, Gd2O3 and Dy2O3.
No dendrite precipitation occurred with yttrium oxide, the most stable of all.
Schuyler et al. (1976) fabricated crucibles of Dy2O3, HREMO (heavy rare earth
mixed oxide), LREMO (light rare earth mixed oxide) and yttria for use as cruci-
ble and mould with titanium alloys. HREMO was the concentrate obtained from
Xenotime, and LREMO materials were concentrates of Basnasite. The crucibles
were reasonably dense (85–97%of theoretical density) and well bonded. The oxygen
pick-up by titanium melt from these crucibles is given in Table 4.4. HREMO cru-
cibles turned out to be second best only to Y2O3 and Y2O3·15Ti. Low reactivity was
displayed by Dy2O3 also. Lowering the apparent oxygen stoichiometry was more
effective in reducing oxygen contamination of melts made in HREMO (by nearly
50%) than in the case of Y2O3.

[Link] Yttria
Yttrium and oxygen ingested in titanium caused Y2O3 precipitation at the grain
boundaries of the Ti matrix on cooling (Lyon et al. 1973). Like zirconia, on contact
with titanium, Y2O3 can be reduced to Y2O3−x. The darkened colour of yttria revealed
this. Melt experiments were performed using a sub-stoichiometric yttria crucible.
Yttrium oxide (Y2O3, yttria) is one of the most widely investigated refractory
materials for titanium metallurgy, but its performance is still not considered com-
pletely satisfactory. Depending on the melting conditions, and possibly the crucible
build, an oxygen contamination of 0.06–1.6 wt.% and a yttrium contamination of
0–1.93 wt.% occurs. There is also formation of yttria inclusions, attributed to the
erosion of the crucibles by the melt (Schaffoner 2020).
Somewhat similar values were given earlier by Griesenauer et al. (1972). The oxy-
gen and yttrium contamination in titanium melted in Y2O3 at 1770°C for 8 min under
argon was 1.05 and 3.6 wt.%, respectively. Replacing the crucible with Y2O3−x at the
same temperature under vacuum brought the contamination level O and Y down to
0.6 and 2.2 wt.%, respectively, even after prolonging the melting to 10 min. Y2O3−x
is more resistant to attack by molten Ti. The 97% dense yttria crucibles experience
thermal shock issues when the metal couples directly with the field in an induction
furnace. The conditions become more manageable if a susceptor, like a molybdenum
crucible, is used.
If titanium melt is to be contained in yttria, the best chance to minimize con-
tamination is provided by the sub-stoichiometric yttria crucible. However, during
166 Metal–Crucible Interactions

cooling, dissolved yttrium scavenges oxygen dissolved in the melt and precipitates
out as yttria particles. Incidentally, the presence of yttria particles results in relative
softening of the titanium matrix.
Lyon et al. (1973) also observed that hypostoichiometric yttria as a crucible leads
to decreased oxygen contamination in the melt. To promote low oxygen stoichiom-
etry in the crucibles, Schuyler et al. (1976) used yttria crucibles re-fired (>1615°C) in
argon by induction heating of a graphite susceptor. The crucibles were coated on the
inside surface with a carbon–acetone slurry and dried. No carbon coating was given
to Y2O3–15Ti crucibles.
Schuyler et al. (1976) summarized his observations as follows. (i)Y2O3is the best
material for limiting contamination (0.16%oxygen) of low-melting titanium alloy
melts. (ii) Lowering the apparent oxygen stoichiometry was not of significant benefit
for Y2O3, whereas it was effective in reducing oxygen contamination of melts made
in hypostoichometric HREMO and ThO2 (by nearly 50%).
Three reports were issued by the Naval Research Laboratory, authored by Chapin
and Friske (1954a, 1954b, 1955), covering oxides, carbon, graphite, carbides, borides
and sulphides as possible crucible materials for melting titanium. The general con-
clusion was that all the materials investigated lacked the required inertness, as all
reacted with Ti to varying degrees of severity.

[Link] Carbon, Graphite and Carbide Crucibles

Carbon and graphite are immediate natural choices for a crucible material for any
melting in a non-oxidizing atmosphere. Chapin and Friske 1954b used crucibles of
high-purity carbon with exceptionally smooth, dense surfaces and of commercially
pure and spectroscopically pure graphite to melt titanium, specifically to check-
whether purity and surface finish could make a difference. They did not. Crucibles
made of the monocarbides of Ti, Zr, V, Nb, Ta and W were also investigated for their
suitability to contain and melt titanium. Each of the carbides was wetted by molten
titanium, resulting in general dissolution of the crucible and contamination of the
melt, mainly with carbon. TiC was found to be the main carbide phase found in the
solidified melt, formed from reduction of the crucible carbide by molten titanium
at the liquid–solid interface. The carbides could not be considered suitable crucible
materials. The results reported by Weber et al. (1957) and Lyon et al. (1973) in later
years were similar to the conclusions of Chapin and Friske (1954).
Schuyler et al. (1976) evaluated SiC for possible use as crucible and mould for tita-
nium alloys. The oxygen contamination of the alloy decreased, but there was a marked
increase in Si and C contents. Slip casting a layer of Y2O3 on the inside of a SiC crucible
was tried with the aim of combining the inertness of Y2O3 and the excellent thermal
shock resistance of SiC. But, Y2O3 reacted with SiC during sintering, which converted
the Y2O3 to a carbide. Further, SiC backup crucibles underwent structural deterioration.

[Link] Borides
Crucibles made of diborides of Ti, Zr and Cr were investigated by Chapin and Friske
(1955) for melting titanium. In all cases, there was a general solution attack by mol-
ten titanium on the crucible, and the melt became severely contaminated with boron
Major Melt–Crucible Systems 167

compounds. Solution of the reaction products raised the melting point of the melt,
making pouring difficult. These borides are not suitable as crucible materials.
Titanium melts quietly in all the boride crucibles with no visible evidence of
reaction between crucible and melt and no detectable vapour evolution. The affinity
between titanium and the borides is so high that the reaction between the crucible
and the metal is initiated in the solid state itself. All the borides were wetted by
molten titanium. All the melts solidified in boride crucibles showed deep meniscuses
and an acute angle of contact between crucible and melt. The melts became viscous
soon after the charge melted and could not be completely emptied by pouring. The
boride crucibles had good thermal shock resistance and did not crack either during
the introduction of the somewhat cooler metal charge into a hot crucible or dur-
ing melting or after pouring. In spite of their apparently excellent high-temperature
properties, the borides lack the necessary chemical inertness toward molten titanium
and are useless as crucible materials.

[Link] Sulphides
There are a limited number of sulphides with high melting points (above 1800°C)
and good high-temperature stability. Two such compounds, CeS and ThS, have been
considered for use as crucibles. The CeS crucibles were made by hydrostatic press-
ing followed by sintering at about 1950°C in vacuum. A CeS crucible was used for
melting titanium.
Successful experimental fusions were reported (Eastman et al. 1951) for titanium
in CeS crucibles. Other than the statement that a sound ingot was obtained with-
out attack on the crucible (although the ingot adhered to the crucible), no details of
metal quality were presented. Chapin and Friske (1955) melted titanium in a CeS
crucible using induction heating and high vacuum. Melting occurred quietly. The
molten metal poured without much difficulty provided it was not superheated exces-
sively and was not held for extended periods. In either of these cases, it became
viscous and difficult to pour. The resulting ingot, strongly yellow-coloured on the
surface, adhered to the crucible and would not come off. All the metal melted in
CeS crucibles was found to contain appreciable amounts of sulphur, ranging from
0.165 to 0.570 wt.%. Inspite of the ingestion of sulphur and cerium, the hardness of
the metal was not markedly increased. This situation is similar to carbon ingestion
into vanadium.
CeS is not sufficiently inert to prevent gradual consumption of the crucible by
direct solution into the molten metal, but this material had the least deleterious influ-
enceon the quality of metal melted in any of the materials evaluated in crucible form.
Following Chapin and Friske (1955)’s report, Schuyler et al. 1976 melted a Ti–2.7Be
alloy in a cerium sulphide crucible. The solidified alloy showed an increase in oxy-
gen by 0.03 wt.%, cerium by 0.56 wt.% and sulphur by 0.19 wt.%.

4.25.7 Crucibles for Titanium Alloys

Almost all ceramic crucibles, including Y2O3, are attacked by molten titanium to a
certain extent. In the melting of TiAl alloys, Y2O3, BaZrO3, Ca-doped BaZrO3 and
168 Metal–Crucible Interactions

AlN are the most resistant to attack, and a final <0.1 wt.%oxygen content in the ingot
can be achieved.

4.25.8 TiAl Alloys
The technologies for manufacturing billets of TiAl-based intermetallic compounds
are melting and casting in a ceramic mould and directional solidification in a
ceramic crucible under inert gas or vacuum. Oxygen contamination is, in general,
hard to control but needs to be controlled. Oxygen gains entry into a melt from
cover gas and especially from insufficiently inert crucible and mould. The elements
that are part of the oxide ceramics, i.e. Y, Zr, Ca, Mg and Si, also diffuse from the
crucible or mould walls into the ingot. Fashu et al. (2020) have summarized, in
Table 4.5, the current status as regards the use of various crucibles for the melting of
titanium–aluminiumalloys.

[Link] Zirconia (Stabilized) Crucible

TiAl alloys were induction melted (1500°C, 2 min) in argon using crucibles of zir-
conia stabilized with yttria, calcia or magnesia (Barbosa and Ribeiro 2000). All
the samples picked up (~0.27 to 0.93 wt.%) zirconium in addition to oxygen. Of the
three types, calcia-stabilized zirconia contaminated the melt least. The α-case was
300–600 µm thick. The contamination was less (α-case 200 µm) when pure calcia
crucibles were used.
TiAl melted and cooled in graphite crucibles was contaminated with 3–7
atom % carbon. The larger value corresponds to regions near the surface. However,
the alloy melted in yttria-stabilized zirconia and cast in graphite mould showed no
sign of carbon contamination.
Lin and Lin (1999a) vacuum arc melted Ti-6Al-4V alloy and cast it centrifugally
in a porous ZrO2 mould. Interfacial reactions between Ti-6Al-4V and ZrO2 mould
were preceded by the penetration of liquid titanium due to capillary and centrifugal
forces through the interconnected pores near the mould surface. Ti leached oxygen
from ZrO2 to form α-Ti(O) solid solution. Zirconium also dissolved in α-Ti(O). The
SiO2 in the binder in the ZrO2 mould readily reacted with titanium to form Ti5Si3.
The stabilizer CaO incorporated into zirconia is released, dissolves in Ti, and reacts
with ZrO2 and Ti(O) to form Ca3Ti2O7 and CaAl4O7 in the reaction zone.
Hypostoichiometric zirconia contaminates Ti melt less, not only in crucibles dur-
ing melting but also in moulds during casting. Nakamura et al. (1988) incorporated
zirconium powder (3–5 wt.%) in the zirconia face coat to serveas a barrier to the
diffusion of oxygen from the mould into titanium. The zirconia face coat, made
hypostoichiometric by vacuum firing, also resulted in a thinner α-case (Calvert 1981).
Gao et al. (2011) investigated the effect of porosity (16–42%) of Y2O3 on the con-
tamination levels of Ti-54Al (at .%) melted (1600°C, 100 min) and solidified in the
crucible. Depth of melt penetration, oxygen pick-up, amount of yttria particles and
also yttrium content increased with increase in porosity and holding time. The oxy-
gen content of the original melt stock was 400 ppm. The final oxygen content in the
alloys made by Gao et al. (2011) using a yttria crucible of low (16%) porosity was
Major Melt–Crucible Systems 169

TABLE 4.5
Contamination of TiAl Alloys during Vacuum Induction Melting in Different
Crucibles
Crucible Holding Interaction Oxygen
material time, min layer, μm content, wt.% References
SiO2 0.5–5 Not 1.7, Si:1.7 Tetsui et al. 2012b
reported
Mullite 15–60 190–840 – Kuang et al. 2001
(3Al2O3·2SiO2)
MgO 18–60 1296–3570 0.95–1.09 Kuang et al. 2000
Al2O3 0.5–60 30–247 0.16–2.5 Eatesami et al. 2009, Kuang et al.
2000, Kuang et al. 2001, Liu et
al. 2005, Tetsui et al. 2010,
Tetsui et al. 2012b, Zhang et al.
2013a, Zhang et al. 2012
ZrO2 (incl. 0.5–60 120–300 0.55–1.29 Barbosa and Ribeiro 2000,
stabilized) Kuang et al. 2001, Tetsui et al.
2010, Tetsui et al. 2012b, Zhang
et al. 2013a
Y2O3·15ZrO2 3 Not 0.38 Tetsui et al. 2012b
reported
CaO 0.5–60 5 0.10–0.71 Barbosa and Ribeiro 2000,
Kuang et al. 2000, Liu et al.
2005, Sakamoto et al. 1992,
Tetsui et al. 2010, Zhang et al.
2013a
Y2O3 0.5–60 8–100 0.06–0.33 Cui et al. 2012, Hockaday and
Bisaka 2010, Kuang et al. 2001,
Liu et al. 2005, Cui et al. 2010,
Tetsui et al. 2010, Tetsui 2012b,
Zhang et al. 2013a
Y2O3-coated 18–60 10–19 0.016–0.21 Kuang et al. 2000
MgO
Y2O3-coated 10–34 Not 0.04–0.13 Schuster and Palm 2006, Van
Al2O3 reported Humbeeck 1999
Y2O3-coated 60–120 15–65 0.21–0.31 Barbosa et al. 2007, Barbosa et
ZrO2 (incl. al. 2003, Barbosa et al. 2005,
stabilized) Barbosa et al. 2006
Y2O3-coated 1 Not 0.19–0.22 Gomes et al. 2013
ZrO2·SiO2 reported
CaZrO3 5–20 300–350 0.13–0.73 Schafföner et al. 2015a, 2015b,
Li et al. 2010, Klotz et al. 2019
CaZrO3–Al2O3 30 150 – Lu et al. 2017
composite
(Continued )
170 Metal–Crucible Interactions

TABLE 4.5 (CONTINUED)

Contamination of TiAl Alloys during Vacuum Induction Melting in
Different Crucibles
Crucible Holding Interaction Oxygen
material time, min layer,μm content, wt.% References
BaZrO3 5 870–2000 0.065–0.13 Chen et al. 2018b, Zhang et al.
2013b
Ca-doped 2 5 0.09 Chen et al. 2018a
BaZrO3
AlN 5–25 1.1–6.4 0.025–0.113 Chapin and Friske 1954b
BN 5–25 250–>500 – Chapin and Friske 1954b
Graphite 0.3–2 Not C:4 Barbosa and Ribeiro 2000, Cegen
reported et al. 2016
Water-cooled Cu 3 0 0.09 Tetsui et al. 2012b

Source: Fashu et al. 2020.

not more than 800 ppm even after a holding time of 100 minutes. If this inertness is
repeated in larger crucibles and bigger batches, and if the crucible survives the ther-
mal shock resistance test, the crucible problem may be seeing the light at the end of
the tunnel for TiAl alloy melting.
The impurities to look for in titanium–aluminium alloys are oxygen and carbon,
and according to industrial standards, the oxygen content should not exceed 0.1 wt.%
(Tetsui et al. 2012b). A summary of the performances of different crucible materials
used to melt TiAl alloys (1650°C, 5 min, Ar) is shown in Figure 4.17A and 4.17B.
Among the oxide crucibles tested for melting titanium alloys, yttria and calcia
were best in terms of chemical inertness, but they are both poor in shock resistance
(Barbosa et al. 2007; Cui et al. 2010; Barbosa and Ribeiro 2005). Y2O3 meets the
requirement of <0.1 wt% oxygen in ingots (Tetsui et al. 2012a, 2012b).

[Link] Yttria
The details of melting and consequent oxygen enrichment in the melt (+) are given
in Table 4.6.
Every sample had dispersed yttria inclusions (20 µm) in the metal matrix. The
average measured oxygen content of the melt charge was 0.08 wt.%. SEM X-ray dot
map micrographs reveal that there was no penetration of the melt into the Y2O3 cru-
cible, even though the latter contained crevices and cracks. The oxygen enrichment
of the cast samples (0.06–0.10 wt.%) obtained by Cui et al. (2010) are consistent with
those reported by Kuang et al. (2000) for TiAl melted in yttria-coated crucibles. The
presence of small yttrium oxide particles dispersed in the metal matrix could be due
to erosion promoted by inductive stirring action in the melt during melting,
Lapin et al. (2011) tested a Y2O3 crucible with Ti46Al8Nb (at .%) alloy (liquidus
1570°C) for directional solidification (DS) in a Bridgman-type apparatus under 10
kPa argon. The overall interaction was found to be diffusion-controlled dissolution
Major Melt–Crucible Systems 171

FIGURE 4.17A Melting time and O pick-up by TiAl alloys from different crucibles I (From
Tetsui et al. 2012b).

of the Y2O3 crucible in the melt, resulting in an increase of oxygen and yttrium
content in γ(TiAl) + α2(Ti3Al) matrix and precipitation of Y2O3 particles in the inter-
dendritic region. The change in oxygen content with temperature and duration of
contact in the DS samples (Lapin et al. 2011) at other temperatures also is given in
Figure 4.18. On the whole, the ingestion of impurities appears controllable and the
Y2O3 crucible useful for DS with this alloy. The crucibles used by Lapin et al. (2011)
were dense (92.8% of TD or theoretical density) with very small (0.7) volume frac-
tion of open porosity. An oxygen content of about 1200 wt​.p​pm in cast TiAl-based
alloys is within the technically acceptable limit.
Barbosa et al. (2005) melted Ti-48Al alloy at 1600°C and cooled it inside CaO-,
MgO- and Y2O3-stabilized ZrO2 crucibles with an inner 200-µm thick layer of Y2O3.
Every sample was found to have a small amount of yttrium oxide inclusions, possibly
eroded from the coating. In the sample melted in the yttria-stablized zirconia cru-
cible that was not coated with yttria, the inclusions were of alumina. Every sample
was contaminated with a small amount of yttrium (average value between 0.14 and
0.16 at.%). No zirconium was detected in solution. The Y2O3 layer effectively isolates
the crucible base material from the melt. The crucibles and coating were intact after
melting. Significantly, the solidified cylindrical ingot of the alloy has a concave top
surface when prepared in the uncoated crucible as compared with the distinct con-
vex top surface of the alloy melted and cast in the coated crucible – a pointer to the
172 Metal–Crucible Interactions

FIGURE 4.17B Melting time and pick-up of other elements by TiAl alloys from different
crucibles II (From Tetsui et al. 2012b).

significant difference in the wetting behaviour. Earlier, Spedding et al. (1970) had
given a sketch representing the cross section of a typical reduction crucible in cal-
ciothermic reduction of rare earth trifluorides. Figure 4.15 highlighted the wetting
behaviour of the reaction products and the shape of solidified ingot.
In a later investigation, Barbosa et al. (2007) applied the Y2O3 layer on a zirconia
crucible by painting followed by a drying operation, and no sintering was done. The
porosity was probably high and adhesion to the zirconia crucible lower than opti-
mum. In spite of this, the absence of Zr in the cast alloy showed that the Y2O3 layer
was an effective barrier between the crucible base material (ZrO2) and the molten
metal due to its limited wettability by TiAl (Kuang et al. (2001).

[Link] Calcia
Sakamoto et al. (1992) prepared Ti-33–38 wt.% A1 alloys by VIM (1550–1620°C,
20 min) in a calcia crucible of cp-titanium sheets and cp-aluminium blocks. The
charge (~7 kg) was coupled directly in the 125 mm dia crucible, and the melting was
done in argon. The behaviour of impurity elements in the molten titanium aluminide
is shown in Figure 4.19. Impurity content (Fe, Si, N and H) did not vary except for
oxygen and calcium. The oxygen content of the alloy charge was 0.034 wt.%, and the
Major Melt–Crucible Systems 173

TABLE 4.6
Contamination from Crucible in TiAl Melting in Argon Atmosphere
Melt- crucible Time,
combination Temperature, °C min Contamination Reference
Ti-47 at % Al-yttria 1600. 20 Oxygen: +0.06 to 0.07 Cui et al. 2010
wt.%
1700 20 Oxygen: +0.08 to 0.10
wt.%Yttrium: +0.15 to
0.19 at .%
Ti-46Al-8Nb (at.%) 1650 210 Oxygen: +0.4 at.% Lapin et al.
- yttria 2011
Ti-48 at% Al-ZrO2 1600 Oxygen:Zirconium: Barboza et al.
stabilized with +zeroYttrium: +0.14 2005
CaO, MgO or yttria to0.16 at .%
with 200 µm yttria
coating
Ti-33–38 wt.% 1550 to 1620 20 Oxygen: 0.10 to 0.13 Sakamoto et al
Al-calcia (0.034 wt.%Calcium: −(decrease) 1992
wt.% O, <0.01
wt.% Ca in the
alloy)

+ represents ingestion, − represents decrease.

calcium content was <0.001 wt.%. Oxygen content increased to 0.1–0.13 wt % in 20

min, and calcium content decreased.

[Link] Graphite
Barbosa and Ribeiro (2000) do not suggest induction melting of TiAl-based alloys
in graphite crucibles because of contamination of products by carbon and carbides.
However, Cegan et al. (2014) noted that a technically acceptable content of carbon-
can be achieved in TiAl-based alloys by using better-quality graphite crucibles.
On induction melting of TiAl at 1500°C for 2 min, Barbosa and Ribeiro (2000)
found the alloy to be contaminated with 3–7 atom % carbon. However, the alloy
melted in yttria-stabilized zirconia and cast in a graphite mould showed no sign of
carbon contamination. Cegan et al. (2015) prepared three different Ti-47Al-8 Nb/Ta
alloys by induction melting (1700°C, 20s, Ar) of the components(Ti, Al, Y, Nb-60Al
and Ta-80Al) in graphite crucibles and centrifugal casting into graphite moulds. The
compositions are all in at.%. The carbon content measured in the as-cast alloys is
relatively low when compared with that (3–7 at .%) reported by Barbosa and Ribeiro
(2000). The lowest carbon content of 460 wt​.p​pm is found in the 8Ta- alloy, which
corresponds to about 0.2 at.%. Cegan et al. (2015) did the melting at 1700°C for 20
seconds compared with 1500°C for 2 min. by Barbosa and Ribeiro (2000). A lower
174 Metal–Crucible Interactions

FIGURE 4.18 Oxygen pick-up during directional solidification of TiAl in yttria (From
Lapin et al. 2011).

carbon ingestion in spite of the much higher temperature used may be traced to the
quality of the graphite crucibles.
Melting in graphite crucibles and casting into graphite moulds can be considered
as a viable method of preparation of TiAl-based alloys containing tantalum and nio-
bium (Cegan et al. 2015).
In a later investigation, Lapin and Klimová (2019) prepared Ti-38 to 45Al alloys
with Nb and Mo additives by VIM at 1700°C under 103 Pa argon for about 3 min in
graphite crucibles followed by casting into graphite moulds. VIM in high-density
graphite crucibles resulted in a reproducible increase of carbon content by (0.9 ± 0.1)
at .%, which can be accommodated in the process by a suitable choice of starting
alloy composition. Nevertheless, Fashu et al. (2020) have remarked that TiAl alloys
are yet to be successfully melted in graphite crucibles keeping contamination levels
sufficiently low as required by industrial standards.

[Link] Nitrides: AlN and BN

Both high-purity AlN and BN have been evaluated as crucibles for melting TiAl
(Ti-46Al-8Nb) alloys (Kartavykh et al. 2009, 2010). The AlN crucible with the sam-
ple was heated to 1670°C (i.e. 100°C higher than the liquidus temperature), held for
up to 25 min and rapidly cooled.
Major Melt–Crucible Systems 175

FIGURE 4.19 Impurity content in TiAl after melting in calcia crucible (From Sakamoto
et al. 1992).

The free surfaces of all solidified melt samples were clean, without temper colours
or signs of oxidation. The solidified casting easily came off from the crucible. On
prolonged (25 min) superheating, micro-precipitates of nitride phases ((Ti,Al)xNy,
NbN) formed in the α2 matrix phase (Ti3Al) in the near-contact zone at a distance of
no more than 300 µm from the ingot–crucible interface. However, Kartavykh et al.
noted that this layer is removable during the finishing of as-cast parts.
The alloy ingots made in these experiments had no oxygen-containing (oxide)
phases. Their maximum oxygen was 1100 wt​.pp​m, which is 1.5–2 times lower than
that obtained in similar experiments performed on Y2O3 crucibles (Kartavykh and
Cherdyntsev 2008). The oxygen ingestion on heating the titanium–aluminiumalloy
in various crucibles is summarized in Figure 4.20.
Kartavykh et al. (2010) also used small pyrolytic boron nitride (pBN) boats to
hold and melt Ti-46Al-8Nb (at .%) alloy. The facility and procedure were the same
as those used for AlN. Boron nitride is wetted by the alloy melt. The melt had reacted
with the boron nitride, forming Ti2AlN and (Ti,Nb)B in the form of needle-like
micro-precipitates at the surface between the alloy and the crucible. The reaction
layer comprising these particles propagates quickly into the alloy bulk, spreading to
the whole sample in about 20 min of holding at temperature. pBN is unsuitable as a
material for crucibles/moulds for titanium-based melts.
176 Metal–Crucible Interactions

FIGURE 4.20 Pick-up of interstitials by TiAl after melting in AlN and other crucibles
(From Kartavykh et al. 2009).

[Link] The Zirconates
It is seen that crucibles of oxides, carbides, borides and sulphides have been tested
for melting titanium alloys. CaO and Y2O3 were best in chemical inertness but
poor in shock resistance. CaO is also hygroscopic. A chemical inertness equal to or
exceeding that of yttria and calcia, and a much better resistance to thermal shock, is
possessed by the zirconates – CaZrO3 and BaZrO3.

[Link].1 Calcium Zirconate
When calcia-stabilized ZrO2 was heated in contact with titanium in solid state and
the interface was examined, CaZrO3 had formed at the interface. Chang and Lin
(2010) surmised that CaZrO3 is a stable phase in contact with titanium alloys and
a potential refractory material for titanium metallurgy. CaZrO3 was evaluated in
recent years in contact with commercially pure titanium (cp-Ti), Ti6Al4V, titanium
aluminide (TiAl) and TiFe. In contact with Ti6Al4V, CaZrO3 showed remarkable
corrosion resistance when used as investment casting moulds. Compared with
moulds based on Y2O3, the zirconate caused a much lower hardness increase and
excellent demoulding (Freitag et al. 2017; Klotz et al. 2019). CaZrO3 moulds were
used by Kim et al. (2001) to cast commercial-purity titanium metal, Ti6Al4V and
γ-TiAl melts. CaZrO3 was comparable to CaO in stability.
Major Melt–Crucible Systems 177

Li et al. (2010) used calcium zirconate (CaZrO3), perhaps for the first time, asa
crucible for melting Ti6Al4V and also NiTi. Only slight corrosion was reported.
SEM and X-ray examination of the solidified melt showed that both alloys formed
an interfacial reaction layer with the ceramic, but there was no transfer of alloy or
crucible components across the interface to the other side, indicating that CaZrO3 is
a promising refractory for melting titanium alloys.
Li et al. (2007, 2013) obtained coarse-grain CaZrO3 by slip casting of CaCO3 and
ZrO2, firing at 1500°C for 3 h and crushing. The product had 12.4% open poros-
ity. Schaffoner et al. (2013) tried both solid-state synthesis and electric arc melting
methods to prepare coarse-grained calcium zirconate. The fused calcium zirconate
had considerably lower porosity than solid-state synthesized material. After crush-
ing and milling of fused CaZrO3, preparation of large crucibles by cold isostatic
pressing resulted in crucibles with 15.9–16.5% porosity.
Schaffoner et al. (2015a, 2015b) tested the corrosion of fused CaZrO3 crucibles in
contact with γ-TiAl melts during VIM. The crucibles were prepared by cold isostatic
pressing, dried and then fired at 1650°C for 6 h. The batch size was approximately
1000 g, and the molten mass was held for 20 min before casting in massive steel
moulds. There was a uniform reduction in diameter of the crucible due to visible cor-
rosion. The melts were contaminated with both zirconium (2.71–9.96 at.%) and oxy-
gen (0.45–0.8 at.%). There was no stoichiometric correlation between the amounts of
zirconium and oxygen taken up by the melt. Interestingly, calcium zirconate ingests
more zirconium into the melt than pure zirconia (2.1–3.1 at .%) crucibles (Friedrich
et al. 2008). The oxygen contamination was greater than what was observed after
melting TiAl in CaO crucibles (<0.35 m.%). under similar conditions. The crucible
was infiltrated by the melt. Beginning from the melt side, a reaction zone 2 mm thick
was also observed. Fused CaZrO3 did not prove to be a satisfactory refractory to melt
pure γ-TiAl.
In contact with cp-Ti, fused CaZrO3 reacted strongly, especially by dissolution
(Schafföner et al. 2015a, 2015b). As mentioned, electrofused raw material contains a
certain amount of c-ZrO2 impurities resulting from evaporation of CaO, and this is
responsible for the marked reactivity of fused calcium zirconate wares. Schaffoner et
al. (2015a) tested crucibles with four different compositions – one plain and the other
three CaO adjusted – in contact with Ti6Al4Vmelts. Melting experiments (1100
g batch, 1670°C, 20 min) were done with Ti6Al4Vin a vacuum induction furnace
under backfilled argon (800 mbar) to suppress the evaporation of volatile elements
such as calcium. The melt was cast in steel moulds.
Dissolution of the zirconia phase by the alloy melt occurred. The calcium zir-
conate phase did not dissolve. The zirconium and oxygen contamination of the melts,
which was highest for the plain fused calcium zirconate crucibles that contained a
substantial amount of c-ZrO2, decreased progressively as the c-ZrO2 proportion of
the calcium zirconate crucibles was decreased by increasing the amount of calcium
hydroxide additive in crucible making. The calcium hydroxide addition cannot be
increased without limit, because after a certain level, the crucibles started disintegrat-
ing. When Ti6Al4Vwas melted in yttria crucibles under similar conditions (Friedrich
et al. 2008), melt contamination by oxygen was 0.74 wt.% and by yttrium was 0.26
178 Metal–Crucible Interactions

wt.%. Lower levels of oxygen contamination (0.43 wt.%) were reached in the com-
position-adjusted fused calcium zirconate crucibles (0.50 wt.% Zr).
A substantial infiltration by the melt occurred in fused calcium zirconate crucible
with no Ca(OH)2 added. The infiltration by the titanium alloy melt decreased to a
hardly detectable level with calcium hydroxide additions. Possibly, the infiltration of
the crucibles occurred by a preferred dissolution of the zirconia phase accompanied
by a subsequent infiltration along the dissolved zirconia phase. It is known that zir-
conia dissolves into titanium even in the solid state (Ruh 1963; Lin et al. 1999b). The
zirconate phase remained unaffected.

[Link].2 Barium Zirconate
Barium zirconate was evaluated as a crucible material for melting titanium, TiAl,
TiNi, Ti2Ni and alsoTi6Al4V and cp-Ti. Melts richer in titanium significantly
increase the corrosion of BaZrO3 (Chen et al. 2019). The corrosion was suppressed
by additions of CaO and Y2O3 to BaZrO3. But, Y2O3 doping led to poor sintering,
which could be offset by adding TiO2. Compared with CaZrO3, crucibles made of
BaZrO3 (Zhang et al. 2013b) did not result in an interaction layer, and the oxygen
contamination was lower than in Y2O3 crucibles.
Chen et al. (2019) carried out pilot-scale melting of TiAl-based alloys using fused
BaZrO3 crucibles. Crucibles (inner diameter of 190 mm and length of 290 mm) were
prepared by cold isostatic pressing using BaZrO3 refractory of less than 3 mm aver-
age particle size. They were then fired at 1750°C for 6 h. About 25 kg of the com-
ponents (titanium sponge, aluminium ingot Al-60 wt.% Nb master alloy to yield
Ti46Al8Nb [at .%]) were charged in a BaZrO3 crucible and melted (1550°C, 30 s) in a
vacuum induction furnace under argon. The alloy melt was cast in a graphite mould.
The objective was to get a homogeneous Ti46Al-8Nb alloy with less than 0.1
wt.% oxygen (Tetsui et al. 2012a, 2012b). The average oxygen concentration of
ingots was about 0.078 wt.% (i.e. a pick-up of approximately 0.05 wt.% oxygen),
and the variation of oxygen content with sampling position in the ingot was mar-
ginal, indicating good homogeneity inspite of a very short holding time (30s)
during induction melting in 25 kg batches using190 mm diameter crucibles. In
experiments carried out on a similar scale using CaO and Y2O3 crucibles, the oxy-
gen content in the alloys was 0.24 and 0.32 wt.%, respectively (Friedrich et al.
2008). BaZrO3 appears to be a less contaminating refractory for induction melting
of TiAl-based alloys (Chen et al. 2019).
The Ca-doped BaZrO3 crucible was prepared starting from BaCO3, ZrO2 and
CaO by cold isostatic pressing followed by solid-state sintering at 1750°C. The
refractory consisted of Ba1−xCaxZrO3 and CaO phases (Chen et al. 2018a, 2018b).
Figure 4.21 shows the oxygen pick-up by the TiAl alloy in the Ca-doped BaZrO3 and
BaZrO3 crucibles, along with the literature data for samples prepared in other oxide
crucibles.
The objective stated by Chen et al. (2018c)was to produce TiAl ingot with a com-
position as close as possible to Ti-46 at.% Al. High-purity raw aluminium pellets
(>99.99%) and sponge titanium (>99.9%) charge were heated in a vacuum induction
furnace at 1600°C for 2 min under argon. The melt was either poured into a graphite
Major Melt–Crucible Systems 179

FIGURE 4.21 Oxygen pick-up by TiAl from calcium-doped barium zirconate (From
Chen et al. 2018c).

mould or solidified in the the Ca-doped BaZrO3 crucible. The oxygen concentration
of the resulting TiAl ingot was 0.09 wt.%, lower than the maximum permitted oxy-
gen concentration (Sakamoto et al. 1992). In comparison, the oxygen content of alloy
melted in a plain BaZrO3 crucible was about 0.13 wt.%.
The oxygen concentration of the TiAl alloy melted in the Ca-doped BaZrO3 cru-
cible (0.09 wt.%) was noticeably lower than for the other conventional crucibles such
as Y2O3, ZrO2 and Al2O3 as well as plain BaZrO3, although the thermochemical
stability of Y2O3 and CaO refractories was higher than that of the Ca-doped BaZrO3
refractory (Chen et al. 2017). A possible explanation could be a difference in the wet-
ting behaviour of these refractories. Ca-doped BaZrO3 is a very promising candidate
crucible for induction melting of the highly reactive TiAl alloys (Chen et al. 2018c).

[Link] CCIM
Cold crucible induction melting (CCIM) in a water-cooled segmented copper crucible
was one of the two methods used by Sakamoto et al. (1992) to prepare Ti-33–38 wt.%
Al alloys by melting together cp-titanium sheets and cp-aluminium blocks. Melting
was carried out in a crucible that could hold about 30 kg of titanium aluminide. The
temperature was 1550–1620°C and holding time 25 min – all under 2.7 × 104 Pa back-
filled argon. The oxygen content of the alloy charge was 0.034 wt.%, and the calcium
content was <0.001 wt.%, which did not change in the titanium aluminide obtained.
As seen in Figure 4.22, this was a melting truly free from crucible contamination.
180 Metal–Crucible Interactions

FIGURE 4.22 Change in oxygen content of TiAl by cold crucible induction melting. (From
Sakamoto et al. 1992).

[Link] EBM
Apart from CCIM, Sakamoto et al. (1992) prepared Ti-33–38 wt.% Al alloys by melt-
ing together cp-titanium and cp-aluminium in an EBM furnace. The oxygen content
of the charge decreased from 0.032 and 0.043 wt.% to <0.01 wt.%. Aluminium con-
tent also simultaneously decreased by 10 wt.%. The aluminium content is eventually
adjusted in EBM by adding Al through the feeder.

4.25.9 NiTi(-X)-Based Shape Memory Alloys (SMA)

The conventional process to produce NiTi alloy is by VIM. The limitations are con-
tamination of the bath by carbon from graphite crucibles or graphite moulds or by
oxygen from the use of MgO and Al2O3 crucibles (Jackson et al. 1972). Ni50Ti melts
at~1310°C. Alloy melting is generally performed between 1450 and 1500°C. These
temperatures are lower than those required for melting TiAl alloys. However, the
need to control composition within a very narrow margin near equiatomic compo-
sition does keep the pressure on the melter (Fashu et al. 2020). Besides, impurity
ingestion can easily make the alloy useless for its intended applications. The shape
memory effect and the related functional and strength properties of NiTi alloys
are strongly dependent on both composition and purity of the alloys, When carbon
Major Melt–Crucible Systems 181

TABLE 4.7
Interstitial Contamination in NiTi(-X) Alloy on Vacuum Induction Melting in
Various Crucibles
Duration Interaction
of layer Impurity
Crucible melting, thickness, content,
material min µm wt.% Reference
SiC 90 Not reported C: 1.129 Chapin and Friske 1954b, Sadrnezhaad
and Raz 2005
Al2O3 90 Not reported O: 1.6 Chapin and Friske 1954b, Sadrnezhaad
and Raz 2005
ZrO2 90 Not reported O: 1.046 Chapin and Friske 1954b, Sadrnezhaad
and Raz 2005
CaZrO3 5 30 Li et al. 2010
BaZrO3 2 20 O: 0.065–0.09 Chen et al. 2018b
High- 0.2–49 2–5 O: 0.02– Frenzel et al. 2004, Kim et al. 2001,
density 0.082C: Kramer 2009, Otubo et al. 2006,
graphite 0.017–0.25 Zhang et al. 2005, 2006

Source: Fashu et al. 2020.

ingested in NiTi alloys reacts with titanium and forms TiC, the Ni ‘content’ increases
and causes a shift of phase transition temperatures. A shift of 0.1 at .% in Ni content
causes a 10 K difference in phase transition temperatures. Oxygen and nitrogen are
also detrimental impurities. The impurity ingestion in NiTi alloys on melting in vari-
ous crucibles is summarized in Table 4.7 (Fashu et al. 2020).
As well as the right composition, a good chemical homogeneity is also needed in
the alloy ingot. Production of SMA components involves hot working of the homo-
geneous alloy and machining to the final shape (Frenzel et al. 2004).
Both BaZrO3 and high-density graphite are suitable crucible materials for melt-
ing the NiTi (-X) alloys (Fashu et al. 2020). In addition, certain on-the-spot tricks
are used, like lining graphite crucibles with titanium pieces while melting nickel–tita-
nium alloys. The arrangement of the charge in the crucible is as shown in Figure 4.23
(Frenzel et al. 2004; Zhang et al. 2005). This arrangement ensures that the protective
TiC layer forms quickly and early on during melting on the graphite crucible (Zhang et
al. 2005), and right from the first melting, the direct contact between Ni and graphite
is avoided. The net result is stopping the carbon transport from the graphite to the melt
and minimum carbon contamination. The maximum acceptable oxygen and carbon
content of NiTi alloys should not exceed 0.05 wt.% (Nayan et al. 2007). Table 4.7
shows the crucible–melt interaction data for NiTi-based SMAs melted using different
crucibles. Melts in ceramic oxide crucibles are not pure enough (Fashu et al. 2020).
Titanium melts at 1670°C, the melting point of nickel is 1455°C, and a NiTi melt
with 50.7 at .% Ni solidifies at 1310°C. Frenzel et al. (2004) prepared a NiTi alloy
182 Metal–Crucible Interactions

FIGURE 4.23 Arrangement of charge in graphite crucible for NiTi melting (From Frenzel
et al. 2014).

from Ni pellets and Ti rods by VIM in a graphite crucible, which was also the sus-
ceptor. The batch size was 1 kg and the target composition Ni-50.7–Ti-49.3 (at .%).
Thick discs (2 mm) were cut from the rod and used for cladding the crucible from
inside in some experiments.
In melting, the target temperature of 1550°C was reached within 8 min, and melt-
ing of Ni and subsequent dissolution of Ti in the Ni melt occurred within the next 1
min. The melt was held for 10s more to ensure sufficient mixing. Frenzel et al. (2004)
noted that in both the loading scenarios, nickel melted first, and then, the Ti rods
slowly sank and dissolved into the melt (in approximately 1 min). The melt was then
poured into a preheated (550°C) Y-shaped steelmould with an Y2O3 coating. Figure
4.24 shows carbon and oxygen concentrations in NiTi ingots processed in crucibles
that were used for the first (1) and second (2) time.
The first set in Figure 4.24 refers to the ingot that was produced using a crucible
filling arrangement where Ni was in direct contact with the graphite. Here, when the
crucible was used for the first time, the carbon content of the NiTi ingot is high. In the
second melting in the same crucible, the residual element contents come down to lower
than benchmark levels. In the second set, titanium discs are in direct contact with
graphite and not nickel. The carbon (and oxygen) contents in both the ingots are far
below the benchmark levels right from the first melting. The carbon–titanium reaction
is inhibited by many factors: (i) the solubility of carbon in solid Ti is very small, (ii) the
direct physical contact between the flat titanium plate lining and the curved inner sur-
face of the graphite crucible cannot be good, and (iii) the reaction is additionally hin-
dered by diffusion barriers formed by the initial interaction. One graphite crucible was
repeatedly used by Frenzel et al. (2004) for as many as 10 batches, and the carbon and
oxygen contamination levels remained quite reproducible and significantly below the
accepted industrial standard. Irrespective of the way the components are arranged in
the crucible, the carbon ingestion into the melt in the second melting in the same cruci-
ble is lower. This is attributed by Frenzel et al. (2004) to the formation of a TiC layer in
Major Melt–Crucible Systems 183

FIGURE 4.24 Carbon and oxygen contents in NiTi melted in graphite crucibles (From
Frenzel et al. 2004).

the inner surface of the crucible by the time the first melt gets over. SEM micrographs
revealed the presence of the reaction layer after the first melt. Significantly, the thick-
ness of the reaction layer does not change much with the number of melts completed in
the crucible. In a campaign of multiple melting, the special charge arrangement needs
to be done for only the very first melting. All subsequent meltings can be done follow-
ing the easier arrangement (no cladding).
Elahinia et al. (2012) also used a graphite crucible in VIM but did not allow the
melt temperature to exceed 1450°C. This was 100°C lower than the temperature used
by Frenzel et al. (2004). The carbon content in the ingot was controlled to between
200 and 500 ppm, and at this level, carbon does not affect the shape memory charac-
teristics of the alloy. Elahinia et al. (2012) used the procedure and arrangement out-
lined by Frenzel et al. (2004) but doubled the batch size to 1 kg. The carbon content
in final products ranges from 0.04% to 0.06% (Otubo et al. 2003b). VIM also ensures
melt homogeneity by the strong electromagnetic stirring.
Otubo et al. (2006), in a classic investigation, directly addressed the effects of
graphite quality and batch size of the NiTi melt on achieving the residual carbon con-
tent acceptable for commercial NiTi SMA products. The ingot produced in a high-
porosity graphite crucible from a charge of Ti plates intercalated with electrolytic Ni
184 Metal–Crucible Interactions

plates presented the highest carbon contamination of 2440 ppm (ppm are in wt.%).
Changing to a low-porosity crucible, the contamination was reduced by a factor of
two (1000–1880 ppm). This decrease is due to a decrease of actual contact area due
to the change from a high-porosity graphite crucible to a low-porosity crucible of
the same apparent contact area. As the melting batch size was scaled up, and in the
process, the specific contact area (area of contact with the graphite per unit weight of
the melt) decreased, the average carbon content was reduced by another factor of two
to reach around 600 ppm. The oxygen content generally remained between 400 and
900 ppm. This reduction in carbon contamination is the consequence of the change
in the ratio between apparent contact area and liquid volume, which decreased from
0.68 to 0.43, representing a reduction of 37%, while the ingot mass increased from
1.5 kg in the second stage to 4.3 kg in the third stage, that is, almost three times.

[Link] Barium Zirconate (Calcia-Doped) Crucibles

Barium zirconate shows good corrosion resistance to TiNi, TiFe and TiAl melts, but
it still exhibited insufficient stability against alloys rich in titanium. This limitation
was addressed by Chen et al. (2018a) by using CaO-doped BaZrO3 crucibles, which
were less corroded by Ti2Ni melt than the BaZrO3 crucible. The CaO-doped BaZrO3
refractory may be potentially the more suitable refractory for induction melting of
the titanium alloys. Chen et al. (2017) also carried out similar experiments with
yttria-doped BaZrO3 to melt TiNi and Ti2Ni alloys. Doping with Y2O3 decreases the
sinterability of the BaZrO3 crucible, but the crucible becomes more resistant to melt
attack. The oxygen content of the TiNi alloy increased from 0.045% in the charge
to 0.058% in the plain crucible but only to 0.051% in the doped crucible. The corre-
sponding values in the case of Ti2Ni were 0.36% and 0.09% oxygen. The contamina-
tion by barium and zirconium was also similarly much less in the alloy melted in the
doped crucible compared with the one melted in the plain barium zirconate crucible.
The extent of erosion of yttria-doped BaZrO3 crucible by both the melts is lower than
that of the plain BaZrO3 crucible. The yttria-doped BaZrO3 appeared more suitable
for induction melting of titanium alloys (Chen et al. 2017).

[Link] VAR and EBM

SMA components have been commercially produced by argon arc melting of pure
components to get a homogeneous alloy, followed by a hot working and machining
to the final shape (Frenzel et al. 2004).
The use of the EBM process to produce NiTi SMA in practice was begun by
Otubo et al. (2003a, 2003b) in Japan. Nickel has a higher vapour pressure than Ti
and can evaporate during EBM. The nickel loss, which was the initial concern in
using EBM, can be overcome by controlling the EB power and the initial composi-
tion of the raw materials. Otubo et al. (2004) have shown that composition control is
possible in NiTi alloys as long as certain precautions are taken during the feed mate-
rial preparation. EBM product has a lower carbon content, ranging from 0.007% to
0.016% compared with 0.04% to 0.06% in the NiTi alloy commercially made by
VIM (Otubo et al. 2003b). Oxygen impurity is also lower (Otubo et al. 2004).
Major Melt–Crucible Systems 185

4.25.10 Hydrogen Storage Alloys

As a class of titanium-based functional materials, hydrogen storage alloys, like SMA,
also have a critical requirement for composition control. They are very sensitive to
contamination by non-metallic impurities, deviations from the target composition
and inhomogeneity.
Fashu et al. (2020) suggest, based on published information, the use of crucibles
made of CaO, BaZrO3 or CaZrO3 for melting TiFe-based alloys. The sensitivity of
CaO refractories towards water vapour can be largely resolved by doping with Y2O3,
which results in the formation of metastable CaY2O4 phase on grain boundaries. One
technique used to manage contamination issues in alloy melting is to minimize the
contact time between the crucible and the reactive components in the melt. The less
reactive components are pre-alloyed, e.g. Mn with FeV, before co-melting them with
more reactive (rare earth metal deoxidizer) and more refractory (titanium, zirco-
nium, chromium) components. In certain practices, Ti- and Zr-containing AB2-type
hydrogen storage alloys are pre-melted in large-capacity (>1000 kg/load) crucible-
free plasma heated skull furnaces (Friedrich 1994).
TiFe alloy melted in a BaZrO3 crucible had better hydrogen storage properties
than one melted in a graphite crucible (Li et al. 2015). The inertness of BaZrO3 was
enhanced after doping with CaO (Chen et al. 2018a) and that of CaZrO3 after add-
ing Al2O3 (Lu et al. 2017, 2019). This was attributed to altered wetting behaviour of
these ceramic crucibles. CaZrO3 doped with Al2O3, Y2O3 and MgO has been shown
to possess higher chemical stability and good tolerance against thermal shock.
The melting of TiFe in a graphite crucible (Li et al. 2015) results in contamination
of the ingot by carbon. During melting, TiC forms and sticks strongly to the crucible
wall. On melting Ti- and Zr-based AB2-type alloys containing Mn, carbides of man-
ganese form even at lower temperatures, below 1300°C. Severe interaction of the
melt with the graphite can even cause the crucible to disintegrate.
Fashu et al. (2020) have observed a satisfactory performance of crucibles made
of Al2O3, Al2O3·SiO2 and low-density graphite lined with multilayer coatings where
the external (facing the melt) layer used was Y2O3. In trial melts, the AB2-type alloy
did not stick to the crucible even after solidification inside, and displayed hydrogen
sorption properties similar to a reference sample made by arc melting.

4.26 SUMMARY
In this chapter, the status as regards the metal–crucible systems in some of the most
important metal processing operations has been summarized. Few metals have been
as closely dependent on availability of crucibles for their extraction as iron. The
reduction process predisposes the as-produced iron, called the hot metal or pig iron,
to be liberally contaminated with a variety of impurities. The metal is still usable as
such for a few applications (after a certain amount of metallurgical tending), but as
the impurities are progressively decreased both in number and in concentration, the
variety of uses increases greatly, and the uses become more sophisticated. ‘Clean
steel’, which has less than 50 ppm of residual impurities (PCSNHO), is the cleanest
186 Metal–Crucible Interactions

metal produced routinely under industrial conditions in tonnage quantities. The pro-
cess metallurgy part of iron and steel is all about impurities control, i.e. removal of
impurities from the metal and protection of the metal from re-ingestion of impuri-
ties during further processing to a finished product. Refractories (crucibles) play a
direct role here. The refractories undergo exemplary physical, thermal and chemical
stresses in the various equipment for the making of iron and steel, and these facilities
have long been the test track for new and improved refractories, not only for ferrous
metallurgy but for all non-ferrous metallurgical operations as well.
Almost all the world’s aluminium is produced by electrolysis in Hall–Héroult
cells. The issues of metal–crucible interaction in the Hall–Héroult cell are mainly
concerned with the interaction of the molten metal and the bath components with the
cathode and the refractory layer beneath the cathode. Even though the temperatures
involved are relatively low, interactions are many, and intensity is high. In the design
and construction of the cell, the objective is to have the 800–850°C isotherms within
the main body of the upper, protective refractory layer (Siljan et al. 2002). The soli-
dus of the advancing melt falls in this temperature window. The most important
purpose of the refractory layer, i.e. as a barrier to bath intrusion and to avoid deterio-
ration of the thermal insulation layers below, will then be served.
Key interactions to be considered in the cell involve the electrolyte, molten alu-
minium and sodium with the cell materials, particularly the cathode carbon block
and the refractory beneath it.
Sodium is generated and also transported through the cathode carbon. As well
as sodium and electrolyte, aluminium also ends up at the other side of the cathode
due to damage to the cathode. The pathways created by the damage allow liquid
aluminium to penetrate down to the top of the refractory lining. Then, a number of
reactions occur between the penetrating materials and the refractory, which result in
the formation of new phases, causing volume expansion and also viscosity changes
in the melt. Typically, the barrier refractory layer is made of aluminosilicates (usu-
ally fireclay), which are prone to infiltration and attack by molten electrolyte compo-
nents. Conditions are chosen such that the reaction among cryolite, sodium fluoride,
alumina, silica and mullite lead to the formation of albite (NaAlSi3O8), which raises
the viscosity so much that melt intrusion into the barrier refractory is halted. In prac-
tice, this (albite-rich melt) is achieved by selecting a refractory with silica content
above 60% and by a controlled input of sodium compounds, e.g. with excess sodium
fluoride.
As-reduced aluminium from the Hall–Héroult cell is purified further in post-
reduction treatments by fluxing (and skimming) and degassing before casting the
99.8–99.85% Al into moulds. The general requirements for refractories in aluminium
melting furnaces are porosity <15%, average pore size <5 μm and low wetting by Al.
The use of anti-wetting additions to refractories in contact with molten aluminium
is a popular approach. Many different materials have been used for this purpose; the
details remain proprietary, and so do the exact mechanism(s) of their action.
The furnaces used for producing molten copper from concentrates and scrap face
stiff challenges to the refractory life in the form of highly aggressive slags, mechani-
cal stresses, batch operation and higher operating temperatures. Mag–chrome is the
Major Melt–Crucible Systems 187

refractory that serves well here, with the composition and type tweaked depending
on the zone of its use. When everything appeared to be settling down, the use of
‘mag–chrome’ refractories began to be scrutinized and questioned as being a source
of a possible environmental villain, hexavalent chromium. In a way, the search for a
‘more’ suitable refractory has begun all over again. Mag–alumina instead of mag–
chrome is a possibility in almost every copper production application. Periclase–spi-
nel or spinel–periclase might play a wider role. Doloma and mag–doloma might be
an option for refining furnaces, where highly basic slags and highly oxidized copper
make the replacement of mag–chrome a work in progress.
Due to their notoriously reactive nature and the need for elevated temperatures
for their production and processing, the task of handling almost every member of
the less common metals group has been challenging. Metallothermic reductions are
generally exothermic reactions and have been extensively used for the preparation of
less common metals.
For a host of technical reasons, the compound chosen is either an oxide or a
fluoride or a chloride, and the reducing agent is aluminium, calcium, magnesium,
sodium, potassium or lithium. Reduction using a metal invariably needs a high tem-
perature, attained either by external heating or by the exothermicity of the reaction
or by a combination of both. An inert containment is a prerequisite to protect the
compound intermediate, the reducing agent and the as-reduced metal from contami-
nation. The seriousness of a reactive melt increases first of all with its reactivity and
also when the melt forms at a higher temperature and in larger quantities and persists
for longer durations. A bomb reactor, which is a refractory-lined steel vessel that is
usually closed during the course of a reduction, is often used. The refractory lining
is principally used to contain the melt while it forms and separates into two layers –
metallic melt as the bottom layer and slag as the top layer. The linings used include
dolomite, magnesia, alumina, and even calcium fluoride and magnesium fluoride
when these fluorides form as a product of metallothermic reduction. In another large
group of reductions, unalloyed tantalum crucibles are the containers (for rare earth
metals) in reactions conducted in vacuum induction furnaces. In relatively lower-
temperature reductions, molybdenum or titanium containers are frequently used,
and the reactor inside may also be lined with molybdenum or tantalum. In certain
sophisticated reduction arrangements, water-cooled copper vessels are used as cru-
cibles, and it is the thin layerof the metal being processed frozen on the water-cooled
vessel that is the de facto container.
Fused salt electrolysis methods score over metallothermic reduction as superior
processes for certain rare metal reductions. The cells are built using graphite, molyb-
denum and tungsten as materials that contact the electrolyte and the reduced metal.
A technically sophisticated method, pioneered by the Reno group, is to collect the
reduced metal in a frozen electrolyte skull. Even yttrium (melting point 1526°C) was
electrowon in a fluoride electrolyte cell operated at 1700°C.
The melting and casting of titanium brings forth every conceivable challenge that
can exist in the melt processing of a reactive metal. Perhaps, only vanadium sur-
passes titanium in being the unsolved problem in this context. The critical process
is alloy melting. The crucible–melt interactions in a VIM operation depend on the
188 Metal–Crucible Interactions

titanium content of the alloy as well as the temperature needed for its melting. In
practice, TiAl(γ) alloys can be melted at temperatures between 1550 and 1750°C,
while Ti6Al4V alloys require 1660–1800°C due to the high Ti content. Until now,
no refractory material has been found to be absolutely inert against these alloys, and
some interactions between melt and crucible and even mould materials have always
occurred during melting and casting. Cold crucible methods have been used to keep
contamination levels low, but they have other issues,such as batch size and superheat.
Among the large number of crucible options examined by numerous investigators,
yttria, calcia, aluminium nitride, graphite, barium zirconate and calcium zirconate
have been the most promising. The usage may involve a monolithic crucible or a
coating. Impurities and porosity need to be closely controlled to derive the maximum
benefit of chemical stability provided by these crucible materials.
In this chapter, building on the matter presented in the earlier chapters, all major
and tangible metal–crucible interaction scenarios have been examined. It is true that
in the field of metals processing, there are literally hundreds of interaction situations,
each unique in its own way and depending on many parameters, only some of which
are quantifiable or quantified. It will be highly desirable if a ready reckoner can be
made to accurately predict the performance expected of a given melt–crucible com-
bination. But, such an effort may lead to a path going nowhere, because the drivers
of each published interaction are not available for complete and accurate listing. The
way to go is, therefore, armed with the information available so far, to try to solve the
emerging problems by analogy and experimentation. This volume is created with the
intention of helping the reader to do this. The quest is on.
References
Adabifiroozjaei, E., Koshy, P., and Sorrell, C.C. (2015). Assessment of non-wetting materi-
als for use in refractories for aluminium melting furnaces. Journal of the Australian
Ceramic Society, 51(1): 139–145.
Afshar, S. and Allaire, C. (1996). The corrosion of refractories by molten aluminum. JOM:
The Journal of The Minerals, Metals & Materials Society, 48(5): 23–27.
Afshar, S. and Allaire, C. (2001). Furnaces: Improving low cement castables by non-wetting
additives. JOM:The Journal of The Minerals, Metals & Materials Society, 53(8): 24–27.
Aguilar-Santillan, J. (2008). Wetting of Al2O3 by molten aluminum: The influence of BaSO4
additions. Journal of Nanomaterials, Vol. 2008, Article ID 629185, 12 pp.
Alangi, N., Mukherjee, J., Anupama, P., Verma, M.K., Chakravarthy, Y., Padmanabhan,
P.V.A., Das, A.K., and Gantayet, L.M. (2011). Liquid uranium corrosion studies of pro-
tective yttria coatings on tantalum substrate. Journal of Nuclear Materials, 410(1–3):
39–45.
Allaire, C. (1992). Refractories for the lining of holding and melting furnaces. In: M.M.
Avedesian et a1. (Ed.), Advanced in production and fabrication of light metals and
metal matrix composites. CIM, Montreal, Canada, pp. 163–174.
Allaire, C. (2000). Refractories for molten aluminium confinement. Canadian Ceramics,
69(1): 14–21.
Allen, N.P., Kubaschewski, O., and Von Goldbeck, O. (1951). The free energy diagram of the
vanadium‐oxygen System. Journal of the Electrochemical Society, 98(11): 417–424.
Baldwin, W.J. (1948). Zircon and zirconia refractories. Chemical Engineering Progress, 44:
875.
Barbosa, J. and Ribeiro, C.S. (2000). Influence of crucible material on the level of contamina-
tion in TiAl using induction melting. International Journal of Cast Metals Research,
12(5): 293–301.
Barbosa, J., Ribeiro, C.S., and Monteiro, A.C. (2003). Processing of γ- TiAl by ceramic
crucible induction melting, and pouring in ceramic shells. Materials Science Forum,
426–432: 1933–1938.
Barbosa, J., Ribeiro, C.S., Teodoro, O.M.N.D., and Monteiro, A.C. (2005). Evaluation of
Y2O3 front layer of ceramic crucibles for vacuum induction melting of TiAl based
alloys, EPD Congress 2005. In: Proceedings of sessions and symposia TMS annual
meeting, TMS, Warrendale, PA.
Barbosa, J., Puga, H., Ribeiro, C.S., Teodoro, O.M.N.D., and Monteiro, A.C. (2006).
Characterisation of metal/mould interface on investment casting of γ-TiAl. International
Journal of Cast Metals Research, 19(6): 331–338.
Barbosa, J., Ribeiro, C.S., and Monteiro, A.C. (2007). Influence of superheating on casting of
γ- TiAl. Intermetallics, 15(7): 945–955.
Bayley, J. and Eckstein, K. (2006). Roman and medieval litharge cakes: Structure and com-
position. In: J. Pérez-Arantegui (Ed.), Proceedings of the 34th international symposium
on archaeometry, Institución Fernando el Católito. CSIC, Zaragoza, pp. 145–153.
Beard, A.P. and Crooks,D.D. (1954). Kilogram scale reductions of vanadium pentoxide to
vanadium metal, J. Electrochem. Soc., 101(12): 597–600.
Beaudry, B.J. and Gschneidner, Jr., K.A. (1978). Preparation and basic properties of rare earth
metals. In: K.A. Gschneidner, Jr. and L. Eyring (Eds.), Handbook on the physics and chem-
istry of rare earths (Vol. 1). North Holland Publishing Company, Amsterdam, pp. 173–232.

189
190 References

Biswas, S. and Sarkar, D. (2020). Introduction to refractories for iron and steelmaking,
Springer International Publishing, Switzerland.
Block, F.E. (1960). Preparation of high-purity yttrium by metallic reduction of yttrium
trichloride, report of Investigations 5588. Bureau of Mines, U.S. Department of the
Interior, Washington, DC.
Bonadia, P., Braulio, M., CAllo, J.B., and Pandolfelli, V.C. (2006). Refractory selection for
long-distance molten-aluminium delivery. American Ceramic Society Bulletin, 85(8):
9301–9309.
Booth, S.H. (1940). Use of refractories in melting copper and copper alloys in an Ajax-Wyatt
induction furnace. Bulletin of the American Ceramic Society, 19: 171.
Bowen, N.L. and Greig, J.W. (1924). The system: Al2O3−SiO2. Journal of the American
Ceramic Society, 7(4): 238–254.
Brace, P.H. (1948). Reactions of molten titanium with certain refractory oxides. Journal of
the Electrochemical Society, 94(4): 170–176.
Braulio, M.A.L., Martinez, A.T., Luz, A.P., Liebske, C., and Pandolfelli, V.C. (2011). Basic
slag attack of spinel-containing refractory castables. Ceramics International, 37(6):
1935–1945.
Bravo, I.M.V., PINA, I.P., and Ransanz, G.R. (2017). Electrofused magnesio-chromite:
Complex refractory material. Macla: Revista de la Sociedad Española de Mineralogía,
22: 89–90.
Brewer, L., Bromley, L.A., Gilles, P.W., and Lofgren, N.L. (1948). The preparation and tests
of refractory sulfide crucibles (September 7), AECD-2253.
Bridges, P.J. and Hauzeur, F. (1991). Advances in the technology of titanium castings. Cast
Metals, 4(3): 152–154.
Brondyke, K.J. (1953). Effect of molten aluminum on alumina‐silica refractories. Journal of
the American Ceramic Society, 36(5): 171–174.
Brook, R.J. (Ed.). (1991). Concise encyclopedia of advanced ceramic materials. Pergamon
Press, Oxford.
Buhr, A., Bruckhausen, R., and Fahndrich, R. (2016). The steel industry in Germany–Trends
in clean steel technology and refractory engineering. Refractories World Forum, 8(1):
57–63.
Bullock, E., Brunetaud, R., Condé, J.F., Keown, S.R., and Pugh, S.F. (Eds.). (1989). Research
and development of high temperature materials for industry. Elsevier Applied Science,
London.
Bunch, B.H. and Hellemans, A. (2004). The history of science and technology: A browser’s
guide to the great discoveries, inventions, and the people who made them, from the
dawn of time to today. Houghton Mifflin Company, Boston, MA.
Calverley, A. and Lever, R.F. (1957). The floating-zone melting of refractory metals by elec-
tron bombardment. Journal of Scientific Instruments, 34(4): 142–147.
Calvert, E.D. (1981). An investment mold for titanium casting, report of investigations 8541.
Bureau of Mines, U.S. Department of the Interior, Washington, DC.
Camel, D., Drevet, B., Brizé, V., Disdier, F., Cierniak, E., and Eustathopoulos, N. (2017).
The crucible/silicon interface in directional solidification of photovoltaic silicon. Acta
Materialia, 129: 415–427.
Canterford, J.H. (1985). Magnesia—An important industrial mineral: A review of process-
ing options and uses. Mineral Processing and Extractive Metallurgy Review, 2(1–2):
57–104.
Carlson, O.N., Burkholder, H.R., Martsching, G.A., and Schmidt, F.A. (1981). Preparation of
high purity vanadium. In: H.Y. Sohn, O.N. Carlson, and J.T. Smith (Eds.), Extractive
metallurgy of refractory metals—110th annual meeting, The Metallurgical Society of
AIME. Warrandale, PA, pp.191–203.
References 191

Carlson, O.N., Schmidt, F.A., and Krupp, W.E. (1966). A process for preparing high—Purity
vanadium. JOM:The Journal of The Minerals, Metals & Materials Society, 18(3):
320–323.
Cegan, T. Szurman, I., Kursa, M., Holešinský, J., and Vontorová, J. (2016). Preparation of TiAl-
based alloys by induction melting in graphite crucibles. Kovove Materialy, 53(2): 69–78.
Chakravarthy, Y., Bhandari, S., Chaturvedi, V., Pragatheeswaran, A., Nagraj, A.,
Thiyagarajan, T.K., Ananthapadmanaban, P.V., and Das, A.K. (2015). Plasma spray
deposition of yttrium oxide on graphite, coating characterization and interaction with
molten uranium. Journal of the European Ceramic Society, 35(2): 787–794.
Chang, Y.W. and Lin, C.C. (2010). Compositional dependence of phase formation mechanisms
at the interface between titanium and calcia‐stabilized zirconia at 1550°C. Journal of
the American Ceramic Society, 93(11): 3893–3901.
Chapin, E.J., and Friske, W.H. (1954a). A Metallurgical Evaluation of Refractory Compounds
for Containing Molten Titanium, Part I-Oxides (No. NRL-4447). Naval Research Lab,
Washington, DC.
Chapin, E.J. and Friske, W.H. (1954b). A metallurgical evaluation of refractory compounds
for containing molten titanium, part II-carbon, graphite, and carbides (No. NRL-
4467). Naval Research Lab, Washington, DC.
Chapin, E.J. and Friske, W.H. (1955). A metallurgical evaluation of refractory compounds
for containing molten titanium. Part III-borides and sulfides (No. NRL-4478). Naval
Research Lab, Washington, DC.
Chen, G., Gao, P., Kang, J., Li, B., Ali, W., Qin, Z., Lu, X., and Li, C. (2017). Improved sta-
bility of BaZrO3 refractory with Y2O3 additive and its interaction with titanium melts.
Journal of Alloys and Compounds, 726: 403–409.
Chen, G., Kang, J., Gao, P., Ali, W., Qin, Z., Lu, X., and Li, C. (2018a). Effect of CaO addi-
tive on the interfacial reaction between the BaZrO3 refractory and titanium enrichment
melt. In: Kim, H., Wesstrom, B., Alam, S., Ouchi, T., Azimi, G., Neelameggham, N.
R., Wang,S., Guan, X. (Eds.) Rare Metal Technology. Springer, Cham, pp. 235–244.
Chen, G., Kang, J., Gao, P., Qin, Z., Lu, X., and Li, C. (2018b). Dissolution of BaZrO3 refrac-
tory in titanium melt. International Journal of Applied Ceramic Technology, 15(6):
1459–1466.
Chen, G., Kang, J., Lan, B., Gao, P., Lu, X., and Li, C. (2018c). Evaluation of Ca-doped
BaZrO3 as the crucible refractory for melting TiAl alloys. Ceramics International,
44(11): 12627–12633.
Chen, G., Lan, B., Xiong, F., Gao, P., Zhang, H., Lu, X., and Li, C. (2019). Pilot-scale experi-
mental evaluation of induction melting of Ti-46Al-8Nb alloy in the fused BaZrO3 cru-
cible. Vacuum, 159: 293–298.
Chindgren, C.J., Bauerle, L.C., and Rosenbaum, J.B. (1963). Modifications in bomb reduction
of vanadium oxide, report of investigations 6284. Bureau of Mines, U.S. Department of
the Interior, Washington, DC.
Clites, P.G. and Calvert, E.D. (1961). Laboratory casting high-temperature metals. JOM: The
Journal of The Minerals, Metals & Materials Society, 13(2): 136–138.
Cooper, A.R. and Kingery, W.D. (1959). Corrosion of refractories by liquid slags and glasses.
In: W.D. Kingery (Ed.), Kinetics of high-temperature processes. John Wiley and Sons,
New York, pp. 85–92.
Cooper, D.E., Snow, K.S., Fulton, J.C., and Tommaney, J.W. (1965). Feasibility experiments
of vacuum arc remelting of 28 in.-diameter ingots in three-electrode AC furnace. JOM:
The Journal of The Minerals, Metals & Materials Society, 17(12): 1368–1373.
Croat, J.J. (1969). The preparation of high purity dysprosium, holmium and erbium by the
lithium reduction of their trichloride salts, No. IS-T-346, Ames Laboratory, ERDA, Iowa
State University, Ames, IA. National Technical Information Service, Springfield, VA.
192 References

Cui, R., Gao, M., Zhang, H., and Gong, S. (2010). Interactions between TiAl alloys and yttria
refractory material in casting process. Journal of Materials Processing Technology,
210(9): 1190–1196.
Cui, R.J., Tang, X.X., Gao, M., Zhang, H., Gong, S.K. (2012). Microstructure and composi-
tion of cast Ti-47Al-2Cr-2Nb alloys produced by yttria crucibles, Mater. Sci. Eng. A
541: 14–21.
Daane, A.H. and Spedding, F.H. (1953). Preparation of yttrium and some heavy rare earth
metals, J. Electrochem. Soc., 100: 442–444.
Daane, A.H., Dennison, D.H. and Spedding, F.H. (1953). The preparation of samarium and
ytterbium metals, J. Am. Chem. Soc., 75: 2272–2273.
Daane, A.H. (1961a). Metallothermic preparation of rare earth metals. In: F.H. Spedding and
A.H. Daane (Eds.), The rare earths. Wiley, New York, pp. 102–112.
Daane, A.H. (1961b). Yttrium. In: C.A. Hampel (Ed.), Rare metals handbook. Reinhold,
London, pp. 653–666.
Daane, A.H. and Spedding, F.H. (1953). Preparation of yttrium and some heavy rare earth
metals. Journal of the Electrochemical Society, 100(10): 442–444.
Darling, A.S. (1990). Non-ferrous metals. In: I. McNeil (Ed.), An encyclopaedia of the history
of technology. Routledge, London, pp. 47–145.
Dennison, D.H., Tschetter, M.J., and Gschneidner, Jr., K.A. (1966a). The solubility of tanta-
lum in eight liquid rare earth metals. Journal of the Less Common Metals, 10: 108–115.
Dennison, D.H., Tschetter, M.J., and Gschneidner, Jr., K.A. (1966b). The solubility of Ta and
W in liquid rare earth metals. Journal of the Less Common Metals, 11: 423–435.
Dunsing, T.W. (2002). Tips on good refractory practice for aluminum-melting furnaces.
Industrial Heating, 69(4): 27–29.
Durov, A.V., Naidich, Y.V., and Kostyuk, B.D. (2005). Investigation of interaction of metal
melts and zirconia. Journal of Materials Science, 40(9–10): 2173–2178.
Eastman, E.D., Brewer, L., Bromley, L.A., Gilles, P.W., and Lofgren, N.I. (1951). Preparation
and tests of refractory sulfide crucibles. Journal of the American Ceramic Society,
34(4): 128–136.
Eastwood, L.W. and Craighead, C.M. (1950). Refractories for melting titanium part I (No.
AF-TR-6218 (pt. 1)). Battelle Memorial Inst., National Technical Information Service,
Springfield, VA.
Eatesami, D., Hadavi, M.M., and Habibollahzade, A. (2009). Melting of γ-TiAl in the alu-
mina crucible. Russian Journal of Non-Ferrous Metals, 50(4): 363–367.
Economos, G. and Kingery, W.D. (1953). Metal‐ceramic interactions: II, Metal‐oxide inter-
facial reactions at elevated temperatures. Journal of the American Ceramic Society,
36(12): 403–409.
Elahinia, M.H., Hashemi, M., Tabesh, M., and Bhaduri, S.B. (2012). Manufacturing and pro-
cessing of NiTi implants: A review. Progress in Materials Science, 57(5): 911–946.
Ellingham, H.J.T. (1944). Reducibility of oxides and sulphides in metallurgical processes.
Journal of the Society of Chemical Industry, 63(5): 125–160.
Emeis, R. (1954). Tiegelfreies Ziehen von Silicium-Einkristallen. Zeitschrift für
Naturforschung A, 9(1): 67–68.
Engel, R. (2015). Refractory considerations. Aluminium International Today, 27(3): 23.
Erb, A., Walker, E., and Flükiger, R. (1995). BaZrO3: The solution for the crucible corrosion
problem during the single crystal growth of high-Tc superconductors REBa2Cu3O7−δ;
RE = Y, Pr, Physica C: Superconductivity, 245(3–4): 245–251.
Fadeev, A.V., Bazhenov, V.E., and Koltygin, A.V. (2015). Improvement in the casting technol-
ogy of blades for aviation gas-turbine engines made of TNM-B1 titanium aluminide
alloy produced by induction crucible melting. Russian Journal of Non-Ferrous Metals,
56(1): 26–32.
References 193

Fahey, N.P., Swinbourne, D.R., Yan, S., and Osborne, J.M. (2004). The solubility of Cr2O3
in calcium ferrite slags at 1573 K. Metallurgical and Materials Transactions B, 35(2):
197–202.
Fashu, S., Lototskyy, M., Davids, M.W., Pickering, L., Linkov, V., Tai, S., Renheng, T.,
Fangming, X., Fursikov, P.V., and Tarasov, B.P. (2020). A review on crucibles for induc-
tion melting of titanium alloys. Materials and Design, 186: 108295.
Field, T.E. (1947). Heat-cast refractory for magnesium melting furnaces, U. S. Pat. 2,416,472
(Feb. 25, 1947).
Finlay, G.R. and Fetterly, G. (1952). Boron nitride — An unusual refractory. Journal of the
American Ceramic Society, 31: 141.
Forrester, R. (2019). History of metallurgy (2nd Ed.). SSRN Electronic Journal. doi:10.2139/
ssrn.2864178.
Freitag, L., Schafföner, S., Lippert, N., Faßauer, C., Aneziris, C.G., Legner, C., and Klotz,
U.E. (2017). Silica-free investment casting molds based on calcium zirconate. Ceramics
International, 43(9): 6807–6814.
Frenzel, J., Zhang, Z., Neuking, K., and Eggeler, G. (2004). High quality vacuum induction
melting of small quantities of NiTi shape memory alloys in graphite crucibles. Journal
of Alloys and Compounds, 385(1–2): 214–223.
Friedrich, B. (1994). Large-scale production and quality assurance of hydrogen storage (bat-
tery) alloys. Journal of Materials Engineering and Performance, 3(1): 37–46.
Friedrich, B., Lochbichler, C., Brenk, J., Vonderstein, C., Schafföner, S., and Aneziris,
C. (2016). Ceramic crucible melting of titanium alloys and intermetallics in VIM,
Presentation, April 2016, Titanium Europe 2016, Paris. [Link] RG​.2​
.1​.3992​.6161
Frueh, C., Poirier, D.R., Maguire, M.C., and Harding, R.A. (1996). Attempts to develop a
ceramic mould for titanium casting—A review. International Journal of Cast Metals
Research, 9(4): 233–239.
Frueh, C., Poirier, D.R., and Maguire, M.C. (1997). The effect of silica-containing binders
on the titanium/face coat reaction. Metallurgical and Materials Transactions B, 28(5):
919–926.
Fruehan, R.J. (Ed.). (1998). The making, shaping and treating of steel; Hubble, D.H., chapter 3
steel plant refractories; Hubble, D. H., Russell, R.O., Vernon, H.L. and Marr, R.J., chap-
ter 4, steelmaking refractories (11th Ed.). The AISE Steel Foundation, Pittsburgh, PA.
Gale, W.K.V. (1990). Ferrous metals. In: I. McNeil (Ed.), An encyclopaedia of the history of
technology. Routledge, London, pp. 146–185.
Gao, M., Cui, R., Ma, L., Zhang, H., Tang, X., and Zhang, H. (2011). Physical erosion of
yttria crucibles in Ti–54Al alloy casting process. Journal of Materials Processing
Technology, 211(12): 2004–2011.
Gayler, M.L.V. (1930). The preparation and properties of chemically pure metals.
Metallwirtscbaft, 9: 677.
Gingerich, K.A. (1968). Gaseous metal nitrides. II. The dissociation energy, heat of subli-
mation, and heat of formation of zirconium mononitride. The Journal of Chemical
Physics, 49(1): 14–18.
Glazunov, S.G. (1970). Precision casting of titanium. In: R.I. Jaffee and N.E. Promisel
(Eds.), The science, technology and application of titanium. Elsevier, Amsterdam, pp.
143–147.
Goldschmidt, H. and Vautin, C. (1898). Aluminium as a heating and reducing agent. Journal
of the Society of Chemical Industry, 6(17): 543–545.
Gomes, F., Puga, H., Barbosa, J., and Ribeiro, C.S. (2011). Effect of melting pressure and super-
heating on chemical composition and contamination of yttria-coated ceramic crucible
induction melted titanium alloys. Journal of Materials Science, 46(14): 4922–4936.
194 References

Gomes, F., Barbosa, J., and Ribeiro, C.S. (2013). Aluminium evaporation during ceramic
crucible induction melting of titanium aluminides. Materials Science Forum, 730:
697–702.
Goto, K., Argent, B.B., and Lee, W.E. (1997). Corrosion of MgO—MgAl2O4 spinel refrac-
tory bricks by calcium aluminosilicate slag. Journal of the American Ceramic Society,
80(2): 461–471.
Gray, P.M.J. (1951). The production of pure cerium metal by electrolytic and thermal reduc-
tion processes. Transactions of the Institution of Mining and Metallurgy, 61: 141–170.
Griesenauer, N.M., Lyon, S.R., and Alexander, C.A. (1972). Vacuum induction melting of
titanium. Journal of Vacuum Science and Technology, 9(6): 1351–1355.
Gupta, C.K. and Krishnamurthy, N. (1992). Extractive metallurgy of vanadium. Elsevier
Science Publishers B. V., The Netherlands.
Hall, F.P. and Insley, H. (1933). Phase diagrams for ceramists. American Ceramic Society,
Columbus, OH.
Hampel, C.A. (1961). Rare metals handbook. Reinhold Publishing Corporation, London.
Harding, R.A. and Wickins, M. (2003). Temperature measurements during induction skull
melting of titanium aluminide. Materials Science and Technology, 19: 1235–1246.
Harding, R.A., Wickins, M., Wang, H., Djambazov, G., and Pericleous, K.A. (2011).
Development of a turbulence-free casting technique for titanium aluminides.
Intermetallics, 19(6): 805–813.
Harvey, F.A. (1945). Refractories. In: D.M. Liddell (Editor-in-Chief), Handbook of nonfer-
rous metallurgy. McGraw-Hill, New York.
Henrie, T.A. and Morrice, E. (1966). V. A high temperature electrowinning cell for rare earths.
JOM: The Journal of The Minerals, Metals & Materials Society, 18: 1207–1208.
Herget, C. (1985). Metallurgical ways to NdFeB alloys. Permanent magnets from cobalt-
reduced NdFeB. (retroactive coverage). In: Eighth international workshop on rare-
earth magnets and their applications and the fourth international symposium on
magnetic anisotropy and coercivity in rare earth--transition metal alloys, 407–422.
Hirschhorn, I.S. (1968). Commercial production of rare earth metals by fused salt electroly-
sis. JOM: The Journal of The Minerals, Metals & Materials Society, 20(3): 19–221.
Hockaday, S.A.C., and Bisaka, K. (2010). Experience and results from running of a 1 kg Ti
scale Kroll reactor. In: Proceedings of the advanced metals initiative: Light metals
conference 2010. Southern African Institute of Mining and Metallurgy, Johannesburg.
pp. 265–280.
Huffine, C.L. and Williams, J.M. (1961). Refining and purification of rare earth metals. In:
F.H. Spedding and A.H. Daane (Eds.), The rare earths. Wiley, New York, pp. 145–162.
Jackson, C.M., Wagner, H.J., and Wasilewski, R.J. (1972). 55-Nitinol—the alloy with a mem-
ory: its physical metallurgy, properties, and applications Report NASA SP-5110.
Jain, S.C., Bose, D.K., and Gupta, C.K. (1971). Pilot-plant production of capacitor-grade
tantalum powder. Bhabha Atomic Research Centre, Trombay, India.
Jamrack, W.D. (1963). Rare metal extraction by chemical engineering techniques. Pergamon,
Oxford.
Johns, H.L., and King, A.G. (1970). Zirconia tailored for thermal shock resistance. Ceram.
Age, 86(5): 29–31.
Jordan, L. and Swanger, W.H. (1930). Preparation of Pure Nickel, RP257. Journal of Research
of the National Bureau of Standards, 5: 1291.
Kamat, G.R. and Gupta, C.K. (1971). Open aluminothermic reduction of columbium (Nb)
pentoxide and purification of the reduced metal. Metallurgical Transactions, 2(10):
2817–2823.
Kamyshnykova, K. and Lapin, J. (2018). Vacuum induction melting and solidification of
TiAl-based alloy in graphite crucibles. Vacuum, 154: 218–226.
References 195

Kaptay, G., Matsushita, T., Mukai, K., and Ohuchi, T. (2004). On different modifications
of the capillary model of penetration of inert liquid metals into porous refractories
and their connection to the pore size distribution of the refractories. Metallurgical and
Materials Transactions B, 35(3): 471–486.
Kartavykh, A.V. and Cherdyntsev, V.V. (2008). Chemical compatibility of a TiAl-Nb melt
with oxygen-free crucible ceramics made of aluminum nitride. Russian Metallurgy
(Metally), 2008(6): 491.
Kartavykh, A.V., Tcherdyntsev, V.V., and Zollinger, J. (2009). TiAl–Nb melt interaction with
AlN refractory crucibles. Materials Chemistry and Physics, 116(1): 300–304.
Kartavykh, A.V., Tcherdyntsev, V.V., and Zollinger, J. (2010). TiAl–Nb melt interaction with
pyrolytic boron nitride crucibles. Materials Chemistry and Physics, 119(3): 347–350.
Kaufmann, A.R. and Gordon, E. (1947). Vacuum melting and casting of beryllium. Metal
Progress, 52: 387.
Keck, P.H. and Golay, M.J. (1953). Crystallization of silicon from a floating liquid zone.
Physical Review, 89(6): 1297.
Kim, S., Hong, T., and Kim, Y. (2001). Evaluation of thermal stability of mold materials for
magnesium investment casting. Materials Transactions, 42(3): 539–542.
Kim, J.H., Kim, H.T., Woo, Y.M., Kim, K.H., Lee, C.B., and Fielding, R.S. (2013). Interaction
studies of ceramic vacuum plasma spraying for the melting crucible materials. Nuclear
Engineering and Technology, 45(5): 683–688.
Klotz, U.E., Legner, C., Bulling, F., Freitag, L., Faßauer, C., Schafföner, S., and Aneziris, C.G.
(2019). Investment casting of titanium alloys with calcium zirconate moulds and cruci-
bles. The International Journal of Advanced Manufacturing Technology, 103(1) 343–353.
Knauft, R.W. (1943). Zircon refractories for aluminum melting furnaces. Metals and Alloys,
18: 1326.
Koger, J.W., Holcombe, C.E., and Banker, J.G. (1976). Coatings on graphite crucibles used in
melting uranium. Thin Solid Films, 39: 297–303.
Komarek, K.L. and Silver, M. (1962). Proc. IAEA Symp., Thermodynamics of Nuclear
Materials. Wien, 749.
Korgul, P., Wilson, D.R., and Lee, W.E. (1997). Microstructural analysis of corroded alumina-
spinel castable refractories. Journal of the European Ceramic Society, 17(1): 77–84.
Kostov, A. and Friedrich, B. (2005). Selection of crucible oxides in molten titanium and
titanium aluminum alloys by thermo-chemistry calculations. Journal of Mining and
Metallurgy, Section B, 41(1): 113–125.
Kostov, A. and Friedrich, B. (2006). Predicting thermodynamic stability of crucible oxides in
molten titanium and titanium alloys. Computational Materials Science, 38(2): 374–385.
Kramer, G.M. (2009). A comparison of chemistry and inclusion distribution and morphol-
ogy versus melting method of NiTi alloys. Journal of Materials Engineering and
Performance, 18(5): 479–483.
Kroll, W.J. (1940). The production of ductile titanium. Transactions of the Electrochemical
Society, 78(1): 35–47.
Kroll, W.J. (1956). The pyrometallurgy of halides. Metallurgical Reviews, 1(1): 291–337.
Kroll, W.J. (1959). The present state of titanium extractive metallurgy. Transactions of the
American Institute of Mining and Metallurgical Engineers, 215(4): 546–553.
Kroll, W.J. and Stephens, W.W. (1950). Pilot plants. Production of malleable zirconium.
Industrial and Engineering Chemistry, 42(2): 395–398.
Kroll, W.J., Schlechten, A.W., and Yerkes, L.A. (1946). Ductile zirconium from zircon sand.
Transactions of the Electrochemical Society, 89(1): 263.
Kroll, W.J., Schlechten, A.W., Carmody, W.R., Yerkes, L.A., Holmes, H.P., and Gilbert, H.L.
(1947). Recent progress in the metallurgy of malleable zirconium. Transactions of the
Electrochemical Society, 92(1): 99–113.
196 References

Kroll, W.J., Anderson, C.T., Holmes, H.P., Yerkes, L.A., and Gilbert, H.L. (1948). Large‐scale
laboratory production of ductile zirconium. Journal of the Electrochemical Society,
94(1): 1.
Kuang, R.J., Harding, P.A., and Campbell, J. (2000). Investigation into refractories as crucible
and mould materials for melting and casting γ -TiAl alloys. Materials Science and
Technology, 16(9): 1007–1016.
Kuang, R.J., Harding, P.A., and Campbell, J. (2001). A study of refractories as crucible and
mould materials for melting and casting γ-TiAl alloys. International Journal of Cast
Metals Research, 13(5): 277–292.
Kubaschewski, O. and Alcock, C.B. (1979). Materials thermochemistry. Pergamon, Oxford.
Kubaschewski, O. and Dench, W.A. (1954). The free-energy diagram of the system titanium
oxygen. Journal of the Institute of Metals, 82: 87–91.
Kuznietz, M., Livne, Z., Cotler, C., and Erez, G. (1988). Diffusion of liquid uranium into
foils of tantalum metal and tantalum-10 wt% tungsten alloy up to 1350°C. Journal of
Nuclear Materials, 152(2–3): 235–245.
Lapin, J. (2009, May). TiAl-based alloys: Present status and future perspectives. In Conference
proceedings, METAL, 19, No. 21.5.2009, 12 pp.
Lapin, J. and Klimová, A. (2019). Vacuum induction melting and casting of TiAl-based
matrix in-situ composites reinforced by carbide particles using graphite crucibles and
moulds. Vacuum, 169: 108930.
Lapin, J., Gabalcová, Z., and Pelachová, T. (2011). Effect of Y2O3 crucible on contamina-
tion of directionally solidified intermetallic Ti–46Al–8Nb alloy. Intermetallics, 19(3):
396–403.
Lee, W.E., and Zhang, S. (1999). Melt corrosion of oxide and oxide–carbonrefractories.
International Materials Reviews, 44(3): 77–104.
Lee, W.E. and Moore, R.E. (1998). Evolution of in situ refractories in the 20th century.
Journal of the American Ceramic Society, 81(6): 1385–1410.
Lee, W.E. and Rainforth, W.M. (1994). Ceramic microstructures, property control by pro-
cessing. Chapman and Hall, London.
Lee, W.E. and Zhang, S. (2004, January). Direct and indirect slag corrosion of oxide and
oxide-C refractories. In: VII international conference on molten slags fluxes and salts.
The South African Institute of Mining and Metallurgy, Johannesburg, Symposium
Series S 36, pp. 309–319.
Li, Z., Lee, W.E., and Zhang, S. (2007). Low‐temperature synthesis of CaZrO3 powder from
molten salts. Journal of the American Ceramic Society, 90(2): 364–368.
Li, C.H., Gao, Y.H., Lu, X.G., Ding, W.Z., Ren, Z.M., and Deng, K. (2010). Interaction
between the ceramic CaZrO3 and the melt of titanium alloys. Advances in Science and
Technology, 70: 136–140.
Li, M., Gehre, P., and Aneziris, C.G. (2013). Investigation of calcium zirconate ceramic syn-
thesized by slip casting and calcination. Journal of the European Ceramic Society,
33(10): 2007–2012.
Li, C.H., He, J., Zhang, Z., Yang, B., Leng, H.Y., Lu, X.G., Li, Z.L., Wu, Z., and Wang, H.B.
(2015). Preparation of TiFe based alloys melted by CaO crucible and its hydrogen stor-
age properties. Journal of Alloys and Compounds, 618: 679–684.
Liu, K., Ma, Y.C., Gao, M., Rao, G.B., Li, Y.Y., Wei, K., Wu,X., and Loretto, M.H.
(2005). Single step centrifugal casting TiAl automotive valves. Intermetallics, 13(9):
925–928.
Lin, K.F. and Lin, C.C. (1998). Interfacial reactions between zirconia and titanium. Scripta
Materialia, 39(10): 1333–1338.
Lin, K.F. and Lin, C.C. (1999a). Interfacial reactions between Ti-6Al-4V alloy and zirconia
mold during casting. Journal of Materials Science, 34(23): 5899–5906.
References 197

Lin, K.F. and Lin, C.C. (1999b). Transmission electron microscope investigation of the inter-
face between titanium and zirconia. Journal of the American Ceramic Society, 82(11):
3179–3185.
Lochbichler, C. and Friedrich, B. (2007). Induction melting using refractaries and direct
deoxidisation of Ti and TiAl Scrap. In: European Metallurgical Conference, EMC
2007, June 11-14, Düsseldorf, Germany.
Long, G. and Foster, L.M. (1959). Aluminum nitride, a refractory for aluminum to 2000oC.
Journal of the American Ceramic Society, 42(2): 53–59.
López, V.H. and Kennedy, A.R. (2006). Flux-assisted wetting and spreading of Al on TiC.
Journal of Colloid and Interface Science, 298(1): 356–362.
Lu, M.W., Lin, K.L., and Lin, C.C. (2017). Effect of the alumina content on the interfa-
cial reactions between titanium and calcia/zirconia/alumina composites, Contributed
Papers from Materials Science and Technology 2017, October 8–12, Pittsburgh.
Lu, M.W., Lin, K.L., and Lin, C.C. (2019). Investigation of the interactions between titanium
and calcium zirconium oxide (CaZrO3) ceramics modified with alumina. Processing
and Application of Ceramics, 13(1): 79–88.
Lyon, S.R., Inouye, S., Alexander, C.A., and Niesz, D.E. (1973). The interaction of titanium
with refractory oxides. In: Titanium science and technology. Springer, Boston, MA,
pp. 271–284.
Malfliet, A., Lotfian, S., Scheunis, L., Petkov, V., Pandelaers, L., Jones, P.T., and Blanpain,
B. (2014). Degradation mechanisms and use of refractory linings in copper produc-
tion processes: A critical review. Journal of the European Ceramic Society, 34(3):
849–876.
Massalski, T.B. (1990a). Binary alloy phase diagrams (Vol. 1). American Society for Metals,
Metals Park, OH.
Massalski, T.B. (1990b). Binary alloy phase diagrams (Vol. 2). American Society for Metals,
Metals Park, OH.
McCollum, J.M. (1989). An improved 85% alumina phosphate-bonded refractory for molten
aluminum contact. Light Metal Age, 47(11): 22–24.
Merrill, T.W. (1958). Ductile vanadium - A new engineering material. JOM: The Journal of
The Minerals, Metals & Materials Society, 10: 618–621.
Mikami, H.M. and Martinet, J.R. (1979). Carbon magnesia bricks in electric arc furnaces
Refractories J. 55(6): 25–32.
Moriarty, J.L. (1968). The industrial preparation of the rare earth metals by metallothermic
reduction. JOM: The Journal of The Minerals, Metals & Materials Society, 20(11):
41–45.
Morrice, E. and Knickerbocker, R.G. (1961). Rare earth electrolytic metals. In: F.H. Spedding
and A.H. Daane (Eds.), The rare earths. John Wiley, New York, pp. 126–144.
Morrice, E. and Wong, M.M. (1979). Fused salt electrowinning and electrorefining of rare
earth and yttrium metals. Minerals Science and Engineering, 11(3): 125–136.
Morrice, E., Porter, B., Brown, E.A., Wyche, C., and Knickerbocker, R.G. (1961).
Electrowinning of cerium group and yttrium group metals, Report of Investigations
5868. Bureau of Mines, U.S. Department of the Interior, Washington, DC.
Morrice, E., Shedd, E.S., and Henrie, T.A. (1968). Direct electrolysis of rare earth oxides to
metals and alloys in fluoride melts, Report of Investigations 7146. Bureau of Mines,
U.S. Department of the Interior, Washington, DC.
Morscheiser, J., Friedrich, B., and Lochbichler, C. (2008). Potential of ceramic crucibles
for melting of titanium-alloys and gamma-titanium aluminide. 51 Internationales
Feuerfestkolloquium, 15, 16.
Mukherjee, T.K. and Gupta, C.K. (1972). Open aluminothermic reduction of vanadium
oxides. Journal of the Less Common Metals, 27(2): 251–254.
198 References

Nakamura, Y., Nakajima, H., Ishioka, S., and Koiwa, M. (1988). Effect of oxygen on diffusion
of manganese in α-titanium. Acta Metallurgica, 36(10): 2787–2795.
Nandy, R.N. and Jogai, R.K. (2012). Selection of proper refractory materials for energy sav-
ing in aluminium melting and holding furnaces. International Journal of Metallurgical
Engineering, 1(6): 117–121.
Nayan, N., Saikrishna, C.N., Ramaiah, K.V., Bhaumik, S.K., Nair, K.S., and Mittal, M.C.
(2007). Vacuum induction melting of NiTi shape memory alloys in graphite crucible.
Materials Science and Engineerin A, 465(1–2): 44–48.
Nolting, H.J., Simmons, C.R., and Klingenberg, J.J. (1960). Preparation and properties of high
purity yttrium metal. Journal of Inorganic and Nuclear Chemistry, 14(3–4): 208–216.
Okabe, T.H., Suzuki, R.O., Oishi, T., and Ono, K. (1991). Thermodynamic properties of
dilute titanium-oxygen solid solution in beta phase. Materials Transactions, JIM, 32(5):
485–488.
Orsten, S. and Oeters, F. (1986). Dissolution of carbon in liquid iron. In: Process technology
Proceedings, 6: 143–155.
Otubo, J., Rigo, O.D., Neto, C.D.M., Kaufman, M.J., and Mei, P.R. (2003a). NiTi shape mem-
ory alloy ingot production by EBM. Journal de Physique IV, 112: 813–820.
Otubo, J., Rigo, O.D., Neto, C.D.M., Kaufman, M.J., and Mei, P.R. (2003b). Scale up of NiTi
shape memory alloy production by EBM. In: Journal de Physique IV (Proceedings),
112: 873–876. EDP Sciences.
Otubo, J., Rigo, O.D., Neto, C.D.M., Kaufman, M.J., and Mei, P.R. (2004). Low carbon con-
tent NiTi shape memory alloy produced by electron beam melting. Materials Research,
7(2): 263–267.
Otubo, J., Rigo, O.D., Neto, C.D.M., and Mei, P.R. (2006). The effects of vacuum induction
melting and electron beam melting techniques on the purity of NiTi shape memory
alloys. Materials Science and Engineering A, 438: 679–682.
Ou, J., Cockcroft, S.L., Maijer, D.M., Yao, L., Reilly, C., and Akhtar, A. (2015). An exami-
nation of the factors influencing the melting of solid titanium in liquid titanium.
International Journal of Heat and Mass Transfer, 86: 221–233.
Pankratz, L.B., Stuve, J.M., and Gokcen, N.A. (1984). Thermodynamic data for mineral tech-
nology. Bureau of Mines Bulletin 677. U.S. Department of the Interior, Washington,
DC.
Perfect, F.H. (1967). Metallothermic reduction of oxides in water-cooled furnaces.
Transactions of the Metallurgical Society of AIME, 239(9): 1282–1286.
Perfect, F.H. (1981). Aluminothermic chromium and chromium alloys, low in nitrogen.
Metallurgical Transactions B, 12(3): 611–613.
Phelps, G.W. and Wachtman, J.B. (2011). Ullmann’s encyclopedia of industrial chemistry.
Wiley-VCH Verrlag GmbH & Co, Weinheim.
Pidgeon, L.M. (1944). New methods for the production of magnesium. The Canadian Institute
of Mining and Metallurgy, 47: 16–34.
Poirier, J. (2015). A review: Influence of refractories on steel quality. Metallurgical Research
and Technology, 112(4): 410.
Rehren, T. (2003). Crucibles as reaction vessels in ancient metallurgy. In: P. Craddock and
J. Lang (Ed.), Mining and metal production through the ages. British Museum Press,
London, pp. 207–215.
Richardson, F.D. (1948). The thermodynamics of substances of interest in iron and steel
making from 0 to 2400oC I-Oxides. The Journal of the Iron and Steel Institute, 160:
261–270.
Richardson, F.D., Jeffes, J.H.E., and Withers, G. (1950). The thermodynamics of substances
of interest in iron and steel making-II-compounds between oxides. The Journal of the
Iron and Steel Institute, 166: 213.
References 199

Richardson, H.K. (1935). Small cast thorium oxide crucibles. Journal of the American
Ceramic Society, 18(1–12): 65–69.
Richerson, D.W. and Lee, W.E. (2018). Modern ceramic engineering: Properties, processing,
and use in design. CRC Press, Boca Raton, FL.
Rigby, A. (2005). Controlling the processing parameters affecting the refractory requirements
for Peirce-Smith converters and anode refining vessels. In: Converter and fire refining
practices, The Minerals Metals and Materials Society, Warrendale, PA, pp. 213–222.
Rostoker, W. (1958). The metallurgy of vanadium. John Wiley, New York.
Routschka, G. (Ed.). (2001). Refractory materials: pocket manual (3rd Ed.). Vulkan-Verlag,
Essen.
Routschka, G. and Granitzki, K.-E. (2005). Refractory ceramics. In: Ullmann’s encyclopedia
of industrial chemistry. Wiley-VCH Verlag GmbH and Co. KGaA, Weinheim.
Ruh, R. (1963). Reactions of zirconia and titanium at elevated temperatures. Journal of the
American Ceramic Society, 46(7): 301–307.
Ruh, R. and Garrett, H.J. (1964). Reactions of zirconia and chromium. Journal of the
American Ceramic Society, 47(12): 627–629.
Sadrnezhad, S.K. and Raz, S.B. (2005). Interaction between refractory crucible materials
and the melted NiTi shape-memory alloy. Metallurgical and Materials Transactions
B, 36(3): 395–403.
Saha, R.L. and Jacob, K.T. (1986). Casting of titanium and its alloys. Defence Science
Journal, 36(2): 121–141.
Saha, R.L., Nandy, T.K., Misra, R.D.K., and Jacob, K.T. (1990). On the evaluation of stabil-
ity of rare earth oxides as face coats for investment casting of titanium. Metallurgical
Transactions B, 21(3): 559–566.
Sakamoto, K., Yoshikawa, K., Kusamichi, T., and Onoye, T. (1992). Changes in oxygen con-
tents of titanium aluminides by vacuum induction, cold crucible induction and electron
beam melting. ISIJ International, 32(5): 616–624.
Samsonov, G.V., Yasinskaya, G.A., and Wei, T.S. (1960). High-melting carbide, boride and
nitride crucibles. Refractories, 1(1–2): 26–29.
Savitskii, E.M. and Burkhanov, G.S. (1970). Physical metallurgy of refractory metals and
alloys, Consultants Bureau, New York.
Schafföner, S. (2020). Reactions of alkaline earth zirconate refractories with titanium alloys.
In: MATEC web of conferences (Vol. 321). EDP Sciences, p. 10012.
Schafföner, S., Aneziris, C.G., Berek, H., Hubálková, J., and Priese, A. (2013). Fused cal-
cium zirconate for refractory applications. Journal of the European Ceramic Society,
33(15–16): 3411–3418.
Schafföner, S., Aneziris, C.G., Berek, H., Hubálková, J., Rotmann, B., and Friedrich, B.
(2015a). Corrosion behavior of calcium zirconate refractories in contact with titanium
aluminide melts. Journal of the European Ceramic Society, 35(3): 1097–1106.
Schafföner, S., Aneziris, C.G., Berek, H., Rotmann, B., and Friedrich, B. (2015b). Investigating
the corrosion resistance of calcium zirconate in contact with titanium alloy melts.
Journal of the European Ceramic Society, 35(1): 259–266.
Schafföner, S., Qin, T., Fruhstorfer, J., Jahn, C., Schmidt, G., Jansen, H., and Aneziris, C.G.
(2018). Refractory castables for titanium metallurgy based on calcium zirconate.
Materials and Design, 148: 78–86.
Schlesinger, M.E. (1996). Refractories for copper production. Mineral Processing and
Extractive Metallurgy Review: An International Journal, 16(2): 125–146.
Schlesinger, M.E. (2006). Aluminum recycling. CRC Press, Boca Raton, FL.
Schmutzler, H.J. and Sandhage, K.H. (1995). Transformation of Ba-Al-Si precursors to
celsian by high-temperature oxidation and annealing. Metallurgical and Materials
Transactions B, 26(1): 135–148.
200 References

Schnabel, M., Buhr, A., Buchel, G., Kockegey-Lorenz, R., and Dutton, J. (2011). Advantages
of calcium hexaluminate in a corrosive environment. Refractories World Forum, 3(4):
87–94.
Schuster, J.C. and Palm, M. (2006). Reassessment of the binary aluminum-titanium phase
diagram, J. Phase Equilibria Diffusion, 27(3): 255–277.
Schuyler, D.R., Petrusha, J.A., Hall, G.S., and Seagle, S.R. (1976). Development of titanium
alloy casting technology (No. 76–311721). AiResearch Manufacturing Company of
Arizona, Phoenix, AZ.
Schwartz, M.A., White, G.D., and Curtis, C.E. (1953). Crucible handbook a compilation of
data on crucibles used for calcining, sintering, melting, and casting, united states atomic
energy commission. ORNL-1354, Oak Ridge National Laboratory, Oak Ridge, TN.
Seybolt, A.U. (1946). A practical furnace for vacuum melting. Metal Progress, 50: 1102.
Shedd, E.S., Marchant, J.D., and Henrie, T.A. (1964). Continuous electrowinning of cerium
metal from cerium oxides, report of Investigations 6362. Bureau of Mines, U.S.
Department of the Interior, Washington, DC.
Shunk, F.A. (1969). Constitution of binary alloys. Second Supplement, McGraw Hill, New
York.
Siljan, O.-J., Schoning, C., and Grande, T. (2002). State-of-the-art alumino-silicate refracto-
ries for al electrolysis cells. JOM: The Journal of The Minerals, Metals & Materials
Society, 54(5): 46–55.
Spedding, F.H. and Daane, A.H. (1952). The preparation of rare earth metals. Journal of the
American Chemical Society, 74(11): 2783–2785.
Spedding, F.H. and Daane, A.H. (Eds.). (1961). The rare earths. Wiley, New York.
Spedding, F.H., Daane, A.H., Wakefield, G., and Dennison, D.H. (1960). Preparation and
properties of high purity scandium metal. Transactions of the Metallurgical Society of
AIME, 218: 608–611.
Spedding, F.H., Beaudry, B.J., Croat, J.J., and Palmer, P.E. (1968). The properties, preparation
and handling of pure rare earth metals. In Materials Technology—An Interamerican
Approach, p. 151, Am. Soc. Mech. Eng., New York.
Spedding, F.H., Beaudry, B.J., Croat, J.J., and Palmer, P.E. (1970). The preparation and prop-
erties of ultrapure metals. In: Les Elements Des Terres Rares (vol. 1, p. 25). Centre
National de la Recherche Scientifique, Paris. Also R&D Report IS-2283, Ames
Laboratory, USAEC, May.
Sundaram, C.V., Taneja, A.K., and Sridhar Rao, C.H. (1992). Technology trends in the extrac-
tive metallurgy of zirconium, titanium, tantalum and niobium. Mineral Processing and
Extractive Metallurgy Review, 10(1): 239–265.
Taylor, L., Boyer, H.E., and Unterweiser, P.M. (Eds.) (1948). Metals handbook, American
Society for Metals, Cleveland, OH.
Tetsui, T., Kobayashi, T., Mori, T., Kishimoto, T., and Harada, H. (2010). Evaluation of yttria
applicability as a crucible for induction melting of TiAl alloy. Materials Transactions,
51(9): 1656–1662.
Tetsui, T., Kobayashi, T., Kishimoto, A., and Harada, H. (2012a). Structural optimization of
an yttria crucible for melting TiAl alloy. Intermetallics, 20(1): 16–23.
Tetsui, T., Kobayashi, T., Kishimoto, A., and Harada, H. (2012b). Structural optimization of
an yttria crucible for melting TiAl alloy. Intermetallics, 20(1): 16–23.
Theuerer, H.C. (1956). Removal of boron from silicon by hydrogen water vapor treatment.
JOM: The Journal of The Minerals, Metals & Materials Society, 8(10): 1316–1319.
Thews, E.R. (1931). The suitability of various refractory materials for lead refining furnaces.
Feuerfest, 7: 84.
Thompson, J.G. and Cleaves, H.E. (1939). Preparation of high-purity iron, RP1226. Journal
of Research of the National Bureau of Standards, 23: 163–177.
References 201

Thornton, C.P. and Rehren, T. (2009). A truly refractory crucible from fourth millen-
nium Tepe Hissar, Northeast Iran. Journal of Archaeological Science, 36(12):
2700–2712.
Tournier, C., Lorrain, B., Le Guyadec, F., Coudurier, L., and Eustathopoulos, N. (1998).
Kinetics of interfacial reactions in molten U/solid Y2O3 system. Journal of Nuclear
Materials, 254(2–3): 215–220.
Turner, D. (1931). Special refractories for metallurgical research. Transactions of the Faraday
Society, 27: 112.
Tylecote, R.F. (1992). A history of metallurgy (2nd Ed.). Maney Publishing, London.
Uwanyuze, S., Kanyo, J., Myrick, S., and Schafföner, S. (2021). A review on alpha case for-
mation and modeling of mass transfer during investment casting of titanium alloys.
Journal of Alloys and Compounds, 865: 158558.
Van Humbeeck, J. (1999). Non-medical applications of shape memory alloys. Materials
Science and Engineering A, 273: 134–148.
Vikulin, V.V., Rusin, M.Y., and Rusanova, L.N. (2006). Refractory materials and products
based on natural wollastonite for making ceramic accessories to be used in aluminium
industry. Advances in Science and Technology, 45: 2272–2277.
Wang, C.T., Baroch, E.F., Worcester, S.A., and Shen, Y.S. (1970). Preparation and properties
of high-purity vanadium and V-15Cr-5Ti. Metallurgical Transactions, 1(6): 1683–1689.
Washburn, E.W. (1921). The dynamics of capillary flow. Physical Review, 17(3): 273.
Weber, B.C., Thompson, W.M., Bielstein, H.O., and Schwartz, M.A. (1957). Ceramic
Crucible for Melting Titanium. Journal of the American Ceramic Society, 40(11):
363–373.
Weimer Alan, W. (Ed.). (1997). Carbide, nitride and boride materials synthesis and pro-
cessing, Chapman and Hall, London.
Wilhelm, H.A. (1955). The preparation of uranium metal by the reduction of uranium tetra-
fluoride with magnesium. In: Proceedings of the 1st international conference on the
peaceful uses of atomic energy energy. Geneva. Paper 817.
Wilhelm, H.A. (1960). Development of uranium metal production in America. Journal of
Chemical Education, 37(2): 56–68.
Wilhelm, H.A. and Peterson, D.T. (1961). Pure metals from chemical processes (No. IS-394).
Ames Lab, Ames, IA.
Wilhelm, H.A., Schmidt, F.A., and Ellis, T.G. (1966). Columbium metal by the alumino-
thermic reduction of Cb2O5. JOM: The Journal of The Minerals, Metals & Materials
Society, 18(12): 1303–1308.
Wilson, R. (2000). A practical approach to continuous casting of copper-based alloys and
precious metals. Institute of Materials, London.
Winkler, O.C. and Bakish, R.A. (Eds.). (1971). Vacuum metallurgy. Elsevier, Amsterdam.
Xiao, P. and Ralph, B. (Eds.). (2009). Advanced ceramic materials, trans-tech, Zurich,
Switzerland.
Yazawa, A. and Kongoli, F. (2001). Liquidus surface of newly defined “ferrous calcium
silicate slag” and its metallurgical implications. High Temperature Materials and
Processes, 20(3–4): 201–208.
Yurkov, A. (2017). Refractories for aluminum - electrolysis and the cast house. Springer
International Publishing, Cham, Switzerland.
Zhang, Z., Frenzel, J., Neuking, K., and Eggeler, G. (2005). On the reaction between NiTi
melts and crucible graphite during vacuum induction melting of NiTi shape memory
alloys. Acta Materialia, 53(14): 3971–3985.
Zhang, Z., Frenzel, J., Neuking, K., and Eggeler, G. (2006). Vacuum induction melting of
ternary NiTiX (X= Cu, Fe, Hf, Zr) shape memory alloys using graphite crucibles.
Materials Transactions, 47(3): 661–669.
202 References

Zhang, H.R., Tang, X.X., Zhou, L., Gao, M., Zhou, C.G., and Zhang, H. (2012). Interactions
between Ni-44Ti-5Al-2Nb-Mo alloy and oxide ceramics during directional solidifica-
tion process. Journal of Materials Science, 47(17): 6451–6458.
Zhang, H., Tang, X., Zhou, C., Zhang, H., and Zhang, S. (2013a). Comparison of directional
solidification of γ-TiAl alloys in conventional Al2O3 and novel Y2O3-coated Al2O3
crucibles. Journal of the European Ceramic Society, 33(5): 925–934.
Zhang, Z., Zhu, K.L., Liu, L.J., Lu, X.G., Wu, G.X., and Li, C.H. (2013b). Preparation of
BaZrO3 crucible and its interfacial reaction with molten titanium alloys. Journal of the
Chinese Ceramic Society, 41: 1272–1283.
Zhang, Y., Fang, Z.Z., Sun, P., Xia, Y., Lefler, H.D. and Zheng, S. (2020). Deoxygenation of
titanium metal. In: Z.Z. Fang, F.H. Froes, and Y. Zhang (Eds.), Extractive metallurgy
of titanium. Elsevier, Amsterdam, pp 181–223.
Zhao, L. and Sahajwalla, V. (2003). Interfacial phenomena during wetting of graphite/alu-
mina mixtures by liquid iron. ISIJ International, 43(1): 1–6.
Index
Acheson, Edward G., 23 burned dolomite, 110
Aggregates and matrix, 23, 25, 56 burned dolomitic lime, 111
Alpha case, 160 burned limestone, 110
Alumina burned magnesite, 110
calcined, 83 graphite, 109
fused, 83 high alumina, 109
tabular, 83 MgO–C bricks, 111
Alumina–zirconia–carbon, 90, 115 micropore carbon, 109
Aluminium electrolysis cell, Hall, 26 pitch bonded magnesia, 110
Aluminium electrolysis cell, Héroult, 26 semi graphitized carbon, 109
Aluminium nitride siliceous lining, 110
crucibles for melting silicon carbide, 109
aluminium, 98 tar bonded dolomite, 110
gallium, 98 thermal solution, cooling, 108, 109
Ti alloys, 98 Bloom, iron, 14
fabrication, 97 Bloomery, 13, 14
synthesis, 97 Blower
Aluminosilicates, 81 steam engine, 7
Aluminothermy, 133, 137; see also Bomb reactor water wheel, 7
closed bomb, 138 Blowpipes, 3
open reactor, 139 Bomb reactor, 7, 21, 28
water cooled reactor, 140 lining
Andalusite, 82 alumina, 138
AOD, VOD vessel refractories burned lime, 136
dolomite, 111 CaF2, 134, 135
dolomite–magnesia, 111 dolomite, 134, 136
magnesia–carbon, 1, 12 magnesia, 138, 139
magnesia–chrome, 112 MgF2, 134
Bonnin, Gousse, 23
Barium zirconate Booster reaction, 134
Ca-doped, 94 calcium–iodine, 145
fabrication, 94 calcium–sulphur, 136
synthesis, 94 Borides
Y-doped, 94 Hall–Héroult cell components, 95
vs. zirconia, 94 vacuum metallizing boats, 95
Basic oxygen furnace, 21 Boron nitride, 95
Bellows boats and crucibles, 96
manually operated, 3, 9 fabrication, 96
mechanical, 15 machinable ceramics, 96
piston, 17 pyrolytic, 96
water wheel, 14 reaction bonded, 96
Beryllia, 88 synthesis, 96
Bessemer, Henry, 19 Bradley, Charles A., 25
Bessemer converter, 20 Brass, 10
Bessemer’s fixed vessel, 19 Bricks, pre-shaped, 24, 79
BF-LD/BOF, 108 Bronze, 10
Blast furnace, 17, 107
degradation of refractories, 108 Calcia–magnesia, 130
refractories Calciothermy, 133, 137; see also Bomb Reactor

203
204 Index

Calcium aluminate cement, 94 top blown rotary converter, 127

Calcium zirconate Cooper crucible, water cooled, 30, 31, 158, 159
fabrication, 94 Copper, 126
fused, 94 Copper chromite coating, 129
synthesis, 94 Copper reactor, water cooled, 29
Cannon, casting, 15 aluminothermic reduction, 29, 140
Capillary Corrosion of refractories, 93; see also Wetting
absorption, 11 direct, 69–70
forces, 55 indirect, 69–71
penetration, 61 selective, 69
pressure, 61 Corrosion tests, 72
Carbon bricks, 87 brick/cavity/cup, 72
micropore, 87 immersion, 72
super-micropore, 87 rotary slag, 73
Castable, 79 rotating finger, 73
Casting, 71 sessile drop, 72
Catalan furnace, 7 Cort, Henry, 19
Cementation, 11; see also Iron and steel Crucible, 1, 2
brass, 11 Cryolite, 115
crucible, clay, 18 Crystal growth
Ceramics, 77 container, 32
modern, 78 floating zone, 33
refractory, 78 Cu2O–MgO eutectic, 129
special, 78 Cupel, 11
traditional, 78 Cupellation, 11; see also Capillary
Characterization techniques Cupola, 18
EDS, 73 Curtis, T.S., 24
SEM/EDS, 73
vacuum fusion, 73 Davy, Sir Humphrey, 7, 25
XRD, 73 Dinas rock, 22
XRS, 73 Directional Solidification of TiAl, 17, 171
Chrome–alumina, 85 in CaO crucible, 172
Chrome–magnesite, 22 in graphite crucible, 174
for BOF, 23 in yttria crucible, 170–171
direct bonded, 23 Doloma, C-bonded, 92
for open hearth furnace, 23 Dolomite
pitch bonded, 23 bricks, 84
tar bonded, 23 refractory grade, 84
Civilization Dolomite lining
Babylonian, 11 Bessemer converter, 22
Egyptian, 11 open hearth furnace, 22
Hittites, 13 Dony, Abbé, 12
Palestine, 11 DRI–EAF, 108
Roman, 6, 9, 11 Drossing, tin, 11
Clay–charcoal lining, 9
Clerke, Sir Clement, 98 Electrolytic reduction, aluminium, 25; see also
Coating on crucibles, 98 Hall–Héroult cell
issues, 98 Electrolytic reduction, rare earths, 148
porosity, 98 cell materials, 148
yttria layer, 57 carbon, 148
techniques, 98, 153, 154 frozen electrolyte skull, 149
Contact angle, 131 graphite, 148
Converting, 126 iron, 148
hearth furnace, 127 molybdenum, 149
Peirce–Smith converter, 127 refractory oxides, 148
rotary furnace, 127 tungsten, 149
Index 205

Electron beam melting, 30, 32 side lining, 116

Electrorefining carbon blocks, 118
aqueous, 10 SiC, nitride bonded, 118
copper, 10 thermal balance, 116
fused salt, 140 Hearth, 9–17, 109
vanadium, 140 Héroult, Paul Louis Toussaint, 26
Elkington, 10 Hesse crucible, 7, 21
Ellingham diagram, 42 High temperature electrowinning cell, 150
borides, 44 Hot pressing, 79, 88
carbides, 43
fluorides, 45 Inclusions in steel, 113
nitrides, 44 endogenous or micro, 113
oxides, 43 exogenous or macro, 113
Induction skull melting, 30, 31, 77
Fabrication methods, ceramics, 88 Industry
Face coat, 159; see also Coating on aerospace, 27
crucibles atomic energy, 27
Faraday, Michael, 7 electronics, 27
FeTi -X alloys, 185 In-situ refractory, 93
arc melting, 185 investment casting, 157, 159
crucibles for melting Iron and steel, 107
BaZrO3, 185 cast iron, 15, 17
CaO, 185 cementation steel, 11, 18
CaZrO3, 185 clean steel, 108
yttria coated Al2O3, 185 crucible steel, 18
yttria coated Al2O3. SiO2, 185 pig iron, 18
yttria coated graphite, 185 wrought iron, 14
Firebrick, 15 Isostatic pressing
Fireclay, 81 cold, 79, 94, 177
Fire refining, copper, 127 hot, 88
Floating zone melting, 77
Forsterite, 84, 128 Kaolinite clay, 15, 81
Forsterite bursting, 131 Kelly, William, 19
Fraunhofer, Joeph von, 7 Kroll process, 106
Free energy of reaction, 46 containers, 141
contamination, 143
Galena, 11 slag-metal separation, 141
Gilchrist, Percy, 20 Kroll reduction, 142
Goldschmidt, Johannes Wilhelm, 28 nickel as container for Nb, Ta, 143
Goldschmidt process, 28 niobium pentachloride, 143
Grain and bond phases, 24 tantalum pentachloride, 143
Graphite, 23, 87 titanium tetrachloride, 157
vanadium chlorides, 143
Haematite, 107 zirconium tetrachloride, 142
Halide metallurgy, 8
Hall, Charles Martin, 25 Ladle furnace, steel, 108
Hall–Héroult cell, 115 Ladle shroud tube, 114
barrier refractory, 116, 117 Lanthanothermy, 148
alumino silicate bricks, 116 Lead silicate, 11
dry barrier mixture, 116 Less common metals, 8, 131
cathode carbon block, 116 Levitation melting, 77
bath penetration, 116 Linz–Donawitz vessel, 19
capillary flow, 116 Litharge, 11; see also Cupellation
intercalation of sodium, 116
pore size, 117 Mag–doloma, 131
ramming paste, 116 Magnesia–alumina
206 Index

CA bonded, 86 Native metals, 2

direct bonded, 86 copper, 3, 9
fused cast, 86 gold, 3
Magnesia–alumina–carbon, 87, 91 iron, 3
Magnesia–alumina–graphite, 87, 91 silver, 11
Magnesia–carbon, 23 Newcomen, John, 7
pitch impregnated, 86 New Jersey Zinc Company, 13
resin bonded, 86 NiTi alloys, see SMA
tar bonded, 86 Northrup, E.F., 30
Magnesia–chrome, 85
direct bonded, 85 Open hearth furnace, 19, 20
fused cast, 85 basic, 21
rebonded, 85 Oxide–carbon–metal composite, 23
silicate bonded (burned), 85 Oxide–carbon refractories, 92
Magnesia–chrome spinel, 90 anti-oxidants, 92
Magnesia refractories, 23 MgO layer, 91, 92
bricks, 84 slag corrosion, 92
caustic calcined, 83 Oxygen potential
dead burned, 83 titanium–oxygen system, 47, 53
fused, 83 vanadium–oxygen system, 53, 54
spinel, 84
Magnesia–wuestite, 130 Perfect, Fred H., 29
Magnesium aluminate spinel, 90, 125 Periclase–spinel lining, 130
Magnetite, 107 Phase diagram
Martin, Pierre E’mile, 20 Al2O3–SiO2, 82
Matte smelting, 126 CaO–MgO, 85
Melt penetration and erosion, 55; see also silica, 80
Corrosion of refractories titanium–oxygen, 48
Metallothermy, 28, 131 yttrium–oxygen, 51
halides, 132 Pidgeon process, 106, 148
heat of reaction, 133 Pig iron, 15, 16, 107
oxides, 132 Pirani, M. von, 32
product form, 133 Plumbago, 23
Metal powder production Porosity, 55, 61
containers Post reduction treatments, aluminium, 119
Inconel, 140 degassing units, 120
mild steel or stainless steel, 140 melting/holding furnace, 119
molybdenum, 140 transportation ladle, 120
nickel tray, 140 Puddling, iron, 19
powdered lime, 140 Pyrometric cone equivalent, 78
sintered alumina, 140
reduction of fluorides, 140 Rammed graphite, 159
reduction of oxides, 140 Rare earth metals reduction
Middle ages, 7 Ames process, 146
Mineralizers, 25 bomb reduction, 143
Mitsubishi process, 130 chloride reduction, 143
Monolithics, 24, 25, 79 crucible contamination
Moulds, 12 carbon, 151
alumina, 159 iron, 151
granite, 12 silicon, 151
graphite, 159 tantalum, 151
silica, 160 tungsten, 151
sillimanite (Al2SiO5), 159 Goldschmidt process, 146
steel, 159 reactors and liners, 144
YSZ, 160 dolomitic oxide, 146
zircon, 160 fluorite, 146
Mullite, 82 molybdenum, 143
Index 207

sintered lime, 145 Siemens, Carl Wilhelm (aka Siemens, Sir

tantalum, 143, 144 William), 20, 30
titanium, 144 Silica
reductant fused, 80
calcium, 146 properties, 81
lithium, 143 sand, 80
sodium, 144 Silicon carbide, 80
vapour reduction, 143 Acheson process, 95
Rare earth metals refining, 152 synthesis, 95
vacuum distillation Silicon nitride, 96
tungsten condenser, 153 Single crystals, 33
vacuum melting Slag line, 129
tantalum crucible, 152 Slags, 16, 27
tungsten crucible, 153 calcium ferrite slag, 130
Rare metals, 8, 27 cuprite slag, 127
Reduction-distillation, 106 fayalite slag, 128
Refractories, steel cladding, 79 FeO-laden basic slag, 127
Refractories for aluminium melting ferrous silicate slag, 131
aluminium penetration, 121 flash smelter slags, 130
aluminosilicates, 120 olivine slag, 131
anti wetting additives, 123 steel ladle slags, 130
barium based, 123 steel plant, 20, 22, 130
boron based, 123 Slag splashing, 93, 102, 111
celsian, 124 Slide gate plates
belly band zone, 122 MgO–Al2O3–C, 115
corundum layer, 122 MgO–C. Al2O3–ZrO2–C, 115
CA6 or bonite, 125 Slip casting, 79, 88; see also Fabrication
high alumina bricks, phosphate SM alloys, NiTi, 180
bonded, 119 Batch size and contamination, 183
melt line corrosion, 122, 123 crucibles for Melting
effect of oxygen, 123 Al2O3, 180
phosphate bond, 123 BaZrO3–CaO doped, 184
pore size control, 123 BaZrO3 - plain, 184
silicon pickup by melt, 121 graphite, 180
sol-gel binders, 123 MgO, 180
wollastonite, 124 TiC layer on crucible, 182
Refractories for copper, 128 graphite quality and contamination, 183
mag–chrome Smelting, beginning of
direct bonded, 128 copper, 5, 10
fused cast, 129 iron, 13
fused grain bricks, 128 lead, 11
RE oxide reduction, 148 tin, 11
Retorts for zinc Sodium aluminate, 125
horizontal, 12 Spinel, 90
vertical, 12 Steatite, 6, 21
RE–Zn alloy, 147 Steel
RH degasser refractories deoxidation, 108
direct bonded magnesia– desulphurization, 108
chromite, 112 inclusions, 108
spinel castable, 112 killing, 108
tabular alumina, 112 puddled, 19
Rohn, Wilhelm, 30 vacuum degassing, 108
Steel making
Sessile drop, 59 primary, 108
Shaft furnace, 9–11, 15 secondary, 108
stone built, 13 Steel plant refractories, 21
Sialon, 126 Submerged nozzles, alumina–graphite, 113
208 Index

Superheat, 31, 157 CaZrO3, 177

Surface tension, 57, 77; see also Contact angle CaZrO3, composition adjusted, 177
CaZrO3, fused, 177
Tabular alumina, 24, 83, 87, 91, 99 CaZrO3 fabrication, 177
Thermal conductivity CaZrO3 preparation, 177
alumina, 67 CeS, 163, 167
fireclay, 67 CrB2, 166, 167
graphite, 68 graphite, 174
magnesia-carbon, 68 HREO(Heavy rare earth oxide), 165
magnesite, 67 HREO-Hypostoichiometric, 165
magnesite–chrome, 67 LREO (Light rare earth oxide), 165
mullite, 67 LREO-hypostoichiometric, 165
silica, 67 MgO, 162, 163
silicon carbide, 68 NbC, 166
Thermal expansion REO (rare earth oxides), 165
carbon–Al, 65 SiC, 163
corundum, 65 TaC, 166
dolomite, 65 ThO2, 162, 163
fireclay, 65 ThO2 hypostoichiometric, 165
fused silica, 65 ThS, 167
graphite, 65 TiB2, 166, 167
mag–chrome, 65 TiC, 166
magnesia, 65 VC, 166
magnesite–carbon–Al/Mg, 65 WC, SiC, 166
magnesite–carbon–Si/Mg, 65 Y2O3.15Ti, 161, 163
magnessite–carbon, 65 Y2O3 doped BaZrO3, 176
mullite, 65 Y2O3 hypostoichiometric, 165
SiC, 65 Y2O3/K 2SiO3 face coat, 161
silica, 65 yttria, 162, 163, 165, 168
zircon, 65 yttria face coat, 171
Thermal shock resistance factor, 68 ZrB2, 166, 167
Thermit process, 28 ZrC, 166
Thomas, Sydney Gichrist, 20, 22 ZrO2, 164
Three age system, 2 ZrO2-hypostoichiometric, 161, 165
bronze age, 6, 11 Zr-ZrO2 face coat, 170
chalcolithic age, 5 Tundish refractories
iron age, 6 magnesia and forsterite, 114
stone age, 2 MgO–Al2O3 spinel, 114
Th–Zn alloy, 137 Tuyere, 3, 4, 9, 11, 13, 14, 16, 21, 109, 126, 127, 128
Titanium, 168
Titanium and alloys melting, 157 Uranium melting, 153
electon beam, 158, 180 crucibles
induction skull, 158, 159, 179 graphite, 153
vacuum arc, 157, 159 tantalum, 153
vacuum induction, 158 yttria, 153
Titanium melting, casting yttria coated, 153
crucibles, moulds, 160 Uranium reduction, 134
Al2O3, 162, 163
AlN, 174 Vacuum arc melting, 30
BaZrO3, 168 Vacuum induction melting, 30
BeO, 162, 163 Vacuum metallurgy, 8
BN, 175 Vanadium, 153
CaO, 162, 163 crucibles for melting, 154
CaO.15Ti, 163 alumina, 155
CaO doped BaZrO3, 168, 178 beryllia, 155
Index 209

calcia, 155 iron-graphite, 60

cerium sulphide, 155 polycrystal zirconia, 59
graphite, 155 single crystal zirconia, 59
MgO, 155 Work of adhesion, 58
SiC, 155 Wrought iron, 14
stabilized ZrO2, 155
ThO2, 155 Young, William Weston, 22
ZrO2, 155 Yttria, 90
Viscosity of melts
effect of composition, 164 Zircon, 88
effect of temperature, 164 Zirconia, 88
fully stabilized, 89
Watt, James, 7 partially stabilized, 89
Wetting, melt-solid, 55, 57 polymorphism, 88
iron-alumina, 60 Zircon-mullite, 88